diff options
| author | Dave Airlie <airlied@redhat.com> | 2018-05-25 03:28:14 +0300 |
|---|---|---|
| committer | Dave Airlie <airlied@redhat.com> | 2018-05-25 03:28:33 +0300 |
| commit | dd41fb8547c2422f3a3a75e7226525e8bba9381c (patch) | |
| tree | 0e94d33f1fc940999d7b05d5e383fb6e13625ac8 /drivers/gpu/drm | |
| parent | ce234ccc03cfee004e168a1ae4b9d0cfb1974a32 (diff) | |
| parent | b8f3439fa5358ac84d29fa2f4afa115500dec74c (diff) | |
| download | linux-dd41fb8547c2422f3a3a75e7226525e8bba9381c.tar.xz | |
Merge branch 'drm-next-4.18' of git://people.freedesktop.org/~agd5f/linux into drm-next
Last feature request for 4.18. Mostly vega20 support.
- Vega20 support
- clock and powergating for VCN
- misc bug fixes
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180524152427.32713-1-alexander.deucher@amd.com
Diffstat (limited to 'drivers/gpu/drm')
137 files changed, 6964 insertions, 2828 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 68e9f584c570..a51c5a960750 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -62,11 +62,13 @@ amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \ amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce_v6_0.o si_dpm.o si_smc.o amdgpu-y += \ - vi.o mxgpu_vi.o nbio_v6_1.o soc15.o emu_soc.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o + vi.o mxgpu_vi.o nbio_v6_1.o soc15.o emu_soc.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o \ + vega20_reg_init.o # add DF block amdgpu-y += \ - df_v1_7.o + df_v1_7.o \ + df_v3_6.o # add GMC block amdgpu-y += \ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 03a2c0be0bf2..a59c07590cee 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1401,6 +1401,8 @@ struct amdgpu_df_funcs { bool enable); void (*get_clockgating_state)(struct amdgpu_device *adev, u32 *flags); + void (*enable_ecc_force_par_wr_rmw)(struct amdgpu_device *adev, + bool enable); }; /* Define the HW IP blocks will be used in driver , add more if necessary */ enum amd_hw_ip_block_type { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c index a0f48cb9b8f0..236915849cfe 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c @@ -322,3 +322,47 @@ int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev) return ret; } + +union gfx_info { + struct atom_gfx_info_v2_4 v24; +}; + +int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev) +{ + struct amdgpu_mode_info *mode_info = &adev->mode_info; + int index; + uint8_t frev, crev; + uint16_t data_offset; + + index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, + gfx_info); + if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + union gfx_info *gfx_info = (union gfx_info *) + (mode_info->atom_context->bios + data_offset); + switch (crev) { + case 4: + adev->gfx.config.max_shader_engines = gfx_info->v24.gc_num_se; + adev->gfx.config.max_cu_per_sh = gfx_info->v24.gc_num_cu_per_sh; + adev->gfx.config.max_sh_per_se = gfx_info->v24.gc_num_sh_per_se; + adev->gfx.config.max_backends_per_se = gfx_info->v24.gc_num_rb_per_se; + adev->gfx.config.max_texture_channel_caches = gfx_info->v24.gc_num_tccs; + adev->gfx.config.max_gprs = le16_to_cpu(gfx_info->v24.gc_num_gprs); + adev->gfx.config.max_gs_threads = gfx_info->v24.gc_num_max_gs_thds; + adev->gfx.config.gs_vgt_table_depth = gfx_info->v24.gc_gs_table_depth; + adev->gfx.config.gs_prim_buffer_depth = + le16_to_cpu(gfx_info->v24.gc_gsprim_buff_depth); + adev->gfx.config.double_offchip_lds_buf = + gfx_info->v24.gc_double_offchip_lds_buffer; + adev->gfx.cu_info.wave_front_size = le16_to_cpu(gfx_info->v24.gc_wave_size); + adev->gfx.cu_info.max_waves_per_simd = le16_to_cpu(gfx_info->v24.gc_max_waves_per_simd); + adev->gfx.cu_info.max_scratch_slots_per_cu = gfx_info->v24.gc_max_scratch_slots_per_cu; + adev->gfx.cu_info.lds_size = le16_to_cpu(gfx_info->v24.gc_lds_size); + return 0; + default: + return -EINVAL; + } + + } + return -EINVAL; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h index 7689c961c4ef..20f158fd3b76 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h @@ -30,5 +30,6 @@ int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev); int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev); int amdgpu_atomfirmware_get_vram_type(struct amdgpu_device *adev); int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev); +int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index 5b3d3bf5b599..e950730f1933 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -400,6 +400,9 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device, case CHIP_VEGA12: strcpy(fw_name, "amdgpu/vega12_smc.bin"); break; + case CHIP_VEGA20: + strcpy(fw_name, "amdgpu/vega20_smc.bin"); + break; default: DRM_ERROR("SMC firmware not supported\n"); return -EINVAL; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index a8e531d604fa..c5bb36275e93 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -173,9 +173,14 @@ static void amdgpu_ctx_do_release(struct kref *ref) ctx = container_of(ref, struct amdgpu_ctx, refcount); - for (i = 0; i < ctx->adev->num_rings; i++) + for (i = 0; i < ctx->adev->num_rings; i++) { + + if (ctx->adev->rings[i] == &ctx->adev->gfx.kiq.ring) + continue; + drm_sched_entity_fini(&ctx->adev->rings[i]->sched, &ctx->rings[i].entity); + } amdgpu_ctx_fini(ref); } @@ -452,12 +457,17 @@ void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr) if (!ctx->adev) return; - for (i = 0; i < ctx->adev->num_rings; i++) + for (i = 0; i < ctx->adev->num_rings; i++) { + + if (ctx->adev->rings[i] == &ctx->adev->gfx.kiq.ring) + continue; + if (kref_read(&ctx->refcount) == 1) drm_sched_entity_do_release(&ctx->adev->rings[i]->sched, &ctx->rings[i].entity); else DRM_ERROR("ctx %p is still alive\n", ctx); + } } } @@ -474,12 +484,17 @@ void amdgpu_ctx_mgr_entity_cleanup(struct amdgpu_ctx_mgr *mgr) if (!ctx->adev) return; - for (i = 0; i < ctx->adev->num_rings; i++) + for (i = 0; i < ctx->adev->num_rings; i++) { + + if (ctx->adev->rings[i] == &ctx->adev->gfx.kiq.ring) + continue; + if (kref_read(&ctx->refcount) == 1) drm_sched_entity_cleanup(&ctx->adev->rings[i]->sched, &ctx->rings[i].entity); else DRM_ERROR("ctx %p is still alive\n", ctx); + } } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 9fb20a53d5b2..290e279abf0d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -86,6 +86,7 @@ static const char *amdgpu_asic_name[] = { "VEGAM", "VEGA10", "VEGA12", + "VEGA20", "RAVEN", "LAST", }; @@ -1387,6 +1388,7 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev) case CHIP_KABINI: case CHIP_MULLINS: #endif + case CHIP_VEGA20: default: return 0; case CHIP_VEGA10: @@ -1521,6 +1523,7 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) #endif case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: case CHIP_RAVEN: if (adev->asic_type == CHIP_RAVEN) adev->family = AMDGPU_FAMILY_RV; @@ -1715,6 +1718,7 @@ static int amdgpu_device_ip_late_set_cg_state(struct amdgpu_device *adev) /* skip CG for VCE/UVD, it's handled specially */ if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD && adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE && + adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCN && adev->ip_blocks[i].version->funcs->set_clockgating_state) { /* enable clockgating to save power */ r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev, @@ -1814,6 +1818,7 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev) if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD && adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE && + adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCN && adev->ip_blocks[i].version->funcs->set_clockgating_state) { /* ungate blocks before hw fini so that we can shutdown the blocks safely */ r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev, @@ -2155,6 +2160,7 @@ bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type) case CHIP_FIJI: case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: #if defined(CONFIG_DRM_AMD_DC_DCN1_0) case CHIP_RAVEN: #endif @@ -3172,7 +3178,6 @@ error: int amdgpu_device_gpu_recover(struct amdgpu_device *adev, struct amdgpu_job *job, bool force) { - struct drm_atomic_state *state = NULL; int i, r, resched; if (!force && !amdgpu_device_ip_check_soft_reset(adev)) { @@ -3195,10 +3200,6 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, /* block TTM */ resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev); - /* store modesetting */ - if (amdgpu_device_has_dc_support(adev)) - state = drm_atomic_helper_suspend(adev->ddev); - /* block all schedulers and reset given job's ring */ for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { struct amdgpu_ring *ring = adev->rings[i]; @@ -3238,10 +3239,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, kthread_unpark(ring->sched.thread); } - if (amdgpu_device_has_dc_support(adev)) { - if (drm_atomic_helper_resume(adev->ddev, state)) - dev_info(adev->dev, "drm resume failed:%d\n", r); - } else { + if (!amdgpu_device_has_dc_support(adev)) { drm_helper_resume_force_mode(adev->ddev); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 739e7e09c8b0..b0bf2f24da48 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -560,6 +560,13 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x69A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12}, {0x1002, 0x69A3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12}, {0x1002, 0x69AF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12}, + /* Vega 20 */ + {0x1002, 0x66A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT}, + {0x1002, 0x66A1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT}, + {0x1002, 0x66A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT}, + {0x1002, 0x66A3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT}, + {0x1002, 0x66A7, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT}, + {0x1002, 0x66AF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT}, /* Raven */ {0x1002, 0x15dd, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RAVEN|AMD_IS_APU}, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index d09fcab2398f..39ec6b8890a1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -376,14 +376,14 @@ int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring, struct amdgpu_device *adev = ring->adev; uint64_t index; - if (ring != &adev->uvd.ring) { + if (ring != &adev->uvd.inst[ring->me].ring) { ring->fence_drv.cpu_addr = &adev->wb.wb[ring->fence_offs]; ring->fence_drv.gpu_addr = adev->wb.gpu_addr + (ring->fence_offs * 4); } else { /* put fence directly behind firmware */ index = ALIGN(adev->uvd.fw->size, 8); - ring->fence_drv.cpu_addr = adev->uvd.cpu_addr + index; - ring->fence_drv.gpu_addr = adev->uvd.gpu_addr + index; + ring->fence_drv.cpu_addr = adev->uvd.inst[ring->me].cpu_addr + index; + ring->fence_drv.gpu_addr = adev->uvd.inst[ring->me].gpu_addr + index; } amdgpu_fence_write(ring, atomic_read(&ring->fence_drv.last_seq)); amdgpu_irq_get(adev, irq_src, irq_type); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index eb4785e51573..91517b166a3b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -286,7 +286,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file struct drm_crtc *crtc; uint32_t ui32 = 0; uint64_t ui64 = 0; - int i, found; + int i, j, found; int ui32_size = sizeof(ui32); if (!info->return_size || !info->return_pointer) @@ -348,7 +348,8 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file break; case AMDGPU_HW_IP_UVD: type = AMD_IP_BLOCK_TYPE_UVD; - ring_mask = adev->uvd.ring.ready ? 1 : 0; + for (i = 0; i < adev->uvd.num_uvd_inst; i++) + ring_mask |= ((adev->uvd.inst[i].ring.ready ? 1 : 0) << i); ib_start_alignment = AMDGPU_GPU_PAGE_SIZE; ib_size_alignment = 16; break; @@ -361,8 +362,11 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file break; case AMDGPU_HW_IP_UVD_ENC: type = AMD_IP_BLOCK_TYPE_UVD; - for (i = 0; i < adev->uvd.num_enc_rings; i++) - ring_mask |= ((adev->uvd.ring_enc[i].ready ? 1 : 0) << i); + for (i = 0; i < adev->uvd.num_uvd_inst; i++) + for (j = 0; j < adev->uvd.num_enc_rings; j++) + ring_mask |= + ((adev->uvd.inst[i].ring_enc[j].ready ? 1 : 0) << + (j + i * adev->uvd.num_enc_rings)); ib_start_alignment = AMDGPU_GPU_PAGE_SIZE; ib_size_alignment = 1; break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index c7d43e064fc7..9f1a5bd39ae8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -52,6 +52,7 @@ static int psp_sw_init(void *handle) switch (adev->asic_type) { case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: psp_v3_1_set_psp_funcs(psp); break; case CHIP_RAVEN: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c index 262c1267249e..8af16e81c7d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c @@ -66,6 +66,8 @@ static int amdgpu_identity_map(struct amdgpu_device *adev, u32 ring, struct amdgpu_ring **out_ring) { + u32 instance; + switch (mapper->hw_ip) { case AMDGPU_HW_IP_GFX: *out_ring = &adev->gfx.gfx_ring[ring]; @@ -77,13 +79,16 @@ static int amdgpu_identity_map(struct amdgpu_device *adev, *out_ring = &adev->sdma.instance[ring].ring; break; case AMDGPU_HW_IP_UVD: - *out_ring = &adev->uvd.ring; + instance = ring; + *out_ring = &adev->uvd.inst[instance].ring; break; case AMDGPU_HW_IP_VCE: *out_ring = &adev->vce.ring[ring]; break; case AMDGPU_HW_IP_UVD_ENC: - *out_ring = &adev->uvd.ring_enc[ring]; + instance = ring / adev->uvd.num_enc_rings; + *out_ring = + &adev->uvd.inst[instance].ring_enc[ring%adev->uvd.num_enc_rings]; break; case AMDGPU_HW_IP_VCN_DEC: *out_ring = &adev->vcn.ring_dec; @@ -240,13 +245,14 @@ int amdgpu_queue_mgr_map(struct amdgpu_device *adev, ip_num_rings = adev->sdma.num_instances; break; case AMDGPU_HW_IP_UVD: - ip_num_rings = 1; + ip_num_rings = adev->uvd.num_uvd_inst; break; case AMDGPU_HW_IP_VCE: ip_num_rings = adev->vce.num_rings; break; case AMDGPU_HW_IP_UVD_ENC: - ip_num_rings = adev->uvd.num_enc_rings; + ip_num_rings = + adev->uvd.num_enc_rings * adev->uvd.num_uvd_inst; break; case AMDGPU_HW_IP_VCN_DEC: ip_num_rings = 1; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c index 49cad08b5c16..c6850b629d0e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c @@ -362,6 +362,7 @@ void amdgpu_ring_fini(struct amdgpu_ring *ring) dma_fence_put(ring->vmid_wait); ring->vmid_wait = NULL; + ring->me = 0; ring->adev->rings[ring->idx] = NULL; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h index 4f8dac2d36a5..1513124c5659 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h @@ -29,7 +29,7 @@ #include <drm/drm_print.h> /* max number of rings */ -#define AMDGPU_MAX_RINGS 18 +#define AMDGPU_MAX_RINGS 21 #define AMDGPU_MAX_GFX_RINGS 1 #define AMDGPU_MAX_COMPUTE_RINGS 8 #define AMDGPU_MAX_VCE_RINGS 3 diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 69a2b25b3696..e93a0a237dc3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -63,16 +63,44 @@ static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev); /* * Global memory. */ + +/** + * amdgpu_ttm_mem_global_init - Initialize and acquire reference to + * memory object + * + * @ref: Object for initialization. + * + * This is called by drm_global_item_ref() when an object is being + * initialized. + */ static int amdgpu_ttm_mem_global_init(struct drm_global_reference *ref) { return ttm_mem_global_init(ref->object); } +/** + * amdgpu_ttm_mem_global_release - Drop reference to a memory object + * + * @ref: Object being removed + * + * This is called by drm_global_item_unref() when an object is being + * released. + */ static void amdgpu_ttm_mem_global_release(struct drm_global_reference *ref) { ttm_mem_global_release(ref->object); } +/** + * amdgpu_ttm_global_init - Initialize global TTM memory reference + * structures. + * + * @adev: AMDGPU device for which the global structures need to be + * registered. + * + * This is called as part of the AMDGPU ttm init from amdgpu_ttm_init() + * during bring up. + */ static int amdgpu_ttm_global_init(struct amdgpu_device *adev) { struct drm_global_reference *global_ref; @@ -80,7 +108,9 @@ static int amdgpu_ttm_global_init(struct amdgpu_device *adev) struct drm_sched_rq *rq; int r; + /* ensure reference is false in case init fails */ adev->mman.mem_global_referenced = false; + global_ref = &adev->mman.mem_global_ref; global_ref->global_type = DRM_GLOBAL_TTM_MEM; global_ref->size = sizeof(struct ttm_mem_global); @@ -146,6 +176,18 @@ static int amdgpu_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags) return 0; } +/** + * amdgpu_init_mem_type - Initialize a memory manager for a specific + * type of memory request. + * + * @bdev: The TTM BO device object (contains a reference to + * amdgpu_device) + * @type: The type of memory requested + * @man: + * + * This is called by ttm_bo_init_mm() when a buffer object is being + * initialized. + */ static int amdgpu_init_mem_type(struct ttm_bo_device *bdev, uint32_t type, struct ttm_mem_type_manager *man) { @@ -161,6 +203,7 @@ static int amdgpu_init_mem_type(struct ttm_bo_device *bdev, uint32_t type, man->default_caching = TTM_PL_FLAG_CACHED; break; case TTM_PL_TT: + /* GTT memory */ man->func = &amdgpu_gtt_mgr_func; man->gpu_offset = adev->gmc.gart_start; man->available_caching = TTM_PL_MASK_CACHING; @@ -193,6 +236,14 @@ static int amdgpu_init_mem_type(struct ttm_bo_device *bdev, uint32_t type, return 0; } +/** + * amdgpu_evict_flags - Compute placement flags + * + * @bo: The buffer object to evict + * @placement: Possible destination(s) for evicted BO + * + * Fill in placement data when ttm_bo_evict() is called + */ static void amdgpu_evict_flags(struct ttm_buffer_object *bo, struct ttm_placement *placement) { @@ -204,12 +255,14 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo, .flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM }; + /* Don't handle scatter gather BOs */ if (bo->type == ttm_bo_type_sg) { placement->num_placement = 0; placement->num_busy_placement = 0; return; } + /* Object isn't an AMDGPU object so ignore */ if (!amdgpu_ttm_bo_is_amdgpu_bo(bo)) { placement->placement = &placements; placement->busy_placement = &placements; @@ -217,10 +270,12 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo, placement->num_busy_placement = 1; return; } + abo = ttm_to_amdgpu_bo(bo); switch (bo->mem.mem_type) { case TTM_PL_VRAM: if (!adev->mman.buffer_funcs_enabled) { + /* Move to system memory */ amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU); } else if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size && !(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) && @@ -238,6 +293,7 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo, abo->placement.busy_placement = &abo->placements[1]; abo->placement.num_busy_placement = 1; } else { + /* Move to GTT memory */ amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT); } break; @@ -248,6 +304,15 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo, *placement = abo->placement; } +/** + * amdgpu_verify_access - Verify access for a mmap call + * + * @bo: The buffer object to map + * @filp: The file pointer from the process performing the mmap + * + * This is called by ttm_bo_mmap() to verify whether a process + * has the right to mmap a BO to their process space. + */ static int amdgpu_verify_access(struct ttm_buffer_object *bo, struct file *filp) { struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo); @@ -265,6 +330,15 @@ static int amdgpu_verify_access(struct ttm_buffer_object *bo, struct file *filp) filp->private_data); } +/** + * amdgpu_move_null - Register memory for a buffer object + * + * @bo: The bo to assign the memory to + * @new_mem: The memory to be assigned. + * + * Assign the memory from new_mem to the memory of the buffer object + * bo. + */ static void amdgpu_move_null(struct ttm_buffer_object *bo, struct ttm_mem_reg *new_mem) { @@ -275,6 +349,10 @@ static void amdgpu_move_null(struct ttm_buffer_object *bo, new_mem->mm_node = NULL; } +/** + * amdgpu_mm_node_addr - Compute the GPU relative offset of a GTT + * buffer. + */ static uint64_t amdgpu_mm_node_addr(struct ttm_buffer_object *bo, struct drm_mm_node *mm_node, struct ttm_mem_reg *mem) @@ -289,9 +367,10 @@ static uint64_t amdgpu_mm_node_addr(struct ttm_buffer_object *bo, } /** - * amdgpu_find_mm_node - Helper function finds the drm_mm_node - * corresponding to @offset. It also modifies the offset to be - * within the drm_mm_node returned + * amdgpu_find_mm_node - Helper function finds the drm_mm_node + * corresponding to @offset. It also modifies + * the offset to be within the drm_mm_node + * returned */ static struct drm_mm_node *amdgpu_find_mm_node(struct ttm_mem_reg *mem, unsigned long *offset) @@ -430,7 +509,12 @@ error: return r; } - +/** + * amdgpu_move_blit - Copy an entire buffer to another buffer + * + * This is a helper called by amdgpu_bo_move() and + * amdgpu_move_vram_ram() to help move buffers to and from VRAM. + */ static int amdgpu_move_blit(struct ttm_buffer_object *bo, bool evict, bool no_wait_gpu, struct ttm_mem_reg *new_mem, @@ -465,6 +549,11 @@ error: return r; } +/** + * amdgpu_move_vram_ram - Copy VRAM buffer to RAM buffer + * + * Called by amdgpu_bo_move(). + */ static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo, bool evict, struct ttm_operation_ctx *ctx, struct ttm_mem_reg *new_mem) @@ -477,6 +566,8 @@ static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo, bool evict, int r; adev = amdgpu_ttm_adev(bo->bdev); + + /* create space/pages for new_mem in GTT space */ tmp_mem = *new_mem; tmp_mem.mm_node = NULL; placement.num_placement = 1; @@ -491,25 +582,36 @@ static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo, bool evict, return r; } + /* set caching flags */ r = ttm_tt_set_placement_caching(bo->ttm, tmp_mem.placement); if (unlikely(r)) { goto out_cleanup; } + /* Bind the memory to the GTT space */ r = ttm_tt_bind(bo->ttm, &tmp_mem, ctx); if (unlikely(r)) { goto out_cleanup; } + + /* blit VRAM to GTT */ r = amdgpu_move_blit(bo, true, ctx->no_wait_gpu, &tmp_mem, old_mem); if (unlikely(r)) { goto out_cleanup; } + + /* move BO (in tmp_mem) to new_mem */ r = ttm_bo_move_ttm(bo, ctx, new_mem); out_cleanup: ttm_bo_mem_put(bo, &tmp_mem); return r; } +/** + * amdgpu_move_ram_vram - Copy buffer from RAM to VRAM + * + * Called by amdgpu_bo_move(). + */ static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo, bool evict, struct ttm_operation_ctx *ctx, struct ttm_mem_reg *new_mem) @@ -522,6 +624,8 @@ static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo, bool evict, int r; adev = amdgpu_ttm_adev(bo->bdev); + + /* make space in GTT for old_mem buffer */ tmp_mem = *new_mem; tmp_mem.mm_node = NULL; placement.num_placement = 1; @@ -535,10 +639,14 @@ static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo, bool evict, if (unlikely(r)) { return r; } + + /* move/bind old memory to GTT space */ r = ttm_bo_move_ttm(bo, ctx, &tmp_mem); if (unlikely(r)) { goto out_cleanup; } + + /* copy to VRAM */ r = amdgpu_move_blit(bo, true, ctx->no_wait_gpu, new_mem, old_mem); if (unlikely(r)) { goto out_cleanup; @@ -548,6 +656,11 @@ out_cleanup: return r; } +/** + * amdgpu_bo_move - Move a buffer object to a new memory location + * + * Called by ttm_bo_handle_move_mem() + */ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict, struct ttm_operation_ctx *ctx, struct ttm_mem_reg *new_mem) @@ -613,6 +726,11 @@ memcpy: return 0; } +/** + * amdgpu_ttm_io_mem_reserve - Reserve a block of memory during a fault + * + * Called by ttm_mem_io_reserve() ultimately via ttm_bo_vm_fault() + */ static int amdgpu_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) { struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type]; @@ -690,6 +808,14 @@ struct amdgpu_ttm_tt { uint32_t last_set_pages; }; +/** + * amdgpu_ttm_tt_get_user_pages - Pin pages of memory pointed to + * by a USERPTR pointer to memory + * + * Called by amdgpu_gem_userptr_ioctl() and amdgpu_cs_parser_bos(). + * This provides a wrapper around the get_user_pages() call to provide + * device accessible pages that back user memory. + */ int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -719,6 +845,7 @@ int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages) } } + /* loop enough times using contiguous pages of memory */ do { unsigned num_pages = ttm->num_pages - pinned; uint64_t userptr = gtt->userptr + pinned * PAGE_SIZE; @@ -757,6 +884,14 @@ release_pages: return r; } +/** + * amdgpu_ttm_tt_set_user_pages - Copy pages in, putting old pages + * as necessary. + * + * Called by amdgpu_cs_list_validate(). This creates the page list + * that backs user memory and will ultimately be mapped into the device + * address space. + */ void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -771,6 +906,11 @@ void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages) } } +/** + * amdgpu_ttm_tt_mark_user_page - Mark pages as dirty + * + * Called while unpinning userptr pages + */ void amdgpu_ttm_tt_mark_user_pages(struct ttm_tt *ttm) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -789,7 +929,12 @@ void amdgpu_ttm_tt_mark_user_pages(struct ttm_tt *ttm) } } -/* prepare the sg table with the user pages */ +/** + * amdgpu_ttm_tt_pin_userptr - prepare the sg table with the + * user pages + * + * Called by amdgpu_ttm_backend_bind() + **/ static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm) { struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev); @@ -801,17 +946,20 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm) enum dma_data_direction direction = write ? DMA_BIDIRECTIONAL : DMA_TO_DEVICE; + /* Allocate an SG array and squash pages into it */ r = sg_alloc_table_from_pages(ttm->sg, ttm->pages, ttm->num_pages, 0, ttm->num_pages << PAGE_SHIFT, GFP_KERNEL); if (r) goto release_sg; + /* Map SG to device */ r = -ENOMEM; nents = dma_map_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction); if (nents != ttm->sg->nents) goto release_sg; + /* convert SG to linear array of pages and dma addresses */ drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages, gtt->ttm.dma_address, ttm->num_pages); @@ -822,6 +970,9 @@ release_sg: return r; } +/** + * amdgpu_ttm_tt_unpin_userptr - Unpin and unmap userptr pages + */ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm) { struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev); @@ -835,9 +986,10 @@ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm) if (!ttm->sg->sgl) return; - /* free the sg table and pages again */ + /* unmap the pages mapped to the device */ dma_unmap_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction); + /* mark the pages as dirty */ amdgpu_ttm_tt_mark_user_pages(ttm); sg_free_table(ttm->sg); @@ -882,6 +1034,12 @@ gart_bind_fail: return r; } +/** + * amdgpu_ttm_backend_bind - Bind GTT memory + * + * Called by ttm_tt_bind() on behalf of ttm_bo_handle_move_mem(). + * This handles binding GTT memory to the device address space. + */ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm, struct ttm_mem_reg *bo_mem) { @@ -912,7 +1070,10 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm, return 0; } + /* compute PTE flags relevant to this BO memory */ flags = amdgpu_ttm_tt_pte_flags(adev, ttm, bo_mem); + + /* bind pages into GART page tables */ gtt->offset = (u64)bo_mem->start << PAGE_SHIFT; r = amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages, ttm->pages, gtt->ttm.dma_address, flags); @@ -923,6 +1084,9 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm, return r; } +/** + * amdgpu_ttm_alloc_gart - Allocate GART memory for buffer object + */ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo) { struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev); @@ -938,6 +1102,7 @@ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo) amdgpu_gtt_mgr_has_gart_addr(&bo->mem)) return 0; + /* allocate GTT space */ tmp = bo->mem; tmp.mm_node = NULL; placement.num_placement = 1; @@ -953,7 +1118,10 @@ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo) if (unlikely(r)) return r; + /* compute PTE flags for this buffer object */ flags = amdgpu_ttm_tt_pte_flags(adev, bo->ttm, &tmp); + + /* Bind pages */ gtt->offset = (u64)tmp.start << PAGE_SHIFT; r = amdgpu_ttm_gart_bind(adev, bo, flags); if (unlikely(r)) { @@ -969,6 +1137,12 @@ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo) return 0; } +/** + * amdgpu_ttm_recover_gart - Rebind GTT pages + * + * Called by amdgpu_gtt_mgr_recover() from amdgpu_device_reset() to + * rebind GTT pages during a GPU reset. + */ int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo) { struct amdgpu_device *adev = amdgpu_ttm_adev(tbo->bdev); @@ -984,12 +1158,19 @@ int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo) return r; } +/** + * amdgpu_ttm_backend_unbind - Unbind GTT mapped pages + * + * Called by ttm_tt_unbind() on behalf of ttm_bo_move_ttm() and + * ttm_tt_destroy(). + */ static int amdgpu_ttm_backend_unbind(struct ttm_tt *ttm) { struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev); struct amdgpu_ttm_tt *gtt = (void *)ttm; int r; + /* if the pages have userptr pinning then clear that first */ if (gtt->userptr) amdgpu_ttm_tt_unpin_userptr(ttm); @@ -1021,6 +1202,13 @@ static struct ttm_backend_func amdgpu_backend_func = { .destroy = &amdgpu_ttm_backend_destroy, }; +/** + * amdgpu_ttm_tt_create - Create a ttm_tt object for a given BO + * + * @bo: The buffer object to create a GTT ttm_tt object around + * + * Called by ttm_tt_create(). + */ static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_buffer_object *bo, uint32_t page_flags) { @@ -1034,6 +1222,8 @@ static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_buffer_object *bo, return NULL; } gtt->ttm.ttm.func = &amdgpu_backend_func; + + /* allocate space for the uninitialized page entries */ if (ttm_sg_tt_init(>t->ttm, bo, page_flags)) { kfree(gtt); return NULL; @@ -1041,6 +1231,12 @@ static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_buffer_object *bo, return >t->ttm.ttm; } +/** + * amdgpu_ttm_tt_populate - Map GTT pages visible to the device + * + * Map the pages of a ttm_tt object to an address space visible + * to the underlying device. + */ static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm, struct ttm_operation_ctx *ctx) { @@ -1048,6 +1244,7 @@ static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm, struct amdgpu_ttm_tt *gtt = (void *)ttm; bool slave = !!(ttm->page_flags & TTM_PAGE_FLAG_SG); + /* user pages are bound by amdgpu_ttm_tt_pin_userptr() */ if (gtt && gtt->userptr) { ttm->sg = kzalloc(sizeof(struct sg_table), GFP_KERNEL); if (!ttm->sg) @@ -1072,9 +1269,17 @@ static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm, } #endif + /* fall back to generic helper to populate the page array + * and map them to the device */ return ttm_populate_and_map_pages(adev->dev, >t->ttm, ctx); } +/** + * amdgpu_ttm_tt_unpopulate - unmap GTT pages and unpopulate page arrays + * + * Unmaps pages of a ttm_tt object from the device address space and + * unpopulates the page array backing it. + */ static void amdgpu_ttm_tt_unpopulate(struct ttm_tt *ttm) { struct amdgpu_device *adev; @@ -1100,9 +1305,21 @@ static void amdgpu_ttm_tt_unpopulate(struct ttm_tt *ttm) } #endif + /* fall back to generic helper to unmap and unpopulate array */ ttm_unmap_and_unpopulate_pages(adev->dev, >t->ttm); } +/** + * amdgpu_ttm_tt_set_userptr - Initialize userptr GTT ttm_tt + * for the current task + * + * @ttm: The ttm_tt object to bind this userptr object to + * @addr: The address in the current tasks VM space to use + * @flags: Requirements of userptr object. + * + * Called by amdgpu_gem_userptr_ioctl() to bind userptr pages + * to current task + */ int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr, uint32_t flags) { @@ -1127,6 +1344,9 @@ int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr, return 0; } +/** + * amdgpu_ttm_tt_get_usermm - Return memory manager for ttm_tt object + */ struct mm_struct *amdgpu_ttm_tt_get_usermm(struct ttm_tt *ttm) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -1140,6 +1360,12 @@ struct mm_struct *amdgpu_ttm_tt_get_usermm(struct ttm_tt *ttm) return gtt->usertask->mm; } +/** + * amdgpu_ttm_tt_affect_userptr - Determine if a ttm_tt object lays + * inside an address range for the + * current task. + * + */ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start, unsigned long end) { @@ -1150,10 +1376,16 @@ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start, if (gtt == NULL || !gtt->userptr) return false; + /* Return false if no part of the ttm_tt object lies within + * the range + */ size = (unsigned long)gtt->ttm.ttm.num_pages * PAGE_SIZE; if (gtt->userptr > end || gtt->userptr + size <= start) return false; + /* Search the lists of tasks that hold this mapping and see + * if current is one of them. If it is return false. + */ spin_lock(>t->guptasklock); list_for_each_entry(entry, >t->guptasks, list) { if (entry->task == current) { @@ -1168,6 +1400,10 @@ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start, return true; } +/** + * amdgpu_ttm_tt_userptr_invalidated - Has the ttm_tt object been + * invalidated? + */ bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm, int *last_invalidated) { @@ -1178,6 +1414,12 @@ bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm, return prev_invalidated != *last_invalidated; } +/** + * amdgpu_ttm_tt_userptr_needs_pages - Have the pages backing this + * ttm_tt object been invalidated + * since the last time they've + * been set? + */ bool amdgpu_ttm_tt_userptr_needs_pages(struct ttm_tt *ttm) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -1188,6 +1430,9 @@ bool amdgpu_ttm_tt_userptr_needs_pages(struct ttm_tt *ttm) return atomic_read(>t->mmu_invalidations) != gtt->last_set_pages; } +/** + * amdgpu_ttm_tt_is_readonly - Is the ttm_tt object read only? + */ bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -1198,6 +1443,12 @@ bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm) return !!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY); } +/** + * amdgpu_ttm_tt_pte_flags - Compute PTE flags for ttm_tt object + * + * @ttm: The ttm_tt object to compute the flags for + * @mem: The memory registry backing this ttm_tt object + */ uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm, struct ttm_mem_reg *mem) { @@ -1222,6 +1473,16 @@ uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm, return flags; } +/** + * amdgpu_ttm_bo_eviction_valuable - Check to see if we can evict + * a buffer object. + * + * Return true if eviction is sensible. Called by + * ttm_mem_evict_first() on behalf of ttm_bo_mem_force_space() + * which tries to evict buffer objects until it can find space + * for a new object and by ttm_bo_force_list_clean() which is + * used to clean out a memory space. + */ static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo, const struct ttm_place *place) { @@ -1268,6 +1529,19 @@ static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo, return ttm_bo_eviction_valuable(bo, place); } +/** + * amdgpu_ttm_access_memory - Read or Write memory that backs a + * buffer object. + * + * @bo: The buffer object to read/write + * @offset: Offset into buffer object + * @buf: Secondary buffer to write/read from + * @len: Length in bytes of access + * @write: true if writing + * + * This is used to access VRAM that backs a buffer object via MMIO + * access for debugging purposes. + */ static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo, unsigned long offset, void *buf, int len, int write) @@ -1444,13 +1718,22 @@ error_create: adev->fw_vram_usage.reserved_bo = NULL; return r; } - +/** + * amdgpu_ttm_init - Init the memory management (ttm) as well as + * various gtt/vram related fields. + * + * This initializes all of the memory space pools that the TTM layer + * will need such as the GTT space (system memory mapped to the device), + * VRAM (on-board memory), and on-chip memories (GDS, GWS, OA) which + * can be mapped per VMID. + */ int amdgpu_ttm_init(struct amdgpu_device *adev) { uint64_t gtt_size; int r; u64 vis_vram_limit; + /* initialize global references for vram/gtt */ r = amdgpu_ttm_global_init(adev); if (r) { return r; @@ -1471,6 +1754,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) /* We opt to avoid OOM on system pages allocations */ adev->mman.bdev.no_retry = true; + /* Initialize VRAM pool with all of VRAM divided into pages */ r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_VRAM, adev->gmc.real_vram_size >> PAGE_SHIFT); if (r) { @@ -1500,6 +1784,10 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) return r; } + /* allocate memory as required for VGA + * This is used for VGA emulation and pre-OS scanout buffers to + * avoid display artifacts while transitioning between pre-OS + * and driver. */ if (adev->gmc.stolen_size) { r = amdgpu_bo_create_kernel(adev, adev->gmc.stolen_size, PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM, @@ -1511,6 +1799,8 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) DRM_INFO("amdgpu: %uM of VRAM memory ready\n", (unsigned) (adev->gmc.real_vram_size / (1024 * 1024))); + /* Compute GTT size, either bsaed on 3/4th the size of RAM size + * or whatever the user passed on module init */ if (amdgpu_gtt_size == -1) { struct sysinfo si; @@ -1521,6 +1811,8 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) } else gtt_size = (uint64_t)amdgpu_gtt_size << 20; + + /* Initialize GTT memory pool */ r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_TT, gtt_size >> PAGE_SHIFT); if (r) { DRM_ERROR("Failed initializing GTT heap.\n"); @@ -1529,6 +1821,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) DRM_INFO("amdgpu: %uM of GTT memory ready.\n", (unsigned)(gtt_size / (1024 * 1024))); + /* Initialize various on-chip memory pools */ adev->gds.mem.total_size = adev->gds.mem.total_size << AMDGPU_GDS_SHIFT; adev->gds.mem.gfx_partition_size = adev->gds.mem.gfx_partition_size << AMDGPU_GDS_SHIFT; adev->gds.mem.cs_partition_size = adev->gds.mem.cs_partition_size << AMDGPU_GDS_SHIFT; @@ -1568,6 +1861,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) } } + /* Register debugfs entries for amdgpu_ttm */ r = amdgpu_ttm_debugfs_init(adev); if (r) { DRM_ERROR("Failed to init debugfs\n"); @@ -1576,11 +1870,19 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) return 0; } +/** + * amdgpu_ttm_late_init - Handle any late initialization for + * amdgpu_ttm + */ void amdgpu_ttm_late_init(struct amdgpu_device *adev) { + /* return the VGA stolen memory (if any) back to VRAM */ amdgpu_bo_free_kernel(&adev->stolen_vga_memory, NULL, NULL); } +/** + * amdgpu_ttm_fini - De-initialize the TTM memory pools + */ void amdgpu_ttm_fini(struct amdgpu_device *adev) { if (!adev->mman.initialized) @@ -1908,6 +2210,11 @@ static const struct drm_info_list amdgpu_ttm_debugfs_list[] = { #endif }; +/** + * amdgpu_ttm_vram_read - Linear read access to VRAM + * + * Accesses VRAM via MMIO for debugging purposes. + */ static ssize_t amdgpu_ttm_vram_read(struct file *f, char __user *buf, size_t size, loff_t *pos) { @@ -1947,6 +2254,11 @@ static ssize_t amdgpu_ttm_vram_read(struct file *f, char __user *buf, return result; } +/** + * amdgpu_ttm_vram_write - Linear write access to VRAM + * + * Accesses VRAM via MMIO for debugging purposes. + */ static ssize_t amdgpu_ttm_vram_write(struct file *f, const char __user *buf, size_t size, loff_t *pos) { @@ -1995,6 +2307,9 @@ static const struct file_operations amdgpu_ttm_vram_fops = { #ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS +/** + * amdgpu_ttm_gtt_read - Linear read access to GTT memory + */ static ssize_t amdgpu_ttm_gtt_read(struct file *f, char __user *buf, size_t size, loff_t *pos) { @@ -2042,6 +2357,13 @@ static const struct file_operations amdgpu_ttm_gtt_fops = { #endif +/** + * amdgpu_iomem_read - Virtual read access to GPU mapped memory + * + * This function is used to read memory that has been mapped to the + * GPU and the known addresses are not physical addresses but instead + * bus addresses (e.g., what you'd put in an IB or ring buffer). + */ static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf, size_t size, loff_t *pos) { @@ -2050,6 +2372,7 @@ static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf, ssize_t result = 0; int r; + /* retrieve the IOMMU domain if any for this device */ dom = iommu_get_domain_for_dev(adev->dev); while (size) { @@ -2062,6 +2385,10 @@ static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf, bytes = bytes < size ? bytes : size; + /* Translate the bus address to a physical address. If + * the domain is NULL it means there is no IOMMU active + * and the address translation is the identity + */ addr = dom ? iommu_iova_to_phys(dom, addr) : addr; pfn = addr >> PAGE_SHIFT; @@ -2086,6 +2413,13 @@ static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf, return result; } +/** + * amdgpu_iomem_write - Virtual write access to GPU mapped memory + * + * This function is used to write memory that has been mapped to the + * GPU and the known addresses are not physical addresses but instead + * bus addresses (e.g., what you'd put in an IB or ring buffer). + */ static ssize_t amdgpu_iomem_write(struct file *f, const char __user *buf, size_t size, loff_t *pos) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c index 75592bd04d6a..f55f72a37ca8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c @@ -307,6 +307,8 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type) return AMDGPU_FW_LOAD_DIRECT; else return AMDGPU_FW_LOAD_PSP; + case CHIP_VEGA20: + return AMDGPU_FW_LOAD_DIRECT; default: DRM_ERROR("Unknown firmware load type\n"); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c index de4d77af02ae..bcf68f80bbf0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c @@ -70,12 +70,14 @@ #define FIRMWARE_VEGA10 "amdgpu/vega10_uvd.bin" #define FIRMWARE_VEGA12 "amdgpu/vega12_uvd.bin" +#define FIRMWARE_VEGA20 "amdgpu/vega20_uvd.bin" -#define mmUVD_GPCOM_VCPU_DATA0_VEGA10 (0x03c4 + 0x7e00) -#define mmUVD_GPCOM_VCPU_DATA1_VEGA10 (0x03c5 + 0x7e00) -#define mmUVD_GPCOM_VCPU_CMD_VEGA10 (0x03c3 + 0x7e00) -#define mmUVD_NO_OP_VEGA10 (0x03ff + 0x7e00) -#define mmUVD_ENGINE_CNTL_VEGA10 (0x03c6 + 0x7e00) +/* These are common relative offsets for all asics, from uvd_7_0_offset.h, */ +#define UVD_GPCOM_VCPU_CMD 0x03c3 +#define UVD_GPCOM_VCPU_DATA0 0x03c4 +#define UVD_GPCOM_VCPU_DATA1 0x03c5 +#define UVD_NO_OP 0x03ff +#define UVD_BASE_SI 0x3800 /** * amdgpu_uvd_cs_ctx - Command submission parser context @@ -114,6 +116,7 @@ MODULE_FIRMWARE(FIRMWARE_VEGAM); MODULE_FIRMWARE(FIRMWARE_VEGA10); MODULE_FIRMWARE(FIRMWARE_VEGA12); +MODULE_FIRMWARE(FIRMWARE_VEGA20); static void amdgpu_uvd_idle_work_handler(struct work_struct *work); @@ -125,9 +128,9 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev) const char *fw_name; const struct common_firmware_header *hdr; unsigned version_major, version_minor, family_id; - int i, r; + int i, j, r; - INIT_DELAYED_WORK(&adev->uvd.idle_work, amdgpu_uvd_idle_work_handler); + INIT_DELAYED_WORK(&adev->uvd.inst->idle_work, amdgpu_uvd_idle_work_handler); switch (adev->asic_type) { #ifdef CONFIG_DRM_AMDGPU_CIK @@ -177,6 +180,9 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev) case CHIP_VEGAM: fw_name = FIRMWARE_VEGAM; break; + case CHIP_VEGA20: + fw_name = FIRMWARE_VEGA20; + break; default: return -EINVAL; } @@ -231,28 +237,30 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev) if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8); - r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, &adev->uvd.vcpu_bo, - &adev->uvd.gpu_addr, &adev->uvd.cpu_addr); - if (r) { - dev_err(adev->dev, "(%d) failed to allocate UVD bo\n", r); - return r; - } + for (j = 0; j < adev->uvd.num_uvd_inst; j++) { - ring = &adev->uvd.ring; - rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_NORMAL]; - r = drm_sched_entity_init(&ring->sched, &adev->uvd.entity, - rq, NULL); - if (r != 0) { - DRM_ERROR("Failed setting up UVD run queue.\n"); - return r; - } + r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, &adev->uvd.inst[j].vcpu_bo, + &adev->uvd.inst[j].gpu_addr, &adev->uvd.inst[j].cpu_addr); + if (r) { + dev_err(adev->dev, "(%d) failed to allocate UVD bo\n", r); + return r; + } - for (i = 0; i < adev->uvd.max_handles; ++i) { - atomic_set(&adev->uvd.handles[i], 0); - adev->uvd.filp[i] = NULL; - } + ring = &adev->uvd.inst[j].ring; + rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_NORMAL]; + r = drm_sched_entity_init(&ring->sched, &adev->uvd.inst[j].entity, + rq, NULL); + if (r != 0) { + DRM_ERROR("Failed setting up UVD(%d) run queue.\n", j); + return r; + } + for (i = 0; i < adev->uvd.max_handles; ++i) { + atomic_set(&adev->uvd.inst[j].handles[i], 0); + adev->uvd.inst[j].filp[i] = NULL; + } + } /* from uvd v5.0 HW addressing capacity increased to 64 bits */ if (!amdgpu_device_ip_block_version_cmp(adev, AMD_IP_BLOCK_TYPE_UVD, 5, 0)) adev->uvd.address_64_bit = true; @@ -279,20 +287,22 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev) int amdgpu_uvd_sw_fini(struct amdgpu_device *adev) { - int i; - kfree(adev->uvd.saved_bo); + int i, j; - drm_sched_entity_fini(&adev->uvd.ring.sched, &adev->uvd.entity); + for (j = 0; j < adev->uvd.num_uvd_inst; ++j) { + kfree(adev->uvd.inst[j].saved_bo); - amdgpu_bo_free_kernel(&adev->uvd.vcpu_bo, - &adev->uvd.gpu_addr, - (void **)&adev->uvd.cpu_addr); + drm_sched_entity_fini(&adev->uvd.inst[j].ring.sched, &adev->uvd.inst[j].entity); - amdgpu_ring_fini(&adev->uvd.ring); + amdgpu_bo_free_kernel(&adev->uvd.inst[j].vcpu_bo, + &adev->uvd.inst[j].gpu_addr, + (void **)&adev->uvd.inst[j].cpu_addr); - for (i = 0; i < AMDGPU_MAX_UVD_ENC_RINGS; ++i) - amdgpu_ring_fini(&adev->uvd.ring_enc[i]); + amdgpu_ring_fini(&adev->uvd.inst[j].ring); + for (i = 0; i < AMDGPU_MAX_UVD_ENC_RINGS; ++i) + amdgpu_ring_fini(&adev->uvd.inst[j].ring_enc[i]); + } release_firmware(adev->uvd.fw); return 0; @@ -302,32 +312,33 @@ int amdgpu_uvd_suspend(struct amdgpu_device *adev) { unsigned size; void *ptr; - int i; - - if (adev->uvd.vcpu_bo == NULL) - return 0; + int i, j; - cancel_delayed_work_sync(&adev->uvd.idle_work); + for (j = 0; j < adev->uvd.num_uvd_inst; ++j) { + if (adev->uvd.inst[j].vcpu_bo == NULL) + continue; - /* only valid for physical mode */ - if (adev->asic_type < CHIP_POLARIS10) { - for (i = 0; i < adev->uvd.max_handles; ++i) - if (atomic_read(&adev->uvd.handles[i])) - break; + cancel_delayed_work_sync(&adev->uvd.inst[j].idle_work); - if (i == adev->uvd.max_handles) - return 0; - } + /* only valid for physical mode */ + if (adev->asic_type < CHIP_POLARIS10) { + for (i = 0; i < adev->uvd.max_handles; ++i) + if (atomic_read(&adev->uvd.inst[j].handles[i])) + break; - size = amdgpu_bo_size(adev->uvd.vcpu_bo); - ptr = adev->uvd.cpu_addr; + if (i == adev->uvd.max_handles) + continue; + } - adev->uvd.saved_bo = kmalloc(size, GFP_KERNEL); - if (!adev->uvd.saved_bo) - return -ENOMEM; + size = amdgpu_bo_size(adev->uvd.inst[j].vcpu_bo); + ptr = adev->uvd.inst[j].cpu_addr; - memcpy_fromio(adev->uvd.saved_bo, ptr, size); + adev->uvd.inst[j].saved_bo = kmalloc(size, GFP_KERNEL); + if (!adev->uvd.inst[j].saved_bo) + return -ENOMEM; + memcpy_fromio(adev->uvd.inst[j].saved_bo, ptr, size); + } return 0; } @@ -335,59 +346,65 @@ int amdgpu_uvd_resume(struct amdgpu_device *adev) { unsigned size; void *ptr; + int i; - if (adev->uvd.vcpu_bo == NULL) - return -EINVAL; + for (i = 0; i < adev->uvd.num_uvd_inst; i++) { + if (adev->uvd.inst[i].vcpu_bo == NULL) + return -EINVAL; - size = amdgpu_bo_size(adev->uvd.vcpu_bo); - ptr = adev->uvd.cpu_addr; + size = amdgpu_bo_size(adev->uvd.inst[i].vcpu_bo); + ptr = adev->uvd.inst[i].cpu_addr; - if (adev->uvd.saved_bo != NULL) { - memcpy_toio(ptr, adev->uvd.saved_bo, size); - kfree(adev->uvd.saved_bo); - adev->uvd.saved_bo = NULL; - } else { - const struct common_firmware_header *hdr; - unsigned offset; - - hdr = (const struct common_firmware_header *)adev->uvd.fw->data; - if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) { - offset = le32_to_cpu(hdr->ucode_array_offset_bytes); - memcpy_toio(adev->uvd.cpu_addr, adev->uvd.fw->data + offset, - le32_to_cpu(hdr->ucode_size_bytes)); - size -= le32_to_cpu(hdr->ucode_size_bytes); - ptr += le32_to_cpu(hdr->ucode_size_bytes); + if (adev->uvd.inst[i].saved_bo != NULL) { + memcpy_toio(ptr, adev->uvd.inst[i].saved_bo, size); + kfree(adev->uvd.inst[i].saved_bo); + adev->uvd.inst[i].saved_bo = NULL; + } else { + const struct common_firmware_header *hdr; + unsigned offset; + + hdr = (const struct common_firmware_header *)adev->uvd.fw->data; + if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) { + offset = le32_to_cpu(hdr->ucode_array_offset_bytes); + memcpy_toio(adev->uvd.inst[i].cpu_addr, adev->uvd.fw->data + offset, + le32_to_cpu(hdr->ucode_size_bytes)); + size -= le32_to_cpu(hdr->ucode_size_bytes); + ptr += le32_to_cpu(hdr->ucode_size_bytes); + } + memset_io(ptr, 0, size); + /* to restore uvd fence seq */ + amdgpu_fence_driver_force_completion(&adev->uvd.inst[i].ring); } - memset_io(ptr, 0, size); - /* to restore uvd fence seq */ - amdgpu_fence_driver_force_completion(&adev->uvd.ring); } - return 0; } void amdgpu_uvd_free_handles(struct amdgpu_device *adev, struct drm_file *filp) { - struct amdgpu_ring *ring = &adev->uvd.ring; - int i, r; + struct amdgpu_ring *ring; + int i, j, r; - for (i = 0; i < adev->uvd.max_handles; ++i) { - uint32_t handle = atomic_read(&adev->uvd.handles[i]); - if (handle != 0 && adev->uvd.filp[i] == filp) { - struct dma_fence *fence; - - r = amdgpu_uvd_get_destroy_msg(ring, handle, - false, &fence); - if (r) { - DRM_ERROR("Error destroying UVD (%d)!\n", r); - continue; - } + for (j = 0; j < adev->uvd.num_uvd_inst; j++) { + ring = &adev->uvd.inst[j].ring; + + for (i = 0; i < adev->uvd.max_handles; ++i) { + uint32_t handle = atomic_read(&adev->uvd.inst[j].handles[i]); + if (handle != 0 && adev->uvd.inst[j].filp[i] == filp) { + struct dma_fence *fence; + + r = amdgpu_uvd_get_destroy_msg(ring, handle, + false, &fence); + if (r) { + DRM_ERROR("Error destroying UVD(%d) %d!\n", j, r); + continue; + } - dma_fence_wait(fence, false); - dma_fence_put(fence); + dma_fence_wait(fence, false); + dma_fence_put(fence); - adev->uvd.filp[i] = NULL; - atomic_set(&adev->uvd.handles[i], 0); + adev->uvd.inst[j].filp[i] = NULL; + atomic_set(&adev->uvd.inst[j].handles[i], 0); + } } } } @@ -662,15 +679,16 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx, void *ptr; long r; int i; + uint32_t ip_instance = ctx->parser->job->ring->me; if (offset & 0x3F) { - DRM_ERROR("UVD messages must be 64 byte aligned!\n"); + DRM_ERROR("UVD(%d) messages must be 64 byte aligned!\n", ip_instance); return -EINVAL; } r = amdgpu_bo_kmap(bo, &ptr); if (r) { - DRM_ERROR("Failed mapping the UVD message (%ld)!\n", r); + DRM_ERROR("Failed mapping the UVD(%d) message (%ld)!\n", ip_instance, r); return r; } @@ -680,7 +698,7 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx, handle = msg[2]; if (handle == 0) { - DRM_ERROR("Invalid UVD handle!\n"); + DRM_ERROR("Invalid UVD(%d) handle!\n", ip_instance); return -EINVAL; } @@ -691,18 +709,18 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx, /* try to alloc a new handle */ for (i = 0; i < adev->uvd.max_handles; ++i) { - if (atomic_read(&adev->uvd.handles[i]) == handle) { - DRM_ERROR("Handle 0x%x already in use!\n", handle); + if (atomic_read(&adev->uvd.inst[ip_instance].handles[i]) == handle) { + DRM_ERROR("(%d)Handle 0x%x already in use!\n", ip_instance, handle); return -EINVAL; } - if (!atomic_cmpxchg(&adev->uvd.handles[i], 0, handle)) { - adev->uvd.filp[i] = ctx->parser->filp; + if (!atomic_cmpxchg(&adev->uvd.inst[ip_instance].handles[i], 0, handle)) { + adev->uvd.inst[ip_instance].filp[i] = ctx->parser->filp; return 0; } } - DRM_ERROR("No more free UVD handles!\n"); + DRM_ERROR("No more free UVD(%d) handles!\n", ip_instance); return -ENOSPC; case 1: @@ -714,27 +732,27 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx, /* validate the handle */ for (i = 0; i < adev->uvd.max_handles; ++i) { - if (atomic_read(&adev->uvd.handles[i]) == handle) { - if (adev->uvd.filp[i] != ctx->parser->filp) { - DRM_ERROR("UVD handle collision detected!\n"); + if (atomic_read(&adev->uvd.inst[ip_instance].handles[i]) == handle) { + if (adev->uvd.inst[ip_instance].filp[i] != ctx->parser->filp) { + DRM_ERROR("UVD(%d) handle collision detected!\n", ip_instance); return -EINVAL; } return 0; } } - DRM_ERROR("Invalid UVD handle 0x%x!\n", handle); + DRM_ERROR("Invalid UVD(%d) handle 0x%x!\n", ip_instance, handle); return -ENOENT; case 2: /* it's a destroy msg, free the handle */ for (i = 0; i < adev->uvd.max_handles; ++i) - atomic_cmpxchg(&adev->uvd.handles[i], handle, 0); + atomic_cmpxchg(&adev->uvd.inst[ip_instance].handles[i], handle, 0); amdgpu_bo_kunmap(bo); return 0; default: - DRM_ERROR("Illegal UVD message type (%d)!\n", msg_type); + DRM_ERROR("Illegal UVD(%d) message type (%d)!\n", ip_instance, msg_type); return -EINVAL; } BUG(); @@ -805,7 +823,7 @@ static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx) } if ((cmd == 0 || cmd == 0x3) && - (start >> 28) != (ctx->parser->adev->uvd.gpu_addr >> 28)) { + (start >> 28) != (ctx->parser->adev->uvd.inst->gpu_addr >> 28)) { DRM_ERROR("msg/fb buffer %LX-%LX out of 256MB segment!\n", start, end); return -EINVAL; @@ -973,6 +991,8 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo, uint64_t addr; long r; int i; + unsigned offset_idx = 0; + unsigned offset[3] = { UVD_BASE_SI, 0, 0 }; amdgpu_bo_kunmap(bo); amdgpu_bo_unpin(bo); @@ -992,17 +1012,16 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo, goto err; if (adev->asic_type >= CHIP_VEGA10) { - data[0] = PACKET0(mmUVD_GPCOM_VCPU_DATA0_VEGA10, 0); - data[1] = PACKET0(mmUVD_GPCOM_VCPU_DATA1_VEGA10, 0); - data[2] = PACKET0(mmUVD_GPCOM_VCPU_CMD_VEGA10, 0); - data[3] = PACKET0(mmUVD_NO_OP_VEGA10, 0); - } else { - data[0] = PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0); - data[1] = PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0); - data[2] = PACKET0(mmUVD_GPCOM_VCPU_CMD, 0); - data[3] = PACKET0(mmUVD_NO_OP, 0); + offset_idx = 1 + ring->me; + offset[1] = adev->reg_offset[UVD_HWIP][0][1]; + offset[2] = adev->reg_offset[UVD_HWIP][1][1]; } + data[0] = PACKET0(offset[offset_idx] + UVD_GPCOM_VCPU_DATA0, 0); + data[1] = PACKET0(offset[offset_idx] + UVD_GPCOM_VCPU_DATA1, 0); + data[2] = PACKET0(offset[offset_idx] + UVD_GPCOM_VCPU_CMD, 0); + data[3] = PACKET0(offset[offset_idx] + UVD_NO_OP, 0); + ib = &job->ibs[0]; addr = amdgpu_bo_gpu_offset(bo); ib->ptr[0] = data[0]; @@ -1038,7 +1057,7 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo, if (r) goto err_free; - r = amdgpu_job_submit(job, ring, &adev->uvd.entity, + r = amdgpu_job_submit(job, ring, &adev->uvd.inst[ring->me].entity, AMDGPU_FENCE_OWNER_UNDEFINED, &f); if (r) goto err_free; @@ -1126,8 +1145,15 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle, static void amdgpu_uvd_idle_work_handler(struct work_struct *work) { struct amdgpu_device *adev = - container_of(work, struct amdgpu_device, uvd.idle_work.work); - unsigned fences = amdgpu_fence_count_emitted(&adev->uvd.ring); + container_of(work, struct amdgpu_device, uvd.inst->idle_work.work); + unsigned fences = 0, i, j; + + for (i = 0; i < adev->uvd.num_uvd_inst; ++i) { + fences += amdgpu_fence_count_emitted(&adev->uvd.inst[i].ring); + for (j = 0; j < adev->uvd.num_enc_rings; ++j) { + fences += amdgpu_fence_count_emitted(&adev->uvd.inst[i].ring_enc[j]); + } + } if (fences == 0) { if (adev->pm.dpm_enabled) { @@ -1141,7 +1167,7 @@ static void amdgpu_uvd_idle_work_handler(struct work_struct *work) AMD_CG_STATE_GATE); } } else { - schedule_delayed_work(&adev->uvd.idle_work, UVD_IDLE_TIMEOUT); + schedule_delayed_work(&adev->uvd.inst->idle_work, UVD_IDLE_TIMEOUT); } } @@ -1153,7 +1179,7 @@ void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring) if (amdgpu_sriov_vf(adev)) return; - set_clocks = !cancel_delayed_work_sync(&adev->uvd.idle_work); + set_clocks = !cancel_delayed_work_sync(&adev->uvd.inst->idle_work); if (set_clocks) { if (adev->pm.dpm_enabled) { amdgpu_dpm_enable_uvd(adev, true); @@ -1170,7 +1196,7 @@ void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring) void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring) { if (!amdgpu_sriov_vf(ring->adev)) - schedule_delayed_work(&ring->adev->uvd.idle_work, UVD_IDLE_TIMEOUT); + schedule_delayed_work(&ring->adev->uvd.inst->idle_work, UVD_IDLE_TIMEOUT); } /** @@ -1184,27 +1210,28 @@ int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout) { struct dma_fence *fence; long r; + uint32_t ip_instance = ring->me; r = amdgpu_uvd_get_create_msg(ring, 1, NULL); if (r) { - DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r); + DRM_ERROR("amdgpu: (%d)failed to get create msg (%ld).\n", ip_instance, r); goto error; } r = amdgpu_uvd_get_destroy_msg(ring, 1, true, &fence); if (r) { - DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r); + DRM_ERROR("amdgpu: (%d)failed to get destroy ib (%ld).\n", ip_instance, r); goto error; } r = dma_fence_wait_timeout(fence, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out.\n"); + DRM_ERROR("amdgpu: (%d)IB test timed out.\n", ip_instance); r = -ETIMEDOUT; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); + DRM_ERROR("amdgpu: (%d)fence wait failed (%ld).\n", ip_instance, r); } else { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + DRM_DEBUG("ib test on (%d)ring %d succeeded\n", ip_instance, ring->idx); r = 0; } @@ -1232,7 +1259,7 @@ uint32_t amdgpu_uvd_used_handles(struct amdgpu_device *adev) * necessarily linear. So we need to count * all non-zero handles. */ - if (atomic_read(&adev->uvd.handles[i])) + if (atomic_read(&adev->uvd.inst->handles[i])) used_handles++; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h index 32ea20b99e53..b1579fba134c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h @@ -31,30 +31,37 @@ #define AMDGPU_UVD_SESSION_SIZE (50*1024) #define AMDGPU_UVD_FIRMWARE_OFFSET 256 +#define AMDGPU_MAX_UVD_INSTANCES 2 + #define AMDGPU_UVD_FIRMWARE_SIZE(adev) \ (AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(((const struct common_firmware_header *)(adev)->uvd.fw->data)->ucode_size_bytes) + \ 8) - AMDGPU_UVD_FIRMWARE_OFFSET) -struct amdgpu_uvd { +struct amdgpu_uvd_inst { struct amdgpu_bo *vcpu_bo; void *cpu_addr; uint64_t gpu_addr; - unsigned fw_version; void *saved_bo; - unsigned max_handles; atomic_t handles[AMDGPU_MAX_UVD_HANDLES]; struct drm_file *filp[AMDGPU_MAX_UVD_HANDLES]; struct delayed_work idle_work; - const struct firmware *fw; /* UVD firmware */ struct amdgpu_ring ring; struct amdgpu_ring ring_enc[AMDGPU_MAX_UVD_ENC_RINGS]; struct amdgpu_irq_src irq; - bool address_64_bit; - bool use_ctx_buf; struct drm_sched_entity entity; struct drm_sched_entity entity_enc; uint32_t srbm_soft_reset; +}; + +struct amdgpu_uvd { + const struct firmware *fw; /* UVD firmware */ + unsigned fw_version; + unsigned max_handles; unsigned num_enc_rings; + uint8_t num_uvd_inst; + bool address_64_bit; + bool use_ctx_buf; + struct amdgpu_uvd_inst inst[AMDGPU_MAX_UVD_INSTANCES]; }; int amdgpu_uvd_sw_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index a86322f5164f..23d960ec1cf2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -57,6 +57,7 @@ #define FIRMWARE_VEGA10 "amdgpu/vega10_vce.bin" #define FIRMWARE_VEGA12 "amdgpu/vega12_vce.bin" +#define FIRMWARE_VEGA20 "amdgpu/vega20_vce.bin" #ifdef CONFIG_DRM_AMDGPU_CIK MODULE_FIRMWARE(FIRMWARE_BONAIRE); @@ -76,6 +77,7 @@ MODULE_FIRMWARE(FIRMWARE_VEGAM); MODULE_FIRMWARE(FIRMWARE_VEGA10); MODULE_FIRMWARE(FIRMWARE_VEGA12); +MODULE_FIRMWARE(FIRMWARE_VEGA20); static void amdgpu_vce_idle_work_handler(struct work_struct *work); @@ -143,6 +145,9 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size) case CHIP_VEGA12: fw_name = FIRMWARE_VEGA12; break; + case CHIP_VEGA20: + fw_name = FIRMWARE_VEGA20; + break; default: return -EINVAL; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index e5d234cf804f..8851bcdfc260 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -205,13 +205,18 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work) struct amdgpu_device *adev = container_of(work, struct amdgpu_device, vcn.idle_work.work); unsigned fences = amdgpu_fence_count_emitted(&adev->vcn.ring_dec); + unsigned i; + + for (i = 0; i < adev->vcn.num_enc_rings; ++i) { + fences += amdgpu_fence_count_emitted(&adev->vcn.ring_enc[i]); + } if (fences == 0) { - if (adev->pm.dpm_enabled) { - /* might be used when with pg/cg + if (adev->pm.dpm_enabled) amdgpu_dpm_enable_uvd(adev, false); - */ - } + else + amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN, + AMD_PG_STATE_GATE); } else { schedule_delayed_work(&adev->vcn.idle_work, VCN_IDLE_TIMEOUT); } @@ -223,9 +228,11 @@ void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring) bool set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work); if (set_clocks && adev->pm.dpm_enabled) { - /* might be used when with pg/cg - amdgpu_dpm_enable_uvd(adev, true); - */ + if (adev->pm.dpm_enabled) + amdgpu_dpm_enable_uvd(adev, true); + else + amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN, + AMD_PG_STATE_UNGATE); } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h index 2fd7db891689..181e6afa9847 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h @@ -45,6 +45,17 @@ #define VCN_ENC_CMD_REG_WRITE 0x0000000b #define VCN_ENC_CMD_REG_WAIT 0x0000000c +enum engine_status_constants { + UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON = 0x2AAAA0, + UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON = 0x00000002, + UVD_STATUS__UVD_BUSY = 0x00000004, + GB_ADDR_CONFIG_DEFAULT = 0x26010011, + UVD_STATUS__IDLE = 0x2, + UVD_STATUS__BUSY = 0x5, + UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF = 0x1, + UVD_STATUS__RBC_BUSY = 0x1, +}; + struct amdgpu_vcn { struct amdgpu_bo *vcpu_bo; void *cpu_addr; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 1a8f4e0dd023..ccba88cc8c54 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -119,9 +119,7 @@ static void amdgpu_vm_bo_base_init(struct amdgpu_vm_bo_base *base, * is currently evicted. add the bo to the evicted list to make sure it * is validated on next vm use to avoid fault. * */ - spin_lock(&vm->status_lock); list_move_tail(&base->vm_status, &vm->evicted); - spin_unlock(&vm->status_lock); } /** @@ -226,24 +224,16 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm, void *param) { struct ttm_bo_global *glob = adev->mman.bdev.glob; - int r; - - spin_lock(&vm->status_lock); - while (!list_empty(&vm->evicted)) { - struct amdgpu_vm_bo_base *bo_base; - struct amdgpu_bo *bo; + struct amdgpu_vm_bo_base *bo_base, *tmp; + int r = 0; - bo_base = list_first_entry(&vm->evicted, - struct amdgpu_vm_bo_base, - vm_status); - spin_unlock(&vm->status_lock); + list_for_each_entry_safe(bo_base, tmp, &vm->evicted, vm_status) { + struct amdgpu_bo *bo = bo_base->bo; - bo = bo_base->bo; - BUG_ON(!bo); if (bo->parent) { r = validate(param, bo); if (r) - return r; + break; spin_lock(&glob->lru_lock); ttm_bo_move_to_lru_tail(&bo->tbo); @@ -252,22 +242,29 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm, spin_unlock(&glob->lru_lock); } - if (bo->tbo.type == ttm_bo_type_kernel && - vm->use_cpu_for_update) { - r = amdgpu_bo_kmap(bo, NULL); - if (r) - return r; - } - - spin_lock(&vm->status_lock); - if (bo->tbo.type != ttm_bo_type_kernel) + if (bo->tbo.type != ttm_bo_type_kernel) { + spin_lock(&vm->moved_lock); list_move(&bo_base->vm_status, &vm->moved); - else + spin_unlock(&vm->moved_lock); + } else { list_move(&bo_base->vm_status, &vm->relocated); + } } - spin_unlock(&vm->status_lock); - return 0; + spin_lock(&glob->lru_lock); + list_for_each_entry(bo_base, &vm->idle, vm_status) { + struct amdgpu_bo *bo = bo_base->bo; + + if (!bo->parent) + continue; + + ttm_bo_move_to_lru_tail(&bo->tbo); + if (bo->shadow) + ttm_bo_move_to_lru_tail(&bo->shadow->tbo); + } + spin_unlock(&glob->lru_lock); + + return r; } /** @@ -279,13 +276,7 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm, */ bool amdgpu_vm_ready(struct amdgpu_vm *vm) { - bool ready; - - spin_lock(&vm->status_lock); - ready = list_empty(&vm->evicted); - spin_unlock(&vm->status_lock); - - return ready; + return list_empty(&vm->evicted); } /** @@ -477,9 +468,7 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev, pt->parent = amdgpu_bo_ref(parent->base.bo); amdgpu_vm_bo_base_init(&entry->base, vm, pt); - spin_lock(&vm->status_lock); list_move(&entry->base.vm_status, &vm->relocated); - spin_unlock(&vm->status_lock); } if (level < AMDGPU_VM_PTB) { @@ -926,10 +915,8 @@ static void amdgpu_vm_invalidate_level(struct amdgpu_device *adev, if (!entry->base.bo) continue; - spin_lock(&vm->status_lock); - if (list_empty(&entry->base.vm_status)) - list_add(&entry->base.vm_status, &vm->relocated); - spin_unlock(&vm->status_lock); + if (!entry->base.moved) + list_move(&entry->base.vm_status, &vm->relocated); amdgpu_vm_invalidate_level(adev, vm, entry, level + 1); } } @@ -959,6 +946,14 @@ restart: params.adev = adev; if (vm->use_cpu_for_update) { + struct amdgpu_vm_bo_base *bo_base; + + list_for_each_entry(bo_base, &vm->relocated, vm_status) { + r = amdgpu_bo_kmap(bo_base->bo, NULL); + if (unlikely(r)) + return r; + } + r = amdgpu_vm_wait_pd(adev, vm, AMDGPU_FENCE_OWNER_VM); if (unlikely(r)) return r; @@ -974,7 +969,6 @@ restart: params.func = amdgpu_vm_do_set_ptes; } - spin_lock(&vm->status_lock); while (!list_empty(&vm->relocated)) { struct amdgpu_vm_bo_base *bo_base, *parent; struct amdgpu_vm_pt *pt, *entry; @@ -983,14 +977,12 @@ restart: bo_base = list_first_entry(&vm->relocated, struct amdgpu_vm_bo_base, vm_status); - list_del_init(&bo_base->vm_status); - spin_unlock(&vm->status_lock); + bo_base->moved = false; + list_move(&bo_base->vm_status, &vm->idle); bo = bo_base->bo->parent; - if (!bo) { - spin_lock(&vm->status_lock); + if (!bo) continue; - } parent = list_first_entry(&bo->va, struct amdgpu_vm_bo_base, bo_list); @@ -999,12 +991,10 @@ restart: amdgpu_vm_update_pde(¶ms, vm, pt, entry); - spin_lock(&vm->status_lock); if (!vm->use_cpu_for_update && (ndw - params.ib->length_dw) < 32) break; } - spin_unlock(&vm->status_lock); if (vm->use_cpu_for_update) { /* Flush HDP */ @@ -1107,9 +1097,7 @@ static void amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p, if (entry->huge) { /* Add the entry to the relocated list to update it. */ entry->huge = false; - spin_lock(&p->vm->status_lock); list_move(&entry->base.vm_status, &p->vm->relocated); - spin_unlock(&p->vm->status_lock); } return; } @@ -1588,18 +1576,22 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, amdgpu_asic_flush_hdp(adev, NULL); } - spin_lock(&vm->status_lock); + spin_lock(&vm->moved_lock); list_del_init(&bo_va->base.vm_status); + spin_unlock(&vm->moved_lock); /* If the BO is not in its preferred location add it back to * the evicted list so that it gets validated again on the * next command submission. */ - if (bo && bo->tbo.resv == vm->root.base.bo->tbo.resv && - !(bo->preferred_domains & - amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type))) - list_add_tail(&bo_va->base.vm_status, &vm->evicted); - spin_unlock(&vm->status_lock); + if (bo && bo->tbo.resv == vm->root.base.bo->tbo.resv) { + uint32_t mem_type = bo->tbo.mem.mem_type; + + if (!(bo->preferred_domains & amdgpu_mem_type_to_domain(mem_type))) + list_add_tail(&bo_va->base.vm_status, &vm->evicted); + else + list_add(&bo_va->base.vm_status, &vm->idle); + } list_splice_init(&bo_va->invalids, &bo_va->valids); bo_va->cleared = clear; @@ -1808,19 +1800,18 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev, int amdgpu_vm_handle_moved(struct amdgpu_device *adev, struct amdgpu_vm *vm) { + struct amdgpu_bo_va *bo_va, *tmp; + struct list_head moved; bool clear; - int r = 0; - - spin_lock(&vm->status_lock); - while (!list_empty(&vm->moved)) { - struct amdgpu_bo_va *bo_va; - struct reservation_object *resv; + int r; - bo_va = list_first_entry(&vm->moved, - struct amdgpu_bo_va, base.vm_status); - spin_unlock(&vm->status_lock); + INIT_LIST_HEAD(&moved); + spin_lock(&vm->moved_lock); + list_splice_init(&vm->moved, &moved); + spin_unlock(&vm->moved_lock); - resv = bo_va->base.bo->tbo.resv; + list_for_each_entry_safe(bo_va, tmp, &moved, base.vm_status) { + struct reservation_object *resv = bo_va->base.bo->tbo.resv; /* Per VM BOs never need to bo cleared in the page tables */ if (resv == vm->root.base.bo->tbo.resv) @@ -1833,17 +1824,19 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev, clear = true; r = amdgpu_vm_bo_update(adev, bo_va, clear); - if (r) + if (r) { + spin_lock(&vm->moved_lock); + list_splice(&moved, &vm->moved); + spin_unlock(&vm->moved_lock); return r; + } if (!clear && resv != vm->root.base.bo->tbo.resv) reservation_object_unlock(resv); - spin_lock(&vm->status_lock); } - spin_unlock(&vm->status_lock); - return r; + return 0; } /** @@ -1902,11 +1895,11 @@ static void amdgpu_vm_bo_insert_map(struct amdgpu_device *adev, if (mapping->flags & AMDGPU_PTE_PRT) amdgpu_vm_prt_get(adev); - if (bo && bo->tbo.resv == vm->root.base.bo->tbo.resv) { - spin_lock(&vm->status_lock); - if (list_empty(&bo_va->base.vm_status)) - list_add(&bo_va->base.vm_status, &vm->moved); - spin_unlock(&vm->status_lock); + if (bo && bo->tbo.resv == vm->root.base.bo->tbo.resv && + !bo_va->base.moved) { + spin_lock(&vm->moved_lock); + list_move(&bo_va->base.vm_status, &vm->moved); + spin_unlock(&vm->moved_lock); } trace_amdgpu_vm_bo_map(bo_va, mapping); } @@ -2216,9 +2209,9 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev, list_del(&bo_va->base.bo_list); - spin_lock(&vm->status_lock); + spin_lock(&vm->moved_lock); list_del(&bo_va->base.vm_status); - spin_unlock(&vm->status_lock); + spin_unlock(&vm->moved_lock); list_for_each_entry_safe(mapping, next, &bo_va->valids, list) { list_del(&mapping->list); @@ -2258,31 +2251,28 @@ void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev, list_for_each_entry(bo_base, &bo->va, bo_list) { struct amdgpu_vm *vm = bo_base->vm; + bool was_moved = bo_base->moved; bo_base->moved = true; if (evicted && bo->tbo.resv == vm->root.base.bo->tbo.resv) { - spin_lock(&bo_base->vm->status_lock); if (bo->tbo.type == ttm_bo_type_kernel) list_move(&bo_base->vm_status, &vm->evicted); else list_move_tail(&bo_base->vm_status, &vm->evicted); - spin_unlock(&bo_base->vm->status_lock); continue; } - if (bo->tbo.type == ttm_bo_type_kernel) { - spin_lock(&bo_base->vm->status_lock); - if (list_empty(&bo_base->vm_status)) - list_add(&bo_base->vm_status, &vm->relocated); - spin_unlock(&bo_base->vm->status_lock); + if (was_moved) continue; - } - spin_lock(&bo_base->vm->status_lock); - if (list_empty(&bo_base->vm_status)) - list_add(&bo_base->vm_status, &vm->moved); - spin_unlock(&bo_base->vm->status_lock); + if (bo->tbo.type == ttm_bo_type_kernel) { + list_move(&bo_base->vm_status, &vm->relocated); + } else { + spin_lock(&bo_base->vm->moved_lock); + list_move(&bo_base->vm_status, &vm->moved); + spin_unlock(&bo_base->vm->moved_lock); + } } } @@ -2391,10 +2381,11 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, vm->va = RB_ROOT_CACHED; for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) vm->reserved_vmid[i] = NULL; - spin_lock_init(&vm->status_lock); INIT_LIST_HEAD(&vm->evicted); INIT_LIST_HEAD(&vm->relocated); + spin_lock_init(&vm->moved_lock); INIT_LIST_HEAD(&vm->moved); + INIT_LIST_HEAD(&vm->idle); INIT_LIST_HEAD(&vm->freed); /* create scheduler entity for page table updates */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 4cf678684a12..061b99a18cb8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -168,9 +168,6 @@ struct amdgpu_vm { /* tree of virtual addresses mapped */ struct rb_root_cached va; - /* protecting invalidated */ - spinlock_t status_lock; - /* BOs who needs a validation */ struct list_head evicted; @@ -179,6 +176,10 @@ struct amdgpu_vm { /* BOs moved, but not yet updated in the PT */ struct list_head moved; + spinlock_t moved_lock; + + /* All BOs of this VM not currently in the state machine */ + struct list_head idle; /* BO mappings freed, but not yet updated in the PT */ struct list_head freed; @@ -187,9 +188,6 @@ struct amdgpu_vm { struct amdgpu_vm_pt root; struct dma_fence *last_update; - /* protecting freed */ - spinlock_t freed_lock; - /* Scheduler entity for page table updates */ struct drm_sched_entity entity; diff --git a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c index de7be3de0f41..dbf2ccd0c744 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c @@ -473,6 +473,7 @@ static int dce_virtual_hw_init(void *handle) break; case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: break; default: DRM_ERROR("Virtual display unsupported ASIC type: 0x%X\n", adev->asic_type); diff --git a/drivers/gpu/drm/amd/amdgpu/df_v1_7.c b/drivers/gpu/drm/amd/amdgpu/df_v1_7.c index 4ffda996660f..9935371db7ce 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v1_7.c +++ b/drivers/gpu/drm/amd/amdgpu/df_v1_7.c @@ -102,6 +102,13 @@ static void df_v1_7_get_clockgating_state(struct amdgpu_device *adev, *flags |= AMD_CG_SUPPORT_DF_MGCG; } +static void df_v1_7_enable_ecc_force_par_wr_rmw(struct amdgpu_device *adev, + bool enable) +{ + WREG32_FIELD15(DF, 0, DF_CS_AON0_CoherentSlaveModeCtrlA0, + ForceParWrRMW, enable); +} + const struct amdgpu_df_funcs df_v1_7_funcs = { .init = df_v1_7_init, .enable_broadcast_mode = df_v1_7_enable_broadcast_mode, @@ -109,4 +116,5 @@ const struct amdgpu_df_funcs df_v1_7_funcs = { .get_hbm_channel_number = df_v1_7_get_hbm_channel_number, .update_medium_grain_clock_gating = df_v1_7_update_medium_grain_clock_gating, .get_clockgating_state = df_v1_7_get_clockgating_state, + .enable_ecc_force_par_wr_rmw = df_v1_7_enable_ecc_force_par_wr_rmw, }; diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c new file mode 100644 index 000000000000..60608b3df881 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c @@ -0,0 +1,116 @@ +/* + * Copyright 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "df_v3_6.h" + +#include "df/df_3_6_default.h" +#include "df/df_3_6_offset.h" +#include "df/df_3_6_sh_mask.h" + +static u32 df_v3_6_channel_number[] = {1, 2, 0, 4, 0, 8, 0, + 16, 32, 0, 0, 0, 2, 4, 8}; + +static void df_v3_6_init(struct amdgpu_device *adev) +{ +} + +static void df_v3_6_enable_broadcast_mode(struct amdgpu_device *adev, + bool enable) +{ + u32 tmp; + + if (enable) { + tmp = RREG32_SOC15(DF, 0, mmFabricConfigAccessControl); + tmp &= ~FabricConfigAccessControl__CfgRegInstAccEn_MASK; + WREG32_SOC15(DF, 0, mmFabricConfigAccessControl, tmp); + } else + WREG32_SOC15(DF, 0, mmFabricConfigAccessControl, + mmFabricConfigAccessControl_DEFAULT); +} + +static u32 df_v3_6_get_fb_channel_number(struct amdgpu_device *adev) +{ + u32 tmp; + + tmp = RREG32_SOC15(DF, 0, mmDF_CS_UMC_AON0_DramBaseAddress0); + tmp &= DF_CS_UMC_AON0_DramBaseAddress0__IntLvNumChan_MASK; + tmp >>= DF_CS_UMC_AON0_DramBaseAddress0__IntLvNumChan__SHIFT; + + return tmp; +} + +static u32 df_v3_6_get_hbm_channel_number(struct amdgpu_device *adev) +{ + int fb_channel_number; + + fb_channel_number = adev->df_funcs->get_fb_channel_number(adev); + if (fb_channel_number > ARRAY_SIZE(df_v3_6_channel_number)) + fb_channel_number = 0; + + return df_v3_6_channel_number[fb_channel_number]; +} + +static void df_v3_6_update_medium_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + u32 tmp; + + /* Put DF on broadcast mode */ + adev->df_funcs->enable_broadcast_mode(adev, true); + + if (enable && (adev->cg_flags & AMD_CG_SUPPORT_DF_MGCG)) { + tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater); + tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK; + tmp |= DF_V3_6_MGCG_ENABLE_15_CYCLE_DELAY; + WREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater, tmp); + } else { + tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater); + tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK; + tmp |= DF_V3_6_MGCG_DISABLE; + WREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater, tmp); + } + + /* Exit broadcast mode */ + adev->df_funcs->enable_broadcast_mode(adev, false); +} + +static void df_v3_6_get_clockgating_state(struct amdgpu_device *adev, + u32 *flags) +{ + u32 tmp; + + /* AMD_CG_SUPPORT_DF_MGCG */ + tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater); + if (tmp & DF_V3_6_MGCG_ENABLE_15_CYCLE_DELAY) + *flags |= AMD_CG_SUPPORT_DF_MGCG; +} + +const struct amdgpu_df_funcs df_v3_6_funcs = { + .init = df_v3_6_init, + .enable_broadcast_mode = df_v3_6_enable_broadcast_mode, + .get_fb_channel_number = df_v3_6_get_fb_channel_number, + .get_hbm_channel_number = df_v3_6_get_hbm_channel_number, + .update_medium_grain_clock_gating = + df_v3_6_update_medium_grain_clock_gating, + .get_clockgating_state = df_v3_6_get_clockgating_state, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.h b/drivers/gpu/drm/amd/amdgpu/df_v3_6.h new file mode 100644 index 000000000000..e79c58e5efcb --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.h @@ -0,0 +1,40 @@ +/* + * Copyright 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __DF_V3_6_H__ +#define __DF_V3_6_H__ + +#include "soc15_common.h" + +enum DF_V3_6_MGCG { + DF_V3_6_MGCG_DISABLE = 0, + DF_V3_6_MGCG_ENABLE_00_CYCLE_DELAY = 1, + DF_V3_6_MGCG_ENABLE_01_CYCLE_DELAY = 2, + DF_V3_6_MGCG_ENABLE_15_CYCLE_DELAY = 13, + DF_V3_6_MGCG_ENABLE_31_CYCLE_DELAY = 14, + DF_V3_6_MGCG_ENABLE_63_CYCLE_DELAY = 15 +}; + +extern const struct amdgpu_df_funcs df_v3_6_funcs; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index fc1911834ab5..d7530fdfaad5 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -27,6 +27,7 @@ #include "amdgpu_gfx.h" #include "soc15.h" #include "soc15d.h" +#include "amdgpu_atomfirmware.h" #include "gc/gc_9_0_offset.h" #include "gc/gc_9_0_sh_mask.h" @@ -63,6 +64,13 @@ MODULE_FIRMWARE("amdgpu/vega12_mec.bin"); MODULE_FIRMWARE("amdgpu/vega12_mec2.bin"); MODULE_FIRMWARE("amdgpu/vega12_rlc.bin"); +MODULE_FIRMWARE("amdgpu/vega20_ce.bin"); +MODULE_FIRMWARE("amdgpu/vega20_pfp.bin"); +MODULE_FIRMWARE("amdgpu/vega20_me.bin"); +MODULE_FIRMWARE("amdgpu/vega20_mec.bin"); +MODULE_FIRMWARE("amdgpu/vega20_mec2.bin"); +MODULE_FIRMWARE("amdgpu/vega20_rlc.bin"); + MODULE_FIRMWARE("amdgpu/raven_ce.bin"); MODULE_FIRMWARE("amdgpu/raven_pfp.bin"); MODULE_FIRMWARE("amdgpu/raven_me.bin"); @@ -72,29 +80,22 @@ MODULE_FIRMWARE("amdgpu/raven_rlc.bin"); static const struct soc15_reg_golden golden_settings_gc_9_0[] = { - SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPC_UTCL1_CNTL, 0x08000000, 0x08000080), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_UTCL1_CNTL, 0x08000000, 0x08000080), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPG_UTCL1_CNTL, 0x08000000, 0x08000080), SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xf00fffff, 0x00000420), SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_GPU_ID, 0x0000000f, 0x00000000), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmIA_UTCL1_CNTL, 0x08000000, 0x08000080), SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3, 0x00000003, 0x82400024), SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x00000001), SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_0, 0x08000000, 0x08000080), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_1, 0x08000000, 0x08000080), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_2, 0x08000000, 0x08000080), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL, 0x08000000, 0x08000080), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_UTCL1_CNTL, 0x08000000, 0x08000080), SOC15_REG_GOLDEN_VALUE(GC, 0, mmSH_MEM_CONFIG, 0x00001000, 0x00001000), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0x0000000f, 0x01000107), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_0, 0x0007ffff, 0x00000800), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_1, 0x0007ffff, 0x00000800), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_0, 0x01ffffff, 0x0000ff87), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_1, 0x01ffffff, 0x0000ff8f), SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQC_CONFIG, 0x03000000, 0x020a2000), SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfffffeef, 0x010b0000), SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x4a2c0e68), SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0xb5d3f197), SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_CACHE_INVALIDATION, 0x3fff3af3, 0x19200000), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff), - SOC15_REG_GOLDEN_VALUE(GC, 0, mmWD_UTCL1_CNTL, 0x08000000, 0x08000080) + SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff) }; static const struct soc15_reg_golden golden_settings_gc_9_0_vg10[] = @@ -108,6 +109,20 @@ static const struct soc15_reg_golden golden_settings_gc_9_0_vg10[] = SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x00001800, 0x00000800) }; +static const struct soc15_reg_golden golden_settings_gc_9_0_vg20[] = +{ + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_2, 0x0f000000, 0x0a000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_3, 0x30000000, 0x10000000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xf3e777ff, 0x22014042), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xf3e777ff, 0x22014042), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0x00003e00, 0x00000400), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xff840000, 0x04040000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_UTCL1_CNTL2, 0x00030000, 0x00030000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0xffff010f, 0x01000107), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0x000b0000, 0x000b0000), + SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x01000000, 0x01000000) +}; + static const struct soc15_reg_golden golden_settings_gc_9_1[] = { SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL, 0xfffdf3cf, 0x00014104), @@ -241,6 +256,14 @@ static void gfx_v9_0_init_golden_registers(struct amdgpu_device *adev) golden_settings_gc_9_2_1_vg12, ARRAY_SIZE(golden_settings_gc_9_2_1_vg12)); break; + case CHIP_VEGA20: + soc15_program_register_sequence(adev, + golden_settings_gc_9_0, + ARRAY_SIZE(golden_settings_gc_9_0)); + soc15_program_register_sequence(adev, + golden_settings_gc_9_0_vg20, + ARRAY_SIZE(golden_settings_gc_9_0_vg20)); + break; case CHIP_RAVEN: soc15_program_register_sequence(adev, golden_settings_gc_9_1, @@ -468,6 +491,9 @@ static int gfx_v9_0_init_microcode(struct amdgpu_device *adev) case CHIP_VEGA12: chip_name = "vega12"; break; + case CHIP_VEGA20: + chip_name = "vega20"; + break; case CHIP_RAVEN: chip_name = "raven"; break; @@ -1088,9 +1114,10 @@ static const struct amdgpu_gfx_funcs gfx_v9_0_gfx_funcs = { .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q }; -static void gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) +static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) { u32 gb_addr_config; + int err; adev->gfx.funcs = &gfx_v9_0_gfx_funcs; @@ -1112,6 +1139,20 @@ static void gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) gb_addr_config = VEGA12_GB_ADDR_CONFIG_GOLDEN; DRM_INFO("fix gfx.config for vega12\n"); break; + case CHIP_VEGA20: + adev->gfx.config.max_hw_contexts = 8; + adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; + adev->gfx.config.sc_prim_fifo_size_backend = 0x100; + adev->gfx.config.sc_hiz_tile_fifo_size = 0x30; + adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0; + gb_addr_config = RREG32_SOC15(GC, 0, mmGB_ADDR_CONFIG); + gb_addr_config &= ~0xf3e777ff; + gb_addr_config |= 0x22014042; + /* check vbios table if gpu info is not available */ + err = amdgpu_atomfirmware_get_gfx_info(adev); + if (err) + return err; + break; case CHIP_RAVEN: adev->gfx.config.max_hw_contexts = 8; adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; @@ -1161,6 +1202,8 @@ static void gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) adev->gfx.config.gb_addr_config, GB_ADDR_CONFIG, PIPE_INTERLEAVE_SIZE)); + + return 0; } static int gfx_v9_0_ngg_create_buf(struct amdgpu_device *adev, @@ -1394,6 +1437,7 @@ static int gfx_v9_0_sw_init(void *handle) switch (adev->asic_type) { case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: case CHIP_RAVEN: adev->gfx.mec.num_mec = 2; break; @@ -1521,7 +1565,9 @@ static int gfx_v9_0_sw_init(void *handle) adev->gfx.ce_ram_size = 0x8000; - gfx_v9_0_gpu_early_init(adev); + r = gfx_v9_0_gpu_early_init(adev); + if (r) + return r; r = gfx_v9_0_ngg_init(adev); if (r) @@ -3688,6 +3734,7 @@ static int gfx_v9_0_set_clockgating_state(void *handle, switch (adev->asic_type) { case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: case CHIP_RAVEN: gfx_v9_0_update_gfx_clock_gating(adev, state == AMD_CG_STATE_GATE ? true : false); @@ -4680,6 +4727,7 @@ static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: case CHIP_RAVEN: adev->gfx.rlc.funcs = &gfx_v9_0_rlc_funcs; break; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 6cccf0e0acd7..3c0a85d4e4ab 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -675,6 +675,7 @@ static int gmc_v9_0_late_init(void *handle) DRM_INFO("ECC is active.\n"); } else if (r == 0) { DRM_INFO("ECC is not present.\n"); + adev->df_funcs->enable_ecc_force_par_wr_rmw(adev, false); } else { DRM_ERROR("gmc_v9_0_ecc_available() failed. r: %d\n", r); return r; @@ -693,10 +694,7 @@ static void gmc_v9_0_vram_gtt_location(struct amdgpu_device *adev, amdgpu_device_vram_location(adev, &adev->gmc, base); amdgpu_device_gart_location(adev, mc); /* base offset of vram pages */ - if (adev->flags & AMD_IS_APU) - adev->vm_manager.vram_base_offset = gfxhub_v1_0_get_mc_fb_offset(adev); - else - adev->vm_manager.vram_base_offset = 0; + adev->vm_manager.vram_base_offset = gfxhub_v1_0_get_mc_fb_offset(adev); } /** @@ -755,6 +753,7 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_VEGA10: /* all engines support GPUVM */ case CHIP_VEGA12: /* all engines support GPUVM */ + case CHIP_VEGA20: default: adev->gmc.gart_size = 512ULL << 20; break; @@ -860,6 +859,7 @@ static int gmc_v9_0_sw_init(void *handle) break; case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: /* * To fulfill 4-level page support, * vm size is 256TB (48bit), maximum size of Vega10, @@ -977,6 +977,7 @@ static void gmc_v9_0_init_golden_registers(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_VEGA10: + case CHIP_VEGA20: soc15_program_register_sequence(adev, golden_settings_mmhub_1_0_0, ARRAY_SIZE(golden_settings_mmhub_1_0_0)); diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c index 43f925773b57..3d53c4413f13 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c @@ -734,6 +734,7 @@ int mmhub_v1_0_set_clockgating(struct amdgpu_device *adev, switch (adev->asic_type) { case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: case CHIP_RAVEN: mmhub_v1_0_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE ? true : false); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c index df34dc79d444..365517c0121e 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c @@ -34,10 +34,19 @@ #define smnCPM_CONTROL 0x11180460 #define smnPCIE_CNTL2 0x11180070 +/* vega20 */ +#define mmRCC_DEV0_EPF0_STRAP0_VG20 0x0011 +#define mmRCC_DEV0_EPF0_STRAP0_VG20_BASE_IDX 2 + static u32 nbio_v7_0_get_rev_id(struct amdgpu_device *adev) { u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0); + if (adev->asic_type == CHIP_VEGA20) + tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0_VG20); + else + tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0); + tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK; tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT; @@ -75,10 +84,14 @@ static void nbio_v7_0_sdma_doorbell_range(struct amdgpu_device *adev, int instan SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE); u32 doorbell_range = RREG32(reg); + u32 range = 2; + + if (adev->asic_type == CHIP_VEGA20) + range = 8; if (use_doorbell) { doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, OFFSET, doorbell_index); - doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 2); + doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, range); } else doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 0); @@ -133,6 +146,9 @@ static void nbio_v7_0_update_medium_grain_clock_gating(struct amdgpu_device *ade { uint32_t def, data; + if (adev->asic_type == CHIP_VEGA20) + return; + /* NBIF_MGCG_CTRL_LCLK */ def = data = RREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK); diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c index 196e75def1f2..0c768e388ace 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c @@ -41,6 +41,9 @@ MODULE_FIRMWARE("amdgpu/vega10_sos.bin"); MODULE_FIRMWARE("amdgpu/vega10_asd.bin"); MODULE_FIRMWARE("amdgpu/vega12_sos.bin"); MODULE_FIRMWARE("amdgpu/vega12_asd.bin"); +MODULE_FIRMWARE("amdgpu/vega20_sos.bin"); +MODULE_FIRMWARE("amdgpu/vega20_asd.bin"); + #define smnMP1_FIRMWARE_FLAGS 0x3010028 diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index 03a36cbe7557..ca53b3fba422 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -42,6 +42,8 @@ MODULE_FIRMWARE("amdgpu/vega10_sdma.bin"); MODULE_FIRMWARE("amdgpu/vega10_sdma1.bin"); MODULE_FIRMWARE("amdgpu/vega12_sdma.bin"); MODULE_FIRMWARE("amdgpu/vega12_sdma1.bin"); +MODULE_FIRMWARE("amdgpu/vega20_sdma.bin"); +MODULE_FIRMWARE("amdgpu/vega20_sdma1.bin"); MODULE_FIRMWARE("amdgpu/raven_sdma.bin"); #define SDMA0_POWER_CNTL__ON_OFF_CONDITION_HOLD_TIME_MASK 0x000000F8L @@ -107,6 +109,28 @@ static const struct soc15_reg_golden golden_settings_sdma_4_1[] = SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003c0) }; +static const struct soc15_reg_golden golden_settings_sdma_4_2[] = +{ + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831d07), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xffffffff, 0x3f000100), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0000773f, 0x00004002), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff0, 0x00403000), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000), + SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003c0), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831d07), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CLK_CTRL, 0xffffffff, 0x3f000100), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0000773f, 0x00004002), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000), + SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_PAGE, 0x000003ff, 0x000003c0) +}; + static const struct soc15_reg_golden golden_settings_sdma_rv1[] = { SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00000002), @@ -139,6 +163,11 @@ static void sdma_v4_0_init_golden_registers(struct amdgpu_device *adev) golden_settings_sdma_vg12, ARRAY_SIZE(golden_settings_sdma_vg12)); break; + case CHIP_VEGA20: + soc15_program_register_sequence(adev, + golden_settings_sdma_4_2, + ARRAY_SIZE(golden_settings_sdma_4_2)); + break; case CHIP_RAVEN: soc15_program_register_sequence(adev, golden_settings_sdma_4_1, @@ -182,6 +211,9 @@ static int sdma_v4_0_init_microcode(struct amdgpu_device *adev) case CHIP_VEGA12: chip_name = "vega12"; break; + case CHIP_VEGA20: + chip_name = "vega20"; + break; case CHIP_RAVEN: chip_name = "raven"; break; @@ -1516,6 +1548,7 @@ static int sdma_v4_0_set_clockgating_state(void *handle, switch (adev->asic_type) { case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: case CHIP_RAVEN: sdma_v4_0_update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE ? true : false); diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index 90065766fffb..68b4a22a8892 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -41,8 +41,6 @@ #include "sdma1/sdma1_4_0_offset.h" #include "hdp/hdp_4_0_offset.h" #include "hdp/hdp_4_0_sh_mask.h" -#include "mp/mp_9_0_offset.h" -#include "mp/mp_9_0_sh_mask.h" #include "smuio/smuio_9_0_offset.h" #include "smuio/smuio_9_0_sh_mask.h" @@ -53,6 +51,7 @@ #include "gfxhub_v1_0.h" #include "mmhub_v1_0.h" #include "df_v1_7.h" +#include "df_v3_6.h" #include "vega10_ih.h" #include "sdma_v4_0.h" #include "uvd_v7_0.h" @@ -489,16 +488,24 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev) case CHIP_RAVEN: vega10_reg_base_init(adev); break; + case CHIP_VEGA20: + vega20_reg_base_init(adev); + break; default: return -EINVAL; } if (adev->flags & AMD_IS_APU) adev->nbio_funcs = &nbio_v7_0_funcs; + else if (adev->asic_type == CHIP_VEGA20) + adev->nbio_funcs = &nbio_v7_0_funcs; else adev->nbio_funcs = &nbio_v6_1_funcs; - adev->df_funcs = &df_v1_7_funcs; + if (adev->asic_type == CHIP_VEGA20) + adev->df_funcs = &df_v3_6_funcs; + else + adev->df_funcs = &df_v1_7_funcs; adev->nbio_funcs->detect_hw_virt(adev); if (amdgpu_sriov_vf(adev)) @@ -507,12 +514,15 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev) switch (adev->asic_type) { case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: amdgpu_device_ip_block_add(adev, &vega10_common_ip_block); amdgpu_device_ip_block_add(adev, &gmc_v9_0_ip_block); amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block); - amdgpu_device_ip_block_add(adev, &psp_v3_1_ip_block); - if (!amdgpu_sriov_vf(adev)) - amdgpu_device_ip_block_add(adev, &pp_smu_ip_block); + if (adev->asic_type != CHIP_VEGA20) { + amdgpu_device_ip_block_add(adev, &psp_v3_1_ip_block); + if (!amdgpu_sriov_vf(adev)) + amdgpu_device_ip_block_add(adev, &pp_smu_ip_block); + } if (adev->enable_virtual_display || amdgpu_sriov_vf(adev)) amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block); #if defined(CONFIG_DRM_AMD_DC) @@ -660,6 +670,27 @@ static int soc15_common_early_init(void *handle) adev->pg_flags = 0; adev->external_rev_id = adev->rev_id + 0x14; break; + case CHIP_VEGA20: + adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | + AMD_CG_SUPPORT_GFX_MGLS | + AMD_CG_SUPPORT_GFX_CGCG | + AMD_CG_SUPPORT_GFX_CGLS | + AMD_CG_SUPPORT_GFX_3D_CGCG | + AMD_CG_SUPPORT_GFX_3D_CGLS | + AMD_CG_SUPPORT_GFX_CP_LS | + AMD_CG_SUPPORT_MC_LS | + AMD_CG_SUPPORT_MC_MGCG | + AMD_CG_SUPPORT_SDMA_MGCG | + AMD_CG_SUPPORT_SDMA_LS | + AMD_CG_SUPPORT_BIF_MGCG | + AMD_CG_SUPPORT_BIF_LS | + AMD_CG_SUPPORT_HDP_MGCG | + AMD_CG_SUPPORT_ROM_MGCG | + AMD_CG_SUPPORT_VCE_MGCG | + AMD_CG_SUPPORT_UVD_MGCG; + adev->pg_flags = 0; + adev->external_rev_id = adev->rev_id + 0x28; + break; case CHIP_RAVEN: adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS | @@ -679,8 +710,10 @@ static int soc15_common_early_init(void *handle) AMD_CG_SUPPORT_MC_MGCG | AMD_CG_SUPPORT_MC_LS | AMD_CG_SUPPORT_SDMA_MGCG | - AMD_CG_SUPPORT_SDMA_LS; - adev->pg_flags = AMD_PG_SUPPORT_SDMA; + AMD_CG_SUPPORT_SDMA_LS | + AMD_CG_SUPPORT_VCN_MGCG; + + adev->pg_flags = AMD_PG_SUPPORT_SDMA | AMD_PG_SUPPORT_VCN; if (adev->powerplay.pp_feature & PP_GFXOFF_MASK) adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG | @@ -872,6 +905,7 @@ static int soc15_common_set_clockgating_state(void *handle, switch (adev->asic_type) { case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: adev->nbio_funcs->update_medium_grain_clock_gating(adev, state == AMD_CG_STATE_GATE ? true : false); adev->nbio_funcs->update_medium_grain_light_sleep(adev, diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.h b/drivers/gpu/drm/amd/amdgpu/soc15.h index f70da8a29f86..1f714b7af520 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.h +++ b/drivers/gpu/drm/amd/amdgpu/soc15.h @@ -55,5 +55,6 @@ void soc15_program_register_sequence(struct amdgpu_device *adev, const u32 array_size); int vega10_reg_base_init(struct amdgpu_device *adev); +int vega20_reg_base_init(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/soc15_common.h b/drivers/gpu/drm/amd/amdgpu/soc15_common.h index def865067edd..0942f492d2e1 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15_common.h +++ b/drivers/gpu/drm/amd/amdgpu/soc15_common.h @@ -47,6 +47,21 @@ #define WREG32_SOC15_OFFSET(ip, inst, reg, offset, value) \ WREG32((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + offset, value) +#define SOC15_WAIT_ON_RREG(ip, inst, reg, expected_value, mask, ret) \ + do { \ + uint32_t tmp_ = RREG32(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg); \ + uint32_t loop = adev->usec_timeout; \ + while ((tmp_ & (mask)) != (expected_value)) { \ + udelay(2); \ + tmp_ = RREG32(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg); \ + loop--; \ + if (!loop) { \ + ret = -ETIMEDOUT; \ + break; \ + } \ + } \ + } while (0) + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c index 87cbb142dd0b..6fed3d7797a8 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c @@ -93,6 +93,7 @@ static void uvd_v4_2_ring_set_wptr(struct amdgpu_ring *ring) static int uvd_v4_2_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + adev->uvd.num_uvd_inst = 1; uvd_v4_2_set_ring_funcs(adev); uvd_v4_2_set_irq_funcs(adev); @@ -107,7 +108,7 @@ static int uvd_v4_2_sw_init(void *handle) int r; /* UVD TRAP */ - r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.irq); + r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.inst->irq); if (r) return r; @@ -119,9 +120,9 @@ static int uvd_v4_2_sw_init(void *handle) if (r) return r; - ring = &adev->uvd.ring; + ring = &adev->uvd.inst->ring; sprintf(ring->name, "uvd"); - r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0); + r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0); return r; } @@ -150,7 +151,7 @@ static void uvd_v4_2_enable_mgcg(struct amdgpu_device *adev, static int uvd_v4_2_hw_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring = &adev->uvd.inst->ring; uint32_t tmp; int r; @@ -208,7 +209,7 @@ done: static int uvd_v4_2_hw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring = &adev->uvd.inst->ring; if (RREG32(mmUVD_STATUS) != 0) uvd_v4_2_stop(adev); @@ -251,7 +252,7 @@ static int uvd_v4_2_resume(void *handle) */ static int uvd_v4_2_start(struct amdgpu_device *adev) { - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring = &adev->uvd.inst->ring; uint32_t rb_bufsz; int i, j, r; u32 tmp; @@ -523,6 +524,18 @@ static void uvd_v4_2_ring_emit_ib(struct amdgpu_ring *ring, amdgpu_ring_write(ring, ib->length_dw); } +static void uvd_v4_2_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) +{ + int i; + + WARN_ON(ring->wptr % 2 || count % 2); + + for (i = 0; i < count / 2; i++) { + amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0)); + amdgpu_ring_write(ring, 0); + } +} + /** * uvd_v4_2_mc_resume - memory controller programming * @@ -536,7 +549,7 @@ static void uvd_v4_2_mc_resume(struct amdgpu_device *adev) uint32_t size; /* programm the VCPU memory controller bits 0-27 */ - addr = (adev->uvd.gpu_addr + AMDGPU_UVD_FIRMWARE_OFFSET) >> 3; + addr = (adev->uvd.inst->gpu_addr + AMDGPU_UVD_FIRMWARE_OFFSET) >> 3; size = AMDGPU_UVD_FIRMWARE_SIZE(adev) >> 3; WREG32(mmUVD_VCPU_CACHE_OFFSET0, addr); WREG32(mmUVD_VCPU_CACHE_SIZE0, size); @@ -553,11 +566,11 @@ static void uvd_v4_2_mc_resume(struct amdgpu_device *adev) WREG32(mmUVD_VCPU_CACHE_SIZE2, size); /* bits 28-31 */ - addr = (adev->uvd.gpu_addr >> 28) & 0xF; + addr = (adev->uvd.inst->gpu_addr >> 28) & 0xF; WREG32(mmUVD_LMI_ADDR_EXT, (addr << 12) | (addr << 0)); /* bits 32-39 */ - addr = (adev->uvd.gpu_addr >> 32) & 0xFF; + addr = (adev->uvd.inst->gpu_addr >> 32) & 0xFF; WREG32(mmUVD_LMI_EXT40_ADDR, addr | (0x9 << 16) | (0x1 << 31)); WREG32(mmUVD_UDEC_ADDR_CONFIG, adev->gfx.config.gb_addr_config); @@ -664,7 +677,7 @@ static int uvd_v4_2_process_interrupt(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) { DRM_DEBUG("IH: UVD TRAP\n"); - amdgpu_fence_process(&adev->uvd.ring); + amdgpu_fence_process(&adev->uvd.inst->ring); return 0; } @@ -732,7 +745,6 @@ static const struct amd_ip_funcs uvd_v4_2_ip_funcs = { static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = { .type = AMDGPU_RING_TYPE_UVD, .align_mask = 0xf, - .nop = PACKET0(mmUVD_NO_OP, 0), .support_64bit_ptrs = false, .get_rptr = uvd_v4_2_ring_get_rptr, .get_wptr = uvd_v4_2_ring_get_wptr, @@ -745,7 +757,7 @@ static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = { .emit_fence = uvd_v4_2_ring_emit_fence, .test_ring = uvd_v4_2_ring_test_ring, .test_ib = amdgpu_uvd_ring_test_ib, - .insert_nop = amdgpu_ring_insert_nop, + .insert_nop = uvd_v4_2_ring_insert_nop, .pad_ib = amdgpu_ring_generic_pad_ib, .begin_use = amdgpu_uvd_ring_begin_use, .end_use = amdgpu_uvd_ring_end_use, @@ -753,7 +765,7 @@ static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = { static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev) { - adev->uvd.ring.funcs = &uvd_v4_2_ring_funcs; + adev->uvd.inst->ring.funcs = &uvd_v4_2_ring_funcs; } static const struct amdgpu_irq_src_funcs uvd_v4_2_irq_funcs = { @@ -763,8 +775,8 @@ static const struct amdgpu_irq_src_funcs uvd_v4_2_irq_funcs = { static void uvd_v4_2_set_irq_funcs(struct amdgpu_device *adev) { - adev->uvd.irq.num_types = 1; - adev->uvd.irq.funcs = &uvd_v4_2_irq_funcs; + adev->uvd.inst->irq.num_types = 1; + adev->uvd.inst->irq.funcs = &uvd_v4_2_irq_funcs; } const struct amdgpu_ip_block_version uvd_v4_2_ip_block = diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c index 6445d55e7d5a..341ee6d55ce8 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c @@ -89,6 +89,7 @@ static void uvd_v5_0_ring_set_wptr(struct amdgpu_ring *ring) static int uvd_v5_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + adev->uvd.num_uvd_inst = 1; uvd_v5_0_set_ring_funcs(adev); uvd_v5_0_set_irq_funcs(adev); @@ -103,7 +104,7 @@ static int uvd_v5_0_sw_init(void *handle) int r; /* UVD TRAP */ - r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.irq); + r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.inst->irq); if (r) return r; @@ -115,9 +116,9 @@ static int uvd_v5_0_sw_init(void *handle) if (r) return r; - ring = &adev->uvd.ring; + ring = &adev->uvd.inst->ring; sprintf(ring->name, "uvd"); - r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0); + r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0); return r; } @@ -144,7 +145,7 @@ static int uvd_v5_0_sw_fini(void *handle) static int uvd_v5_0_hw_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring = &adev->uvd.inst->ring; uint32_t tmp; int r; @@ -204,7 +205,7 @@ done: static int uvd_v5_0_hw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring = &adev->uvd.inst->ring; if (RREG32(mmUVD_STATUS) != 0) uvd_v5_0_stop(adev); @@ -253,9 +254,9 @@ static void uvd_v5_0_mc_resume(struct amdgpu_device *adev) /* programm memory controller bits 0-27 */ WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW, - lower_32_bits(adev->uvd.gpu_addr)); + lower_32_bits(adev->uvd.inst->gpu_addr)); WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH, - upper_32_bits(adev->uvd.gpu_addr)); + upper_32_bits(adev->uvd.inst->gpu_addr)); offset = AMDGPU_UVD_FIRMWARE_OFFSET; size = AMDGPU_UVD_FIRMWARE_SIZE(adev); @@ -287,7 +288,7 @@ static void uvd_v5_0_mc_resume(struct amdgpu_device *adev) */ static int uvd_v5_0_start(struct amdgpu_device *adev) { - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring = &adev->uvd.inst->ring; uint32_t rb_bufsz, tmp; uint32_t lmi_swap_cntl; uint32_t mp_swap_cntl; @@ -540,6 +541,18 @@ static void uvd_v5_0_ring_emit_ib(struct amdgpu_ring *ring, amdgpu_ring_write(ring, ib->length_dw); } +static void uvd_v5_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) +{ + int i; + + WARN_ON(ring->wptr % 2 || count % 2); + + for (i = 0; i < count / 2; i++) { + amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0)); + amdgpu_ring_write(ring, 0); + } +} + static bool uvd_v5_0_is_idle(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -586,7 +599,7 @@ static int uvd_v5_0_process_interrupt(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) { DRM_DEBUG("IH: UVD TRAP\n"); - amdgpu_fence_process(&adev->uvd.ring); + amdgpu_fence_process(&adev->uvd.inst->ring); return 0; } @@ -840,7 +853,6 @@ static const struct amd_ip_funcs uvd_v5_0_ip_funcs = { static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = { .type = AMDGPU_RING_TYPE_UVD, .align_mask = 0xf, - .nop = PACKET0(mmUVD_NO_OP, 0), .support_64bit_ptrs = false, .get_rptr = uvd_v5_0_ring_get_rptr, .get_wptr = uvd_v5_0_ring_get_wptr, @@ -853,7 +865,7 @@ static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = { .emit_fence = uvd_v5_0_ring_emit_fence, .test_ring = uvd_v5_0_ring_test_ring, .test_ib = amdgpu_uvd_ring_test_ib, - .insert_nop = amdgpu_ring_insert_nop, + .insert_nop = uvd_v5_0_ring_insert_nop, .pad_ib = amdgpu_ring_generic_pad_ib, .begin_use = amdgpu_uvd_ring_begin_use, .end_use = amdgpu_uvd_ring_end_use, @@ -861,7 +873,7 @@ static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = { static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev) { - adev->uvd.ring.funcs = &uvd_v5_0_ring_funcs; + adev->uvd.inst->ring.funcs = &uvd_v5_0_ring_funcs; } static const struct amdgpu_irq_src_funcs uvd_v5_0_irq_funcs = { @@ -871,8 +883,8 @@ static const struct amdgpu_irq_src_funcs uvd_v5_0_irq_funcs = { static void uvd_v5_0_set_irq_funcs(struct amdgpu_device *adev) { - adev->uvd.irq.num_types = 1; - adev->uvd.irq.funcs = &uvd_v5_0_irq_funcs; + adev->uvd.inst->irq.num_types = 1; + adev->uvd.inst->irq.funcs = &uvd_v5_0_irq_funcs; } const struct amdgpu_ip_block_version uvd_v5_0_ip_block = diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c index ca6ab56357b5..bfddf97dd13e 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c @@ -91,7 +91,7 @@ static uint64_t uvd_v6_0_enc_ring_get_rptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - if (ring == &adev->uvd.ring_enc[0]) + if (ring == &adev->uvd.inst->ring_enc[0]) return RREG32(mmUVD_RB_RPTR); else return RREG32(mmUVD_RB_RPTR2); @@ -121,7 +121,7 @@ static uint64_t uvd_v6_0_enc_ring_get_wptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - if (ring == &adev->uvd.ring_enc[0]) + if (ring == &adev->uvd.inst->ring_enc[0]) return RREG32(mmUVD_RB_WPTR); else return RREG32(mmUVD_RB_WPTR2); @@ -152,7 +152,7 @@ static void uvd_v6_0_enc_ring_set_wptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - if (ring == &adev->uvd.ring_enc[0]) + if (ring == &adev->uvd.inst->ring_enc[0]) WREG32(mmUVD_RB_WPTR, lower_32_bits(ring->wptr)); else @@ -375,6 +375,7 @@ error: static int uvd_v6_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + adev->uvd.num_uvd_inst = 1; if (!(adev->flags & AMD_IS_APU) && (RREG32_SMC(ixCC_HARVEST_FUSES) & CC_HARVEST_FUSES__UVD_DISABLE_MASK)) @@ -399,14 +400,14 @@ static int uvd_v6_0_sw_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; /* UVD TRAP */ - r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.irq); + r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.inst->irq); if (r) return r; /* UVD ENC TRAP */ if (uvd_v6_0_enc_support(adev)) { for (i = 0; i < adev->uvd.num_enc_rings; ++i) { - r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 119, &adev->uvd.irq); + r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 119, &adev->uvd.inst->irq); if (r) return r; } @@ -418,17 +419,17 @@ static int uvd_v6_0_sw_init(void *handle) if (!uvd_v6_0_enc_support(adev)) { for (i = 0; i < adev->uvd.num_enc_rings; ++i) - adev->uvd.ring_enc[i].funcs = NULL; + adev->uvd.inst->ring_enc[i].funcs = NULL; - adev->uvd.irq.num_types = 1; + adev->uvd.inst->irq.num_types = 1; adev->uvd.num_enc_rings = 0; DRM_INFO("UVD ENC is disabled\n"); } else { struct drm_sched_rq *rq; - ring = &adev->uvd.ring_enc[0]; + ring = &adev->uvd.inst->ring_enc[0]; rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_NORMAL]; - r = drm_sched_entity_init(&ring->sched, &adev->uvd.entity_enc, + r = drm_sched_entity_init(&ring->sched, &adev->uvd.inst->entity_enc, rq, NULL); if (r) { DRM_ERROR("Failed setting up UVD ENC run queue.\n"); @@ -440,17 +441,17 @@ static int uvd_v6_0_sw_init(void *handle) if (r) return r; - ring = &adev->uvd.ring; + ring = &adev->uvd.inst->ring; sprintf(ring->name, "uvd"); - r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0); + r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0); if (r) return r; if (uvd_v6_0_enc_support(adev)) { for (i = 0; i < adev->uvd.num_enc_rings; ++i) { - ring = &adev->uvd.ring_enc[i]; + ring = &adev->uvd.inst->ring_enc[i]; sprintf(ring->name, "uvd_enc%d", i); - r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0); + r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0); if (r) return r; } @@ -469,10 +470,10 @@ static int uvd_v6_0_sw_fini(void *handle) return r; if (uvd_v6_0_enc_support(adev)) { - drm_sched_entity_fini(&adev->uvd.ring_enc[0].sched, &adev->uvd.entity_enc); + drm_sched_entity_fini(&adev->uvd.inst->ring_enc[0].sched, &adev->uvd.inst->entity_enc); for (i = 0; i < adev->uvd.num_enc_rings; ++i) - amdgpu_ring_fini(&adev->uvd.ring_enc[i]); + amdgpu_ring_fini(&adev->uvd.inst->ring_enc[i]); } return amdgpu_uvd_sw_fini(adev); @@ -488,7 +489,7 @@ static int uvd_v6_0_sw_fini(void *handle) static int uvd_v6_0_hw_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring = &adev->uvd.inst->ring; uint32_t tmp; int i, r; @@ -532,7 +533,7 @@ static int uvd_v6_0_hw_init(void *handle) if (uvd_v6_0_enc_support(adev)) { for (i = 0; i < adev->uvd.num_enc_rings; ++i) { - ring = &adev->uvd.ring_enc[i]; + ring = &adev->uvd.inst->ring_enc[i]; ring->ready = true; r = amdgpu_ring_test_ring(ring); if (r) { @@ -563,7 +564,7 @@ done: static int uvd_v6_0_hw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring = &adev->uvd.inst->ring; if (RREG32(mmUVD_STATUS) != 0) uvd_v6_0_stop(adev); @@ -611,9 +612,9 @@ static void uvd_v6_0_mc_resume(struct amdgpu_device *adev) /* programm memory controller bits 0-27 */ WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW, - lower_32_bits(adev->uvd.gpu_addr)); + lower_32_bits(adev->uvd.inst->gpu_addr)); WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH, - upper_32_bits(adev->uvd.gpu_addr)); + upper_32_bits(adev->uvd.inst->gpu_addr)); offset = AMDGPU_UVD_FIRMWARE_OFFSET; size = AMDGPU_UVD_FIRMWARE_SIZE(adev); @@ -726,7 +727,7 @@ static void cz_set_uvd_clock_gating_branches(struct amdgpu_device *adev, */ static int uvd_v6_0_start(struct amdgpu_device *adev) { - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring = &adev->uvd.inst->ring; uint32_t rb_bufsz, tmp; uint32_t lmi_swap_cntl; uint32_t mp_swap_cntl; @@ -866,14 +867,14 @@ static int uvd_v6_0_start(struct amdgpu_device *adev) WREG32_FIELD(UVD_RBC_RB_CNTL, RB_NO_FETCH, 0); if (uvd_v6_0_enc_support(adev)) { - ring = &adev->uvd.ring_enc[0]; + ring = &adev->uvd.inst->ring_enc[0]; WREG32(mmUVD_RB_RPTR, lower_32_bits(ring->wptr)); WREG32(mmUVD_RB_WPTR, lower_32_bits(ring->wptr)); WREG32(mmUVD_RB_BASE_LO, ring->gpu_addr); WREG32(mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr)); WREG32(mmUVD_RB_SIZE, ring->ring_size / 4); - ring = &adev->uvd.ring_enc[1]; + ring = &adev->uvd.inst->ring_enc[1]; WREG32(mmUVD_RB_RPTR2, lower_32_bits(ring->wptr)); WREG32(mmUVD_RB_WPTR2, lower_32_bits(ring->wptr)); WREG32(mmUVD_RB_BASE_LO2, ring->gpu_addr); @@ -1099,6 +1100,18 @@ static void uvd_v6_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring) amdgpu_ring_write(ring, 0xE); } +static void uvd_v6_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) +{ + int i; + + WARN_ON(ring->wptr % 2 || count % 2); + + for (i = 0; i < count / 2; i++) { + amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0)); + amdgpu_ring_write(ring, 0); + } +} + static void uvd_v6_0_enc_ring_emit_pipeline_sync(struct amdgpu_ring *ring) { uint32_t seq = ring->fence_drv.sync_seq; @@ -1158,10 +1171,10 @@ static bool uvd_v6_0_check_soft_reset(void *handle) srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_UVD, 1); if (srbm_soft_reset) { - adev->uvd.srbm_soft_reset = srbm_soft_reset; + adev->uvd.inst->srbm_soft_reset = srbm_soft_reset; return true; } else { - adev->uvd.srbm_soft_reset = 0; + adev->uvd.inst->srbm_soft_reset = 0; return false; } } @@ -1170,7 +1183,7 @@ static int uvd_v6_0_pre_soft_reset(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (!adev->uvd.srbm_soft_reset) + if (!adev->uvd.inst->srbm_soft_reset) return 0; uvd_v6_0_stop(adev); @@ -1182,9 +1195,9 @@ static int uvd_v6_0_soft_reset(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; u32 srbm_soft_reset; - if (!adev->uvd.srbm_soft_reset) + if (!adev->uvd.inst->srbm_soft_reset) return 0; - srbm_soft_reset = adev->uvd.srbm_soft_reset; + srbm_soft_reset = adev->uvd.inst->srbm_soft_reset; if (srbm_soft_reset) { u32 tmp; @@ -1212,7 +1225,7 @@ static int uvd_v6_0_post_soft_reset(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (!adev->uvd.srbm_soft_reset) + if (!adev->uvd.inst->srbm_soft_reset) return 0; mdelay(5); @@ -1238,17 +1251,17 @@ static int uvd_v6_0_process_interrupt(struct amdgpu_device *adev, switch (entry->src_id) { case 124: - amdgpu_fence_process(&adev->uvd.ring); + amdgpu_fence_process(&adev->uvd.inst->ring); break; case 119: if (likely(uvd_v6_0_enc_support(adev))) - amdgpu_fence_process(&adev->uvd.ring_enc[0]); + amdgpu_fence_process(&adev->uvd.inst->ring_enc[0]); else int_handled = false; break; case 120: if (likely(uvd_v6_0_enc_support(adev))) - amdgpu_fence_process(&adev->uvd.ring_enc[1]); + amdgpu_fence_process(&adev->uvd.inst->ring_enc[1]); else int_handled = false; break; @@ -1531,7 +1544,6 @@ static const struct amd_ip_funcs uvd_v6_0_ip_funcs = { static const struct amdgpu_ring_funcs uvd_v6_0_ring_phys_funcs = { .type = AMDGPU_RING_TYPE_UVD, .align_mask = 0xf, - .nop = PACKET0(mmUVD_NO_OP, 0), .support_64bit_ptrs = false, .get_rptr = uvd_v6_0_ring_get_rptr, .get_wptr = uvd_v6_0_ring_get_wptr, @@ -1547,7 +1559,7 @@ static const struct amdgpu_ring_funcs uvd_v6_0_ring_phys_funcs = { .emit_hdp_flush = uvd_v6_0_ring_emit_hdp_flush, .test_ring = uvd_v6_0_ring_test_ring, .test_ib = amdgpu_uvd_ring_test_ib, - .insert_nop = amdgpu_ring_insert_nop, + .insert_nop = uvd_v6_0_ring_insert_nop, .pad_ib = amdgpu_ring_generic_pad_ib, .begin_use = amdgpu_uvd_ring_begin_use, .end_use = amdgpu_uvd_ring_end_use, @@ -1612,10 +1624,10 @@ static const struct amdgpu_ring_funcs uvd_v6_0_enc_ring_vm_funcs = { static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev) { if (adev->asic_type >= CHIP_POLARIS10) { - adev->uvd.ring.funcs = &uvd_v6_0_ring_vm_funcs; + adev->uvd.inst->ring.funcs = &uvd_v6_0_ring_vm_funcs; DRM_INFO("UVD is enabled in VM mode\n"); } else { - adev->uvd.ring.funcs = &uvd_v6_0_ring_phys_funcs; + adev->uvd.inst->ring.funcs = &uvd_v6_0_ring_phys_funcs; DRM_INFO("UVD is enabled in physical mode\n"); } } @@ -1625,7 +1637,7 @@ static void uvd_v6_0_set_enc_ring_funcs(struct amdgpu_device *adev) int i; for (i = 0; i < adev->uvd.num_enc_rings; ++i) - adev->uvd.ring_enc[i].funcs = &uvd_v6_0_enc_ring_vm_funcs; + adev->uvd.inst->ring_enc[i].funcs = &uvd_v6_0_enc_ring_vm_funcs; DRM_INFO("UVD ENC is enabled in VM mode\n"); } @@ -1638,11 +1650,11 @@ static const struct amdgpu_irq_src_funcs uvd_v6_0_irq_funcs = { static void uvd_v6_0_set_irq_funcs(struct amdgpu_device *adev) { if (uvd_v6_0_enc_support(adev)) - adev->uvd.irq.num_types = adev->uvd.num_enc_rings + 1; + adev->uvd.inst->irq.num_types = adev->uvd.num_enc_rings + 1; else - adev->uvd.irq.num_types = 1; + adev->uvd.inst->irq.num_types = 1; - adev->uvd.irq.funcs = &uvd_v6_0_irq_funcs; + adev->uvd.inst->irq.funcs = &uvd_v6_0_irq_funcs; } const struct amdgpu_ip_block_version uvd_v6_0_ip_block = diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c index 0ca63d588670..57d32f21b3a6 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c @@ -40,6 +40,8 @@ #include "mmhub/mmhub_1_0_offset.h" #include "mmhub/mmhub_1_0_sh_mask.h" +#define UVD7_MAX_HW_INSTANCES_VEGA20 2 + static void uvd_v7_0_set_ring_funcs(struct amdgpu_device *adev); static void uvd_v7_0_set_enc_ring_funcs(struct amdgpu_device *adev); static void uvd_v7_0_set_irq_funcs(struct amdgpu_device *adev); @@ -47,6 +49,11 @@ static int uvd_v7_0_start(struct amdgpu_device *adev); static void uvd_v7_0_stop(struct amdgpu_device *adev); static int uvd_v7_0_sriov_start(struct amdgpu_device *adev); +static int amdgpu_ih_clientid_uvds[] = { + SOC15_IH_CLIENTID_UVD, + SOC15_IH_CLIENTID_UVD1 +}; + /** * uvd_v7_0_ring_get_rptr - get read pointer * @@ -58,7 +65,7 @@ static uint64_t uvd_v7_0_ring_get_rptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - return RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR); + return RREG32_SOC15(UVD, ring->me, mmUVD_RBC_RB_RPTR); } /** @@ -72,10 +79,10 @@ static uint64_t uvd_v7_0_enc_ring_get_rptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - if (ring == &adev->uvd.ring_enc[0]) - return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR); + if (ring == &adev->uvd.inst[ring->me].ring_enc[0]) + return RREG32_SOC15(UVD, ring->me, mmUVD_RB_RPTR); else - return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2); + return RREG32_SOC15(UVD, ring->me, mmUVD_RB_RPTR2); } /** @@ -89,7 +96,7 @@ static uint64_t uvd_v7_0_ring_get_wptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - return RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR); + return RREG32_SOC15(UVD, ring->me, mmUVD_RBC_RB_WPTR); } /** @@ -106,10 +113,10 @@ static uint64_t uvd_v7_0_enc_ring_get_wptr(struct amdgpu_ring *ring) if (ring->use_doorbell) return adev->wb.wb[ring->wptr_offs]; - if (ring == &adev->uvd.ring_enc[0]) - return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR); + if (ring == &adev->uvd.inst[ring->me].ring_enc[0]) + return RREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR); else - return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2); + return RREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR2); } /** @@ -123,7 +130,7 @@ static void uvd_v7_0_ring_set_wptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr)); + WREG32_SOC15(UVD, ring->me, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr)); } /** @@ -144,11 +151,11 @@ static void uvd_v7_0_enc_ring_set_wptr(struct amdgpu_ring *ring) return; } - if (ring == &adev->uvd.ring_enc[0]) - WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, + if (ring == &adev->uvd.inst[ring->me].ring_enc[0]) + WREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR, lower_32_bits(ring->wptr)); else - WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, + WREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr)); } @@ -170,8 +177,8 @@ static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring) r = amdgpu_ring_alloc(ring, 16); if (r) { - DRM_ERROR("amdgpu: uvd enc failed to lock ring %d (%d).\n", - ring->idx, r); + DRM_ERROR("amdgpu: uvd enc failed to lock (%d)ring %d (%d).\n", + ring->me, ring->idx, r); return r; } amdgpu_ring_write(ring, HEVC_ENC_CMD_END); @@ -184,11 +191,11 @@ static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring) } if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); + DRM_DEBUG("(%d)ring test on %d succeeded in %d usecs\n", + ring->me, ring->idx, i); } else { - DRM_ERROR("amdgpu: ring %d test failed\n", - ring->idx); + DRM_ERROR("amdgpu: (%d)ring %d test failed\n", + ring->me, ring->idx); r = -ETIMEDOUT; } @@ -342,24 +349,24 @@ static int uvd_v7_0_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = uvd_v7_0_enc_get_create_msg(ring, 1, NULL); if (r) { - DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r); + DRM_ERROR("amdgpu: (%d)failed to get create msg (%ld).\n", ring->me, r); goto error; } r = uvd_v7_0_enc_get_destroy_msg(ring, 1, true, &fence); if (r) { - DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r); + DRM_ERROR("amdgpu: (%d)failed to get destroy ib (%ld).\n", ring->me, r); goto error; } r = dma_fence_wait_timeout(fence, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out.\n"); + DRM_ERROR("amdgpu: (%d)IB test timed out.\n", ring->me); r = -ETIMEDOUT; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); + DRM_ERROR("amdgpu: (%d)fence wait failed (%ld).\n", ring->me, r); } else { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + DRM_DEBUG("ib test on (%d)ring %d succeeded\n", ring->me, ring->idx); r = 0; } error: @@ -370,6 +377,10 @@ error: static int uvd_v7_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + if (adev->asic_type == CHIP_VEGA20) + adev->uvd.num_uvd_inst = UVD7_MAX_HW_INSTANCES_VEGA20; + else + adev->uvd.num_uvd_inst = 1; if (amdgpu_sriov_vf(adev)) adev->uvd.num_enc_rings = 1; @@ -386,19 +397,21 @@ static int uvd_v7_0_sw_init(void *handle) { struct amdgpu_ring *ring; struct drm_sched_rq *rq; - int i, r; + int i, j, r; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - /* UVD TRAP */ - r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_UVD, 124, &adev->uvd.irq); - if (r) - return r; - - /* UVD ENC TRAP */ - for (i = 0; i < adev->uvd.num_enc_rings; ++i) { - r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_UVD, i + 119, &adev->uvd.irq); + for (j = 0; j < adev->uvd.num_uvd_inst; j++) { + /* UVD TRAP */ + r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_uvds[j], 124, &adev->uvd.inst[j].irq); if (r) return r; + + /* UVD ENC TRAP */ + for (i = 0; i < adev->uvd.num_enc_rings; ++i) { + r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_uvds[j], i + 119, &adev->uvd.inst[j].irq); + if (r) + return r; + } } r = amdgpu_uvd_sw_init(adev); @@ -415,43 +428,48 @@ static int uvd_v7_0_sw_init(void *handle) DRM_INFO("PSP loading UVD firmware\n"); } - ring = &adev->uvd.ring_enc[0]; - rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_NORMAL]; - r = drm_sched_entity_init(&ring->sched, &adev->uvd.entity_enc, - rq, NULL); - if (r) { - DRM_ERROR("Failed setting up UVD ENC run queue.\n"); - return r; + for (j = 0; j < adev->uvd.num_uvd_inst; j++) { + ring = &adev->uvd.inst[j].ring_enc[0]; + rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_NORMAL]; + r = drm_sched_entity_init(&ring->sched, &adev->uvd.inst[j].entity_enc, + rq, NULL); + if (r) { + DRM_ERROR("(%d)Failed setting up UVD ENC run queue.\n", j); + return r; + } } r = amdgpu_uvd_resume(adev); if (r) return r; - if (!amdgpu_sriov_vf(adev)) { - ring = &adev->uvd.ring; - sprintf(ring->name, "uvd"); - r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0); - if (r) - return r; - } - for (i = 0; i < adev->uvd.num_enc_rings; ++i) { - ring = &adev->uvd.ring_enc[i]; - sprintf(ring->name, "uvd_enc%d", i); - if (amdgpu_sriov_vf(adev)) { - ring->use_doorbell = true; - - /* currently only use the first enconding ring for - * sriov, so set unused location for other unused rings. - */ - if (i == 0) - ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING0_1 * 2; - else - ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING2_3 * 2 + 1; + for (j = 0; j < adev->uvd.num_uvd_inst; j++) { + if (!amdgpu_sriov_vf(adev)) { + ring = &adev->uvd.inst[j].ring; + sprintf(ring->name, "uvd<%d>", j); + r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst[j].irq, 0); + if (r) + return r; + } + + for (i = 0; i < adev->uvd.num_enc_rings; ++i) { + ring = &adev->uvd.inst[j].ring_enc[i]; + sprintf(ring->name, "uvd_enc%d<%d>", i, j); + if (amdgpu_sriov_vf(adev)) { + ring->use_doorbell = true; + + /* currently only use the first enconding ring for + * sriov, so set unused location for other unused rings. + */ + if (i == 0) + ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING0_1 * 2; + else + ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING2_3 * 2 + 1; + } + r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst[j].irq, 0); + if (r) + return r; } - r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0); - if (r) - return r; } r = amdgpu_virt_alloc_mm_table(adev); @@ -463,7 +481,7 @@ static int uvd_v7_0_sw_init(void *handle) static int uvd_v7_0_sw_fini(void *handle) { - int i, r; + int i, j, r; struct amdgpu_device *adev = (struct amdgpu_device *)handle; amdgpu_virt_free_mm_table(adev); @@ -472,11 +490,12 @@ static int uvd_v7_0_sw_fini(void *handle) if (r) return r; - drm_sched_entity_fini(&adev->uvd.ring_enc[0].sched, &adev->uvd.entity_enc); - - for (i = 0; i < adev->uvd.num_enc_rings; ++i) - amdgpu_ring_fini(&adev->uvd.ring_enc[i]); + for (j = 0; j < adev->uvd.num_uvd_inst; ++j) { + drm_sched_entity_fini(&adev->uvd.inst[j].ring_enc[0].sched, &adev->uvd.inst[j].entity_enc); + for (i = 0; i < adev->uvd.num_enc_rings; ++i) + amdgpu_ring_fini(&adev->uvd.inst[j].ring_enc[i]); + } return amdgpu_uvd_sw_fini(adev); } @@ -490,9 +509,9 @@ static int uvd_v7_0_sw_fini(void *handle) static int uvd_v7_0_hw_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring; uint32_t tmp; - int i, r; + int i, j, r; if (amdgpu_sriov_vf(adev)) r = uvd_v7_0_sriov_start(adev); @@ -501,57 +520,60 @@ static int uvd_v7_0_hw_init(void *handle) if (r) goto done; - if (!amdgpu_sriov_vf(adev)) { - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; - goto done; - } - - r = amdgpu_ring_alloc(ring, 10); - if (r) { - DRM_ERROR("amdgpu: ring failed to lock UVD ring (%d).\n", r); - goto done; + for (j = 0; j < adev->uvd.num_uvd_inst; ++j) { + ring = &adev->uvd.inst[j].ring; + + if (!amdgpu_sriov_vf(adev)) { + ring->ready = true; + r = amdgpu_ring_test_ring(ring); + if (r) { + ring->ready = false; + goto done; + } + + r = amdgpu_ring_alloc(ring, 10); + if (r) { + DRM_ERROR("amdgpu: (%d)ring failed to lock UVD ring (%d).\n", j, r); + goto done; + } + + tmp = PACKET0(SOC15_REG_OFFSET(UVD, j, + mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL), 0); + amdgpu_ring_write(ring, tmp); + amdgpu_ring_write(ring, 0xFFFFF); + + tmp = PACKET0(SOC15_REG_OFFSET(UVD, j, + mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL), 0); + amdgpu_ring_write(ring, tmp); + amdgpu_ring_write(ring, 0xFFFFF); + + tmp = PACKET0(SOC15_REG_OFFSET(UVD, j, + mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL), 0); + amdgpu_ring_write(ring, tmp); + amdgpu_ring_write(ring, 0xFFFFF); + + /* Clear timeout status bits */ + amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, j, + mmUVD_SEMA_TIMEOUT_STATUS), 0)); + amdgpu_ring_write(ring, 0x8); + + amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, j, + mmUVD_SEMA_CNTL), 0)); + amdgpu_ring_write(ring, 3); + + amdgpu_ring_commit(ring); } - tmp = PACKET0(SOC15_REG_OFFSET(UVD, 0, - mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL), 0); - amdgpu_ring_write(ring, tmp); - amdgpu_ring_write(ring, 0xFFFFF); - - tmp = PACKET0(SOC15_REG_OFFSET(UVD, 0, - mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL), 0); - amdgpu_ring_write(ring, tmp); - amdgpu_ring_write(ring, 0xFFFFF); - - tmp = PACKET0(SOC15_REG_OFFSET(UVD, 0, - mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL), 0); - amdgpu_ring_write(ring, tmp); - amdgpu_ring_write(ring, 0xFFFFF); - - /* Clear timeout status bits */ - amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0, - mmUVD_SEMA_TIMEOUT_STATUS), 0)); - amdgpu_ring_write(ring, 0x8); - - amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0, - mmUVD_SEMA_CNTL), 0)); - amdgpu_ring_write(ring, 3); - - amdgpu_ring_commit(ring); - } - - for (i = 0; i < adev->uvd.num_enc_rings; ++i) { - ring = &adev->uvd.ring_enc[i]; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; - goto done; + for (i = 0; i < adev->uvd.num_enc_rings; ++i) { + ring = &adev->uvd.inst[j].ring_enc[i]; + ring->ready = true; + r = amdgpu_ring_test_ring(ring); + if (r) { + ring->ready = false; + goto done; + } } } - done: if (!r) DRM_INFO("UVD and UVD ENC initialized successfully.\n"); @@ -569,7 +591,7 @@ done: static int uvd_v7_0_hw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct amdgpu_ring *ring = &adev->uvd.ring; + int i; if (!amdgpu_sriov_vf(adev)) uvd_v7_0_stop(adev); @@ -578,7 +600,8 @@ static int uvd_v7_0_hw_fini(void *handle) DRM_DEBUG("For SRIOV client, shouldn't do anything.\n"); } - ring->ready = false; + for (i = 0; i < adev->uvd.num_uvd_inst; ++i) + adev->uvd.inst[i].ring.ready = false; return 0; } @@ -618,48 +641,51 @@ static void uvd_v7_0_mc_resume(struct amdgpu_device *adev) { uint32_t size = AMDGPU_UVD_FIRMWARE_SIZE(adev); uint32_t offset; + int i; - if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) { - WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW, - lower_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr)); - WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH, - upper_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr)); - offset = 0; - } else { - WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW, - lower_32_bits(adev->uvd.gpu_addr)); - WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH, - upper_32_bits(adev->uvd.gpu_addr)); - offset = size; - } + for (i = 0; i < adev->uvd.num_uvd_inst; ++i) { + if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) { + WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW, + lower_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr)); + WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH, + upper_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr)); + offset = 0; + } else { + WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW, + lower_32_bits(adev->uvd.inst[i].gpu_addr)); + WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH, + upper_32_bits(adev->uvd.inst[i].gpu_addr)); + offset = size; + } - WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0, - AMDGPU_UVD_FIRMWARE_OFFSET >> 3); - WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE0, size); - - WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW, - lower_32_bits(adev->uvd.gpu_addr + offset)); - WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH, - upper_32_bits(adev->uvd.gpu_addr + offset)); - WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET1, (1 << 21)); - WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_UVD_HEAP_SIZE); - - WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW, - lower_32_bits(adev->uvd.gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE)); - WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH, - upper_32_bits(adev->uvd.gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE)); - WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2, (2 << 21)); - WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE2, - AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40)); - - WREG32_SOC15(UVD, 0, mmUVD_UDEC_ADDR_CONFIG, - adev->gfx.config.gb_addr_config); - WREG32_SOC15(UVD, 0, mmUVD_UDEC_DB_ADDR_CONFIG, - adev->gfx.config.gb_addr_config); - WREG32_SOC15(UVD, 0, mmUVD_UDEC_DBW_ADDR_CONFIG, - adev->gfx.config.gb_addr_config); - - WREG32_SOC15(UVD, 0, mmUVD_GP_SCRATCH4, adev->uvd.max_handles); + WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET0, + AMDGPU_UVD_FIRMWARE_OFFSET >> 3); + WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_SIZE0, size); + + WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW, + lower_32_bits(adev->uvd.inst[i].gpu_addr + offset)); + WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH, + upper_32_bits(adev->uvd.inst[i].gpu_addr + offset)); + WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET1, (1 << 21)); + WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_UVD_HEAP_SIZE); + + WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW, + lower_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE)); + WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH, + upper_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE)); + WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET2, (2 << 21)); + WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_SIZE2, + AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40)); + + WREG32_SOC15(UVD, i, mmUVD_UDEC_ADDR_CONFIG, + adev->gfx.config.gb_addr_config); + WREG32_SOC15(UVD, i, mmUVD_UDEC_DB_ADDR_CONFIG, + adev->gfx.config.gb_addr_config); + WREG32_SOC15(UVD, i, mmUVD_UDEC_DBW_ADDR_CONFIG, + adev->gfx.config.gb_addr_config); + + WREG32_SOC15(UVD, i, mmUVD_GP_SCRATCH4, adev->uvd.max_handles); + } } static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev, @@ -669,6 +695,7 @@ static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev, uint64_t addr = table->gpu_addr; struct mmsch_v1_0_init_header *header = (struct mmsch_v1_0_init_header *)table->cpu_addr; uint32_t size; + int i; size = header->header_size + header->vce_table_size + header->uvd_table_size; @@ -688,11 +715,12 @@ static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev, /* 4, set resp to zero */ WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP, 0); - WDOORBELL32(adev->uvd.ring_enc[0].doorbell_index, 0); - adev->wb.wb[adev->uvd.ring_enc[0].wptr_offs] = 0; - adev->uvd.ring_enc[0].wptr = 0; - adev->uvd.ring_enc[0].wptr_old = 0; - + for (i = 0; i < adev->uvd.num_uvd_inst; ++i) { + WDOORBELL32(adev->uvd.inst[i].ring_enc[0].doorbell_index, 0); + adev->wb.wb[adev->uvd.inst[i].ring_enc[0].wptr_offs] = 0; + adev->uvd.inst[i].ring_enc[0].wptr = 0; + adev->uvd.inst[i].ring_enc[0].wptr_old = 0; + } /* 5, kick off the initialization and wait until VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero */ WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_HOST, 0x10000001); @@ -725,6 +753,7 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev) struct mmsch_v1_0_cmd_end end = { {0} }; uint32_t *init_table = adev->virt.mm_table.cpu_addr; struct mmsch_v1_0_init_header *header = (struct mmsch_v1_0_init_header *)init_table; + uint8_t i = 0; direct_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_WRITE; direct_rd_mod_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE; @@ -742,120 +771,121 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev) init_table += header->uvd_table_offset; - ring = &adev->uvd.ring; - ring->wptr = 0; - size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4); - - MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), - 0xFFFFFFFF, 0x00000004); - /* mc resume*/ - if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) { - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), - lower_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr)); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), - upper_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr)); - offset = 0; - } else { - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), - lower_32_bits(adev->uvd.gpu_addr)); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), - upper_32_bits(adev->uvd.gpu_addr)); - offset = size; + for (i = 0; i < adev->uvd.num_uvd_inst; ++i) { + ring = &adev->uvd.inst[i].ring; + ring->wptr = 0; + size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4); + + MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS), + 0xFFFFFFFF, 0x00000004); + /* mc resume*/ + if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) { + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), + lower_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr)); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), + upper_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr)); + offset = 0; + } else { + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), + lower_32_bits(adev->uvd.inst[i].gpu_addr)); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), + upper_32_bits(adev->uvd.inst[i].gpu_addr)); + offset = size; + } + + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET0), + AMDGPU_UVD_FIRMWARE_OFFSET >> 3); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE0), size); + + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), + lower_32_bits(adev->uvd.inst[i].gpu_addr + offset)); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), + upper_32_bits(adev->uvd.inst[i].gpu_addr + offset)); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET1), (1 << 21)); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE1), AMDGPU_UVD_HEAP_SIZE); + + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW), + lower_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE)); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH), + upper_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE)); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET2), (2 << 21)); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE2), + AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40)); + + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_GP_SCRATCH4), adev->uvd.max_handles); + /* mc resume end*/ + + /* disable clock gating */ + MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_CGC_CTRL), + ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK, 0); + + /* disable interupt */ + MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_MASTINT_EN), + ~UVD_MASTINT_EN__VCPU_EN_MASK, 0); + + /* stall UMC and register bus before resetting VCPU */ + MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2), + ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, + UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); + + /* put LMI, VCPU, RBC etc... into reset */ + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_SOFT_RESET), + (uint32_t)(UVD_SOFT_RESET__LMI_SOFT_RESET_MASK | + UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK | + UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK | + UVD_SOFT_RESET__RBC_SOFT_RESET_MASK | + UVD_SOFT_RESET__CSM_SOFT_RESET_MASK | + UVD_SOFT_RESET__CXW_SOFT_RESET_MASK | + UVD_SOFT_RESET__TAP_SOFT_RESET_MASK | + UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK)); + + /* initialize UVD memory controller */ + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL), + (uint32_t)((0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) | + UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK | + UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK | + UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK | + UVD_LMI_CTRL__REQ_MODE_MASK | + 0x00100000L)); + + /* take all subblocks out of reset, except VCPU */ + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_SOFT_RESET), + UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); + + /* enable VCPU clock */ + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CNTL), + UVD_VCPU_CNTL__CLK_EN_MASK); + + /* enable master interrupt */ + MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_MASTINT_EN), + ~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK), + (UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK)); + + /* clear the bit 4 of UVD_STATUS */ + MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS), + ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT), 0); + + /* force RBC into idle state */ + size = order_base_2(ring->ring_size); + tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, size); + tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RBC_RB_CNTL), tmp); + + ring = &adev->uvd.inst[i].ring_enc[0]; + ring->wptr = 0; + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_LO), ring->gpu_addr); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_HI), upper_32_bits(ring->gpu_addr)); + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RB_SIZE), ring->ring_size / 4); + + /* boot up the VCPU */ + MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_SOFT_RESET), 0); + + /* enable UMC */ + MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2), + ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0); + + MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS), 0x02, 0x02); } - - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0), - AMDGPU_UVD_FIRMWARE_OFFSET >> 3); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_SIZE0), size); - - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), - lower_32_bits(adev->uvd.gpu_addr + offset)); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), - upper_32_bits(adev->uvd.gpu_addr + offset)); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_OFFSET1), (1 << 21)); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_SIZE1), AMDGPU_UVD_HEAP_SIZE); - - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW), - lower_32_bits(adev->uvd.gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE)); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH), - upper_32_bits(adev->uvd.gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE)); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2), (2 << 21)); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_SIZE2), - AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40)); - - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_GP_SCRATCH4), adev->uvd.max_handles); - /* mc resume end*/ - - /* disable clock gating */ - MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_CGC_CTRL), - ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK, 0); - - /* disable interupt */ - MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), - ~UVD_MASTINT_EN__VCPU_EN_MASK, 0); - - /* stall UMC and register bus before resetting VCPU */ - MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), - ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, - UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); - - /* put LMI, VCPU, RBC etc... into reset */ - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), - (uint32_t)(UVD_SOFT_RESET__LMI_SOFT_RESET_MASK | - UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK | - UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK | - UVD_SOFT_RESET__RBC_SOFT_RESET_MASK | - UVD_SOFT_RESET__CSM_SOFT_RESET_MASK | - UVD_SOFT_RESET__CXW_SOFT_RESET_MASK | - UVD_SOFT_RESET__TAP_SOFT_RESET_MASK | - UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK)); - - /* initialize UVD memory controller */ - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL), - (uint32_t)((0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) | - UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK | - UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK | - UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK | - UVD_LMI_CTRL__REQ_MODE_MASK | - 0x00100000L)); - - /* take all subblocks out of reset, except VCPU */ - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), - UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); - - /* enable VCPU clock */ - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CNTL), - UVD_VCPU_CNTL__CLK_EN_MASK); - - /* enable master interrupt */ - MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), - ~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK), - (UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK)); - - /* clear the bit 4 of UVD_STATUS */ - MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), - ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT), 0); - - /* force RBC into idle state */ - size = order_base_2(ring->ring_size); - tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, size); - tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), tmp); - - ring = &adev->uvd.ring_enc[0]; - ring->wptr = 0; - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_BASE_LO), ring->gpu_addr); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_BASE_HI), upper_32_bits(ring->gpu_addr)); - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_SIZE), ring->ring_size / 4); - - /* boot up the VCPU */ - MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0); - - /* enable UMC */ - MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), - ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0); - - MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0x02, 0x02); - /* add end packet */ memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end)); table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4; @@ -874,15 +904,17 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev) */ static int uvd_v7_0_start(struct amdgpu_device *adev) { - struct amdgpu_ring *ring = &adev->uvd.ring; + struct amdgpu_ring *ring; uint32_t rb_bufsz, tmp; uint32_t lmi_swap_cntl; uint32_t mp_swap_cntl; - int i, j, r; + int i, j, k, r; - /* disable DPG */ - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS), 0, - ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + for (k = 0; k < adev->uvd.num_uvd_inst; ++k) { + /* disable DPG */ + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_POWER_STATUS), 0, + ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + } /* disable byte swapping */ lmi_swap_cntl = 0; @@ -890,157 +922,159 @@ static int uvd_v7_0_start(struct amdgpu_device *adev) uvd_v7_0_mc_resume(adev); - /* disable clock gating */ - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_CGC_CTRL), 0, - ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK); - - /* disable interupt */ - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), 0, - ~UVD_MASTINT_EN__VCPU_EN_MASK); - - /* stall UMC and register bus before resetting VCPU */ - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), - UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, - ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); - mdelay(1); - - /* put LMI, VCPU, RBC etc... into reset */ - WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET, - UVD_SOFT_RESET__LMI_SOFT_RESET_MASK | - UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK | - UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK | - UVD_SOFT_RESET__RBC_SOFT_RESET_MASK | - UVD_SOFT_RESET__CSM_SOFT_RESET_MASK | - UVD_SOFT_RESET__CXW_SOFT_RESET_MASK | - UVD_SOFT_RESET__TAP_SOFT_RESET_MASK | - UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK); - mdelay(5); + for (k = 0; k < adev->uvd.num_uvd_inst; ++k) { + ring = &adev->uvd.inst[k].ring; + /* disable clock gating */ + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_CGC_CTRL), 0, + ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK); - /* initialize UVD memory controller */ - WREG32_SOC15(UVD, 0, mmUVD_LMI_CTRL, - (0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) | - UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK | - UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK | - UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK | - UVD_LMI_CTRL__REQ_MODE_MASK | - 0x00100000L); + /* disable interupt */ + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_MASTINT_EN), 0, + ~UVD_MASTINT_EN__VCPU_EN_MASK); + + /* stall UMC and register bus before resetting VCPU */ + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_LMI_CTRL2), + UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, + ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); + mdelay(1); + + /* put LMI, VCPU, RBC etc... into reset */ + WREG32_SOC15(UVD, k, mmUVD_SOFT_RESET, + UVD_SOFT_RESET__LMI_SOFT_RESET_MASK | + UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK | + UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK | + UVD_SOFT_RESET__RBC_SOFT_RESET_MASK | + UVD_SOFT_RESET__CSM_SOFT_RESET_MASK | + UVD_SOFT_RESET__CXW_SOFT_RESET_MASK | + UVD_SOFT_RESET__TAP_SOFT_RESET_MASK | + UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK); + mdelay(5); + + /* initialize UVD memory controller */ + WREG32_SOC15(UVD, k, mmUVD_LMI_CTRL, + (0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) | + UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK | + UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK | + UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK | + UVD_LMI_CTRL__REQ_MODE_MASK | + 0x00100000L); #ifdef __BIG_ENDIAN - /* swap (8 in 32) RB and IB */ - lmi_swap_cntl = 0xa; - mp_swap_cntl = 0; + /* swap (8 in 32) RB and IB */ + lmi_swap_cntl = 0xa; + mp_swap_cntl = 0; #endif - WREG32_SOC15(UVD, 0, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl); - WREG32_SOC15(UVD, 0, mmUVD_MP_SWAP_CNTL, mp_swap_cntl); - - WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA0, 0x40c2040); - WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA1, 0x0); - WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB0, 0x40c2040); - WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB1, 0x0); - WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_ALU, 0); - WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUX, 0x88); - - /* take all subblocks out of reset, except VCPU */ - WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET, - UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); - mdelay(5); + WREG32_SOC15(UVD, k, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl); + WREG32_SOC15(UVD, k, mmUVD_MP_SWAP_CNTL, mp_swap_cntl); - /* enable VCPU clock */ - WREG32_SOC15(UVD, 0, mmUVD_VCPU_CNTL, - UVD_VCPU_CNTL__CLK_EN_MASK); + WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXA0, 0x40c2040); + WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXA1, 0x0); + WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXB0, 0x40c2040); + WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXB1, 0x0); + WREG32_SOC15(UVD, k, mmUVD_MPC_SET_ALU, 0); + WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUX, 0x88); - /* enable UMC */ - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0, - ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); + /* take all subblocks out of reset, except VCPU */ + WREG32_SOC15(UVD, k, mmUVD_SOFT_RESET, + UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); + mdelay(5); + + /* enable VCPU clock */ + WREG32_SOC15(UVD, k, mmUVD_VCPU_CNTL, + UVD_VCPU_CNTL__CLK_EN_MASK); + + /* enable UMC */ + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_LMI_CTRL2), 0, + ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); - /* boot up the VCPU */ - WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET, 0); - mdelay(10); + /* boot up the VCPU */ + WREG32_SOC15(UVD, k, mmUVD_SOFT_RESET, 0); + mdelay(10); - for (i = 0; i < 10; ++i) { - uint32_t status; + for (i = 0; i < 10; ++i) { + uint32_t status; - for (j = 0; j < 100; ++j) { - status = RREG32_SOC15(UVD, 0, mmUVD_STATUS); + for (j = 0; j < 100; ++j) { + status = RREG32_SOC15(UVD, k, mmUVD_STATUS); + if (status & 2) + break; + mdelay(10); + } + r = 0; if (status & 2) break; + + DRM_ERROR("UVD(%d) not responding, trying to reset the VCPU!!!\n", k); + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_SOFT_RESET), + UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK, + ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); mdelay(10); + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_SOFT_RESET), 0, + ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); + mdelay(10); + r = -1; } - r = 0; - if (status & 2) - break; - - DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n"); - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), - UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK, - ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); - mdelay(10); - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0, - ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); - mdelay(10); - r = -1; - } - if (r) { - DRM_ERROR("UVD not responding, giving up!!!\n"); - return r; - } - /* enable master interrupt */ - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), - (UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK), - ~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK)); - - /* clear the bit 4 of UVD_STATUS */ - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0, - ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT)); - - /* force RBC into idle state */ - rb_bufsz = order_base_2(ring->ring_size); - tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz); - tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1); - tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1); - tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0); - tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1); - tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1); - WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp); - - /* set the write pointer delay */ - WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL, 0); - - /* set the wb address */ - WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR, - (upper_32_bits(ring->gpu_addr) >> 2)); - - /* programm the RB_BASE for ring buffer */ - WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, - lower_32_bits(ring->gpu_addr)); - WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, - upper_32_bits(ring->gpu_addr)); - - /* Initialize the ring buffer's read and write pointers */ - WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR, 0); - - ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR); - WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR, - lower_32_bits(ring->wptr)); - - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), 0, - ~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK); - - ring = &adev->uvd.ring_enc[0]; - WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr)); - WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr)); - WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr); - WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr)); - WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4); + if (r) { + DRM_ERROR("UVD(%d) not responding, giving up!!!\n", k); + return r; + } + /* enable master interrupt */ + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_MASTINT_EN), + (UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK), + ~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK)); - ring = &adev->uvd.ring_enc[1]; - WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr)); - WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr)); - WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr); - WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr)); - WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4); + /* clear the bit 4 of UVD_STATUS */ + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_STATUS), 0, + ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT)); + /* force RBC into idle state */ + rb_bufsz = order_base_2(ring->ring_size); + tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz); + tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1); + tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1); + tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0); + tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1); + tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1); + WREG32_SOC15(UVD, k, mmUVD_RBC_RB_CNTL, tmp); + + /* set the write pointer delay */ + WREG32_SOC15(UVD, k, mmUVD_RBC_RB_WPTR_CNTL, 0); + + /* set the wb address */ + WREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR_ADDR, + (upper_32_bits(ring->gpu_addr) >> 2)); + + /* programm the RB_BASE for ring buffer */ + WREG32_SOC15(UVD, k, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW, + lower_32_bits(ring->gpu_addr)); + WREG32_SOC15(UVD, k, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH, + upper_32_bits(ring->gpu_addr)); + + /* Initialize the ring buffer's read and write pointers */ + WREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR, 0); + + ring->wptr = RREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR); + WREG32_SOC15(UVD, k, mmUVD_RBC_RB_WPTR, + lower_32_bits(ring->wptr)); + + WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_RBC_RB_CNTL), 0, + ~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK); + + ring = &adev->uvd.inst[k].ring_enc[0]; + WREG32_SOC15(UVD, k, mmUVD_RB_RPTR, lower_32_bits(ring->wptr)); + WREG32_SOC15(UVD, k, mmUVD_RB_WPTR, lower_32_bits(ring->wptr)); + WREG32_SOC15(UVD, k, mmUVD_RB_BASE_LO, ring->gpu_addr); + WREG32_SOC15(UVD, k, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr)); + WREG32_SOC15(UVD, k, mmUVD_RB_SIZE, ring->ring_size / 4); + + ring = &adev->uvd.inst[k].ring_enc[1]; + WREG32_SOC15(UVD, k, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr)); + WREG32_SOC15(UVD, k, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr)); + WREG32_SOC15(UVD, k, mmUVD_RB_BASE_LO2, ring->gpu_addr); + WREG32_SOC15(UVD, k, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr)); + WREG32_SOC15(UVD, k, mmUVD_RB_SIZE2, ring->ring_size / 4); + } return 0; } @@ -1053,26 +1087,30 @@ static int uvd_v7_0_start(struct amdgpu_device *adev) */ static void uvd_v7_0_stop(struct amdgpu_device *adev) { - /* force RBC into idle state */ - WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, 0x11010101); - - /* Stall UMC and register bus before resetting VCPU */ - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), - UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, - ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); - mdelay(1); - - /* put VCPU into reset */ - WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET, - UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); - mdelay(5); + uint8_t i = 0; - /* disable VCPU clock */ - WREG32_SOC15(UVD, 0, mmUVD_VCPU_CNTL, 0x0); + for (i = 0; i < adev->uvd.num_uvd_inst; ++i) { + /* force RBC into idle state */ + WREG32_SOC15(UVD, i, mmUVD_RBC_RB_CNTL, 0x11010101); + + /* Stall UMC and register bus before resetting VCPU */ + WREG32_P(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2), + UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, + ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); + mdelay(1); + + /* put VCPU into reset */ + WREG32_SOC15(UVD, i, mmUVD_SOFT_RESET, + UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); + mdelay(5); - /* Unstall UMC and register bus */ - WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0, - ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); + /* disable VCPU clock */ + WREG32_SOC15(UVD, i, mmUVD_VCPU_CNTL, 0x0); + + /* Unstall UMC and register bus */ + WREG32_P(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2), 0, + ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); + } } /** @@ -1091,26 +1129,26 @@ static void uvd_v7_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_CONTEXT_ID), 0)); amdgpu_ring_write(ring, seq); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0)); amdgpu_ring_write(ring, addr & 0xffffffff); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0)); amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0)); amdgpu_ring_write(ring, 0); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0)); amdgpu_ring_write(ring, 0); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0)); amdgpu_ring_write(ring, 0); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0)); amdgpu_ring_write(ring, 2); } @@ -1159,30 +1197,30 @@ static int uvd_v7_0_ring_test_ring(struct amdgpu_ring *ring) unsigned i; int r; - WREG32_SOC15(UVD, 0, mmUVD_CONTEXT_ID, 0xCAFEDEAD); + WREG32_SOC15(UVD, ring->me, mmUVD_CONTEXT_ID, 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", - ring->idx, r); + DRM_ERROR("amdgpu: (%d)cp failed to lock ring %d (%d).\n", + ring->me, ring->idx, r); return r; } amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_CONTEXT_ID), 0)); amdgpu_ring_write(ring, 0xDEADBEEF); amdgpu_ring_commit(ring); for (i = 0; i < adev->usec_timeout; i++) { - tmp = RREG32_SOC15(UVD, 0, mmUVD_CONTEXT_ID); + tmp = RREG32_SOC15(UVD, ring->me, mmUVD_CONTEXT_ID); if (tmp == 0xDEADBEEF) break; DRM_UDELAY(1); } if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); + DRM_DEBUG("(%d)ring test on %d succeeded in %d usecs\n", + ring->me, ring->idx, i); } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); + DRM_ERROR("(%d)amdgpu: ring %d test failed (0x%08X)\n", + ring->me, ring->idx, tmp); r = -EINVAL; } return r; @@ -1203,17 +1241,17 @@ static void uvd_v7_0_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_device *adev = ring->adev; amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_VMID), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_VMID), 0)); amdgpu_ring_write(ring, vmid); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_64BIT_BAR_LOW), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_64BIT_BAR_LOW), 0)); amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr)); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH), 0)); amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr)); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_IB_SIZE), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_RBC_IB_SIZE), 0)); amdgpu_ring_write(ring, ib->length_dw); } @@ -1241,13 +1279,13 @@ static void uvd_v7_0_ring_emit_wreg(struct amdgpu_ring *ring, struct amdgpu_device *adev = ring->adev; amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0)); amdgpu_ring_write(ring, reg << 2); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0)); amdgpu_ring_write(ring, val); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0)); amdgpu_ring_write(ring, 8); } @@ -1257,16 +1295,16 @@ static void uvd_v7_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg, struct amdgpu_device *adev = ring->adev; amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0)); amdgpu_ring_write(ring, reg << 2); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0)); amdgpu_ring_write(ring, val); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GP_SCRATCH8), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GP_SCRATCH8), 0)); amdgpu_ring_write(ring, mask); amdgpu_ring_write(ring, - PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0)); + PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0)); amdgpu_ring_write(ring, 12); } @@ -1287,12 +1325,15 @@ static void uvd_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring, static void uvd_v7_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) { - int i; struct amdgpu_device *adev = ring->adev; + int i; - for (i = 0; i < count; i++) - amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0)); + WARN_ON(ring->wptr % 2 || count % 2); + for (i = 0; i < count / 2; i++) { + amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_NO_OP), 0)); + amdgpu_ring_write(ring, 0); + } } static void uvd_v7_0_enc_ring_insert_end(struct amdgpu_ring *ring) @@ -1359,16 +1400,16 @@ static bool uvd_v7_0_check_soft_reset(void *handle) if (REG_GET_FIELD(tmp, SRBM_STATUS, UVD_RQ_PENDING) || REG_GET_FIELD(tmp, SRBM_STATUS, UVD_BUSY) || - (RREG32_SOC15(UVD, 0, mmUVD_STATUS) & + (RREG32_SOC15(UVD, ring->me, mmUVD_STATUS) & AMDGPU_UVD_STATUS_BUSY_MASK)) srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_UVD, 1); if (srbm_soft_reset) { - adev->uvd.srbm_soft_reset = srbm_soft_reset; + adev->uvd.inst[ring->me].srbm_soft_reset = srbm_soft_reset; return true; } else { - adev->uvd.srbm_soft_reset = 0; + adev->uvd.inst[ring->me].srbm_soft_reset = 0; return false; } } @@ -1377,7 +1418,7 @@ static int uvd_v7_0_pre_soft_reset(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (!adev->uvd.srbm_soft_reset) + if (!adev->uvd.inst[ring->me].srbm_soft_reset) return 0; uvd_v7_0_stop(adev); @@ -1389,9 +1430,9 @@ static int uvd_v7_0_soft_reset(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; u32 srbm_soft_reset; - if (!adev->uvd.srbm_soft_reset) + if (!adev->uvd.inst[ring->me].srbm_soft_reset) return 0; - srbm_soft_reset = adev->uvd.srbm_soft_reset; + srbm_soft_reset = adev->uvd.inst[ring->me].srbm_soft_reset; if (srbm_soft_reset) { u32 tmp; @@ -1419,7 +1460,7 @@ static int uvd_v7_0_post_soft_reset(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (!adev->uvd.srbm_soft_reset) + if (!adev->uvd.inst[ring->me].srbm_soft_reset) return 0; mdelay(5); @@ -1441,17 +1482,32 @@ static int uvd_v7_0_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { + uint32_t ip_instance; + + switch (entry->client_id) { + case SOC15_IH_CLIENTID_UVD: + ip_instance = 0; + break; + case SOC15_IH_CLIENTID_UVD1: + ip_instance = 1; + break; + default: + DRM_ERROR("Unhandled client id: %d\n", entry->client_id); + return 0; + } + DRM_DEBUG("IH: UVD TRAP\n"); + switch (entry->src_id) { case 124: - amdgpu_fence_process(&adev->uvd.ring); + amdgpu_fence_process(&adev->uvd.inst[ip_instance].ring); break; case 119: - amdgpu_fence_process(&adev->uvd.ring_enc[0]); + amdgpu_fence_process(&adev->uvd.inst[ip_instance].ring_enc[0]); break; case 120: if (!amdgpu_sriov_vf(adev)) - amdgpu_fence_process(&adev->uvd.ring_enc[1]); + amdgpu_fence_process(&adev->uvd.inst[ip_instance].ring_enc[1]); break; default: DRM_ERROR("Unhandled interrupt: %d %d\n", @@ -1467,9 +1523,9 @@ static void uvd_v7_0_set_sw_clock_gating(struct amdgpu_device *adev) { uint32_t data, data1, data2, suvd_flags; - data = RREG32_SOC15(UVD, 0, mmUVD_CGC_CTRL); - data1 = RREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_GATE); - data2 = RREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_CTRL); + data = RREG32_SOC15(UVD, ring->me, mmUVD_CGC_CTRL); + data1 = RREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_GATE); + data2 = RREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_CTRL); data &= ~(UVD_CGC_CTRL__CLK_OFF_DELAY_MASK | UVD_CGC_CTRL__CLK_GATE_DLY_TIMER_MASK); @@ -1513,18 +1569,18 @@ static void uvd_v7_0_set_sw_clock_gating(struct amdgpu_device *adev) UVD_SUVD_CGC_CTRL__SDB_MODE_MASK); data1 |= suvd_flags; - WREG32_SOC15(UVD, 0, mmUVD_CGC_CTRL, data); - WREG32_SOC15(UVD, 0, mmUVD_CGC_GATE, 0); - WREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_GATE, data1); - WREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_CTRL, data2); + WREG32_SOC15(UVD, ring->me, mmUVD_CGC_CTRL, data); + WREG32_SOC15(UVD, ring->me, mmUVD_CGC_GATE, 0); + WREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_GATE, data1); + WREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_CTRL, data2); } static void uvd_v7_0_set_hw_clock_gating(struct amdgpu_device *adev) { uint32_t data, data1, cgc_flags, suvd_flags; - data = RREG32_SOC15(UVD, 0, mmUVD_CGC_GATE); - data1 = RREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_GATE); + data = RREG32_SOC15(UVD, ring->me, mmUVD_CGC_GATE); + data1 = RREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_GATE); cgc_flags = UVD_CGC_GATE__SYS_MASK | UVD_CGC_GATE__UDEC_MASK | @@ -1556,8 +1612,8 @@ static void uvd_v7_0_set_hw_clock_gating(struct amdgpu_device *adev) data |= cgc_flags; data1 |= suvd_flags; - WREG32_SOC15(UVD, 0, mmUVD_CGC_GATE, data); - WREG32_SOC15(UVD, 0, mmUVD_SUVD_CGC_GATE, data1); + WREG32_SOC15(UVD, ring->me, mmUVD_CGC_GATE, data); + WREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_GATE, data1); } static void uvd_v7_0_set_bypass_mode(struct amdgpu_device *adev, bool enable) @@ -1616,7 +1672,7 @@ static int uvd_v7_0_set_powergating_state(void *handle, if (!(adev->pg_flags & AMD_PG_SUPPORT_UVD)) return 0; - WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, UVD_POWER_STATUS__UVD_PG_EN_MASK); + WREG32_SOC15(UVD, ring->me, mmUVD_POWER_STATUS, UVD_POWER_STATUS__UVD_PG_EN_MASK); if (state == AMD_PG_STATE_GATE) { uvd_v7_0_stop(adev); @@ -1657,7 +1713,6 @@ const struct amd_ip_funcs uvd_v7_0_ip_funcs = { static const struct amdgpu_ring_funcs uvd_v7_0_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_UVD, .align_mask = 0xf, - .nop = PACKET0(0x81ff, 0), .support_64bit_ptrs = false, .vmhub = AMDGPU_MMHUB, .get_rptr = uvd_v7_0_ring_get_rptr, @@ -1719,18 +1774,27 @@ static const struct amdgpu_ring_funcs uvd_v7_0_enc_ring_vm_funcs = { static void uvd_v7_0_set_ring_funcs(struct amdgpu_device *adev) { - adev->uvd.ring.funcs = &uvd_v7_0_ring_vm_funcs; - DRM_INFO("UVD is enabled in VM mode\n"); + int i; + + for (i = 0; i < adev->uvd.num_uvd_inst; i++) { + adev->uvd.inst[i].ring.funcs = &uvd_v7_0_ring_vm_funcs; + adev->uvd.inst[i].ring.me = i; + DRM_INFO("UVD(%d) is enabled in VM mode\n", i); + } } static void uvd_v7_0_set_enc_ring_funcs(struct amdgpu_device *adev) { - int i; + int i, j; - for (i = 0; i < adev->uvd.num_enc_rings; ++i) - adev->uvd.ring_enc[i].funcs = &uvd_v7_0_enc_ring_vm_funcs; + for (j = 0; j < adev->uvd.num_uvd_inst; j++) { + for (i = 0; i < adev->uvd.num_enc_rings; ++i) { + adev->uvd.inst[j].ring_enc[i].funcs = &uvd_v7_0_enc_ring_vm_funcs; + adev->uvd.inst[j].ring_enc[i].me = j; + } - DRM_INFO("UVD ENC is enabled in VM mode\n"); + DRM_INFO("UVD(%d) ENC is enabled in VM mode\n", j); + } } static const struct amdgpu_irq_src_funcs uvd_v7_0_irq_funcs = { @@ -1740,8 +1804,12 @@ static const struct amdgpu_irq_src_funcs uvd_v7_0_irq_funcs = { static void uvd_v7_0_set_irq_funcs(struct amdgpu_device *adev) { - adev->uvd.irq.num_types = adev->uvd.num_enc_rings + 1; - adev->uvd.irq.funcs = &uvd_v7_0_irq_funcs; + int i; + + for (i = 0; i < adev->uvd.num_uvd_inst; i++) { + adev->uvd.inst[i].irq.num_types = adev->uvd.num_enc_rings + 1; + adev->uvd.inst[i].irq.funcs = &uvd_v7_0_irq_funcs; + } } const struct amdgpu_ip_block_version uvd_v7_0_ip_block = diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c index 0501746b6c2c..110b294ebed3 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c @@ -35,7 +35,6 @@ #include "mmhub/mmhub_9_1_offset.h" #include "mmhub/mmhub_9_1_sh_mask.h" -static int vcn_v1_0_start(struct amdgpu_device *adev); static int vcn_v1_0_stop(struct amdgpu_device *adev); static void vcn_v1_0_set_dec_ring_funcs(struct amdgpu_device *adev); static void vcn_v1_0_set_enc_ring_funcs(struct amdgpu_device *adev); @@ -146,10 +145,6 @@ static int vcn_v1_0_hw_init(void *handle) struct amdgpu_ring *ring = &adev->vcn.ring_dec; int i, r; - r = vcn_v1_0_start(adev); - if (r) - goto done; - ring->ready = true; r = amdgpu_ring_test_ring(ring); if (r) { @@ -185,11 +180,9 @@ static int vcn_v1_0_hw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; struct amdgpu_ring *ring = &adev->vcn.ring_dec; - int r; - r = vcn_v1_0_stop(adev); - if (r) - return r; + if (RREG32_SOC15(VCN, 0, mmUVD_STATUS)) + vcn_v1_0_stop(adev); ring->ready = false; @@ -288,14 +281,14 @@ static void vcn_v1_0_mc_resume(struct amdgpu_device *adev) * * Disable clock gating for VCN block */ -static void vcn_v1_0_disable_clock_gating(struct amdgpu_device *adev, bool sw) +static void vcn_v1_0_disable_clock_gating(struct amdgpu_device *adev) { uint32_t data; /* JPEG disable CGC */ data = RREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL); - if (sw) + if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG) data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT; else data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE_MASK; @@ -310,7 +303,7 @@ static void vcn_v1_0_disable_clock_gating(struct amdgpu_device *adev, bool sw) /* UVD disable CGC */ data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL); - if (sw) + if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG) data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT; else data &= ~ UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK; @@ -415,13 +408,13 @@ static void vcn_v1_0_disable_clock_gating(struct amdgpu_device *adev, bool sw) * * Enable clock gating for VCN block */ -static void vcn_v1_0_enable_clock_gating(struct amdgpu_device *adev, bool sw) +static void vcn_v1_0_enable_clock_gating(struct amdgpu_device *adev) { uint32_t data = 0; /* enable JPEG CGC */ data = RREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL); - if (sw) + if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG) data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT; else data |= 0 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT; @@ -435,7 +428,7 @@ static void vcn_v1_0_enable_clock_gating(struct amdgpu_device *adev, bool sw) /* enable UVD CGC */ data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL); - if (sw) + if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG) data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT; else data |= 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT; @@ -480,6 +473,94 @@ static void vcn_v1_0_enable_clock_gating(struct amdgpu_device *adev, bool sw) WREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL, data); } +static void vcn_1_0_disable_static_power_gating(struct amdgpu_device *adev) +{ + uint32_t data = 0; + int ret; + + if (adev->pg_flags & AMD_PG_SUPPORT_VCN) { + data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDW_PWR_CONFIG__SHIFT); + + WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data); + SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON, 0xFFFFFF, ret); + } else { + data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT + | 1 << UVD_PGFSM_CONFIG__UVDW_PWR_CONFIG__SHIFT); + WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data); + SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, 0, 0xFFFFFFFF, ret); + } + + /* polling UVD_PGFSM_STATUS to confirm UVDM_PWR_STATUS , UVDU_PWR_STATUS are 0 (power on) */ + + data = RREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS); + data &= ~0x103; + if (adev->pg_flags & AMD_PG_SUPPORT_VCN) + data |= UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON | UVD_POWER_STATUS__UVD_PG_EN_MASK; + + WREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS, data); +} + +static void vcn_1_0_enable_static_power_gating(struct amdgpu_device *adev) +{ + uint32_t data = 0; + int ret; + + if (adev->pg_flags & AMD_PG_SUPPORT_VCN) { + /* Before power off, this indicator has to be turned on */ + data = RREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS); + data &= ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK; + data |= UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF; + WREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS, data); + + + data = (2 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT + | 2 << UVD_PGFSM_CONFIG__UVDW_PWR_CONFIG__SHIFT); + + WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data); + + data = (2 << UVD_PGFSM_STATUS__UVDM_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDU_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDF_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDC_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDB_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDIL_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDIR_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDTD_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDTE_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDE_PWR_STATUS__SHIFT + | 2 << UVD_PGFSM_STATUS__UVDW_PWR_STATUS__SHIFT); + SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, data, 0xFFFFFFFF, ret); + } +} + /** * vcn_v1_0_start - start VCN block * @@ -499,8 +580,9 @@ static int vcn_v1_0_start(struct amdgpu_device *adev) vcn_v1_0_mc_resume(adev); + vcn_1_0_disable_static_power_gating(adev); /* disable clock gating */ - vcn_v1_0_disable_clock_gating(adev, true); + vcn_v1_0_disable_clock_gating(adev); /* disable interupt */ WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), 0, @@ -680,16 +762,45 @@ static int vcn_v1_0_stop(struct amdgpu_device *adev) WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0, ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); - /* enable clock gating */ - vcn_v1_0_enable_clock_gating(adev, true); + WREG32_SOC15(VCN, 0, mmUVD_STATUS, 0); + vcn_v1_0_enable_clock_gating(adev); + vcn_1_0_enable_static_power_gating(adev); return 0; } +bool vcn_v1_0_is_idle(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + return (RREG32_SOC15(VCN, 0, mmUVD_STATUS) == 0x2); +} + +int vcn_v1_0_wait_for_idle(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + int ret = 0; + + SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_STATUS, 0x2, 0x2, ret); + + return ret; +} + static int vcn_v1_0_set_clockgating_state(void *handle, enum amd_clockgating_state state) { - /* needed for driver unload*/ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + bool enable = (state == AMD_CG_STATE_GATE) ? true : false; + + if (enable) { + /* wait for STATUS to clear */ + if (vcn_v1_0_is_idle(handle)) + return -EBUSY; + vcn_v1_0_enable_clock_gating(adev); + } else { + /* disable HW gating and enable Sw gating */ + vcn_v1_0_disable_clock_gating(adev); + } return 0; } @@ -1048,16 +1159,36 @@ static int vcn_v1_0_process_interrupt(struct amdgpu_device *adev, return 0; } -static void vcn_v1_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) +static void vcn_v1_0_dec_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) { - int i; struct amdgpu_device *adev = ring->adev; + int i; - for (i = 0; i < count; i++) - amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0)); + WARN_ON(ring->wptr % 2 || count % 2); + for (i = 0; i < count / 2; i++) { + amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0)); + amdgpu_ring_write(ring, 0); + } } +static int vcn_v1_0_set_powergating_state(void *handle, + enum amd_powergating_state state) +{ + /* This doesn't actually powergate the VCN block. + * That's done in the dpm code via the SMC. This + * just re-inits the block as necessary. The actual + * gating still happens in the dpm code. We should + * revisit this when there is a cleaner line between + * the smc and the hw blocks + */ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (state == AMD_PG_STATE_GATE) + return vcn_v1_0_stop(adev); + else + return vcn_v1_0_start(adev); +} static const struct amd_ip_funcs vcn_v1_0_ip_funcs = { .name = "vcn_v1_0", @@ -1069,20 +1200,19 @@ static const struct amd_ip_funcs vcn_v1_0_ip_funcs = { .hw_fini = vcn_v1_0_hw_fini, .suspend = vcn_v1_0_suspend, .resume = vcn_v1_0_resume, - .is_idle = NULL /* vcn_v1_0_is_idle */, - .wait_for_idle = NULL /* vcn_v1_0_wait_for_idle */, + .is_idle = vcn_v1_0_is_idle, + .wait_for_idle = vcn_v1_0_wait_for_idle, .check_soft_reset = NULL /* vcn_v1_0_check_soft_reset */, .pre_soft_reset = NULL /* vcn_v1_0_pre_soft_reset */, .soft_reset = NULL /* vcn_v1_0_soft_reset */, .post_soft_reset = NULL /* vcn_v1_0_post_soft_reset */, .set_clockgating_state = vcn_v1_0_set_clockgating_state, - .set_powergating_state = NULL /* vcn_v1_0_set_powergating_state */, + .set_powergating_state = vcn_v1_0_set_powergating_state, }; static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_VCN_DEC, .align_mask = 0xf, - .nop = PACKET0(0x81ff, 0), .support_64bit_ptrs = false, .vmhub = AMDGPU_MMHUB, .get_rptr = vcn_v1_0_dec_ring_get_rptr, @@ -1101,7 +1231,7 @@ static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = { .emit_vm_flush = vcn_v1_0_dec_ring_emit_vm_flush, .test_ring = amdgpu_vcn_dec_ring_test_ring, .test_ib = amdgpu_vcn_dec_ring_test_ib, - .insert_nop = vcn_v1_0_ring_insert_nop, + .insert_nop = vcn_v1_0_dec_ring_insert_nop, .insert_start = vcn_v1_0_dec_ring_insert_start, .insert_end = vcn_v1_0_dec_ring_insert_end, .pad_ib = amdgpu_ring_generic_pad_ib, diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c b/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c new file mode 100644 index 000000000000..52778de93ab0 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c @@ -0,0 +1,53 @@ +/* + * Copyright 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "soc15.h" + +#include "soc15_common.h" +#include "soc15_hw_ip.h" +#include "vega20_ip_offset.h" + +int vega20_reg_base_init(struct amdgpu_device *adev) +{ + /* HW has more IP blocks, only initialized the blocke beend by our driver */ + uint32_t i; + for (i = 0 ; i < MAX_INSTANCE ; ++i) { + adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i])); + adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i])); + adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i])); + adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i])); + adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i])); + adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i])); + adev->reg_offset[UVD_HWIP][i] = (uint32_t *)(&(UVD_BASE.instance[i])); + adev->reg_offset[VCE_HWIP][i] = (uint32_t *)(&(VCE_BASE.instance[i])); + adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i])); + adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DCE_BASE.instance[i])); + adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i])); + adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(SDMA0_BASE.instance[i])); + adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(SDMA1_BASE.instance[i])); + adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i])); + } + return 0; +} + + diff --git a/drivers/gpu/drm/amd/display/Kconfig b/drivers/gpu/drm/amd/display/Kconfig index e6ca72c0d347..d5d4586e6176 100644 --- a/drivers/gpu/drm/amd/display/Kconfig +++ b/drivers/gpu/drm/amd/display/Kconfig @@ -34,10 +34,4 @@ config DEBUG_KERNEL_DC if you want to hit kdgb_break in assert. -config DRM_AMD_DC_VEGAM - bool "VEGAM support" - depends on DRM_AMD_DC - help - Choose this option if you want to have - VEGAM support for display engine endmenu diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index f2f54a9df56f..1ce10bc2d37b 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -911,6 +911,7 @@ amdgpu_dm_update_connector_after_detect(struct amdgpu_dm_connector *aconnector) drm_mode_connector_update_edid_property(connector, NULL); aconnector->num_modes = 0; aconnector->dc_sink = NULL; + aconnector->edid = NULL; } mutex_unlock(&dev->mode_config.mutex); @@ -1115,6 +1116,7 @@ static int dce110_register_irq_handlers(struct amdgpu_device *adev) if (adev->asic_type == CHIP_VEGA10 || adev->asic_type == CHIP_VEGA12 || + adev->asic_type == CHIP_VEGA20 || adev->asic_type == CHIP_RAVEN) client_id = SOC15_IH_CLIENTID_DCE; @@ -1513,11 +1515,10 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev) case CHIP_POLARIS11: case CHIP_POLARIS10: case CHIP_POLARIS12: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case CHIP_VEGAM: -#endif case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: if (dce110_register_irq_handlers(dm->adev)) { DRM_ERROR("DM: Failed to initialize IRQ\n"); goto fail; @@ -1708,9 +1709,7 @@ static int dm_early_init(void *handle) adev->mode_info.plane_type = dm_plane_type_default; break; case CHIP_POLARIS10: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case CHIP_VEGAM: -#endif adev->mode_info.num_crtc = 6; adev->mode_info.num_hpd = 6; adev->mode_info.num_dig = 6; @@ -1718,6 +1717,7 @@ static int dm_early_init(void *handle) break; case CHIP_VEGA10: case CHIP_VEGA12: + case CHIP_VEGA20: adev->mode_info.num_crtc = 6; adev->mode_info.num_hpd = 6; adev->mode_info.num_dig = 6; @@ -1966,6 +1966,7 @@ static int fill_plane_attributes_from_fb(struct amdgpu_device *adev, if (adev->asic_type == CHIP_VEGA10 || adev->asic_type == CHIP_VEGA12 || + adev->asic_type == CHIP_VEGA20 || adev->asic_type == CHIP_RAVEN) { /* Fill GFX9 params */ plane_state->tiling_info.gfx9.num_pipes = diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c index e3d90e918d1b..b329393307e5 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c @@ -88,9 +88,9 @@ static void __drm_lut_to_dc_gamma(struct drm_color_lut *lut, g = drm_color_lut_extract(lut[i].green, 16); b = drm_color_lut_extract(lut[i].blue, 16); - gamma->entries.red[i] = dal_fixed31_32_from_int(r); - gamma->entries.green[i] = dal_fixed31_32_from_int(g); - gamma->entries.blue[i] = dal_fixed31_32_from_int(b); + gamma->entries.red[i] = dc_fixpt_from_int(r); + gamma->entries.green[i] = dc_fixpt_from_int(g); + gamma->entries.blue[i] = dc_fixpt_from_int(b); } return; } @@ -101,9 +101,9 @@ static void __drm_lut_to_dc_gamma(struct drm_color_lut *lut, g = drm_color_lut_extract(lut[i].green, 16); b = drm_color_lut_extract(lut[i].blue, 16); - gamma->entries.red[i] = dal_fixed31_32_from_fraction(r, MAX_DRM_LUT_VALUE); - gamma->entries.green[i] = dal_fixed31_32_from_fraction(g, MAX_DRM_LUT_VALUE); - gamma->entries.blue[i] = dal_fixed31_32_from_fraction(b, MAX_DRM_LUT_VALUE); + gamma->entries.red[i] = dc_fixpt_from_fraction(r, MAX_DRM_LUT_VALUE); + gamma->entries.green[i] = dc_fixpt_from_fraction(g, MAX_DRM_LUT_VALUE); + gamma->entries.blue[i] = dc_fixpt_from_fraction(b, MAX_DRM_LUT_VALUE); } } @@ -208,7 +208,7 @@ void amdgpu_dm_set_ctm(struct dm_crtc_state *crtc) for (i = 0; i < 12; i++) { /* Skip 4th element */ if (i % 4 == 3) { - stream->gamut_remap_matrix.matrix[i] = dal_fixed31_32_zero; + stream->gamut_remap_matrix.matrix[i] = dc_fixpt_zero; continue; } diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index ca0b08bfa2cf..bd449351803f 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -330,11 +330,6 @@ bool dm_helpers_dp_mst_send_payload_allocation( return true; } -bool dm_helpers_dc_conn_log(struct dc_context *ctx, struct log_entry *entry, enum dc_log_type event) -{ - return true; -} - void dm_dtn_log_begin(struct dc_context *ctx) {} diff --git a/drivers/gpu/drm/amd/display/dc/basics/Makefile b/drivers/gpu/drm/amd/display/dc/basics/Makefile index bca33bd9a0d2..b49ea96b5dae 100644 --- a/drivers/gpu/drm/amd/display/dc/basics/Makefile +++ b/drivers/gpu/drm/amd/display/dc/basics/Makefile @@ -24,7 +24,7 @@ # It provides the general basic services required by other DAL # subcomponents. -BASICS = conversion.o fixpt31_32.o fixpt32_32.o \ +BASICS = conversion.o fixpt31_32.o \ logger.o log_helpers.o vector.o AMD_DAL_BASICS = $(addprefix $(AMDDALPATH)/dc/basics/,$(BASICS)) diff --git a/drivers/gpu/drm/amd/display/dc/basics/conversion.c b/drivers/gpu/drm/amd/display/dc/basics/conversion.c index 310964915a83..50b47f11875c 100644 --- a/drivers/gpu/drm/amd/display/dc/basics/conversion.c +++ b/drivers/gpu/drm/amd/display/dc/basics/conversion.c @@ -41,22 +41,22 @@ uint16_t fixed_point_to_int_frac( uint16_t result; - uint16_t d = (uint16_t)dal_fixed31_32_floor( - dal_fixed31_32_abs( + uint16_t d = (uint16_t)dc_fixpt_floor( + dc_fixpt_abs( arg)); if (d <= (uint16_t)(1 << integer_bits) - (1 / (uint16_t)divisor)) - numerator = (uint16_t)dal_fixed31_32_round( - dal_fixed31_32_mul_int( + numerator = (uint16_t)dc_fixpt_round( + dc_fixpt_mul_int( arg, divisor)); else { - numerator = dal_fixed31_32_floor( - dal_fixed31_32_sub( - dal_fixed31_32_from_int( + numerator = dc_fixpt_floor( + dc_fixpt_sub( + dc_fixpt_from_int( 1LL << integer_bits), - dal_fixed31_32_recip( - dal_fixed31_32_from_int( + dc_fixpt_recip( + dc_fixpt_from_int( divisor)))); } @@ -66,8 +66,8 @@ uint16_t fixed_point_to_int_frac( result = (uint16_t)( (1 << (integer_bits + fractional_bits + 1)) + numerator); - if ((result != 0) && dal_fixed31_32_lt( - arg, dal_fixed31_32_zero)) + if ((result != 0) && dc_fixpt_lt( + arg, dc_fixpt_zero)) result |= 1 << (integer_bits + fractional_bits); return result; @@ -84,15 +84,15 @@ void convert_float_matrix( uint32_t buffer_size) { const struct fixed31_32 min_2_13 = - dal_fixed31_32_from_fraction(S2D13_MIN, DIVIDER); + dc_fixpt_from_fraction(S2D13_MIN, DIVIDER); const struct fixed31_32 max_2_13 = - dal_fixed31_32_from_fraction(S2D13_MAX, DIVIDER); + dc_fixpt_from_fraction(S2D13_MAX, DIVIDER); uint32_t i; for (i = 0; i < buffer_size; ++i) { uint32_t reg_value = fixed_point_to_int_frac( - dal_fixed31_32_clamp( + dc_fixpt_clamp( flt[i], min_2_13, max_2_13), diff --git a/drivers/gpu/drm/amd/display/dc/basics/fixpt31_32.c b/drivers/gpu/drm/amd/display/dc/basics/fixpt31_32.c index 7191c3213743..e61dd97d0928 100644 --- a/drivers/gpu/drm/amd/display/dc/basics/fixpt31_32.c +++ b/drivers/gpu/drm/amd/display/dc/basics/fixpt31_32.c @@ -64,9 +64,7 @@ static inline unsigned long long complete_integer_division_u64( #define GET_FRACTIONAL_PART(x) \ (FRACTIONAL_PART_MASK & (x)) -struct fixed31_32 dal_fixed31_32_from_fraction( - long long numerator, - long long denominator) +struct fixed31_32 dc_fixpt_from_fraction(long long numerator, long long denominator) { struct fixed31_32 res; @@ -118,63 +116,7 @@ struct fixed31_32 dal_fixed31_32_from_fraction( return res; } -struct fixed31_32 dal_fixed31_32_from_int_nonconst( - long long arg) -{ - struct fixed31_32 res; - - ASSERT((LONG_MIN <= arg) && (arg <= LONG_MAX)); - - res.value = arg << FIXED31_32_BITS_PER_FRACTIONAL_PART; - - return res; -} - -struct fixed31_32 dal_fixed31_32_shl( - struct fixed31_32 arg, - unsigned char shift) -{ - struct fixed31_32 res; - - ASSERT(((arg.value >= 0) && (arg.value <= LLONG_MAX >> shift)) || - ((arg.value < 0) && (arg.value >= LLONG_MIN >> shift))); - - res.value = arg.value << shift; - - return res; -} - -struct fixed31_32 dal_fixed31_32_add( - struct fixed31_32 arg1, - struct fixed31_32 arg2) -{ - struct fixed31_32 res; - - ASSERT(((arg1.value >= 0) && (LLONG_MAX - arg1.value >= arg2.value)) || - ((arg1.value < 0) && (LLONG_MIN - arg1.value <= arg2.value))); - - res.value = arg1.value + arg2.value; - - return res; -} - -struct fixed31_32 dal_fixed31_32_sub( - struct fixed31_32 arg1, - struct fixed31_32 arg2) -{ - struct fixed31_32 res; - - ASSERT(((arg2.value >= 0) && (LLONG_MIN + arg2.value <= arg1.value)) || - ((arg2.value < 0) && (LLONG_MAX + arg2.value >= arg1.value))); - - res.value = arg1.value - arg2.value; - - return res; -} - -struct fixed31_32 dal_fixed31_32_mul( - struct fixed31_32 arg1, - struct fixed31_32 arg2) +struct fixed31_32 dc_fixpt_mul(struct fixed31_32 arg1, struct fixed31_32 arg2) { struct fixed31_32 res; @@ -213,7 +155,7 @@ struct fixed31_32 dal_fixed31_32_mul( tmp = arg1_fra * arg2_fra; tmp = (tmp >> FIXED31_32_BITS_PER_FRACTIONAL_PART) + - (tmp >= (unsigned long long)dal_fixed31_32_half.value); + (tmp >= (unsigned long long)dc_fixpt_half.value); ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); @@ -225,8 +167,7 @@ struct fixed31_32 dal_fixed31_32_mul( return res; } -struct fixed31_32 dal_fixed31_32_sqr( - struct fixed31_32 arg) +struct fixed31_32 dc_fixpt_sqr(struct fixed31_32 arg) { struct fixed31_32 res; @@ -257,7 +198,7 @@ struct fixed31_32 dal_fixed31_32_sqr( tmp = arg_fra * arg_fra; tmp = (tmp >> FIXED31_32_BITS_PER_FRACTIONAL_PART) + - (tmp >= (unsigned long long)dal_fixed31_32_half.value); + (tmp >= (unsigned long long)dc_fixpt_half.value); ASSERT(tmp <= (unsigned long long)(LLONG_MAX - res.value)); @@ -266,8 +207,7 @@ struct fixed31_32 dal_fixed31_32_sqr( return res; } -struct fixed31_32 dal_fixed31_32_recip( - struct fixed31_32 arg) +struct fixed31_32 dc_fixpt_recip(struct fixed31_32 arg) { /* * @note @@ -276,41 +216,40 @@ struct fixed31_32 dal_fixed31_32_recip( ASSERT(arg.value); - return dal_fixed31_32_from_fraction( - dal_fixed31_32_one.value, + return dc_fixpt_from_fraction( + dc_fixpt_one.value, arg.value); } -struct fixed31_32 dal_fixed31_32_sinc( - struct fixed31_32 arg) +struct fixed31_32 dc_fixpt_sinc(struct fixed31_32 arg) { struct fixed31_32 square; - struct fixed31_32 res = dal_fixed31_32_one; + struct fixed31_32 res = dc_fixpt_one; int n = 27; struct fixed31_32 arg_norm = arg; - if (dal_fixed31_32_le( - dal_fixed31_32_two_pi, - dal_fixed31_32_abs(arg))) { - arg_norm = dal_fixed31_32_sub( + if (dc_fixpt_le( + dc_fixpt_two_pi, + dc_fixpt_abs(arg))) { + arg_norm = dc_fixpt_sub( arg_norm, - dal_fixed31_32_mul_int( - dal_fixed31_32_two_pi, + dc_fixpt_mul_int( + dc_fixpt_two_pi, (int)div64_s64( arg_norm.value, - dal_fixed31_32_two_pi.value))); + dc_fixpt_two_pi.value))); } - square = dal_fixed31_32_sqr(arg_norm); + square = dc_fixpt_sqr(arg_norm); do { - res = dal_fixed31_32_sub( - dal_fixed31_32_one, - dal_fixed31_32_div_int( - dal_fixed31_32_mul( + res = dc_fixpt_sub( + dc_fixpt_one, + dc_fixpt_div_int( + dc_fixpt_mul( square, res), n * (n - 1))); @@ -319,37 +258,35 @@ struct fixed31_32 dal_fixed31_32_sinc( } while (n > 2); if (arg.value != arg_norm.value) - res = dal_fixed31_32_div( - dal_fixed31_32_mul(res, arg_norm), + res = dc_fixpt_div( + dc_fixpt_mul(res, arg_norm), arg); return res; } -struct fixed31_32 dal_fixed31_32_sin( - struct fixed31_32 arg) +struct fixed31_32 dc_fixpt_sin(struct fixed31_32 arg) { - return dal_fixed31_32_mul( + return dc_fixpt_mul( arg, - dal_fixed31_32_sinc(arg)); + dc_fixpt_sinc(arg)); } -struct fixed31_32 dal_fixed31_32_cos( - struct fixed31_32 arg) +struct fixed31_32 dc_fixpt_cos(struct fixed31_32 arg) { /* TODO implement argument normalization */ - const struct fixed31_32 square = dal_fixed31_32_sqr(arg); + const struct fixed31_32 square = dc_fixpt_sqr(arg); - struct fixed31_32 res = dal_fixed31_32_one; + struct fixed31_32 res = dc_fixpt_one; int n = 26; do { - res = dal_fixed31_32_sub( - dal_fixed31_32_one, - dal_fixed31_32_div_int( - dal_fixed31_32_mul( + res = dc_fixpt_sub( + dc_fixpt_one, + dc_fixpt_div_int( + dc_fixpt_mul( square, res), n * (n - 1))); @@ -367,37 +304,35 @@ struct fixed31_32 dal_fixed31_32_cos( * * Calculated as Taylor series. */ -static struct fixed31_32 fixed31_32_exp_from_taylor_series( - struct fixed31_32 arg) +static struct fixed31_32 fixed31_32_exp_from_taylor_series(struct fixed31_32 arg) { unsigned int n = 9; - struct fixed31_32 res = dal_fixed31_32_from_fraction( + struct fixed31_32 res = dc_fixpt_from_fraction( n + 2, n + 1); /* TODO find correct res */ - ASSERT(dal_fixed31_32_lt(arg, dal_fixed31_32_one)); + ASSERT(dc_fixpt_lt(arg, dc_fixpt_one)); do - res = dal_fixed31_32_add( - dal_fixed31_32_one, - dal_fixed31_32_div_int( - dal_fixed31_32_mul( + res = dc_fixpt_add( + dc_fixpt_one, + dc_fixpt_div_int( + dc_fixpt_mul( arg, res), n)); while (--n != 1); - return dal_fixed31_32_add( - dal_fixed31_32_one, - dal_fixed31_32_mul( + return dc_fixpt_add( + dc_fixpt_one, + dc_fixpt_mul( arg, res)); } -struct fixed31_32 dal_fixed31_32_exp( - struct fixed31_32 arg) +struct fixed31_32 dc_fixpt_exp(struct fixed31_32 arg) { /* * @brief @@ -406,44 +341,43 @@ struct fixed31_32 dal_fixed31_32_exp( * where m = round(x / ln(2)), r = x - m * ln(2) */ - if (dal_fixed31_32_le( - dal_fixed31_32_ln2_div_2, - dal_fixed31_32_abs(arg))) { - int m = dal_fixed31_32_round( - dal_fixed31_32_div( + if (dc_fixpt_le( + dc_fixpt_ln2_div_2, + dc_fixpt_abs(arg))) { + int m = dc_fixpt_round( + dc_fixpt_div( arg, - dal_fixed31_32_ln2)); + dc_fixpt_ln2)); - struct fixed31_32 r = dal_fixed31_32_sub( + struct fixed31_32 r = dc_fixpt_sub( arg, - dal_fixed31_32_mul_int( - dal_fixed31_32_ln2, + dc_fixpt_mul_int( + dc_fixpt_ln2, m)); ASSERT(m != 0); - ASSERT(dal_fixed31_32_lt( - dal_fixed31_32_abs(r), - dal_fixed31_32_one)); + ASSERT(dc_fixpt_lt( + dc_fixpt_abs(r), + dc_fixpt_one)); if (m > 0) - return dal_fixed31_32_shl( + return dc_fixpt_shl( fixed31_32_exp_from_taylor_series(r), (unsigned char)m); else - return dal_fixed31_32_div_int( + return dc_fixpt_div_int( fixed31_32_exp_from_taylor_series(r), 1LL << -m); } else if (arg.value != 0) return fixed31_32_exp_from_taylor_series(arg); else - return dal_fixed31_32_one; + return dc_fixpt_one; } -struct fixed31_32 dal_fixed31_32_log( - struct fixed31_32 arg) +struct fixed31_32 dc_fixpt_log(struct fixed31_32 arg) { - struct fixed31_32 res = dal_fixed31_32_neg(dal_fixed31_32_one); + struct fixed31_32 res = dc_fixpt_neg(dc_fixpt_one); /* TODO improve 1st estimation */ struct fixed31_32 error; @@ -453,15 +387,15 @@ struct fixed31_32 dal_fixed31_32_log( /* TODO if arg is zero, return -INF */ do { - struct fixed31_32 res1 = dal_fixed31_32_add( - dal_fixed31_32_sub( + struct fixed31_32 res1 = dc_fixpt_add( + dc_fixpt_sub( res, - dal_fixed31_32_one), - dal_fixed31_32_div( + dc_fixpt_one), + dc_fixpt_div( arg, - dal_fixed31_32_exp(res))); + dc_fixpt_exp(res))); - error = dal_fixed31_32_sub( + error = dc_fixpt_sub( res, res1); @@ -472,66 +406,11 @@ struct fixed31_32 dal_fixed31_32_log( return res; } -struct fixed31_32 dal_fixed31_32_pow( - struct fixed31_32 arg1, - struct fixed31_32 arg2) -{ - return dal_fixed31_32_exp( - dal_fixed31_32_mul( - dal_fixed31_32_log(arg1), - arg2)); -} - -int dal_fixed31_32_floor( - struct fixed31_32 arg) -{ - unsigned long long arg_value = abs_i64(arg.value); - - if (arg.value >= 0) - return (int)GET_INTEGER_PART(arg_value); - else - return -(int)GET_INTEGER_PART(arg_value); -} - -int dal_fixed31_32_round( - struct fixed31_32 arg) -{ - unsigned long long arg_value = abs_i64(arg.value); - - const long long summand = dal_fixed31_32_half.value; - - ASSERT(LLONG_MAX - (long long)arg_value >= summand); - - arg_value += summand; - - if (arg.value >= 0) - return (int)GET_INTEGER_PART(arg_value); - else - return -(int)GET_INTEGER_PART(arg_value); -} - -int dal_fixed31_32_ceil( - struct fixed31_32 arg) -{ - unsigned long long arg_value = abs_i64(arg.value); - - const long long summand = dal_fixed31_32_one.value - - dal_fixed31_32_epsilon.value; - - ASSERT(LLONG_MAX - (long long)arg_value >= summand); - - arg_value += summand; - - if (arg.value >= 0) - return (int)GET_INTEGER_PART(arg_value); - else - return -(int)GET_INTEGER_PART(arg_value); -} /* this function is a generic helper to translate fixed point value to * specified integer format that will consist of integer_bits integer part and * fractional_bits fractional part. For example it is used in - * dal_fixed31_32_u2d19 to receive 2 bits integer part and 19 bits fractional + * dc_fixpt_u2d19 to receive 2 bits integer part and 19 bits fractional * part in 32 bits. It is used in hw programming (scaler) */ @@ -570,35 +449,30 @@ static inline unsigned int clamp_ux_dy( return min_clamp; } -unsigned int dal_fixed31_32_u2d19( - struct fixed31_32 arg) +unsigned int dc_fixpt_u2d19(struct fixed31_32 arg) { return ux_dy(arg.value, 2, 19); } -unsigned int dal_fixed31_32_u0d19( - struct fixed31_32 arg) +unsigned int dc_fixpt_u0d19(struct fixed31_32 arg) { return ux_dy(arg.value, 0, 19); } -unsigned int dal_fixed31_32_clamp_u0d14( - struct fixed31_32 arg) +unsigned int dc_fixpt_clamp_u0d14(struct fixed31_32 arg) { return clamp_ux_dy(arg.value, 0, 14, 1); } -unsigned int dal_fixed31_32_clamp_u0d10( - struct fixed31_32 arg) +unsigned int dc_fixpt_clamp_u0d10(struct fixed31_32 arg) { return clamp_ux_dy(arg.value, 0, 10, 1); } -int dal_fixed31_32_s4d19( - struct fixed31_32 arg) +int dc_fixpt_s4d19(struct fixed31_32 arg) { if (arg.value < 0) - return -(int)ux_dy(dal_fixed31_32_abs(arg).value, 4, 19); + return -(int)ux_dy(dc_fixpt_abs(arg).value, 4, 19); else return ux_dy(arg.value, 4, 19); } diff --git a/drivers/gpu/drm/amd/display/dc/basics/fixpt32_32.c b/drivers/gpu/drm/amd/display/dc/basics/fixpt32_32.c deleted file mode 100644 index 4d3aaa82a07b..000000000000 --- a/drivers/gpu/drm/amd/display/dc/basics/fixpt32_32.c +++ /dev/null @@ -1,161 +0,0 @@ -/* - * Copyright 2012-15 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#include "dm_services.h" -#include "include/fixed32_32.h" - -static uint64_t u64_div(uint64_t n, uint64_t d) -{ - uint32_t i = 0; - uint64_t r; - uint64_t q = div64_u64_rem(n, d, &r); - - for (i = 0; i < 32; ++i) { - uint64_t sbit = q & (1ULL<<63); - - r <<= 1; - r |= sbit ? 1 : 0; - q <<= 1; - if (r >= d) { - r -= d; - q |= 1; - } - } - - if (2*r >= d) - q += 1; - return q; -} - -struct fixed32_32 dal_fixed32_32_from_fraction(uint32_t n, uint32_t d) -{ - struct fixed32_32 fx; - - fx.value = u64_div((uint64_t)n << 32, (uint64_t)d << 32); - return fx; -} - -struct fixed32_32 dal_fixed32_32_add( - struct fixed32_32 lhs, - struct fixed32_32 rhs) -{ - struct fixed32_32 fx = {lhs.value + rhs.value}; - - ASSERT(fx.value >= rhs.value); - return fx; -} - -struct fixed32_32 dal_fixed32_32_add_int(struct fixed32_32 lhs, uint32_t rhs) -{ - struct fixed32_32 fx = {lhs.value + ((uint64_t)rhs << 32)}; - - ASSERT(fx.value >= (uint64_t)rhs << 32); - return fx; - -} -struct fixed32_32 dal_fixed32_32_sub( - struct fixed32_32 lhs, - struct fixed32_32 rhs) -{ - struct fixed32_32 fx; - - ASSERT(lhs.value >= rhs.value); - fx.value = lhs.value - rhs.value; - return fx; -} - -struct fixed32_32 dal_fixed32_32_sub_int(struct fixed32_32 lhs, uint32_t rhs) -{ - struct fixed32_32 fx; - - ASSERT(lhs.value >= ((uint64_t)rhs<<32)); - fx.value = lhs.value - ((uint64_t)rhs<<32); - return fx; -} - -struct fixed32_32 dal_fixed32_32_mul( - struct fixed32_32 lhs, - struct fixed32_32 rhs) -{ - struct fixed32_32 fx; - uint64_t lhs_int = lhs.value>>32; - uint64_t lhs_frac = (uint32_t)lhs.value; - uint64_t rhs_int = rhs.value>>32; - uint64_t rhs_frac = (uint32_t)rhs.value; - uint64_t ahbh = lhs_int * rhs_int; - uint64_t ahbl = lhs_int * rhs_frac; - uint64_t albh = lhs_frac * rhs_int; - uint64_t albl = lhs_frac * rhs_frac; - - ASSERT((ahbh>>32) == 0); - - fx.value = (ahbh<<32) + ahbl + albh + (albl>>32); - return fx; - -} - -struct fixed32_32 dal_fixed32_32_mul_int(struct fixed32_32 lhs, uint32_t rhs) -{ - struct fixed32_32 fx; - uint64_t lhsi = (lhs.value>>32) * (uint64_t)rhs; - uint64_t lhsf; - - ASSERT((lhsi>>32) == 0); - lhsf = ((uint32_t)lhs.value) * (uint64_t)rhs; - ASSERT((lhsi<<32) + lhsf >= lhsf); - fx.value = (lhsi<<32) + lhsf; - return fx; -} - -struct fixed32_32 dal_fixed32_32_div( - struct fixed32_32 lhs, - struct fixed32_32 rhs) -{ - struct fixed32_32 fx; - - fx.value = u64_div(lhs.value, rhs.value); - return fx; -} - -struct fixed32_32 dal_fixed32_32_div_int(struct fixed32_32 lhs, uint32_t rhs) -{ - struct fixed32_32 fx; - - fx.value = u64_div(lhs.value, (uint64_t)rhs << 32); - return fx; -} - -uint32_t dal_fixed32_32_ceil(struct fixed32_32 v) -{ - ASSERT((uint32_t)v.value ? (v.value >> 32) + 1 >= 1 : true); - return (v.value>>32) + ((uint32_t)v.value ? 1 : 0); -} - -uint32_t dal_fixed32_32_round(struct fixed32_32 v) -{ - ASSERT(v.value + (1ULL<<31) >= (1ULL<<31)); - return (v.value + (1ULL<<31))>>32; -} - diff --git a/drivers/gpu/drm/amd/display/dc/basics/log_helpers.c b/drivers/gpu/drm/amd/display/dc/basics/log_helpers.c index 854678a0c54b..021451549ff7 100644 --- a/drivers/gpu/drm/amd/display/dc/basics/log_helpers.c +++ b/drivers/gpu/drm/amd/display/dc/basics/log_helpers.c @@ -94,7 +94,6 @@ void dc_conn_log(struct dc_context *ctx, dm_logger_append(&entry, "%2.2X ", hex_data[i]); dm_logger_append(&entry, "^\n"); - dm_helpers_dc_conn_log(ctx, &entry, event); fail: dm_logger_close(&entry); diff --git a/drivers/gpu/drm/amd/display/dc/basics/logger.c b/drivers/gpu/drm/amd/display/dc/basics/logger.c index 31bee054f43a..738a818d58d1 100644 --- a/drivers/gpu/drm/amd/display/dc/basics/logger.c +++ b/drivers/gpu/drm/amd/display/dc/basics/logger.c @@ -61,7 +61,7 @@ static const struct dc_log_type_info log_type_info_tbl[] = { {LOG_EVENT_UNDERFLOW, "Underflow"}, {LOG_IF_TRACE, "InterfaceTrace"}, {LOG_DTN, "DTN"}, - {LOG_PROFILING, "Profiling"} + {LOG_DISPLAYSTATS, "DisplayStats"} }; @@ -402,3 +402,4 @@ cleanup: entry->max_buf_bytes = 0; } } + diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c index 10a5807a7e8b..b8cef7af3c4a 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c @@ -1330,6 +1330,9 @@ static enum bp_result bios_parser_get_firmware_info( case 2: result = get_firmware_info_v3_2(bp, info); break; + case 3: + result = get_firmware_info_v3_2(bp, info); + break; default: break; } diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.c b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.c index be066c49b984..253bbb1eea60 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.c +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper.c @@ -51,9 +51,7 @@ bool dal_bios_parser_init_cmd_tbl_helper( return true; case DCE_VERSION_11_2: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case DCE_VERSION_11_22: -#endif *h = dal_cmd_tbl_helper_dce112_get_table(); return true; diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c index 9b9e06995805..bbbcef566c55 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c @@ -52,9 +52,7 @@ bool dal_bios_parser_init_cmd_tbl_helper2( return true; case DCE_VERSION_11_2: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case DCE_VERSION_11_22: -#endif *h = dal_cmd_tbl_helper_dce112_get_table2(); return true; #if defined(CONFIG_DRM_AMD_DC_DCN1_0) diff --git a/drivers/gpu/drm/amd/display/dc/calcs/custom_float.c b/drivers/gpu/drm/amd/display/dc/calcs/custom_float.c index 7243c37f569e..31d167bc548f 100644 --- a/drivers/gpu/drm/amd/display/dc/calcs/custom_float.c +++ b/drivers/gpu/drm/amd/display/dc/calcs/custom_float.c @@ -36,41 +36,41 @@ static bool build_custom_float( uint32_t exp_offset = (1 << (format->exponenta_bits - 1)) - 1; const struct fixed31_32 mantissa_constant_plus_max_fraction = - dal_fixed31_32_from_fraction( + dc_fixpt_from_fraction( (1LL << (format->mantissa_bits + 1)) - 1, 1LL << format->mantissa_bits); struct fixed31_32 mantiss; - if (dal_fixed31_32_eq( + if (dc_fixpt_eq( value, - dal_fixed31_32_zero)) { + dc_fixpt_zero)) { *negative = false; *mantissa = 0; *exponenta = 0; return true; } - if (dal_fixed31_32_lt( + if (dc_fixpt_lt( value, - dal_fixed31_32_zero)) { + dc_fixpt_zero)) { *negative = format->sign; - value = dal_fixed31_32_neg(value); + value = dc_fixpt_neg(value); } else { *negative = false; } - if (dal_fixed31_32_lt( + if (dc_fixpt_lt( value, - dal_fixed31_32_one)) { + dc_fixpt_one)) { uint32_t i = 1; do { - value = dal_fixed31_32_shl(value, 1); + value = dc_fixpt_shl(value, 1); ++i; - } while (dal_fixed31_32_lt( + } while (dc_fixpt_lt( value, - dal_fixed31_32_one)); + dc_fixpt_one)); --i; @@ -81,15 +81,15 @@ static bool build_custom_float( } *exponenta = exp_offset - i; - } else if (dal_fixed31_32_le( + } else if (dc_fixpt_le( mantissa_constant_plus_max_fraction, value)) { uint32_t i = 1; do { - value = dal_fixed31_32_shr(value, 1); + value = dc_fixpt_shr(value, 1); ++i; - } while (dal_fixed31_32_lt( + } while (dc_fixpt_lt( mantissa_constant_plus_max_fraction, value)); @@ -98,23 +98,23 @@ static bool build_custom_float( *exponenta = exp_offset; } - mantiss = dal_fixed31_32_sub( + mantiss = dc_fixpt_sub( value, - dal_fixed31_32_one); + dc_fixpt_one); - if (dal_fixed31_32_lt( + if (dc_fixpt_lt( mantiss, - dal_fixed31_32_zero) || - dal_fixed31_32_lt( - dal_fixed31_32_one, + dc_fixpt_zero) || + dc_fixpt_lt( + dc_fixpt_one, mantiss)) - mantiss = dal_fixed31_32_zero; + mantiss = dc_fixpt_zero; else - mantiss = dal_fixed31_32_shl( + mantiss = dc_fixpt_shl( mantiss, format->mantissa_bits); - *mantissa = dal_fixed31_32_floor(mantiss); + *mantissa = dc_fixpt_floor(mantiss); return true; } diff --git a/drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c b/drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c index 4ee3c26f7c13..2c4e8f0cb2dc 100644 --- a/drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/calcs/dce_calcs.c @@ -59,10 +59,8 @@ static enum bw_calcs_version bw_calcs_version_from_asic_id(struct hw_asic_id asi return BW_CALCS_VERSION_POLARIS10; if (ASIC_REV_IS_POLARIS11_M(asic_id.hw_internal_rev)) return BW_CALCS_VERSION_POLARIS11; -#if defined(CONFIG_DRM_AMD_DC_VEGAM) if (ASIC_REV_IS_VEGAM(asic_id.hw_internal_rev)) return BW_CALCS_VERSION_VEGAM; -#endif return BW_CALCS_VERSION_INVALID; case FAMILY_AI: @@ -2151,11 +2149,9 @@ void bw_calcs_init(struct bw_calcs_dceip *bw_dceip, dceip.mcifwr_all_surfaces_burst_time = bw_int_to_fixed(0); /* todo: this is a bug*/ break; case BW_CALCS_VERSION_POLARIS10: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) /* TODO: Treat VEGAM the same as P10 for now * Need to tune the para for VEGAM if needed */ case BW_CALCS_VERSION_VEGAM: -#endif vbios.memory_type = bw_def_gddr5; vbios.dram_channel_width_in_bits = 32; vbios.number_of_dram_channels = asic_id.vram_width / vbios.dram_channel_width_in_bits; diff --git a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c index a102c192328d..49a4ea45466d 100644 --- a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c @@ -873,14 +873,14 @@ bool dcn_validate_bandwidth( } if (pipe->plane_state->rotation % 2 == 0) { - ASSERT(pipe->plane_res.scl_data.ratios.horz.value != dal_fixed31_32_one.value + ASSERT(pipe->plane_res.scl_data.ratios.horz.value != dc_fixpt_one.value || v->scaler_rec_out_width[input_idx] == v->viewport_width[input_idx]); - ASSERT(pipe->plane_res.scl_data.ratios.vert.value != dal_fixed31_32_one.value + ASSERT(pipe->plane_res.scl_data.ratios.vert.value != dc_fixpt_one.value || v->scaler_recout_height[input_idx] == v->viewport_height[input_idx]); } else { - ASSERT(pipe->plane_res.scl_data.ratios.horz.value != dal_fixed31_32_one.value + ASSERT(pipe->plane_res.scl_data.ratios.horz.value != dc_fixpt_one.value || v->scaler_recout_height[input_idx] == v->viewport_width[input_idx]); - ASSERT(pipe->plane_res.scl_data.ratios.vert.value != dal_fixed31_32_one.value + ASSERT(pipe->plane_res.scl_data.ratios.vert.value != dc_fixpt_one.value || v->scaler_rec_out_width[input_idx] == v->viewport_height[input_idx]); } v->dcc_enable[input_idx] = pipe->plane_state->dcc.enable ? dcn_bw_yes : dcn_bw_no; diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c index b44cf52090a5..2fa521812d23 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c @@ -469,6 +469,13 @@ static void link_disconnect_sink(struct dc_link *link) link->dpcd_sink_count = 0; } +static void link_disconnect_remap(struct dc_sink *prev_sink, struct dc_link *link) +{ + dc_sink_release(link->local_sink); + link->local_sink = prev_sink; +} + + static bool detect_dp( struct dc_link *link, struct display_sink_capability *sink_caps, @@ -551,6 +558,17 @@ static bool detect_dp( return true; } +static bool is_same_edid(struct dc_edid *old_edid, struct dc_edid *new_edid) +{ + if (old_edid->length != new_edid->length) + return false; + + if (new_edid->length == 0) + return false; + + return (memcmp(old_edid->raw_edid, new_edid->raw_edid, new_edid->length) == 0); +} + bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason) { struct dc_sink_init_data sink_init_data = { 0 }; @@ -558,9 +576,13 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason) uint8_t i; bool converter_disable_audio = false; struct audio_support *aud_support = &link->dc->res_pool->audio_support; + bool same_edid = false; enum dc_edid_status edid_status; struct dc_context *dc_ctx = link->ctx; struct dc_sink *sink = NULL; + struct dc_sink *prev_sink = NULL; + struct dpcd_caps prev_dpcd_caps; + bool same_dpcd = true; enum dc_connection_type new_connection_type = dc_connection_none; DC_LOGGER_INIT(link->ctx->logger); if (link->connector_signal == SIGNAL_TYPE_VIRTUAL) @@ -575,6 +597,11 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason) link->local_sink) return true; + prev_sink = link->local_sink; + if (prev_sink != NULL) { + dc_sink_retain(prev_sink); + memcpy(&prev_dpcd_caps, &link->dpcd_caps, sizeof(struct dpcd_caps)); + } link_disconnect_sink(link); if (new_connection_type != dc_connection_none) { @@ -616,14 +643,25 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason) link, &sink_caps, &converter_disable_audio, - aud_support, reason)) + aud_support, reason)) { + if (prev_sink != NULL) + dc_sink_release(prev_sink); return false; + } + // Check if dpcp block is the same + if (prev_sink != NULL) { + if (memcmp(&link->dpcd_caps, &prev_dpcd_caps, sizeof(struct dpcd_caps))) + same_dpcd = false; + } /* Active dongle downstream unplug */ if (link->type == dc_connection_active_dongle && link->dpcd_caps.sink_count. - bits.SINK_COUNT == 0) + bits.SINK_COUNT == 0) { + if (prev_sink != NULL) + dc_sink_release(prev_sink); return true; + } if (link->type == dc_connection_mst_branch) { LINK_INFO("link=%d, mst branch is now Connected\n", @@ -631,9 +669,11 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason) /* Need to setup mst link_cap struct here * otherwise dc_link_detect() will leave mst link_cap * empty which leads to allocate_mst_payload() has "0" - * pbn_per_slot value leading to exception on dal_fixed31_32_div() + * pbn_per_slot value leading to exception on dc_fixpt_div() */ link->verified_link_cap = link->reported_link_cap; + if (prev_sink != NULL) + dc_sink_release(prev_sink); return false; } @@ -643,6 +683,8 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason) default: DC_ERROR("Invalid connector type! signal:%d\n", link->connector_signal); + if (prev_sink != NULL) + dc_sink_release(prev_sink); return false; } /* switch() */ @@ -665,6 +707,8 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason) sink = dc_sink_create(&sink_init_data); if (!sink) { DC_ERROR("Failed to create sink!\n"); + if (prev_sink != NULL) + dc_sink_release(prev_sink); return false; } @@ -688,22 +732,33 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason) break; } - if (link->connector_signal == SIGNAL_TYPE_DISPLAY_PORT && - sink_caps.transaction_type == - DDC_TRANSACTION_TYPE_I2C_OVER_AUX) { - /* - * TODO debug why Dell 2413 doesn't like - * two link trainings - */ + // Check if edid is the same + if ((prev_sink != NULL) && ((edid_status == EDID_THE_SAME) || (edid_status == EDID_OK))) + same_edid = is_same_edid(&prev_sink->dc_edid, &sink->dc_edid); - /* deal with non-mst cases */ - dp_hbr_verify_link_cap(link, &link->reported_link_cap); - } + // If both edid and dpcd are the same, then discard new sink and revert back to original sink + if ((same_edid) && (same_dpcd)) { + link_disconnect_remap(prev_sink, link); + sink = prev_sink; + prev_sink = NULL; + } else { + if (link->connector_signal == SIGNAL_TYPE_DISPLAY_PORT && + sink_caps.transaction_type == + DDC_TRANSACTION_TYPE_I2C_OVER_AUX) { + /* + * TODO debug why Dell 2413 doesn't like + * two link trainings + */ + + /* deal with non-mst cases */ + dp_hbr_verify_link_cap(link, &link->reported_link_cap); + } - /* HDMI-DVI Dongle */ - if (sink->sink_signal == SIGNAL_TYPE_HDMI_TYPE_A && - !sink->edid_caps.edid_hdmi) - sink->sink_signal = SIGNAL_TYPE_DVI_SINGLE_LINK; + /* HDMI-DVI Dongle */ + if (sink->sink_signal == SIGNAL_TYPE_HDMI_TYPE_A && + !sink->edid_caps.edid_hdmi) + sink->sink_signal = SIGNAL_TYPE_DVI_SINGLE_LINK; + } /* Connectivity log: detection */ for (i = 0; i < sink->dc_edid.length / EDID_BLOCK_SIZE; i++) { @@ -762,10 +817,14 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason) sink_caps.signal = SIGNAL_TYPE_NONE; } - LINK_INFO("link=%d, dc_sink_in=%p is now %s\n", + LINK_INFO("link=%d, dc_sink_in=%p is now %s prev_sink=%p dpcd same=%d edid same=%d\n", link->link_index, sink, (sink_caps.signal == SIGNAL_TYPE_NONE ? - "Disconnected":"Connected")); + "Disconnected":"Connected"), prev_sink, + same_dpcd, same_edid); + + if (prev_sink != NULL) + dc_sink_release(prev_sink); return true; } @@ -2059,10 +2118,10 @@ static struct fixed31_32 get_pbn_per_slot(struct dc_stream_state *stream) &stream->sink->link->cur_link_settings; uint32_t link_rate_in_mbps = link_settings->link_rate * LINK_RATE_REF_FREQ_IN_MHZ; - struct fixed31_32 mbps = dal_fixed31_32_from_int( + struct fixed31_32 mbps = dc_fixpt_from_int( link_rate_in_mbps * link_settings->lane_count); - return dal_fixed31_32_div_int(mbps, 54); + return dc_fixpt_div_int(mbps, 54); } static int get_color_depth(enum dc_color_depth color_depth) @@ -2103,7 +2162,7 @@ static struct fixed31_32 get_pbn_from_timing(struct pipe_ctx *pipe_ctx) numerator = 64 * PEAK_FACTOR_X1000; denominator = 54 * 8 * 1000 * 1000; kbps *= numerator; - peak_kbps = dal_fixed31_32_from_fraction(kbps, denominator); + peak_kbps = dc_fixpt_from_fraction(kbps, denominator); return peak_kbps; } @@ -2230,7 +2289,7 @@ static enum dc_status allocate_mst_payload(struct pipe_ctx *pipe_ctx) /* slot X.Y for only current stream */ pbn_per_slot = get_pbn_per_slot(stream); pbn = get_pbn_from_timing(pipe_ctx); - avg_time_slots_per_mtp = dal_fixed31_32_div(pbn, pbn_per_slot); + avg_time_slots_per_mtp = dc_fixpt_div(pbn, pbn_per_slot); stream_encoder->funcs->set_mst_bandwidth( stream_encoder, @@ -2247,7 +2306,7 @@ static enum dc_status deallocate_mst_payload(struct pipe_ctx *pipe_ctx) struct link_encoder *link_encoder = link->link_enc; struct stream_encoder *stream_encoder = pipe_ctx->stream_res.stream_enc; struct dp_mst_stream_allocation_table proposed_table = {0}; - struct fixed31_32 avg_time_slots_per_mtp = dal_fixed31_32_from_int(0); + struct fixed31_32 avg_time_slots_per_mtp = dc_fixpt_from_int(0); uint8_t i; bool mst_mode = (link->type == dc_connection_mst_branch); DC_LOGGER_INIT(link->ctx->logger); diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_hwss.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_hwss.c index 7c866a7d5e77..82cd1d6e6e59 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_hwss.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_hwss.c @@ -11,8 +11,6 @@ #include "dc_link_dp.h" #include "dc_link_ddc.h" #include "dm_helpers.h" -#include "dce/dce_link_encoder.h" -#include "dce/dce_stream_encoder.h" #include "dpcd_defs.h" enum dc_status core_link_read_dpcd( diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index 9eb731fb5251..751f3ac9d921 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -79,10 +79,8 @@ enum dce_version resource_parse_asic_id(struct hw_asic_id asic_id) ASIC_REV_IS_POLARIS12_V(asic_id.hw_internal_rev)) { dc_version = DCE_VERSION_11_2; } -#if defined(CONFIG_DRM_AMD_DC_VEGAM) if (ASIC_REV_IS_VEGAM(asic_id.hw_internal_rev)) dc_version = DCE_VERSION_11_22; -#endif break; case FAMILY_AI: dc_version = DCE_VERSION_12_0; @@ -129,9 +127,7 @@ struct resource_pool *dc_create_resource_pool( num_virtual_links, dc, asic_id); break; case DCE_VERSION_11_2: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case DCE_VERSION_11_22: -#endif res_pool = dce112_create_resource_pool( num_virtual_links, dc); break; @@ -500,9 +496,9 @@ static void calculate_viewport(struct pipe_ctx *pipe_ctx) data->viewport_c.x = data->viewport.x / vpc_div; data->viewport_c.y = data->viewport.y / vpc_div; data->inits.h_c = (data->viewport.x % vpc_div) != 0 ? - dal_fixed31_32_half : dal_fixed31_32_zero; + dc_fixpt_half : dc_fixpt_zero; data->inits.v_c = (data->viewport.y % vpc_div) != 0 ? - dal_fixed31_32_half : dal_fixed31_32_zero; + dc_fixpt_half : dc_fixpt_zero; /* Round up, assume original video size always even dimensions */ data->viewport_c.width = (data->viewport.width + vpc_div - 1) / vpc_div; data->viewport_c.height = (data->viewport.height + vpc_div - 1) / vpc_div; @@ -631,10 +627,10 @@ static void calculate_scaling_ratios(struct pipe_ctx *pipe_ctx) pipe_ctx->plane_state->rotation == ROTATION_ANGLE_270) rect_swap_helper(&surf_src); - pipe_ctx->plane_res.scl_data.ratios.horz = dal_fixed31_32_from_fraction( + pipe_ctx->plane_res.scl_data.ratios.horz = dc_fixpt_from_fraction( surf_src.width, plane_state->dst_rect.width); - pipe_ctx->plane_res.scl_data.ratios.vert = dal_fixed31_32_from_fraction( + pipe_ctx->plane_res.scl_data.ratios.vert = dc_fixpt_from_fraction( surf_src.height, plane_state->dst_rect.height); @@ -656,6 +652,14 @@ static void calculate_scaling_ratios(struct pipe_ctx *pipe_ctx) pipe_ctx->plane_res.scl_data.ratios.horz_c.value /= 2; pipe_ctx->plane_res.scl_data.ratios.vert_c.value /= 2; } + pipe_ctx->plane_res.scl_data.ratios.horz = dc_fixpt_truncate( + pipe_ctx->plane_res.scl_data.ratios.horz, 19); + pipe_ctx->plane_res.scl_data.ratios.vert = dc_fixpt_truncate( + pipe_ctx->plane_res.scl_data.ratios.vert, 19); + pipe_ctx->plane_res.scl_data.ratios.horz_c = dc_fixpt_truncate( + pipe_ctx->plane_res.scl_data.ratios.horz_c, 19); + pipe_ctx->plane_res.scl_data.ratios.vert_c = dc_fixpt_truncate( + pipe_ctx->plane_res.scl_data.ratios.vert_c, 19); } static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *recout_skip) @@ -692,32 +696,33 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r * init_bot = init + scaling_ratio * init_c = init + truncated_vp_c_offset(from calculate viewport) */ - data->inits.h = dal_fixed31_32_div_int( - dal_fixed31_32_add_int(data->ratios.horz, data->taps.h_taps + 1), 2); + data->inits.h = dc_fixpt_truncate(dc_fixpt_div_int( + dc_fixpt_add_int(data->ratios.horz, data->taps.h_taps + 1), 2), 19); + + data->inits.h_c = dc_fixpt_truncate(dc_fixpt_add(data->inits.h_c, dc_fixpt_div_int( + dc_fixpt_add_int(data->ratios.horz_c, data->taps.h_taps_c + 1), 2)), 19); - data->inits.h_c = dal_fixed31_32_add(data->inits.h_c, dal_fixed31_32_div_int( - dal_fixed31_32_add_int(data->ratios.horz_c, data->taps.h_taps_c + 1), 2)); + data->inits.v = dc_fixpt_truncate(dc_fixpt_div_int( + dc_fixpt_add_int(data->ratios.vert, data->taps.v_taps + 1), 2), 19); - data->inits.v = dal_fixed31_32_div_int( - dal_fixed31_32_add_int(data->ratios.vert, data->taps.v_taps + 1), 2); + data->inits.v_c = dc_fixpt_truncate(dc_fixpt_add(data->inits.v_c, dc_fixpt_div_int( + dc_fixpt_add_int(data->ratios.vert_c, data->taps.v_taps_c + 1), 2)), 19); - data->inits.v_c = dal_fixed31_32_add(data->inits.v_c, dal_fixed31_32_div_int( - dal_fixed31_32_add_int(data->ratios.vert_c, data->taps.v_taps_c + 1), 2)); /* Adjust for viewport end clip-off */ if ((data->viewport.x + data->viewport.width) < (src.x + src.width) && !flip_horz_scan_dir) { int vp_clip = src.x + src.width - data->viewport.width - data->viewport.x; - int int_part = dal_fixed31_32_floor( - dal_fixed31_32_sub(data->inits.h, data->ratios.horz)); + int int_part = dc_fixpt_floor( + dc_fixpt_sub(data->inits.h, data->ratios.horz)); int_part = int_part > 0 ? int_part : 0; data->viewport.width += int_part < vp_clip ? int_part : vp_clip; } if ((data->viewport.y + data->viewport.height) < (src.y + src.height) && !flip_vert_scan_dir) { int vp_clip = src.y + src.height - data->viewport.height - data->viewport.y; - int int_part = dal_fixed31_32_floor( - dal_fixed31_32_sub(data->inits.v, data->ratios.vert)); + int int_part = dc_fixpt_floor( + dc_fixpt_sub(data->inits.v, data->ratios.vert)); int_part = int_part > 0 ? int_part : 0; data->viewport.height += int_part < vp_clip ? int_part : vp_clip; @@ -725,8 +730,8 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r if ((data->viewport_c.x + data->viewport_c.width) < (src.x + src.width) / vpc_div && !flip_horz_scan_dir) { int vp_clip = (src.x + src.width) / vpc_div - data->viewport_c.width - data->viewport_c.x; - int int_part = dal_fixed31_32_floor( - dal_fixed31_32_sub(data->inits.h_c, data->ratios.horz_c)); + int int_part = dc_fixpt_floor( + dc_fixpt_sub(data->inits.h_c, data->ratios.horz_c)); int_part = int_part > 0 ? int_part : 0; data->viewport_c.width += int_part < vp_clip ? int_part : vp_clip; @@ -734,8 +739,8 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r if ((data->viewport_c.y + data->viewport_c.height) < (src.y + src.height) / vpc_div && !flip_vert_scan_dir) { int vp_clip = (src.y + src.height) / vpc_div - data->viewport_c.height - data->viewport_c.y; - int int_part = dal_fixed31_32_floor( - dal_fixed31_32_sub(data->inits.v_c, data->ratios.vert_c)); + int int_part = dc_fixpt_floor( + dc_fixpt_sub(data->inits.v_c, data->ratios.vert_c)); int_part = int_part > 0 ? int_part : 0; data->viewport_c.height += int_part < vp_clip ? int_part : vp_clip; @@ -745,9 +750,9 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r if (data->viewport.x && !flip_horz_scan_dir) { int int_part; - data->inits.h = dal_fixed31_32_add(data->inits.h, dal_fixed31_32_mul_int( + data->inits.h = dc_fixpt_add(data->inits.h, dc_fixpt_mul_int( data->ratios.horz, recout_skip->width)); - int_part = dal_fixed31_32_floor(data->inits.h) - data->viewport.x; + int_part = dc_fixpt_floor(data->inits.h) - data->viewport.x; if (int_part < data->taps.h_taps) { int int_adj = data->viewport.x >= (data->taps.h_taps - int_part) ? (data->taps.h_taps - int_part) : data->viewport.x; @@ -760,15 +765,15 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r int_part = data->taps.h_taps; } data->inits.h.value &= 0xffffffff; - data->inits.h = dal_fixed31_32_add_int(data->inits.h, int_part); + data->inits.h = dc_fixpt_add_int(data->inits.h, int_part); } if (data->viewport_c.x && !flip_horz_scan_dir) { int int_part; - data->inits.h_c = dal_fixed31_32_add(data->inits.h_c, dal_fixed31_32_mul_int( + data->inits.h_c = dc_fixpt_add(data->inits.h_c, dc_fixpt_mul_int( data->ratios.horz_c, recout_skip->width)); - int_part = dal_fixed31_32_floor(data->inits.h_c) - data->viewport_c.x; + int_part = dc_fixpt_floor(data->inits.h_c) - data->viewport_c.x; if (int_part < data->taps.h_taps_c) { int int_adj = data->viewport_c.x >= (data->taps.h_taps_c - int_part) ? (data->taps.h_taps_c - int_part) : data->viewport_c.x; @@ -781,15 +786,15 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r int_part = data->taps.h_taps_c; } data->inits.h_c.value &= 0xffffffff; - data->inits.h_c = dal_fixed31_32_add_int(data->inits.h_c, int_part); + data->inits.h_c = dc_fixpt_add_int(data->inits.h_c, int_part); } if (data->viewport.y && !flip_vert_scan_dir) { int int_part; - data->inits.v = dal_fixed31_32_add(data->inits.v, dal_fixed31_32_mul_int( + data->inits.v = dc_fixpt_add(data->inits.v, dc_fixpt_mul_int( data->ratios.vert, recout_skip->height)); - int_part = dal_fixed31_32_floor(data->inits.v) - data->viewport.y; + int_part = dc_fixpt_floor(data->inits.v) - data->viewport.y; if (int_part < data->taps.v_taps) { int int_adj = data->viewport.y >= (data->taps.v_taps - int_part) ? (data->taps.v_taps - int_part) : data->viewport.y; @@ -802,15 +807,15 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r int_part = data->taps.v_taps; } data->inits.v.value &= 0xffffffff; - data->inits.v = dal_fixed31_32_add_int(data->inits.v, int_part); + data->inits.v = dc_fixpt_add_int(data->inits.v, int_part); } if (data->viewport_c.y && !flip_vert_scan_dir) { int int_part; - data->inits.v_c = dal_fixed31_32_add(data->inits.v_c, dal_fixed31_32_mul_int( + data->inits.v_c = dc_fixpt_add(data->inits.v_c, dc_fixpt_mul_int( data->ratios.vert_c, recout_skip->height)); - int_part = dal_fixed31_32_floor(data->inits.v_c) - data->viewport_c.y; + int_part = dc_fixpt_floor(data->inits.v_c) - data->viewport_c.y; if (int_part < data->taps.v_taps_c) { int int_adj = data->viewport_c.y >= (data->taps.v_taps_c - int_part) ? (data->taps.v_taps_c - int_part) : data->viewport_c.y; @@ -823,12 +828,12 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct view *r int_part = data->taps.v_taps_c; } data->inits.v_c.value &= 0xffffffff; - data->inits.v_c = dal_fixed31_32_add_int(data->inits.v_c, int_part); + data->inits.v_c = dc_fixpt_add_int(data->inits.v_c, int_part); } /* Interlaced inits based on final vert inits */ - data->inits.v_bot = dal_fixed31_32_add(data->inits.v, data->ratios.vert); - data->inits.v_c_bot = dal_fixed31_32_add(data->inits.v_c, data->ratios.vert_c); + data->inits.v_bot = dc_fixpt_add(data->inits.v, data->ratios.vert); + data->inits.v_c_bot = dc_fixpt_add(data->inits.v_c, data->ratios.vert_c); if (pipe_ctx->plane_state->rotation == ROTATION_ANGLE_90 || pipe_ctx->plane_state->rotation == ROTATION_ANGLE_270) { diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index cd4f4341cb53..9cfde0ccf4e9 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -75,6 +75,7 @@ struct dc_caps { bool dynamic_audio; bool is_apu; bool dual_link_dvi; + bool post_blend_color_processing; }; struct dc_dcc_surface_param { @@ -203,6 +204,7 @@ struct dc_debug { bool clock_trace; bool validation_trace; bool bandwidth_calcs_trace; + int max_downscale_src_width; /* stutter efficiency related */ bool disable_stutter; @@ -239,6 +241,8 @@ struct dc_debug { bool az_endpoint_mute_only; bool always_use_regamma; bool p010_mpo_support; + bool recovery_enabled; + }; struct dc_state; struct resource_pool; @@ -499,18 +503,18 @@ struct dc_surface_update { struct dc_plane_state *surface; /* isr safe update parameters. null means no updates */ - struct dc_flip_addrs *flip_addr; - struct dc_plane_info *plane_info; - struct dc_scaling_info *scaling_info; + const struct dc_flip_addrs *flip_addr; + const struct dc_plane_info *plane_info; + const struct dc_scaling_info *scaling_info; /* following updates require alloc/sleep/spin that is not isr safe, * null means no updates */ - struct dc_gamma *gamma; - struct dc_transfer_func *in_transfer_func; + const struct dc_gamma *gamma; + const struct dc_transfer_func *in_transfer_func; - struct dc_csc_transform *input_csc_color_matrix; - struct fixed31_32 *coeff_reduction_factor; + const struct dc_csc_transform *input_csc_color_matrix; + const struct fixed31_32 *coeff_reduction_factor; }; /* diff --git a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h index 2726b02e006b..90bccd5ccaa2 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h @@ -26,6 +26,8 @@ #ifndef DC_DP_TYPES_H #define DC_DP_TYPES_H +#include "os_types.h" + enum dc_lane_count { LANE_COUNT_UNKNOWN = 0, LANE_COUNT_ONE = 1, diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index 9defe3b17617..76df2534c4a4 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -25,7 +25,7 @@ #ifndef DC_TYPES_H_ #define DC_TYPES_H_ -#include "fixed32_32.h" +#include "os_types.h" #include "fixed31_32.h" #include "irq_types.h" #include "dc_dp_types.h" diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c index fe92a1222803..29294db1a96b 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c @@ -26,7 +26,7 @@ #include "dce_abm.h" #include "dm_services.h" #include "reg_helper.h" -#include "fixed32_32.h" +#include "fixed31_32.h" #include "dc.h" #include "atom.h" diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.c b/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.c index 223db98a568a..599c7ab6befe 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_clock_source.c @@ -590,9 +590,7 @@ static uint32_t dce110_get_pix_clk_dividers( pll_settings, pix_clk_params); break; case DCE_VERSION_11_2: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case DCE_VERSION_11_22: -#endif case DCE_VERSION_12_0: #if defined(CONFIG_DRM_AMD_DC_DCN1_0) case DCN_VERSION_1_0: @@ -659,12 +657,12 @@ static uint32_t dce110_get_d_to_pixel_rate_in_hz( return 0; } - pix_rate = dal_fixed31_32_from_int(clk_src->ref_freq_khz); - pix_rate = dal_fixed31_32_mul_int(pix_rate, 1000); - pix_rate = dal_fixed31_32_mul_int(pix_rate, phase); - pix_rate = dal_fixed31_32_div_int(pix_rate, modulo); + pix_rate = dc_fixpt_from_int(clk_src->ref_freq_khz); + pix_rate = dc_fixpt_mul_int(pix_rate, 1000); + pix_rate = dc_fixpt_mul_int(pix_rate, phase); + pix_rate = dc_fixpt_div_int(pix_rate, modulo); - return dal_fixed31_32_round(pix_rate); + return dc_fixpt_round(pix_rate); } else { return dce110_get_dp_pixel_rate_from_combo_phy_pll(cs, pix_clk_params, pll_settings); } @@ -713,12 +711,12 @@ static bool calculate_ss( const struct spread_spectrum_data *ss_data, struct delta_sigma_data *ds_data) { - struct fixed32_32 fb_div; - struct fixed32_32 ss_amount; - struct fixed32_32 ss_nslip_amount; - struct fixed32_32 ss_ds_frac_amount; - struct fixed32_32 ss_step_size; - struct fixed32_32 modulation_time; + struct fixed31_32 fb_div; + struct fixed31_32 ss_amount; + struct fixed31_32 ss_nslip_amount; + struct fixed31_32 ss_ds_frac_amount; + struct fixed31_32 ss_step_size; + struct fixed31_32 modulation_time; if (ds_data == NULL) return false; @@ -733,42 +731,42 @@ static bool calculate_ss( /* compute SS_AMOUNT_FBDIV & SS_AMOUNT_NFRAC_SLIP & SS_AMOUNT_DSFRAC*/ /* 6 decimal point support in fractional feedback divider */ - fb_div = dal_fixed32_32_from_fraction( + fb_div = dc_fixpt_from_fraction( pll_settings->fract_feedback_divider, 1000000); - fb_div = dal_fixed32_32_add_int(fb_div, pll_settings->feedback_divider); + fb_div = dc_fixpt_add_int(fb_div, pll_settings->feedback_divider); ds_data->ds_frac_amount = 0; /*spreadSpectrumPercentage is in the unit of .01%, * so have to divided by 100 * 100*/ - ss_amount = dal_fixed32_32_mul( - fb_div, dal_fixed32_32_from_fraction(ss_data->percentage, + ss_amount = dc_fixpt_mul( + fb_div, dc_fixpt_from_fraction(ss_data->percentage, 100 * ss_data->percentage_divider)); - ds_data->feedback_amount = dal_fixed32_32_floor(ss_amount); + ds_data->feedback_amount = dc_fixpt_floor(ss_amount); - ss_nslip_amount = dal_fixed32_32_sub(ss_amount, - dal_fixed32_32_from_int(ds_data->feedback_amount)); - ss_nslip_amount = dal_fixed32_32_mul_int(ss_nslip_amount, 10); - ds_data->nfrac_amount = dal_fixed32_32_floor(ss_nslip_amount); + ss_nslip_amount = dc_fixpt_sub(ss_amount, + dc_fixpt_from_int(ds_data->feedback_amount)); + ss_nslip_amount = dc_fixpt_mul_int(ss_nslip_amount, 10); + ds_data->nfrac_amount = dc_fixpt_floor(ss_nslip_amount); - ss_ds_frac_amount = dal_fixed32_32_sub(ss_nslip_amount, - dal_fixed32_32_from_int(ds_data->nfrac_amount)); - ss_ds_frac_amount = dal_fixed32_32_mul_int(ss_ds_frac_amount, 65536); - ds_data->ds_frac_amount = dal_fixed32_32_floor(ss_ds_frac_amount); + ss_ds_frac_amount = dc_fixpt_sub(ss_nslip_amount, + dc_fixpt_from_int(ds_data->nfrac_amount)); + ss_ds_frac_amount = dc_fixpt_mul_int(ss_ds_frac_amount, 65536); + ds_data->ds_frac_amount = dc_fixpt_floor(ss_ds_frac_amount); /* compute SS_STEP_SIZE_DSFRAC */ - modulation_time = dal_fixed32_32_from_fraction( + modulation_time = dc_fixpt_from_fraction( pll_settings->reference_freq * 1000, pll_settings->reference_divider * ss_data->modulation_freq_hz); if (ss_data->flags.CENTER_SPREAD) - modulation_time = dal_fixed32_32_div_int(modulation_time, 4); + modulation_time = dc_fixpt_div_int(modulation_time, 4); else - modulation_time = dal_fixed32_32_div_int(modulation_time, 2); + modulation_time = dc_fixpt_div_int(modulation_time, 2); - ss_step_size = dal_fixed32_32_div(ss_amount, modulation_time); + ss_step_size = dc_fixpt_div(ss_amount, modulation_time); /* SS_STEP_SIZE_DSFRAC_DEC = Int(SS_STEP_SIZE * 2 ^ 16 * 10)*/ - ss_step_size = dal_fixed32_32_mul_int(ss_step_size, 65536 * 10); - ds_data->ds_frac_size = dal_fixed32_32_floor(ss_step_size); + ss_step_size = dc_fixpt_mul_int(ss_step_size, 65536 * 10); + ds_data->ds_frac_size = dc_fixpt_floor(ss_step_size); return true; } @@ -982,9 +980,7 @@ static bool dce110_program_pix_clk( break; case DCE_VERSION_11_2: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case DCE_VERSION_11_22: -#endif case DCE_VERSION_12_0: #if defined(CONFIG_DRM_AMD_DC_DCN1_0) case DCN_VERSION_1_0: diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c b/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c index 78e6beb6cf26..8a581c67bf2d 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c @@ -26,7 +26,7 @@ #include "dce_clocks.h" #include "dm_services.h" #include "reg_helper.h" -#include "fixed32_32.h" +#include "fixed31_32.h" #include "bios_parser_interface.h" #include "dc.h" #include "dmcu.h" @@ -35,7 +35,7 @@ #endif #include "core_types.h" #include "dc_types.h" - +#include "dal_asic_id.h" #define TO_DCE_CLOCKS(clocks)\ container_of(clocks, struct dce_disp_clk, base) @@ -228,19 +228,19 @@ static int dce_clocks_get_dp_ref_freq(struct display_clock *clk) generated according to average value (case as with previous ASICs) */ if (clk_dce->ss_on_dprefclk && clk_dce->dprefclk_ss_divider != 0) { - struct fixed32_32 ss_percentage = dal_fixed32_32_div_int( - dal_fixed32_32_from_fraction( + struct fixed31_32 ss_percentage = dc_fixpt_div_int( + dc_fixpt_from_fraction( clk_dce->dprefclk_ss_percentage, clk_dce->dprefclk_ss_divider), 200); - struct fixed32_32 adj_dp_ref_clk_khz; + struct fixed31_32 adj_dp_ref_clk_khz; - ss_percentage = dal_fixed32_32_sub(dal_fixed32_32_one, + ss_percentage = dc_fixpt_sub(dc_fixpt_one, ss_percentage); adj_dp_ref_clk_khz = - dal_fixed32_32_mul_int( + dc_fixpt_mul_int( ss_percentage, dp_ref_clk_khz); - dp_ref_clk_khz = dal_fixed32_32_floor(adj_dp_ref_clk_khz); + dp_ref_clk_khz = dc_fixpt_floor(adj_dp_ref_clk_khz); } return dp_ref_clk_khz; @@ -256,19 +256,19 @@ static int dce_clocks_get_dp_ref_freq_wrkaround(struct display_clock *clk) int dp_ref_clk_khz = 600000; if (clk_dce->ss_on_dprefclk && clk_dce->dprefclk_ss_divider != 0) { - struct fixed32_32 ss_percentage = dal_fixed32_32_div_int( - dal_fixed32_32_from_fraction( + struct fixed31_32 ss_percentage = dc_fixpt_div_int( + dc_fixpt_from_fraction( clk_dce->dprefclk_ss_percentage, clk_dce->dprefclk_ss_divider), 200); - struct fixed32_32 adj_dp_ref_clk_khz; + struct fixed31_32 adj_dp_ref_clk_khz; - ss_percentage = dal_fixed32_32_sub(dal_fixed32_32_one, + ss_percentage = dc_fixpt_sub(dc_fixpt_one, ss_percentage); adj_dp_ref_clk_khz = - dal_fixed32_32_mul_int( + dc_fixpt_mul_int( ss_percentage, dp_ref_clk_khz); - dp_ref_clk_khz = dal_fixed32_32_floor(adj_dp_ref_clk_khz); + dp_ref_clk_khz = dc_fixpt_floor(adj_dp_ref_clk_khz); } return dp_ref_clk_khz; @@ -413,9 +413,12 @@ static int dce112_set_clock( /*VBIOS will determine DPREFCLK frequency, so we don't set it*/ dce_clk_params.target_clock_frequency = 0; dce_clk_params.clock_type = DCECLOCK_TYPE_DPREFCLK; - dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = + if (!ASICREV_IS_VEGA20_P(clk->ctx->asic_id.hw_internal_rev)) + dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = (dce_clk_params.pll_id == CLOCK_SOURCE_COMBO_DISPLAY_PLL0); + else + dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = false; bp->funcs->set_dce_clock(bp, &dce_clk_params); diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c index 2ee3d9bf1062..a576b8bbb3cd 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c @@ -28,7 +28,7 @@ #include "dce_dmcu.h" #include "dm_services.h" #include "reg_helper.h" -#include "fixed32_32.h" +#include "fixed31_32.h" #include "dc.h" #define TO_DCE_DMCU(dmcu)\ diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.c b/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.c index d737e911971b..5d9506b3d46b 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_ipp.c @@ -195,13 +195,13 @@ static void dce_ipp_program_input_lut( for (i = 0; i < gamma->num_entries; i++) { REG_SET(DC_LUT_SEQ_COLOR, 0, DC_LUT_SEQ_COLOR, - dal_fixed31_32_round( + dc_fixpt_round( gamma->entries.red[i])); REG_SET(DC_LUT_SEQ_COLOR, 0, DC_LUT_SEQ_COLOR, - dal_fixed31_32_round( + dc_fixpt_round( gamma->entries.green[i])); REG_SET(DC_LUT_SEQ_COLOR, 0, DC_LUT_SEQ_COLOR, - dal_fixed31_32_round( + dc_fixpt_round( gamma->entries.blue[i])); } diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters.c b/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters.c index 6243450b41b7..48862bebf29e 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_scl_filters.c @@ -1014,11 +1014,11 @@ static const uint16_t filter_8tap_64p_183[264] = { const uint16_t *get_filter_3tap_16p(struct fixed31_32 ratio) { - if (ratio.value < dal_fixed31_32_one.value) + if (ratio.value < dc_fixpt_one.value) return filter_3tap_16p_upscale; - else if (ratio.value < dal_fixed31_32_from_fraction(4, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(4, 3).value) return filter_3tap_16p_117; - else if (ratio.value < dal_fixed31_32_from_fraction(5, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(5, 3).value) return filter_3tap_16p_150; else return filter_3tap_16p_183; @@ -1026,11 +1026,11 @@ const uint16_t *get_filter_3tap_16p(struct fixed31_32 ratio) const uint16_t *get_filter_3tap_64p(struct fixed31_32 ratio) { - if (ratio.value < dal_fixed31_32_one.value) + if (ratio.value < dc_fixpt_one.value) return filter_3tap_64p_upscale; - else if (ratio.value < dal_fixed31_32_from_fraction(4, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(4, 3).value) return filter_3tap_64p_117; - else if (ratio.value < dal_fixed31_32_from_fraction(5, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(5, 3).value) return filter_3tap_64p_150; else return filter_3tap_64p_183; @@ -1038,11 +1038,11 @@ const uint16_t *get_filter_3tap_64p(struct fixed31_32 ratio) const uint16_t *get_filter_4tap_16p(struct fixed31_32 ratio) { - if (ratio.value < dal_fixed31_32_one.value) + if (ratio.value < dc_fixpt_one.value) return filter_4tap_16p_upscale; - else if (ratio.value < dal_fixed31_32_from_fraction(4, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(4, 3).value) return filter_4tap_16p_117; - else if (ratio.value < dal_fixed31_32_from_fraction(5, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(5, 3).value) return filter_4tap_16p_150; else return filter_4tap_16p_183; @@ -1050,11 +1050,11 @@ const uint16_t *get_filter_4tap_16p(struct fixed31_32 ratio) const uint16_t *get_filter_4tap_64p(struct fixed31_32 ratio) { - if (ratio.value < dal_fixed31_32_one.value) + if (ratio.value < dc_fixpt_one.value) return filter_4tap_64p_upscale; - else if (ratio.value < dal_fixed31_32_from_fraction(4, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(4, 3).value) return filter_4tap_64p_117; - else if (ratio.value < dal_fixed31_32_from_fraction(5, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(5, 3).value) return filter_4tap_64p_150; else return filter_4tap_64p_183; @@ -1062,11 +1062,11 @@ const uint16_t *get_filter_4tap_64p(struct fixed31_32 ratio) const uint16_t *get_filter_5tap_64p(struct fixed31_32 ratio) { - if (ratio.value < dal_fixed31_32_one.value) + if (ratio.value < dc_fixpt_one.value) return filter_5tap_64p_upscale; - else if (ratio.value < dal_fixed31_32_from_fraction(4, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(4, 3).value) return filter_5tap_64p_117; - else if (ratio.value < dal_fixed31_32_from_fraction(5, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(5, 3).value) return filter_5tap_64p_150; else return filter_5tap_64p_183; @@ -1074,11 +1074,11 @@ const uint16_t *get_filter_5tap_64p(struct fixed31_32 ratio) const uint16_t *get_filter_6tap_64p(struct fixed31_32 ratio) { - if (ratio.value < dal_fixed31_32_one.value) + if (ratio.value < dc_fixpt_one.value) return filter_6tap_64p_upscale; - else if (ratio.value < dal_fixed31_32_from_fraction(4, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(4, 3).value) return filter_6tap_64p_117; - else if (ratio.value < dal_fixed31_32_from_fraction(5, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(5, 3).value) return filter_6tap_64p_150; else return filter_6tap_64p_183; @@ -1086,11 +1086,11 @@ const uint16_t *get_filter_6tap_64p(struct fixed31_32 ratio) const uint16_t *get_filter_7tap_64p(struct fixed31_32 ratio) { - if (ratio.value < dal_fixed31_32_one.value) + if (ratio.value < dc_fixpt_one.value) return filter_7tap_64p_upscale; - else if (ratio.value < dal_fixed31_32_from_fraction(4, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(4, 3).value) return filter_7tap_64p_117; - else if (ratio.value < dal_fixed31_32_from_fraction(5, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(5, 3).value) return filter_7tap_64p_150; else return filter_7tap_64p_183; @@ -1098,11 +1098,11 @@ const uint16_t *get_filter_7tap_64p(struct fixed31_32 ratio) const uint16_t *get_filter_8tap_64p(struct fixed31_32 ratio) { - if (ratio.value < dal_fixed31_32_one.value) + if (ratio.value < dc_fixpt_one.value) return filter_8tap_64p_upscale; - else if (ratio.value < dal_fixed31_32_from_fraction(4, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(4, 3).value) return filter_8tap_64p_117; - else if (ratio.value < dal_fixed31_32_from_fraction(5, 3).value) + else if (ratio.value < dc_fixpt_from_fraction(5, 3).value) return filter_8tap_64p_150; else return filter_8tap_64p_183; diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c index e265a0abe361..0a6d483dc046 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c @@ -683,11 +683,11 @@ static void dce110_stream_encoder_set_mst_bandwidth( struct fixed31_32 avg_time_slots_per_mtp) { struct dce110_stream_encoder *enc110 = DCE110STRENC_FROM_STRENC(enc); - uint32_t x = dal_fixed31_32_floor( + uint32_t x = dc_fixpt_floor( avg_time_slots_per_mtp); - uint32_t y = dal_fixed31_32_ceil( - dal_fixed31_32_shl( - dal_fixed31_32_sub_int( + uint32_t y = dc_fixpt_ceil( + dc_fixpt_shl( + dc_fixpt_sub_int( avg_time_slots_per_mtp, x), 26)); diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c b/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c index 832c5daada35..a02e719d7794 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_transform.c @@ -41,7 +41,7 @@ #define DC_LOGGER \ xfm_dce->base.ctx->logger -#define IDENTITY_RATIO(ratio) (dal_fixed31_32_u2d19(ratio) == (1 << 19)) +#define IDENTITY_RATIO(ratio) (dc_fixpt_u2d19(ratio) == (1 << 19)) #define GAMUT_MATRIX_SIZE 12 #define SCL_PHASES 16 @@ -256,27 +256,27 @@ static void calculate_inits( struct fixed31_32 v_init; inits->h_int_scale_ratio = - dal_fixed31_32_u2d19(data->ratios.horz) << 5; + dc_fixpt_u2d19(data->ratios.horz) << 5; inits->v_int_scale_ratio = - dal_fixed31_32_u2d19(data->ratios.vert) << 5; + dc_fixpt_u2d19(data->ratios.vert) << 5; h_init = - dal_fixed31_32_div_int( - dal_fixed31_32_add( + dc_fixpt_div_int( + dc_fixpt_add( data->ratios.horz, - dal_fixed31_32_from_int(data->taps.h_taps + 1)), + dc_fixpt_from_int(data->taps.h_taps + 1)), 2); - inits->h_init.integer = dal_fixed31_32_floor(h_init); - inits->h_init.fraction = dal_fixed31_32_u0d19(h_init) << 5; + inits->h_init.integer = dc_fixpt_floor(h_init); + inits->h_init.fraction = dc_fixpt_u0d19(h_init) << 5; v_init = - dal_fixed31_32_div_int( - dal_fixed31_32_add( + dc_fixpt_div_int( + dc_fixpt_add( data->ratios.vert, - dal_fixed31_32_from_int(data->taps.v_taps + 1)), + dc_fixpt_from_int(data->taps.v_taps + 1)), 2); - inits->v_init.integer = dal_fixed31_32_floor(v_init); - inits->v_init.fraction = dal_fixed31_32_u0d19(v_init) << 5; + inits->v_init.integer = dc_fixpt_floor(v_init); + inits->v_init.fraction = dc_fixpt_u0d19(v_init) << 5; } static void program_scl_ratios_inits( diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c index 2288d0aa773b..a92fb0aa2ff3 100644 --- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c @@ -509,19 +509,19 @@ dce110_translate_regamma_to_hw_format(const struct dc_transfer_func *output_tf, rgb_resulted[hw_points - 1].green = output_tf->tf_pts.green[start_index]; rgb_resulted[hw_points - 1].blue = output_tf->tf_pts.blue[start_index]; - arr_points[0].x = dal_fixed31_32_pow(dal_fixed31_32_from_int(2), - dal_fixed31_32_from_int(region_start)); - arr_points[1].x = dal_fixed31_32_pow(dal_fixed31_32_from_int(2), - dal_fixed31_32_from_int(region_end)); + arr_points[0].x = dc_fixpt_pow(dc_fixpt_from_int(2), + dc_fixpt_from_int(region_start)); + arr_points[1].x = dc_fixpt_pow(dc_fixpt_from_int(2), + dc_fixpt_from_int(region_end)); y_r = rgb_resulted[0].red; y_g = rgb_resulted[0].green; y_b = rgb_resulted[0].blue; - y1_min = dal_fixed31_32_min(y_r, dal_fixed31_32_min(y_g, y_b)); + y1_min = dc_fixpt_min(y_r, dc_fixpt_min(y_g, y_b)); arr_points[0].y = y1_min; - arr_points[0].slope = dal_fixed31_32_div(arr_points[0].y, + arr_points[0].slope = dc_fixpt_div(arr_points[0].y, arr_points[0].x); y_r = rgb_resulted[hw_points - 1].red; @@ -531,21 +531,21 @@ dce110_translate_regamma_to_hw_format(const struct dc_transfer_func *output_tf, /* see comment above, m_arrPoints[1].y should be the Y value for the * region end (m_numOfHwPoints), not last HW point(m_numOfHwPoints - 1) */ - y3_max = dal_fixed31_32_max(y_r, dal_fixed31_32_max(y_g, y_b)); + y3_max = dc_fixpt_max(y_r, dc_fixpt_max(y_g, y_b)); arr_points[1].y = y3_max; - arr_points[1].slope = dal_fixed31_32_zero; + arr_points[1].slope = dc_fixpt_zero; if (output_tf->tf == TRANSFER_FUNCTION_PQ) { /* for PQ, we want to have a straight line from last HW X point, * and the slope to be such that we hit 1.0 at 10000 nits. */ - const struct fixed31_32 end_value = dal_fixed31_32_from_int(125); + const struct fixed31_32 end_value = dc_fixpt_from_int(125); - arr_points[1].slope = dal_fixed31_32_div( - dal_fixed31_32_sub(dal_fixed31_32_one, arr_points[1].y), - dal_fixed31_32_sub(end_value, arr_points[1].x)); + arr_points[1].slope = dc_fixpt_div( + dc_fixpt_sub(dc_fixpt_one, arr_points[1].y), + dc_fixpt_sub(end_value, arr_points[1].x)); } regamma_params->hw_points_num = hw_points; @@ -569,16 +569,16 @@ dce110_translate_regamma_to_hw_format(const struct dc_transfer_func *output_tf, i = 1; while (i != hw_points + 1) { - if (dal_fixed31_32_lt(rgb_plus_1->red, rgb->red)) + if (dc_fixpt_lt(rgb_plus_1->red, rgb->red)) rgb_plus_1->red = rgb->red; - if (dal_fixed31_32_lt(rgb_plus_1->green, rgb->green)) + if (dc_fixpt_lt(rgb_plus_1->green, rgb->green)) rgb_plus_1->green = rgb->green; - if (dal_fixed31_32_lt(rgb_plus_1->blue, rgb->blue)) + if (dc_fixpt_lt(rgb_plus_1->blue, rgb->blue)) rgb_plus_1->blue = rgb->blue; - rgb->delta_red = dal_fixed31_32_sub(rgb_plus_1->red, rgb->red); - rgb->delta_green = dal_fixed31_32_sub(rgb_plus_1->green, rgb->green); - rgb->delta_blue = dal_fixed31_32_sub(rgb_plus_1->blue, rgb->blue); + rgb->delta_red = dc_fixpt_sub(rgb_plus_1->red, rgb->red); + rgb->delta_green = dc_fixpt_sub(rgb_plus_1->green, rgb->green); + rgb->delta_blue = dc_fixpt_sub(rgb_plus_1->blue, rgb->blue); ++rgb_plus_1; ++rgb; @@ -2269,74 +2269,6 @@ static void program_gamut_remap(struct pipe_ctx *pipe_ctx) pipe_ctx->plane_res.xfm->funcs->transform_set_gamut_remap(pipe_ctx->plane_res.xfm, &adjust); } - -/** - * TODO REMOVE, USE UPDATE INSTEAD - */ -static void set_plane_config( - const struct dc *dc, - struct pipe_ctx *pipe_ctx, - struct resource_context *res_ctx) -{ - struct mem_input *mi = pipe_ctx->plane_res.mi; - struct dc_plane_state *plane_state = pipe_ctx->plane_state; - struct xfm_grph_csc_adjustment adjust; - struct out_csc_color_matrix tbl_entry; - unsigned int i; - - memset(&adjust, 0, sizeof(adjust)); - memset(&tbl_entry, 0, sizeof(tbl_entry)); - adjust.gamut_adjust_type = GRAPHICS_GAMUT_ADJUST_TYPE_BYPASS; - - dce_enable_fe_clock(dc->hwseq, mi->inst, true); - - set_default_colors(pipe_ctx); - if (pipe_ctx->stream->csc_color_matrix.enable_adjustment == true) { - tbl_entry.color_space = - pipe_ctx->stream->output_color_space; - - for (i = 0; i < 12; i++) - tbl_entry.regval[i] = - pipe_ctx->stream->csc_color_matrix.matrix[i]; - - pipe_ctx->plane_res.xfm->funcs->opp_set_csc_adjustment - (pipe_ctx->plane_res.xfm, &tbl_entry); - } - - if (pipe_ctx->stream->gamut_remap_matrix.enable_remap == true) { - adjust.gamut_adjust_type = GRAPHICS_GAMUT_ADJUST_TYPE_SW; - - for (i = 0; i < CSC_TEMPERATURE_MATRIX_SIZE; i++) - adjust.temperature_matrix[i] = - pipe_ctx->stream->gamut_remap_matrix.matrix[i]; - } - - pipe_ctx->plane_res.xfm->funcs->transform_set_gamut_remap(pipe_ctx->plane_res.xfm, &adjust); - - pipe_ctx->plane_res.scl_data.lb_params.alpha_en = pipe_ctx->bottom_pipe != 0; - program_scaler(dc, pipe_ctx); - - program_surface_visibility(dc, pipe_ctx); - - mi->funcs->mem_input_program_surface_config( - mi, - plane_state->format, - &plane_state->tiling_info, - &plane_state->plane_size, - plane_state->rotation, - NULL, - false); - if (mi->funcs->set_blank) - mi->funcs->set_blank(mi, pipe_ctx->plane_state->visible); - - if (dc->config.gpu_vm_support) - mi->funcs->mem_input_program_pte_vm( - pipe_ctx->plane_res.mi, - plane_state->format, - &plane_state->tiling_info, - plane_state->rotation); -} - static void update_plane_addr(const struct dc *dc, struct pipe_ctx *pipe_ctx) { @@ -3023,7 +2955,6 @@ static const struct hw_sequencer_funcs dce110_funcs = { .init_hw = init_hw, .apply_ctx_to_hw = dce110_apply_ctx_to_hw, .apply_ctx_for_surface = dce110_apply_ctx_for_surface, - .set_plane_config = set_plane_config, .update_plane_addr = update_plane_addr, .update_pending_status = dce110_update_pending_status, .set_input_transfer_func = dce110_set_input_transfer_func, diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.c index 8ba3c12fc608..a7dce060204f 100644 --- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.c +++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_transform_v.c @@ -373,13 +373,13 @@ static void calculate_inits( struct rect *chroma_viewport) { inits->h_int_scale_ratio_luma = - dal_fixed31_32_u2d19(data->ratios.horz) << 5; + dc_fixpt_u2d19(data->ratios.horz) << 5; inits->v_int_scale_ratio_luma = - dal_fixed31_32_u2d19(data->ratios.vert) << 5; + dc_fixpt_u2d19(data->ratios.vert) << 5; inits->h_int_scale_ratio_chroma = - dal_fixed31_32_u2d19(data->ratios.horz_c) << 5; + dc_fixpt_u2d19(data->ratios.horz_c) << 5; inits->v_int_scale_ratio_chroma = - dal_fixed31_32_u2d19(data->ratios.vert_c) << 5; + dc_fixpt_u2d19(data->ratios.vert_c) << 5; inits->h_init_luma.integer = 1; inits->v_init_luma.integer = 1; diff --git a/drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c b/drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c index fda01574d1ba..2d58daccc005 100644 --- a/drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c @@ -814,14 +814,25 @@ static void bw_calcs_data_update_from_pplib(struct dc *dc) dm_pp_notify_wm_clock_changes(dc->ctx, &clk_ranges); } +static uint32_t read_pipe_fuses(struct dc_context *ctx) +{ + uint32_t value = dm_read_reg_soc15(ctx, mmCC_DC_PIPE_DIS, 0); + /* VG20 support max 6 pipes */ + value = value & 0x3f; + return value; +} + static bool construct( uint8_t num_virtual_links, struct dc *dc, struct dce110_resource_pool *pool) { unsigned int i; + int j; struct dc_context *ctx = dc->ctx; struct irq_service_init_data irq_init_data; + bool harvest_enabled = ASICREV_IS_VEGA20_P(ctx->asic_id.hw_internal_rev); + uint32_t pipe_fuses; ctx->dc_bios->regs = &bios_regs; @@ -915,28 +926,41 @@ static bool construct( if (!pool->base.irqs) goto irqs_create_fail; + /* retrieve valid pipe fuses */ + if (harvest_enabled) + pipe_fuses = read_pipe_fuses(ctx); + + /* index to valid pipe resource */ + j = 0; for (i = 0; i < pool->base.pipe_count; i++) { - pool->base.timing_generators[i] = + if (harvest_enabled) { + if ((pipe_fuses & (1 << i)) != 0) { + dm_error("DC: skip invalid pipe %d!\n", i); + continue; + } + } + + pool->base.timing_generators[j] = dce120_timing_generator_create( ctx, i, &dce120_tg_offsets[i]); - if (pool->base.timing_generators[i] == NULL) { + if (pool->base.timing_generators[j] == NULL) { BREAK_TO_DEBUGGER(); dm_error("DC: failed to create tg!\n"); goto controller_create_fail; } - pool->base.mis[i] = dce120_mem_input_create(ctx, i); + pool->base.mis[j] = dce120_mem_input_create(ctx, i); - if (pool->base.mis[i] == NULL) { + if (pool->base.mis[j] == NULL) { BREAK_TO_DEBUGGER(); dm_error( "DC: failed to create memory input!\n"); goto controller_create_fail; } - pool->base.ipps[i] = dce120_ipp_create(ctx, i); + pool->base.ipps[j] = dce120_ipp_create(ctx, i); if (pool->base.ipps[i] == NULL) { BREAK_TO_DEBUGGER(); dm_error( @@ -944,7 +968,7 @@ static bool construct( goto controller_create_fail; } - pool->base.transforms[i] = dce120_transform_create(ctx, i); + pool->base.transforms[j] = dce120_transform_create(ctx, i); if (pool->base.transforms[i] == NULL) { BREAK_TO_DEBUGGER(); dm_error( @@ -952,16 +976,23 @@ static bool construct( goto res_create_fail; } - pool->base.opps[i] = dce120_opp_create( + pool->base.opps[j] = dce120_opp_create( ctx, i); - if (pool->base.opps[i] == NULL) { + if (pool->base.opps[j] == NULL) { BREAK_TO_DEBUGGER(); dm_error( "DC: failed to create output pixel processor!\n"); } + + /* check next valid pipe */ + j++; } + /* valid pipe num */ + pool->base.pipe_count = j; + pool->base.timing_generator_count = j; + if (!resource_construct(num_virtual_links, dc, &pool->base, &res_create_funcs)) goto res_create_fail; diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/Makefile b/drivers/gpu/drm/amd/display/dc/dcn10/Makefile index 5c69743a4b4f..84f52c63d95c 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dcn10/Makefile @@ -26,7 +26,7 @@ DCN10 = dcn10_resource.o dcn10_ipp.o dcn10_hw_sequencer.o \ dcn10_dpp.o dcn10_opp.o dcn10_optc.o \ dcn10_hubp.o dcn10_mpc.o \ dcn10_dpp_dscl.o dcn10_dpp_cm.o dcn10_cm_common.o \ - dcn10_hubbub.o dcn10_stream_encoder.o + dcn10_hubbub.o dcn10_stream_encoder.o dcn10_link_encoder.o AMD_DAL_DCN10 = $(addprefix $(AMDDALPATH)/dc/dcn10/,$(DCN10)) diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c index 96d5878e9ccd..5d95a997fd9f 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c @@ -169,7 +169,7 @@ bool cm_helper_convert_to_custom_float( } if (fixpoint == true) - arr_points[1].custom_float_y = dal_fixed31_32_clamp_u0d14(arr_points[1].y); + arr_points[1].custom_float_y = dc_fixpt_clamp_u0d14(arr_points[1].y); else if (!convert_to_custom_float_format(arr_points[1].y, &fmt, &arr_points[1].custom_float_y)) { BREAK_TO_DEBUGGER(); @@ -327,19 +327,19 @@ bool cm_helper_translate_curve_to_hw_format( rgb_resulted[hw_points - 1].green = output_tf->tf_pts.green[start_index]; rgb_resulted[hw_points - 1].blue = output_tf->tf_pts.blue[start_index]; - arr_points[0].x = dal_fixed31_32_pow(dal_fixed31_32_from_int(2), - dal_fixed31_32_from_int(region_start)); - arr_points[1].x = dal_fixed31_32_pow(dal_fixed31_32_from_int(2), - dal_fixed31_32_from_int(region_end)); + arr_points[0].x = dc_fixpt_pow(dc_fixpt_from_int(2), + dc_fixpt_from_int(region_start)); + arr_points[1].x = dc_fixpt_pow(dc_fixpt_from_int(2), + dc_fixpt_from_int(region_end)); y_r = rgb_resulted[0].red; y_g = rgb_resulted[0].green; y_b = rgb_resulted[0].blue; - y1_min = dal_fixed31_32_min(y_r, dal_fixed31_32_min(y_g, y_b)); + y1_min = dc_fixpt_min(y_r, dc_fixpt_min(y_g, y_b)); arr_points[0].y = y1_min; - arr_points[0].slope = dal_fixed31_32_div(arr_points[0].y, arr_points[0].x); + arr_points[0].slope = dc_fixpt_div(arr_points[0].y, arr_points[0].x); y_r = rgb_resulted[hw_points - 1].red; y_g = rgb_resulted[hw_points - 1].green; y_b = rgb_resulted[hw_points - 1].blue; @@ -347,22 +347,22 @@ bool cm_helper_translate_curve_to_hw_format( /* see comment above, m_arrPoints[1].y should be the Y value for the * region end (m_numOfHwPoints), not last HW point(m_numOfHwPoints - 1) */ - y3_max = dal_fixed31_32_max(y_r, dal_fixed31_32_max(y_g, y_b)); + y3_max = dc_fixpt_max(y_r, dc_fixpt_max(y_g, y_b)); arr_points[1].y = y3_max; - arr_points[1].slope = dal_fixed31_32_zero; + arr_points[1].slope = dc_fixpt_zero; if (output_tf->tf == TRANSFER_FUNCTION_PQ) { /* for PQ, we want to have a straight line from last HW X point, * and the slope to be such that we hit 1.0 at 10000 nits. */ const struct fixed31_32 end_value = - dal_fixed31_32_from_int(125); + dc_fixpt_from_int(125); - arr_points[1].slope = dal_fixed31_32_div( - dal_fixed31_32_sub(dal_fixed31_32_one, arr_points[1].y), - dal_fixed31_32_sub(end_value, arr_points[1].x)); + arr_points[1].slope = dc_fixpt_div( + dc_fixpt_sub(dc_fixpt_one, arr_points[1].y), + dc_fixpt_sub(end_value, arr_points[1].x)); } lut_params->hw_points_num = hw_points; @@ -386,24 +386,24 @@ bool cm_helper_translate_curve_to_hw_format( i = 1; while (i != hw_points + 1) { - if (dal_fixed31_32_lt(rgb_plus_1->red, rgb->red)) + if (dc_fixpt_lt(rgb_plus_1->red, rgb->red)) rgb_plus_1->red = rgb->red; - if (dal_fixed31_32_lt(rgb_plus_1->green, rgb->green)) + if (dc_fixpt_lt(rgb_plus_1->green, rgb->green)) rgb_plus_1->green = rgb->green; - if (dal_fixed31_32_lt(rgb_plus_1->blue, rgb->blue)) + if (dc_fixpt_lt(rgb_plus_1->blue, rgb->blue)) rgb_plus_1->blue = rgb->blue; - rgb->delta_red = dal_fixed31_32_sub(rgb_plus_1->red, rgb->red); - rgb->delta_green = dal_fixed31_32_sub(rgb_plus_1->green, rgb->green); - rgb->delta_blue = dal_fixed31_32_sub(rgb_plus_1->blue, rgb->blue); + rgb->delta_red = dc_fixpt_sub(rgb_plus_1->red, rgb->red); + rgb->delta_green = dc_fixpt_sub(rgb_plus_1->green, rgb->green); + rgb->delta_blue = dc_fixpt_sub(rgb_plus_1->blue, rgb->blue); if (fixpoint == true) { - rgb->delta_red_reg = dal_fixed31_32_clamp_u0d10(rgb->delta_red); - rgb->delta_green_reg = dal_fixed31_32_clamp_u0d10(rgb->delta_green); - rgb->delta_blue_reg = dal_fixed31_32_clamp_u0d10(rgb->delta_blue); - rgb->red_reg = dal_fixed31_32_clamp_u0d14(rgb->red); - rgb->green_reg = dal_fixed31_32_clamp_u0d14(rgb->green); - rgb->blue_reg = dal_fixed31_32_clamp_u0d14(rgb->blue); + rgb->delta_red_reg = dc_fixpt_clamp_u0d10(rgb->delta_red); + rgb->delta_green_reg = dc_fixpt_clamp_u0d10(rgb->delta_green); + rgb->delta_blue_reg = dc_fixpt_clamp_u0d10(rgb->delta_blue); + rgb->red_reg = dc_fixpt_clamp_u0d14(rgb->red); + rgb->green_reg = dc_fixpt_clamp_u0d14(rgb->green); + rgb->blue_reg = dc_fixpt_clamp_u0d14(rgb->blue); } ++rgb_plus_1; @@ -489,19 +489,19 @@ bool cm_helper_translate_curve_to_degamma_hw_format( rgb_resulted[hw_points - 1].green = output_tf->tf_pts.green[start_index]; rgb_resulted[hw_points - 1].blue = output_tf->tf_pts.blue[start_index]; - arr_points[0].x = dal_fixed31_32_pow(dal_fixed31_32_from_int(2), - dal_fixed31_32_from_int(region_start)); - arr_points[1].x = dal_fixed31_32_pow(dal_fixed31_32_from_int(2), - dal_fixed31_32_from_int(region_end)); + arr_points[0].x = dc_fixpt_pow(dc_fixpt_from_int(2), + dc_fixpt_from_int(region_start)); + arr_points[1].x = dc_fixpt_pow(dc_fixpt_from_int(2), + dc_fixpt_from_int(region_end)); y_r = rgb_resulted[0].red; y_g = rgb_resulted[0].green; y_b = rgb_resulted[0].blue; - y1_min = dal_fixed31_32_min(y_r, dal_fixed31_32_min(y_g, y_b)); + y1_min = dc_fixpt_min(y_r, dc_fixpt_min(y_g, y_b)); arr_points[0].y = y1_min; - arr_points[0].slope = dal_fixed31_32_div(arr_points[0].y, arr_points[0].x); + arr_points[0].slope = dc_fixpt_div(arr_points[0].y, arr_points[0].x); y_r = rgb_resulted[hw_points - 1].red; y_g = rgb_resulted[hw_points - 1].green; y_b = rgb_resulted[hw_points - 1].blue; @@ -509,22 +509,22 @@ bool cm_helper_translate_curve_to_degamma_hw_format( /* see comment above, m_arrPoints[1].y should be the Y value for the * region end (m_numOfHwPoints), not last HW point(m_numOfHwPoints - 1) */ - y3_max = dal_fixed31_32_max(y_r, dal_fixed31_32_max(y_g, y_b)); + y3_max = dc_fixpt_max(y_r, dc_fixpt_max(y_g, y_b)); arr_points[1].y = y3_max; - arr_points[1].slope = dal_fixed31_32_zero; + arr_points[1].slope = dc_fixpt_zero; if (output_tf->tf == TRANSFER_FUNCTION_PQ) { /* for PQ, we want to have a straight line from last HW X point, * and the slope to be such that we hit 1.0 at 10000 nits. */ const struct fixed31_32 end_value = - dal_fixed31_32_from_int(125); + dc_fixpt_from_int(125); - arr_points[1].slope = dal_fixed31_32_div( - dal_fixed31_32_sub(dal_fixed31_32_one, arr_points[1].y), - dal_fixed31_32_sub(end_value, arr_points[1].x)); + arr_points[1].slope = dc_fixpt_div( + dc_fixpt_sub(dc_fixpt_one, arr_points[1].y), + dc_fixpt_sub(end_value, arr_points[1].x)); } lut_params->hw_points_num = hw_points; @@ -548,16 +548,16 @@ bool cm_helper_translate_curve_to_degamma_hw_format( i = 1; while (i != hw_points + 1) { - if (dal_fixed31_32_lt(rgb_plus_1->red, rgb->red)) + if (dc_fixpt_lt(rgb_plus_1->red, rgb->red)) rgb_plus_1->red = rgb->red; - if (dal_fixed31_32_lt(rgb_plus_1->green, rgb->green)) + if (dc_fixpt_lt(rgb_plus_1->green, rgb->green)) rgb_plus_1->green = rgb->green; - if (dal_fixed31_32_lt(rgb_plus_1->blue, rgb->blue)) + if (dc_fixpt_lt(rgb_plus_1->blue, rgb->blue)) rgb_plus_1->blue = rgb->blue; - rgb->delta_red = dal_fixed31_32_sub(rgb_plus_1->red, rgb->red); - rgb->delta_green = dal_fixed31_32_sub(rgb_plus_1->green, rgb->green); - rgb->delta_blue = dal_fixed31_32_sub(rgb_plus_1->blue, rgb->blue); + rgb->delta_red = dc_fixpt_sub(rgb_plus_1->red, rgb->red); + rgb->delta_green = dc_fixpt_sub(rgb_plus_1->green, rgb->green); + rgb->delta_blue = dc_fixpt_sub(rgb_plus_1->blue, rgb->blue); ++rgb_plus_1; ++rgb; diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp.c index 8c4d9e523331..46a35c7f01df 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp.c @@ -130,7 +130,7 @@ void dpp_set_gamut_remap_bypass(struct dcn10_dpp *dpp) /* Gamut remap in bypass */ } -#define IDENTITY_RATIO(ratio) (dal_fixed31_32_u2d19(ratio) == (1 << 19)) +#define IDENTITY_RATIO(ratio) (dc_fixpt_u2d19(ratio) == (1 << 19)) bool dpp_get_optimal_number_of_taps( @@ -152,6 +152,11 @@ bool dpp_get_optimal_number_of_taps( scl_data->format == PIXEL_FORMAT_FP16) return false; + if (scl_data->viewport.width > scl_data->h_active && + dpp->ctx->dc->debug.max_downscale_src_width != 0 && + scl_data->viewport.width > dpp->ctx->dc->debug.max_downscale_src_width) + return false; + /* TODO: add lb check */ /* No support for programming ratio of 4, drop to 3.99999.. */ diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c index 4f373c97804f..116977eb24e2 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_cm.c @@ -811,13 +811,13 @@ void dpp1_program_input_lut( REG_UPDATE(CM_IGAM_LUT_RW_INDEX, CM_IGAM_LUT_RW_INDEX, 0); for (i = 0; i < gamma->num_entries; i++) { REG_SET(CM_IGAM_LUT_SEQ_COLOR, 0, CM_IGAM_LUT_SEQ_COLOR, - dal_fixed31_32_round( + dc_fixpt_round( gamma->entries.red[i])); REG_SET(CM_IGAM_LUT_SEQ_COLOR, 0, CM_IGAM_LUT_SEQ_COLOR, - dal_fixed31_32_round( + dc_fixpt_round( gamma->entries.green[i])); REG_SET(CM_IGAM_LUT_SEQ_COLOR, 0, CM_IGAM_LUT_SEQ_COLOR, - dal_fixed31_32_round( + dc_fixpt_round( gamma->entries.blue[i])); } // Power off LUT memory diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_dscl.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_dscl.c index 3eb824debf43..4ddd6273d5a5 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_dscl.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_dpp_dscl.c @@ -169,7 +169,7 @@ static enum dscl_mode_sel dpp1_dscl_get_dscl_mode( const struct scaler_data *data, bool dbg_always_scale) { - const long long one = dal_fixed31_32_one.value; + const long long one = dc_fixpt_one.value; if (dpp_base->caps->dscl_data_proc_format == DSCL_DATA_PRCESSING_FIXED_FORMAT) { /* DSCL is processing data in fixed format */ @@ -464,8 +464,8 @@ static enum lb_memory_config dpp1_dscl_find_lb_memory_config(struct dcn10_dpp *d int num_part_y, num_part_c; int vtaps = scl_data->taps.v_taps; int vtaps_c = scl_data->taps.v_taps_c; - int ceil_vratio = dal_fixed31_32_ceil(scl_data->ratios.vert); - int ceil_vratio_c = dal_fixed31_32_ceil(scl_data->ratios.vert_c); + int ceil_vratio = dc_fixpt_ceil(scl_data->ratios.vert); + int ceil_vratio_c = dc_fixpt_ceil(scl_data->ratios.vert_c); enum lb_memory_config mem_cfg = LB_MEMORY_CONFIG_0; if (dpp->base.ctx->dc->debug.use_max_lb) @@ -565,52 +565,52 @@ static void dpp1_dscl_set_manual_ratio_init( uint32_t init_int = 0; REG_SET(SCL_HORZ_FILTER_SCALE_RATIO, 0, - SCL_H_SCALE_RATIO, dal_fixed31_32_u2d19(data->ratios.horz) << 5); + SCL_H_SCALE_RATIO, dc_fixpt_u2d19(data->ratios.horz) << 5); REG_SET(SCL_VERT_FILTER_SCALE_RATIO, 0, - SCL_V_SCALE_RATIO, dal_fixed31_32_u2d19(data->ratios.vert) << 5); + SCL_V_SCALE_RATIO, dc_fixpt_u2d19(data->ratios.vert) << 5); REG_SET(SCL_HORZ_FILTER_SCALE_RATIO_C, 0, - SCL_H_SCALE_RATIO_C, dal_fixed31_32_u2d19(data->ratios.horz_c) << 5); + SCL_H_SCALE_RATIO_C, dc_fixpt_u2d19(data->ratios.horz_c) << 5); REG_SET(SCL_VERT_FILTER_SCALE_RATIO_C, 0, - SCL_V_SCALE_RATIO_C, dal_fixed31_32_u2d19(data->ratios.vert_c) << 5); + SCL_V_SCALE_RATIO_C, dc_fixpt_u2d19(data->ratios.vert_c) << 5); /* * 0.24 format for fraction, first five bits zeroed */ - init_frac = dal_fixed31_32_u0d19(data->inits.h) << 5; - init_int = dal_fixed31_32_floor(data->inits.h); + init_frac = dc_fixpt_u0d19(data->inits.h) << 5; + init_int = dc_fixpt_floor(data->inits.h); REG_SET_2(SCL_HORZ_FILTER_INIT, 0, SCL_H_INIT_FRAC, init_frac, SCL_H_INIT_INT, init_int); - init_frac = dal_fixed31_32_u0d19(data->inits.h_c) << 5; - init_int = dal_fixed31_32_floor(data->inits.h_c); + init_frac = dc_fixpt_u0d19(data->inits.h_c) << 5; + init_int = dc_fixpt_floor(data->inits.h_c); REG_SET_2(SCL_HORZ_FILTER_INIT_C, 0, SCL_H_INIT_FRAC_C, init_frac, SCL_H_INIT_INT_C, init_int); - init_frac = dal_fixed31_32_u0d19(data->inits.v) << 5; - init_int = dal_fixed31_32_floor(data->inits.v); + init_frac = dc_fixpt_u0d19(data->inits.v) << 5; + init_int = dc_fixpt_floor(data->inits.v); REG_SET_2(SCL_VERT_FILTER_INIT, 0, SCL_V_INIT_FRAC, init_frac, SCL_V_INIT_INT, init_int); - init_frac = dal_fixed31_32_u0d19(data->inits.v_bot) << 5; - init_int = dal_fixed31_32_floor(data->inits.v_bot); + init_frac = dc_fixpt_u0d19(data->inits.v_bot) << 5; + init_int = dc_fixpt_floor(data->inits.v_bot); REG_SET_2(SCL_VERT_FILTER_INIT_BOT, 0, SCL_V_INIT_FRAC_BOT, init_frac, SCL_V_INIT_INT_BOT, init_int); - init_frac = dal_fixed31_32_u0d19(data->inits.v_c) << 5; - init_int = dal_fixed31_32_floor(data->inits.v_c); + init_frac = dc_fixpt_u0d19(data->inits.v_c) << 5; + init_int = dc_fixpt_floor(data->inits.v_c); REG_SET_2(SCL_VERT_FILTER_INIT_C, 0, SCL_V_INIT_FRAC_C, init_frac, SCL_V_INIT_INT_C, init_int); - init_frac = dal_fixed31_32_u0d19(data->inits.v_c_bot) << 5; - init_int = dal_fixed31_32_floor(data->inits.v_c_bot); + init_frac = dc_fixpt_u0d19(data->inits.v_c_bot) << 5; + init_int = dc_fixpt_floor(data->inits.v_c_bot); REG_SET_2(SCL_VERT_FILTER_INIT_BOT_C, 0, SCL_V_INIT_FRAC_BOT_C, init_frac, SCL_V_INIT_INT_BOT_C, init_int); diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c index b9fb14a3224b..943143efbb82 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.c @@ -476,6 +476,14 @@ void hubbub1_toggle_watermark_change_req(struct hubbub *hubbub) DCHUBBUB_ARB_WATERMARK_CHANGE_REQUEST, watermark_change_req); } +void hubbub1_soft_reset(struct hubbub *hubbub, bool reset) +{ + uint32_t reset_en = reset ? 1 : 0; + + REG_UPDATE(DCHUBBUB_SOFT_RESET, + DCHUBBUB_GLOBAL_SOFT_RESET, reset_en); +} + static bool hubbub1_dcc_support_swizzle( enum swizzle_mode_values swizzle, unsigned int bytes_per_element, diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.h index f479f54e5bb2..6315a0e6b0d6 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.h +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.h @@ -48,7 +48,8 @@ SR(DCHUBBUB_ARB_DF_REQ_OUTSTAND),\ SR(DCHUBBUB_GLOBAL_TIMER_CNTL), \ SR(DCHUBBUB_TEST_DEBUG_INDEX), \ - SR(DCHUBBUB_TEST_DEBUG_DATA) + SR(DCHUBBUB_TEST_DEBUG_DATA),\ + SR(DCHUBBUB_SOFT_RESET) #define HUBBUB_SR_WATERMARK_REG_LIST()\ SR(DCHUBBUB_ARB_ALLOW_SR_ENTER_WATERMARK_A),\ @@ -105,6 +106,7 @@ struct dcn_hubbub_registers { uint32_t DCHUBBUB_SDPIF_AGP_BOT; uint32_t DCHUBBUB_SDPIF_AGP_TOP; uint32_t DCHUBBUB_CRC_CTRL; + uint32_t DCHUBBUB_SOFT_RESET; }; /* set field name */ @@ -114,6 +116,7 @@ struct dcn_hubbub_registers { #define HUBBUB_MASK_SH_LIST_DCN(mask_sh)\ HUBBUB_SF(DCHUBBUB_GLOBAL_TIMER_CNTL, DCHUBBUB_GLOBAL_TIMER_ENABLE, mask_sh), \ + HUBBUB_SF(DCHUBBUB_SOFT_RESET, DCHUBBUB_GLOBAL_SOFT_RESET, mask_sh), \ HUBBUB_SF(DCHUBBUB_ARB_WATERMARK_CHANGE_CNTL, DCHUBBUB_ARB_WATERMARK_CHANGE_REQUEST, mask_sh), \ HUBBUB_SF(DCHUBBUB_ARB_WATERMARK_CHANGE_CNTL, DCHUBBUB_ARB_WATERMARK_CHANGE_DONE_INTERRUPT_DISABLE, mask_sh), \ HUBBUB_SF(DCHUBBUB_ARB_DRAM_STATE_CNTL, DCHUBBUB_ARB_ALLOW_SELF_REFRESH_FORCE_VALUE, mask_sh), \ @@ -143,6 +146,7 @@ struct dcn_hubbub_registers { type DCHUBBUB_ARB_SAT_LEVEL;\ type DCHUBBUB_ARB_MIN_REQ_OUTSTAND;\ type DCHUBBUB_GLOBAL_TIMER_REFDIV;\ + type DCHUBBUB_GLOBAL_SOFT_RESET; \ type SDPIF_FB_TOP;\ type SDPIF_FB_BASE;\ type SDPIF_FB_OFFSET;\ @@ -201,6 +205,7 @@ void hubbub1_toggle_watermark_change_req( void hubbub1_wm_read_state(struct hubbub *hubbub, struct dcn_hubbub_wm *wm); +void hubbub1_soft_reset(struct hubbub *hubbub, bool reset); void hubbub1_construct(struct hubbub *hubbub, struct dc_context *ctx, const struct dcn_hubbub_registers *hubbub_regs, diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.c index 0cbc83edd37f..d2ab78b35a7a 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.c @@ -78,6 +78,27 @@ static void hubp1_disconnect(struct hubp *hubp) CURSOR_ENABLE, 0); } +static void hubp1_disable_control(struct hubp *hubp, bool disable_hubp) +{ + struct dcn10_hubp *hubp1 = TO_DCN10_HUBP(hubp); + uint32_t disable = disable_hubp ? 1 : 0; + + REG_UPDATE(DCHUBP_CNTL, + HUBP_DISABLE, disable); +} + +static unsigned int hubp1_get_underflow_status(struct hubp *hubp) +{ + uint32_t hubp_underflow = 0; + struct dcn10_hubp *hubp1 = TO_DCN10_HUBP(hubp); + + REG_GET(DCHUBP_CNTL, + HUBP_UNDERFLOW_STATUS, + &hubp_underflow); + + return hubp_underflow; +} + static void hubp1_set_hubp_blank_en(struct hubp *hubp, bool blank) { struct dcn10_hubp *hubp1 = TO_DCN10_HUBP(hubp); @@ -1054,8 +1075,8 @@ void hubp1_cursor_set_position( ASSERT(param->h_scale_ratio.value); if (param->h_scale_ratio.value) - dst_x_offset = dal_fixed31_32_floor(dal_fixed31_32_div( - dal_fixed31_32_from_int(dst_x_offset), + dst_x_offset = dc_fixpt_floor(dc_fixpt_div( + dc_fixpt_from_int(dst_x_offset), param->h_scale_ratio)); if (src_x_offset >= (int)param->viewport_width) @@ -1117,6 +1138,9 @@ static struct hubp_funcs dcn10_hubp_funcs = { .hubp_clk_cntl = hubp1_clk_cntl, .hubp_vtg_sel = hubp1_vtg_sel, .hubp_read_state = hubp1_read_state, + .hubp_disable_control = hubp1_disable_control, + .hubp_get_underflow_status = hubp1_get_underflow_status, + }; /*****************************************/ diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.h index fe9b8c4a91ca..af384034398f 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubp.h @@ -253,6 +253,7 @@ HUBP_SF(HUBP0_DCHUBP_CNTL, HUBP_UNDERFLOW_STATUS, mask_sh),\ HUBP_SF(HUBP0_DCHUBP_CNTL, HUBP_NO_OUTSTANDING_REQ, mask_sh),\ HUBP_SF(HUBP0_DCHUBP_CNTL, HUBP_VTG_SEL, mask_sh),\ + HUBP_SF(HUBP0_DCHUBP_CNTL, HUBP_DISABLE, mask_sh),\ HUBP_SF(HUBP0_DCSURF_ADDR_CONFIG, NUM_PIPES, mask_sh),\ HUBP_SF(HUBP0_DCSURF_ADDR_CONFIG, NUM_BANKS, mask_sh),\ HUBP_SF(HUBP0_DCSURF_ADDR_CONFIG, PIPE_INTERLEAVE, mask_sh),\ @@ -421,6 +422,7 @@ #define DCN_HUBP_REG_FIELD_LIST(type) \ type HUBP_BLANK_EN;\ + type HUBP_DISABLE;\ type HUBP_TTU_DISABLE;\ type HUBP_NO_OUTSTANDING_REQ;\ type HUBP_VTG_SEL;\ @@ -723,4 +725,5 @@ void hubp1_read_state(struct hubp *hubp); enum cursor_pitch hubp1_get_cursor_pitch(unsigned int pitch); + #endif diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c index 572fa601a0eb..f8e0576af6e0 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c @@ -127,24 +127,26 @@ static void dcn10_log_hubp_states(struct dc *dc) hubp->funcs->hubp_read_state(hubp); - DTN_INFO("[%2d]: %5xh %6xh %5d %6d %2xh %2xh %6xh" - " %6d %8d %7d %8xh", - hubp->inst, - s->pixel_format, - s->inuse_addr_hi, - s->viewport_width, - s->viewport_height, - s->rotation_angle, - s->h_mirror_en, - s->sw_mode, - s->dcc_en, - s->blank_en, - s->ttu_disable, - s->underflow_status); - DTN_INFO_MICRO_SEC(s->min_ttu_vblank); - DTN_INFO_MICRO_SEC(s->qos_level_low_wm); - DTN_INFO_MICRO_SEC(s->qos_level_high_wm); - DTN_INFO("\n"); + if (!s->blank_en) { + DTN_INFO("[%2d]: %5xh %6xh %5d %6d %2xh %2xh %6xh" + " %6d %8d %7d %8xh", + hubp->inst, + s->pixel_format, + s->inuse_addr_hi, + s->viewport_width, + s->viewport_height, + s->rotation_angle, + s->h_mirror_en, + s->sw_mode, + s->dcc_en, + s->blank_en, + s->ttu_disable, + s->underflow_status); + DTN_INFO_MICRO_SEC(s->min_ttu_vblank); + DTN_INFO_MICRO_SEC(s->qos_level_low_wm); + DTN_INFO_MICRO_SEC(s->qos_level_high_wm); + DTN_INFO("\n"); + } } DTN_INFO("\n=========RQ========\n"); @@ -155,16 +157,17 @@ static void dcn10_log_hubp_states(struct dc *dc) struct dcn_hubp_state *s = &(TO_DCN10_HUBP(pool->hubps[i])->state); struct _vcs_dpi_display_rq_regs_st *rq_regs = &s->rq_regs; - DTN_INFO("[%2d]: %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh\n", - i, rq_regs->drq_expansion_mode, rq_regs->prq_expansion_mode, rq_regs->mrq_expansion_mode, - rq_regs->crq_expansion_mode, rq_regs->plane1_base_address, rq_regs->rq_regs_l.chunk_size, - rq_regs->rq_regs_l.min_chunk_size, rq_regs->rq_regs_l.meta_chunk_size, - rq_regs->rq_regs_l.min_meta_chunk_size, rq_regs->rq_regs_l.dpte_group_size, - rq_regs->rq_regs_l.mpte_group_size, rq_regs->rq_regs_l.swath_height, - rq_regs->rq_regs_l.pte_row_height_linear, rq_regs->rq_regs_c.chunk_size, rq_regs->rq_regs_c.min_chunk_size, - rq_regs->rq_regs_c.meta_chunk_size, rq_regs->rq_regs_c.min_meta_chunk_size, - rq_regs->rq_regs_c.dpte_group_size, rq_regs->rq_regs_c.mpte_group_size, - rq_regs->rq_regs_c.swath_height, rq_regs->rq_regs_c.pte_row_height_linear); + if (!s->blank_en) + DTN_INFO("[%2d]: %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh\n", + pool->hubps[i]->inst, rq_regs->drq_expansion_mode, rq_regs->prq_expansion_mode, rq_regs->mrq_expansion_mode, + rq_regs->crq_expansion_mode, rq_regs->plane1_base_address, rq_regs->rq_regs_l.chunk_size, + rq_regs->rq_regs_l.min_chunk_size, rq_regs->rq_regs_l.meta_chunk_size, + rq_regs->rq_regs_l.min_meta_chunk_size, rq_regs->rq_regs_l.dpte_group_size, + rq_regs->rq_regs_l.mpte_group_size, rq_regs->rq_regs_l.swath_height, + rq_regs->rq_regs_l.pte_row_height_linear, rq_regs->rq_regs_c.chunk_size, rq_regs->rq_regs_c.min_chunk_size, + rq_regs->rq_regs_c.meta_chunk_size, rq_regs->rq_regs_c.min_meta_chunk_size, + rq_regs->rq_regs_c.dpte_group_size, rq_regs->rq_regs_c.mpte_group_size, + rq_regs->rq_regs_c.swath_height, rq_regs->rq_regs_c.pte_row_height_linear); } DTN_INFO("========DLG========\n"); @@ -179,27 +182,28 @@ static void dcn10_log_hubp_states(struct dc *dc) struct dcn_hubp_state *s = &(TO_DCN10_HUBP(pool->hubps[i])->state); struct _vcs_dpi_display_dlg_regs_st *dlg_regs = &s->dlg_attr; - DTN_INFO("[%2d]: %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh" - "% 8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh" - " %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh\n", - i, dlg_regs->refcyc_h_blank_end, dlg_regs->dlg_vblank_end, dlg_regs->min_dst_y_next_start, - dlg_regs->refcyc_per_htotal, dlg_regs->refcyc_x_after_scaler, dlg_regs->dst_y_after_scaler, - dlg_regs->dst_y_prefetch, dlg_regs->dst_y_per_vm_vblank, dlg_regs->dst_y_per_row_vblank, - dlg_regs->dst_y_per_vm_flip, dlg_regs->dst_y_per_row_flip, dlg_regs->ref_freq_to_pix_freq, - dlg_regs->vratio_prefetch, dlg_regs->vratio_prefetch_c, dlg_regs->refcyc_per_pte_group_vblank_l, - dlg_regs->refcyc_per_pte_group_vblank_c, dlg_regs->refcyc_per_meta_chunk_vblank_l, - dlg_regs->refcyc_per_meta_chunk_vblank_c, dlg_regs->refcyc_per_pte_group_flip_l, - dlg_regs->refcyc_per_pte_group_flip_c, dlg_regs->refcyc_per_meta_chunk_flip_l, - dlg_regs->refcyc_per_meta_chunk_flip_c, dlg_regs->dst_y_per_pte_row_nom_l, - dlg_regs->dst_y_per_pte_row_nom_c, dlg_regs->refcyc_per_pte_group_nom_l, - dlg_regs->refcyc_per_pte_group_nom_c, dlg_regs->dst_y_per_meta_row_nom_l, - dlg_regs->dst_y_per_meta_row_nom_c, dlg_regs->refcyc_per_meta_chunk_nom_l, - dlg_regs->refcyc_per_meta_chunk_nom_c, dlg_regs->refcyc_per_line_delivery_pre_l, - dlg_regs->refcyc_per_line_delivery_pre_c, dlg_regs->refcyc_per_line_delivery_l, - dlg_regs->refcyc_per_line_delivery_c, dlg_regs->chunk_hdl_adjust_cur0, dlg_regs->dst_y_offset_cur1, - dlg_regs->chunk_hdl_adjust_cur1, dlg_regs->vready_after_vcount0, dlg_regs->dst_y_delta_drq_limit, - dlg_regs->xfc_reg_transfer_delay, dlg_regs->xfc_reg_precharge_delay, - dlg_regs->xfc_reg_remote_surface_flip_latency); + if (!s->blank_en) + DTN_INFO("[%2d]: %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh" + "% 8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh" + " %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh\n", + pool->hubps[i]->inst, dlg_regs->refcyc_h_blank_end, dlg_regs->dlg_vblank_end, dlg_regs->min_dst_y_next_start, + dlg_regs->refcyc_per_htotal, dlg_regs->refcyc_x_after_scaler, dlg_regs->dst_y_after_scaler, + dlg_regs->dst_y_prefetch, dlg_regs->dst_y_per_vm_vblank, dlg_regs->dst_y_per_row_vblank, + dlg_regs->dst_y_per_vm_flip, dlg_regs->dst_y_per_row_flip, dlg_regs->ref_freq_to_pix_freq, + dlg_regs->vratio_prefetch, dlg_regs->vratio_prefetch_c, dlg_regs->refcyc_per_pte_group_vblank_l, + dlg_regs->refcyc_per_pte_group_vblank_c, dlg_regs->refcyc_per_meta_chunk_vblank_l, + dlg_regs->refcyc_per_meta_chunk_vblank_c, dlg_regs->refcyc_per_pte_group_flip_l, + dlg_regs->refcyc_per_pte_group_flip_c, dlg_regs->refcyc_per_meta_chunk_flip_l, + dlg_regs->refcyc_per_meta_chunk_flip_c, dlg_regs->dst_y_per_pte_row_nom_l, + dlg_regs->dst_y_per_pte_row_nom_c, dlg_regs->refcyc_per_pte_group_nom_l, + dlg_regs->refcyc_per_pte_group_nom_c, dlg_regs->dst_y_per_meta_row_nom_l, + dlg_regs->dst_y_per_meta_row_nom_c, dlg_regs->refcyc_per_meta_chunk_nom_l, + dlg_regs->refcyc_per_meta_chunk_nom_c, dlg_regs->refcyc_per_line_delivery_pre_l, + dlg_regs->refcyc_per_line_delivery_pre_c, dlg_regs->refcyc_per_line_delivery_l, + dlg_regs->refcyc_per_line_delivery_c, dlg_regs->chunk_hdl_adjust_cur0, dlg_regs->dst_y_offset_cur1, + dlg_regs->chunk_hdl_adjust_cur1, dlg_regs->vready_after_vcount0, dlg_regs->dst_y_delta_drq_limit, + dlg_regs->xfc_reg_transfer_delay, dlg_regs->xfc_reg_precharge_delay, + dlg_regs->xfc_reg_remote_surface_flip_latency); } DTN_INFO("========TTU========\n"); @@ -210,14 +214,15 @@ static void dcn10_log_hubp_states(struct dc *dc) struct dcn_hubp_state *s = &(TO_DCN10_HUBP(pool->hubps[i])->state); struct _vcs_dpi_display_ttu_regs_st *ttu_regs = &s->ttu_attr; - DTN_INFO("[%2d]: %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh\n", - i, ttu_regs->qos_level_low_wm, ttu_regs->qos_level_high_wm, ttu_regs->min_ttu_vblank, - ttu_regs->qos_level_flip, ttu_regs->refcyc_per_req_delivery_pre_l, ttu_regs->refcyc_per_req_delivery_l, - ttu_regs->refcyc_per_req_delivery_pre_c, ttu_regs->refcyc_per_req_delivery_c, ttu_regs->refcyc_per_req_delivery_cur0, - ttu_regs->refcyc_per_req_delivery_pre_cur0, ttu_regs->refcyc_per_req_delivery_cur1, - ttu_regs->refcyc_per_req_delivery_pre_cur1, ttu_regs->qos_level_fixed_l, ttu_regs->qos_ramp_disable_l, - ttu_regs->qos_level_fixed_c, ttu_regs->qos_ramp_disable_c, ttu_regs->qos_level_fixed_cur0, - ttu_regs->qos_ramp_disable_cur0, ttu_regs->qos_level_fixed_cur1, ttu_regs->qos_ramp_disable_cur1); + if (!s->blank_en) + DTN_INFO("[%2d]: %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh %8xh\n", + pool->hubps[i]->inst, ttu_regs->qos_level_low_wm, ttu_regs->qos_level_high_wm, ttu_regs->min_ttu_vblank, + ttu_regs->qos_level_flip, ttu_regs->refcyc_per_req_delivery_pre_l, ttu_regs->refcyc_per_req_delivery_l, + ttu_regs->refcyc_per_req_delivery_pre_c, ttu_regs->refcyc_per_req_delivery_c, ttu_regs->refcyc_per_req_delivery_cur0, + ttu_regs->refcyc_per_req_delivery_pre_cur0, ttu_regs->refcyc_per_req_delivery_cur1, + ttu_regs->refcyc_per_req_delivery_pre_cur1, ttu_regs->qos_level_fixed_l, ttu_regs->qos_ramp_disable_l, + ttu_regs->qos_level_fixed_c, ttu_regs->qos_ramp_disable_c, ttu_regs->qos_level_fixed_cur0, + ttu_regs->qos_ramp_disable_cur0, ttu_regs->qos_level_fixed_cur1, ttu_regs->qos_ramp_disable_cur1); } DTN_INFO("\n"); } @@ -321,6 +326,12 @@ void dcn10_log_hw_state(struct dc *dc) s.h_total, s.v_total, s.underflow_occurred_status); + + // Clear underflow for debug purposes + // We want to keep underflow sticky bit on for the longevity tests outside of test environment. + // This function is called only from Windows or Diags test environment, hence it's safe to clear + // it from here without affecting the original intent. + tg->funcs->clear_optc_underflow(tg); } DTN_INFO("\n"); @@ -747,6 +758,90 @@ static void reset_back_end_for_pipe( pipe_ctx->pipe_idx, pipe_ctx->stream_res.tg->inst); } +static bool dcn10_hw_wa_force_recovery(struct dc *dc) +{ + struct hubp *hubp ; + unsigned int i; + bool need_recover = true; + + if (!dc->debug.recovery_enabled) + return false; + + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = + &dc->current_state->res_ctx.pipe_ctx[i]; + if (pipe_ctx != NULL) { + hubp = pipe_ctx->plane_res.hubp; + if (hubp != NULL) { + if (hubp->funcs->hubp_get_underflow_status(hubp) != 0) { + /* one pipe underflow, we will reset all the pipes*/ + need_recover = true; + } + } + } + } + if (!need_recover) + return false; + /* + DCHUBP_CNTL:HUBP_BLANK_EN=1 + DCHUBBUB_SOFT_RESET:DCHUBBUB_GLOBAL_SOFT_RESET=1 + DCHUBP_CNTL:HUBP_DISABLE=1 + DCHUBP_CNTL:HUBP_DISABLE=0 + DCHUBBUB_SOFT_RESET:DCHUBBUB_GLOBAL_SOFT_RESET=0 + DCSURF_PRIMARY_SURFACE_ADDRESS + DCHUBP_CNTL:HUBP_BLANK_EN=0 + */ + + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = + &dc->current_state->res_ctx.pipe_ctx[i]; + if (pipe_ctx != NULL) { + hubp = pipe_ctx->plane_res.hubp; + /*DCHUBP_CNTL:HUBP_BLANK_EN=1*/ + if (hubp != NULL) + hubp->funcs->set_hubp_blank_en(hubp, true); + } + } + /*DCHUBBUB_SOFT_RESET:DCHUBBUB_GLOBAL_SOFT_RESET=1*/ + hubbub1_soft_reset(dc->res_pool->hubbub, true); + + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = + &dc->current_state->res_ctx.pipe_ctx[i]; + if (pipe_ctx != NULL) { + hubp = pipe_ctx->plane_res.hubp; + /*DCHUBP_CNTL:HUBP_DISABLE=1*/ + if (hubp != NULL) + hubp->funcs->hubp_disable_control(hubp, true); + } + } + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = + &dc->current_state->res_ctx.pipe_ctx[i]; + if (pipe_ctx != NULL) { + hubp = pipe_ctx->plane_res.hubp; + /*DCHUBP_CNTL:HUBP_DISABLE=0*/ + if (hubp != NULL) + hubp->funcs->hubp_disable_control(hubp, true); + } + } + /*DCHUBBUB_SOFT_RESET:DCHUBBUB_GLOBAL_SOFT_RESET=0*/ + hubbub1_soft_reset(dc->res_pool->hubbub, false); + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = + &dc->current_state->res_ctx.pipe_ctx[i]; + if (pipe_ctx != NULL) { + hubp = pipe_ctx->plane_res.hubp; + /*DCHUBP_CNTL:HUBP_BLANK_EN=0*/ + if (hubp != NULL) + hubp->funcs->set_hubp_blank_en(hubp, true); + } + } + return true; + +} + + static void dcn10_verify_allow_pstate_change_high(struct dc *dc) { static bool should_log_hw_state; /* prevent hw state log by default */ @@ -755,13 +850,17 @@ static void dcn10_verify_allow_pstate_change_high(struct dc *dc) if (should_log_hw_state) { dcn10_log_hw_state(dc); } - BREAK_TO_DEBUGGER(); + if (dcn10_hw_wa_force_recovery(dc)) { + /*check again*/ + if (!hubbub1_verify_allow_pstate_change_high(dc->res_pool->hubbub)) + BREAK_TO_DEBUGGER(); + } } } /* trigger HW to start disconnect plane from stream on the next vsync */ -static void plane_atomic_disconnect(struct dc *dc, struct pipe_ctx *pipe_ctx) +void hwss1_plane_atomic_disconnect(struct dc *dc, struct pipe_ctx *pipe_ctx) { struct hubp *hubp = pipe_ctx->plane_res.hubp; int dpp_id = pipe_ctx->plane_res.dpp->inst; @@ -944,7 +1043,7 @@ static void dcn10_init_hw(struct dc *dc) dc->res_pool->opps[i]->mpcc_disconnect_pending[pipe_ctx->plane_res.mpcc_inst] = true; pipe_ctx->stream_res.opp = dc->res_pool->opps[i]; - plane_atomic_disconnect(dc, pipe_ctx); + hwss1_plane_atomic_disconnect(dc, pipe_ctx); } for (i = 0; i < dc->res_pool->pipe_count; i++) { @@ -1685,22 +1784,22 @@ static uint16_t fixed_point_to_int_frac( uint16_t result; - uint16_t d = (uint16_t)dal_fixed31_32_floor( - dal_fixed31_32_abs( + uint16_t d = (uint16_t)dc_fixpt_floor( + dc_fixpt_abs( arg)); if (d <= (uint16_t)(1 << integer_bits) - (1 / (uint16_t)divisor)) - numerator = (uint16_t)dal_fixed31_32_floor( - dal_fixed31_32_mul_int( + numerator = (uint16_t)dc_fixpt_floor( + dc_fixpt_mul_int( arg, divisor)); else { - numerator = dal_fixed31_32_floor( - dal_fixed31_32_sub( - dal_fixed31_32_from_int( + numerator = dc_fixpt_floor( + dc_fixpt_sub( + dc_fixpt_from_int( 1LL << integer_bits), - dal_fixed31_32_recip( - dal_fixed31_32_from_int( + dc_fixpt_recip( + dc_fixpt_from_int( divisor)))); } @@ -1710,8 +1809,8 @@ static uint16_t fixed_point_to_int_frac( result = (uint16_t)( (1 << (integer_bits + fractional_bits + 1)) + numerator); - if ((result != 0) && dal_fixed31_32_lt( - arg, dal_fixed31_32_zero)) + if ((result != 0) && dc_fixpt_lt( + arg, dc_fixpt_zero)) result |= 1 << (integer_bits + fractional_bits); return result; @@ -1725,8 +1824,8 @@ void build_prescale_params(struct dc_bias_and_scale *bias_and_scale, && plane_state->input_csc_color_matrix.enable_adjustment && plane_state->coeff_reduction_factor.value != 0) { bias_and_scale->scale_blue = fixed_point_to_int_frac( - dal_fixed31_32_mul(plane_state->coeff_reduction_factor, - dal_fixed31_32_from_fraction(256, 255)), + dc_fixpt_mul(plane_state->coeff_reduction_factor, + dc_fixpt_from_fraction(256, 255)), 2, 13); bias_and_scale->scale_red = bias_and_scale->scale_blue; @@ -1995,7 +2094,7 @@ static void dcn10_blank_pixel_data( static void set_hdr_multiplier(struct pipe_ctx *pipe_ctx) { - struct fixed31_32 multiplier = dal_fixed31_32_from_fraction( + struct fixed31_32 multiplier = dc_fixpt_from_fraction( pipe_ctx->plane_state->sdr_white_level, 80); uint32_t hw_mult = 0x1f000; // 1.0 default multiplier struct custom_float_format fmt; @@ -2179,7 +2278,7 @@ static void dcn10_apply_ctx_for_surface( old_pipe_ctx->plane_state && old_pipe_ctx->stream_res.tg == tg) { - plane_atomic_disconnect(dc, old_pipe_ctx); + hwss1_plane_atomic_disconnect(dc, old_pipe_ctx); removed_pipe[i] = true; DC_LOG_DC( @@ -2487,15 +2586,6 @@ static void set_static_screen_control(struct pipe_ctx **pipe_ctx, set_static_screen_control(pipe_ctx[i]->stream_res.tg, value); } -static void set_plane_config( - const struct dc *dc, - struct pipe_ctx *pipe_ctx, - struct resource_context *res_ctx) -{ - /* TODO */ - program_gamut_remap(pipe_ctx); -} - static void dcn10_config_stereo_parameters( struct dc_stream_state *stream, struct crtc_stereo_flags *flags) { @@ -2673,7 +2763,6 @@ static const struct hw_sequencer_funcs dcn10_funcs = { .init_hw = dcn10_init_hw, .apply_ctx_to_hw = dce110_apply_ctx_to_hw, .apply_ctx_for_surface = dcn10_apply_ctx_for_surface, - .set_plane_config = set_plane_config, .update_plane_addr = dcn10_update_plane_addr, .update_dchub = dcn10_update_dchub, .update_pending_status = dcn10_update_pending_status, diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h index 6c526b5095d9..44f734b73f9e 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h @@ -37,4 +37,6 @@ extern void fill_display_configs( bool is_rgb_cspace(enum dc_color_space output_color_space); +void hwss1_plane_atomic_disconnect(struct dc *dc, struct pipe_ctx *pipe_ctx); + #endif /* __DC_HWSS_DCN10_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.c new file mode 100644 index 000000000000..21fa40ac0786 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.c @@ -0,0 +1,1362 @@ +/* + * Copyright 2012-15 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#include "reg_helper.h" + +#include "core_types.h" +#include "link_encoder.h" +#include "dcn10_link_encoder.h" +#include "stream_encoder.h" +#include "i2caux_interface.h" +#include "dc_bios_types.h" + +#include "gpio_service_interface.h" + +#define CTX \ + enc10->base.ctx +#define DC_LOGGER \ + enc10->base.ctx->logger + +#define REG(reg)\ + (enc10->link_regs->reg) + +#undef FN +#define FN(reg_name, field_name) \ + enc10->link_shift->field_name, enc10->link_mask->field_name + + +/* + * @brief + * Trigger Source Select + * ASIC-dependent, actual values for register programming + */ +#define DCN10_DIG_FE_SOURCE_SELECT_INVALID 0x0 +#define DCN10_DIG_FE_SOURCE_SELECT_DIGA 0x1 +#define DCN10_DIG_FE_SOURCE_SELECT_DIGB 0x2 +#define DCN10_DIG_FE_SOURCE_SELECT_DIGC 0x4 +#define DCN10_DIG_FE_SOURCE_SELECT_DIGD 0x08 +#define DCN10_DIG_FE_SOURCE_SELECT_DIGE 0x10 +#define DCN10_DIG_FE_SOURCE_SELECT_DIGF 0x20 +#define DCN10_DIG_FE_SOURCE_SELECT_DIGG 0x40 + +enum { + DP_MST_UPDATE_MAX_RETRY = 50 +}; + + + +static void aux_initialize(struct dcn10_link_encoder *enc10); + + +static const struct link_encoder_funcs dcn10_lnk_enc_funcs = { + .validate_output_with_stream = + dcn10_link_encoder_validate_output_with_stream, + .hw_init = dcn10_link_encoder_hw_init, + .setup = dcn10_link_encoder_setup, + .enable_tmds_output = dcn10_link_encoder_enable_tmds_output, + .enable_dp_output = dcn10_link_encoder_enable_dp_output, + .enable_dp_mst_output = dcn10_link_encoder_enable_dp_mst_output, + .disable_output = dcn10_link_encoder_disable_output, + .dp_set_lane_settings = dcn10_link_encoder_dp_set_lane_settings, + .dp_set_phy_pattern = dcn10_link_encoder_dp_set_phy_pattern, + .update_mst_stream_allocation_table = + dcn10_link_encoder_update_mst_stream_allocation_table, + .psr_program_dp_dphy_fast_training = + dcn10_psr_program_dp_dphy_fast_training, + .psr_program_secondary_packet = dcn10_psr_program_secondary_packet, + .connect_dig_be_to_fe = dcn10_link_encoder_connect_dig_be_to_fe, + .enable_hpd = dcn10_link_encoder_enable_hpd, + .disable_hpd = dcn10_link_encoder_disable_hpd, + .is_dig_enabled = dcn10_is_dig_enabled, + .destroy = dcn10_link_encoder_destroy +}; + +static enum bp_result link_transmitter_control( + struct dcn10_link_encoder *enc10, + struct bp_transmitter_control *cntl) +{ + enum bp_result result; + struct dc_bios *bp = enc10->base.ctx->dc_bios; + + result = bp->funcs->transmitter_control(bp, cntl); + + return result; +} + +static void enable_phy_bypass_mode( + struct dcn10_link_encoder *enc10, + bool enable) +{ + /* This register resides in DP back end block; + * transmitter is used for the offset + */ + REG_UPDATE(DP_DPHY_CNTL, DPHY_BYPASS, enable); + +} + +static void disable_prbs_symbols( + struct dcn10_link_encoder *enc10, + bool disable) +{ + /* This register resides in DP back end block; + * transmitter is used for the offset + */ + REG_UPDATE_4(DP_DPHY_CNTL, + DPHY_ATEST_SEL_LANE0, disable, + DPHY_ATEST_SEL_LANE1, disable, + DPHY_ATEST_SEL_LANE2, disable, + DPHY_ATEST_SEL_LANE3, disable); +} + +static void disable_prbs_mode( + struct dcn10_link_encoder *enc10) +{ + REG_UPDATE(DP_DPHY_PRBS_CNTL, DPHY_PRBS_EN, 0); +} + +static void program_pattern_symbols( + struct dcn10_link_encoder *enc10, + uint16_t pattern_symbols[8]) +{ + /* This register resides in DP back end block; + * transmitter is used for the offset + */ + REG_SET_3(DP_DPHY_SYM0, 0, + DPHY_SYM1, pattern_symbols[0], + DPHY_SYM2, pattern_symbols[1], + DPHY_SYM3, pattern_symbols[2]); + + /* This register resides in DP back end block; + * transmitter is used for the offset + */ + REG_SET_3(DP_DPHY_SYM1, 0, + DPHY_SYM4, pattern_symbols[3], + DPHY_SYM5, pattern_symbols[4], + DPHY_SYM6, pattern_symbols[5]); + + /* This register resides in DP back end block; + * transmitter is used for the offset + */ + REG_SET_2(DP_DPHY_SYM2, 0, + DPHY_SYM7, pattern_symbols[6], + DPHY_SYM8, pattern_symbols[7]); +} + +static void set_dp_phy_pattern_d102( + struct dcn10_link_encoder *enc10) +{ + /* Disable PHY Bypass mode to setup the test pattern */ + enable_phy_bypass_mode(enc10, false); + + /* For 10-bit PRBS or debug symbols + * please use the following sequence: + * + * Enable debug symbols on the lanes + */ + disable_prbs_symbols(enc10, true); + + /* Disable PRBS mode */ + disable_prbs_mode(enc10); + + /* Program debug symbols to be output */ + { + uint16_t pattern_symbols[8] = { + 0x2AA, 0x2AA, 0x2AA, 0x2AA, + 0x2AA, 0x2AA, 0x2AA, 0x2AA + }; + + program_pattern_symbols(enc10, pattern_symbols); + } + + /* Enable phy bypass mode to enable the test pattern */ + + enable_phy_bypass_mode(enc10, true); +} + +static void set_link_training_complete( + struct dcn10_link_encoder *enc10, + bool complete) +{ + /* This register resides in DP back end block; + * transmitter is used for the offset + */ + REG_UPDATE(DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, complete); + +} + +void dcn10_link_encoder_set_dp_phy_pattern_training_pattern( + struct link_encoder *enc, + uint32_t index) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + /* Write Training Pattern */ + + REG_WRITE(DP_DPHY_TRAINING_PATTERN_SEL, index); + + /* Set HW Register Training Complete to false */ + + set_link_training_complete(enc10, false); + + /* Disable PHY Bypass mode to output Training Pattern */ + + enable_phy_bypass_mode(enc10, false); + + /* Disable PRBS mode */ + disable_prbs_mode(enc10); +} + +static void setup_panel_mode( + struct dcn10_link_encoder *enc10, + enum dp_panel_mode panel_mode) +{ + uint32_t value; + + ASSERT(REG(DP_DPHY_INTERNAL_CTRL)); + value = REG_READ(DP_DPHY_INTERNAL_CTRL); + + switch (panel_mode) { + case DP_PANEL_MODE_EDP: + value = 0x1; + break; + case DP_PANEL_MODE_SPECIAL: + value = 0x11; + break; + default: + value = 0x0; + break; + } + + REG_WRITE(DP_DPHY_INTERNAL_CTRL, value); +} + +static void set_dp_phy_pattern_symbol_error( + struct dcn10_link_encoder *enc10) +{ + /* Disable PHY Bypass mode to setup the test pattern */ + enable_phy_bypass_mode(enc10, false); + + /* program correct panel mode*/ + setup_panel_mode(enc10, DP_PANEL_MODE_DEFAULT); + + /* A PRBS23 pattern is used for most DP electrical measurements. */ + + /* Enable PRBS symbols on the lanes */ + disable_prbs_symbols(enc10, false); + + /* For PRBS23 Set bit DPHY_PRBS_SEL=1 and Set bit DPHY_PRBS_EN=1 */ + REG_UPDATE_2(DP_DPHY_PRBS_CNTL, + DPHY_PRBS_SEL, 1, + DPHY_PRBS_EN, 1); + + /* Enable phy bypass mode to enable the test pattern */ + enable_phy_bypass_mode(enc10, true); +} + +static void set_dp_phy_pattern_prbs7( + struct dcn10_link_encoder *enc10) +{ + /* Disable PHY Bypass mode to setup the test pattern */ + enable_phy_bypass_mode(enc10, false); + + /* A PRBS7 pattern is used for most DP electrical measurements. */ + + /* Enable PRBS symbols on the lanes */ + disable_prbs_symbols(enc10, false); + + /* For PRBS7 Set bit DPHY_PRBS_SEL=0 and Set bit DPHY_PRBS_EN=1 */ + REG_UPDATE_2(DP_DPHY_PRBS_CNTL, + DPHY_PRBS_SEL, 0, + DPHY_PRBS_EN, 1); + + /* Enable phy bypass mode to enable the test pattern */ + enable_phy_bypass_mode(enc10, true); +} + +static void set_dp_phy_pattern_80bit_custom( + struct dcn10_link_encoder *enc10, + const uint8_t *pattern) +{ + /* Disable PHY Bypass mode to setup the test pattern */ + enable_phy_bypass_mode(enc10, false); + + /* Enable debug symbols on the lanes */ + + disable_prbs_symbols(enc10, true); + + /* Enable PHY bypass mode to enable the test pattern */ + /* TODO is it really needed ? */ + + enable_phy_bypass_mode(enc10, true); + + /* Program 80 bit custom pattern */ + { + uint16_t pattern_symbols[8]; + + pattern_symbols[0] = + ((pattern[1] & 0x03) << 8) | pattern[0]; + pattern_symbols[1] = + ((pattern[2] & 0x0f) << 6) | ((pattern[1] >> 2) & 0x3f); + pattern_symbols[2] = + ((pattern[3] & 0x3f) << 4) | ((pattern[2] >> 4) & 0x0f); + pattern_symbols[3] = + (pattern[4] << 2) | ((pattern[3] >> 6) & 0x03); + pattern_symbols[4] = + ((pattern[6] & 0x03) << 8) | pattern[5]; + pattern_symbols[5] = + ((pattern[7] & 0x0f) << 6) | ((pattern[6] >> 2) & 0x3f); + pattern_symbols[6] = + ((pattern[8] & 0x3f) << 4) | ((pattern[7] >> 4) & 0x0f); + pattern_symbols[7] = + (pattern[9] << 2) | ((pattern[8] >> 6) & 0x03); + + program_pattern_symbols(enc10, pattern_symbols); + } + + /* Enable phy bypass mode to enable the test pattern */ + + enable_phy_bypass_mode(enc10, true); +} + +static void set_dp_phy_pattern_hbr2_compliance_cp2520_2( + struct dcn10_link_encoder *enc10, + unsigned int cp2520_pattern) +{ + + /* previously there is a register DP_HBR2_EYE_PATTERN + * that is enabled to get the pattern. + * But it does not work with the latest spec change, + * so we are programming the following registers manually. + * + * The following settings have been confirmed + * by Nick Chorney and Sandra Liu + */ + + /* Disable PHY Bypass mode to setup the test pattern */ + + enable_phy_bypass_mode(enc10, false); + + /* Setup DIG encoder in DP SST mode */ + enc10->base.funcs->setup(&enc10->base, SIGNAL_TYPE_DISPLAY_PORT); + + /* ensure normal panel mode. */ + setup_panel_mode(enc10, DP_PANEL_MODE_DEFAULT); + + /* no vbid after BS (SR) + * DP_LINK_FRAMING_CNTL changed history Sandra Liu + * 11000260 / 11000104 / 110000FC + */ + REG_UPDATE_3(DP_LINK_FRAMING_CNTL, + DP_IDLE_BS_INTERVAL, 0xFC, + DP_VBID_DISABLE, 1, + DP_VID_ENHANCED_FRAME_MODE, 1); + + /* swap every BS with SR */ + REG_UPDATE(DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, 0); + + /* select cp2520 patterns */ + if (REG(DP_DPHY_HBR2_PATTERN_CONTROL)) + REG_UPDATE(DP_DPHY_HBR2_PATTERN_CONTROL, + DP_DPHY_HBR2_PATTERN_CONTROL, cp2520_pattern); + else + /* pre-DCE11 can only generate CP2520 pattern 2 */ + ASSERT(cp2520_pattern == 2); + + /* set link training complete */ + set_link_training_complete(enc10, true); + + /* disable video stream */ + REG_UPDATE(DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, 0); + + /* Disable PHY Bypass mode to setup the test pattern */ + enable_phy_bypass_mode(enc10, false); +} + +static void set_dp_phy_pattern_passthrough_mode( + struct dcn10_link_encoder *enc10, + enum dp_panel_mode panel_mode) +{ + /* program correct panel mode */ + setup_panel_mode(enc10, panel_mode); + + /* restore LINK_FRAMING_CNTL and DPHY_SCRAMBLER_BS_COUNT + * in case we were doing HBR2 compliance pattern before + */ + REG_UPDATE_3(DP_LINK_FRAMING_CNTL, + DP_IDLE_BS_INTERVAL, 0x2000, + DP_VBID_DISABLE, 0, + DP_VID_ENHANCED_FRAME_MODE, 1); + + REG_UPDATE(DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, 0x1FF); + + /* set link training complete */ + set_link_training_complete(enc10, true); + + /* Disable PHY Bypass mode to setup the test pattern */ + enable_phy_bypass_mode(enc10, false); + + /* Disable PRBS mode */ + disable_prbs_mode(enc10); +} + +/* return value is bit-vector */ +static uint8_t get_frontend_source( + enum engine_id engine) +{ + switch (engine) { + case ENGINE_ID_DIGA: + return DCN10_DIG_FE_SOURCE_SELECT_DIGA; + case ENGINE_ID_DIGB: + return DCN10_DIG_FE_SOURCE_SELECT_DIGB; + case ENGINE_ID_DIGC: + return DCN10_DIG_FE_SOURCE_SELECT_DIGC; + case ENGINE_ID_DIGD: + return DCN10_DIG_FE_SOURCE_SELECT_DIGD; + case ENGINE_ID_DIGE: + return DCN10_DIG_FE_SOURCE_SELECT_DIGE; + case ENGINE_ID_DIGF: + return DCN10_DIG_FE_SOURCE_SELECT_DIGF; + case ENGINE_ID_DIGG: + return DCN10_DIG_FE_SOURCE_SELECT_DIGG; + default: + ASSERT_CRITICAL(false); + return DCN10_DIG_FE_SOURCE_SELECT_INVALID; + } +} + +static void configure_encoder( + struct dcn10_link_encoder *enc10, + const struct dc_link_settings *link_settings) +{ + /* set number of lanes */ + + REG_SET(DP_CONFIG, 0, + DP_UDI_LANES, link_settings->lane_count - LANE_COUNT_ONE); + + /* setup scrambler */ + REG_UPDATE(DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_ADVANCE, 1); +} + +void dcn10_psr_program_dp_dphy_fast_training(struct link_encoder *enc, + bool exit_link_training_required) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + + if (exit_link_training_required) + REG_UPDATE(DP_DPHY_FAST_TRAINING, + DPHY_RX_FAST_TRAINING_CAPABLE, 1); + else { + REG_UPDATE(DP_DPHY_FAST_TRAINING, + DPHY_RX_FAST_TRAINING_CAPABLE, 0); + /*In DCE 11, we are able to pre-program a Force SR register + * to be able to trigger SR symbol after 5 idle patterns + * transmitted. Upon PSR Exit, DMCU can trigger + * DPHY_LOAD_BS_COUNT_START = 1. Upon writing 1 to + * DPHY_LOAD_BS_COUNT_START and the internal counter + * reaches DPHY_LOAD_BS_COUNT, the next BS symbol will be + * replaced by SR symbol once. + */ + + REG_UPDATE(DP_DPHY_BS_SR_SWAP_CNTL, DPHY_LOAD_BS_COUNT, 0x5); + } +} + +void dcn10_psr_program_secondary_packet(struct link_encoder *enc, + unsigned int sdp_transmit_line_num_deadline) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + + REG_UPDATE_2(DP_SEC_CNTL1, + DP_SEC_GSP0_LINE_NUM, sdp_transmit_line_num_deadline, + DP_SEC_GSP0_PRIORITY, 1); +} + +bool dcn10_is_dig_enabled(struct link_encoder *enc) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + uint32_t value; + + REG_GET(DIG_BE_EN_CNTL, DIG_ENABLE, &value); + return value; +} + +static void link_encoder_disable(struct dcn10_link_encoder *enc10) +{ + /* reset training pattern */ + REG_SET(DP_DPHY_TRAINING_PATTERN_SEL, 0, + DPHY_TRAINING_PATTERN_SEL, 0); + + /* reset training complete */ + REG_UPDATE(DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, 0); + + /* reset panel mode */ + setup_panel_mode(enc10, DP_PANEL_MODE_DEFAULT); +} + +static void hpd_initialize( + struct dcn10_link_encoder *enc10) +{ + /* Associate HPD with DIG_BE */ + enum hpd_source_id hpd_source = enc10->base.hpd_source; + + REG_UPDATE(DIG_BE_CNTL, DIG_HPD_SELECT, hpd_source); +} + +bool dcn10_link_encoder_validate_dvi_output( + const struct dcn10_link_encoder *enc10, + enum signal_type connector_signal, + enum signal_type signal, + const struct dc_crtc_timing *crtc_timing) +{ + uint32_t max_pixel_clock = TMDS_MAX_PIXEL_CLOCK; + + if (signal == SIGNAL_TYPE_DVI_DUAL_LINK) + max_pixel_clock *= 2; + + /* This handles the case of HDMI downgrade to DVI we don't want to + * we don't want to cap the pixel clock if the DDI is not DVI. + */ + if (connector_signal != SIGNAL_TYPE_DVI_DUAL_LINK && + connector_signal != SIGNAL_TYPE_DVI_SINGLE_LINK) + max_pixel_clock = enc10->base.features.max_hdmi_pixel_clock; + + /* DVI only support RGB pixel encoding */ + if (crtc_timing->pixel_encoding != PIXEL_ENCODING_RGB) + return false; + + /*connect DVI via adpater's HDMI connector*/ + if ((connector_signal == SIGNAL_TYPE_DVI_SINGLE_LINK || + connector_signal == SIGNAL_TYPE_HDMI_TYPE_A) && + signal != SIGNAL_TYPE_HDMI_TYPE_A && + crtc_timing->pix_clk_khz > TMDS_MAX_PIXEL_CLOCK) + return false; + if (crtc_timing->pix_clk_khz < TMDS_MIN_PIXEL_CLOCK) + return false; + + if (crtc_timing->pix_clk_khz > max_pixel_clock) + return false; + + /* DVI supports 6/8bpp single-link and 10/16bpp dual-link */ + switch (crtc_timing->display_color_depth) { + case COLOR_DEPTH_666: + case COLOR_DEPTH_888: + break; + case COLOR_DEPTH_101010: + case COLOR_DEPTH_161616: + if (signal != SIGNAL_TYPE_DVI_DUAL_LINK) + return false; + break; + default: + return false; + } + + return true; +} + +static bool dcn10_link_encoder_validate_hdmi_output( + const struct dcn10_link_encoder *enc10, + const struct dc_crtc_timing *crtc_timing, + int adjusted_pix_clk_khz) +{ + enum dc_color_depth max_deep_color = + enc10->base.features.max_hdmi_deep_color; + + if (max_deep_color < crtc_timing->display_color_depth) + return false; + + if (crtc_timing->display_color_depth < COLOR_DEPTH_888) + return false; + if (adjusted_pix_clk_khz < TMDS_MIN_PIXEL_CLOCK) + return false; + + if ((adjusted_pix_clk_khz == 0) || + (adjusted_pix_clk_khz > enc10->base.features.max_hdmi_pixel_clock)) + return false; + + /* DCE11 HW does not support 420 */ + if (!enc10->base.features.ycbcr420_supported && + crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420) + return false; + + if (!enc10->base.features.flags.bits.HDMI_6GB_EN && + adjusted_pix_clk_khz >= 300000) + return false; + return true; +} + +bool dcn10_link_encoder_validate_dp_output( + const struct dcn10_link_encoder *enc10, + const struct dc_crtc_timing *crtc_timing) +{ + /* default RGB only */ + if (crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB) + return true; + + if (enc10->base.features.flags.bits.IS_YCBCR_CAPABLE) + return true; + + /* for DCE 8.x or later DP Y-only feature, + * we need ASIC cap + FeatureSupportDPYonly, not support 666 + */ + if (crtc_timing->flags.Y_ONLY && + enc10->base.features.flags.bits.IS_YCBCR_CAPABLE && + crtc_timing->display_color_depth != COLOR_DEPTH_666) + return true; + + return false; +} + +void dcn10_link_encoder_construct( + struct dcn10_link_encoder *enc10, + const struct encoder_init_data *init_data, + const struct encoder_feature_support *enc_features, + const struct dcn10_link_enc_registers *link_regs, + const struct dcn10_link_enc_aux_registers *aux_regs, + const struct dcn10_link_enc_hpd_registers *hpd_regs, + const struct dcn10_link_enc_shift *link_shift, + const struct dcn10_link_enc_mask *link_mask) +{ + struct bp_encoder_cap_info bp_cap_info = {0}; + const struct dc_vbios_funcs *bp_funcs = init_data->ctx->dc_bios->funcs; + enum bp_result result = BP_RESULT_OK; + + enc10->base.funcs = &dcn10_lnk_enc_funcs; + enc10->base.ctx = init_data->ctx; + enc10->base.id = init_data->encoder; + + enc10->base.hpd_source = init_data->hpd_source; + enc10->base.connector = init_data->connector; + + enc10->base.preferred_engine = ENGINE_ID_UNKNOWN; + + enc10->base.features = *enc_features; + + enc10->base.transmitter = init_data->transmitter; + + /* set the flag to indicate whether driver poll the I2C data pin + * while doing the DP sink detect + */ + +/* if (dal_adapter_service_is_feature_supported(as, + FEATURE_DP_SINK_DETECT_POLL_DATA_PIN)) + enc10->base.features.flags.bits. + DP_SINK_DETECT_POLL_DATA_PIN = true;*/ + + enc10->base.output_signals = + SIGNAL_TYPE_DVI_SINGLE_LINK | + SIGNAL_TYPE_DVI_DUAL_LINK | + SIGNAL_TYPE_LVDS | + SIGNAL_TYPE_DISPLAY_PORT | + SIGNAL_TYPE_DISPLAY_PORT_MST | + SIGNAL_TYPE_EDP | + SIGNAL_TYPE_HDMI_TYPE_A; + + /* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE. + * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY. + * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer + * DIG is per UNIPHY and used by SST DP, eDP, HDMI, DVI and LVDS. + * Prefer DIG assignment is decided by board design. + * For DCE 8.0, there are only max 6 UNIPHYs, we assume board design + * and VBIOS will filter out 7 UNIPHY for DCE 8.0. + * By this, adding DIGG should not hurt DCE 8.0. + * This will let DCE 8.1 share DCE 8.0 as much as possible + */ + + enc10->link_regs = link_regs; + enc10->aux_regs = aux_regs; + enc10->hpd_regs = hpd_regs; + enc10->link_shift = link_shift; + enc10->link_mask = link_mask; + + switch (enc10->base.transmitter) { + case TRANSMITTER_UNIPHY_A: + enc10->base.preferred_engine = ENGINE_ID_DIGA; + break; + case TRANSMITTER_UNIPHY_B: + enc10->base.preferred_engine = ENGINE_ID_DIGB; + break; + case TRANSMITTER_UNIPHY_C: + enc10->base.preferred_engine = ENGINE_ID_DIGC; + break; + case TRANSMITTER_UNIPHY_D: + enc10->base.preferred_engine = ENGINE_ID_DIGD; + break; + case TRANSMITTER_UNIPHY_E: + enc10->base.preferred_engine = ENGINE_ID_DIGE; + break; + case TRANSMITTER_UNIPHY_F: + enc10->base.preferred_engine = ENGINE_ID_DIGF; + break; + case TRANSMITTER_UNIPHY_G: + enc10->base.preferred_engine = ENGINE_ID_DIGG; + break; + default: + ASSERT_CRITICAL(false); + enc10->base.preferred_engine = ENGINE_ID_UNKNOWN; + } + + /* default to one to mirror Windows behavior */ + enc10->base.features.flags.bits.HDMI_6GB_EN = 1; + + result = bp_funcs->get_encoder_cap_info(enc10->base.ctx->dc_bios, + enc10->base.id, &bp_cap_info); + + /* Override features with DCE-specific values */ + if (result == BP_RESULT_OK) { + enc10->base.features.flags.bits.IS_HBR2_CAPABLE = + bp_cap_info.DP_HBR2_EN; + enc10->base.features.flags.bits.IS_HBR3_CAPABLE = + bp_cap_info.DP_HBR3_EN; + enc10->base.features.flags.bits.HDMI_6GB_EN = bp_cap_info.HDMI_6GB_EN; + } else { + DC_LOG_WARNING("%s: Failed to get encoder_cap_info from VBIOS with error code %d!\n", + __func__, + result); + } +} + +bool dcn10_link_encoder_validate_output_with_stream( + struct link_encoder *enc, + const struct dc_stream_state *stream) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + bool is_valid; + + switch (stream->signal) { + case SIGNAL_TYPE_DVI_SINGLE_LINK: + case SIGNAL_TYPE_DVI_DUAL_LINK: + is_valid = dcn10_link_encoder_validate_dvi_output( + enc10, + stream->sink->link->connector_signal, + stream->signal, + &stream->timing); + break; + case SIGNAL_TYPE_HDMI_TYPE_A: + is_valid = dcn10_link_encoder_validate_hdmi_output( + enc10, + &stream->timing, + stream->phy_pix_clk); + break; + case SIGNAL_TYPE_DISPLAY_PORT: + case SIGNAL_TYPE_DISPLAY_PORT_MST: + is_valid = dcn10_link_encoder_validate_dp_output( + enc10, &stream->timing); + break; + case SIGNAL_TYPE_EDP: + is_valid = (stream->timing.pixel_encoding == PIXEL_ENCODING_RGB) ? true : false; + break; + case SIGNAL_TYPE_VIRTUAL: + is_valid = true; + break; + default: + is_valid = false; + break; + } + + return is_valid; +} + +void dcn10_link_encoder_hw_init( + struct link_encoder *enc) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + struct bp_transmitter_control cntl = { 0 }; + enum bp_result result; + + cntl.action = TRANSMITTER_CONTROL_INIT; + cntl.engine_id = ENGINE_ID_UNKNOWN; + cntl.transmitter = enc10->base.transmitter; + cntl.connector_obj_id = enc10->base.connector; + cntl.lanes_number = LANE_COUNT_FOUR; + cntl.coherent = false; + cntl.hpd_sel = enc10->base.hpd_source; + + if (enc10->base.connector.id == CONNECTOR_ID_EDP) + cntl.signal = SIGNAL_TYPE_EDP; + + result = link_transmitter_control(enc10, &cntl); + + if (result != BP_RESULT_OK) { + DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n", + __func__); + BREAK_TO_DEBUGGER(); + return; + } + + if (enc10->base.connector.id == CONNECTOR_ID_LVDS) { + cntl.action = TRANSMITTER_CONTROL_BACKLIGHT_BRIGHTNESS; + + result = link_transmitter_control(enc10, &cntl); + + ASSERT(result == BP_RESULT_OK); + + } + aux_initialize(enc10); + + /* reinitialize HPD. + * hpd_initialize() will pass DIG_FE id to HW context. + * All other routine within HW context will use fe_engine_offset + * as DIG_FE id even caller pass DIG_FE id. + * So this routine must be called first. + */ + hpd_initialize(enc10); +} + +void dcn10_link_encoder_destroy(struct link_encoder **enc) +{ + kfree(TO_DCN10_LINK_ENC(*enc)); + *enc = NULL; +} + +void dcn10_link_encoder_setup( + struct link_encoder *enc, + enum signal_type signal) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + + switch (signal) { + case SIGNAL_TYPE_EDP: + case SIGNAL_TYPE_DISPLAY_PORT: + /* DP SST */ + REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 0); + break; + case SIGNAL_TYPE_LVDS: + /* LVDS */ + REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 1); + break; + case SIGNAL_TYPE_DVI_SINGLE_LINK: + case SIGNAL_TYPE_DVI_DUAL_LINK: + /* TMDS-DVI */ + REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 2); + break; + case SIGNAL_TYPE_HDMI_TYPE_A: + /* TMDS-HDMI */ + REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 3); + break; + case SIGNAL_TYPE_DISPLAY_PORT_MST: + /* DP MST */ + REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 5); + break; + default: + ASSERT_CRITICAL(false); + /* invalid mode ! */ + break; + } + +} + +/* TODO: still need depth or just pass in adjusted pixel clock? */ +void dcn10_link_encoder_enable_tmds_output( + struct link_encoder *enc, + enum clock_source_id clock_source, + enum dc_color_depth color_depth, + enum signal_type signal, + uint32_t pixel_clock) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + struct bp_transmitter_control cntl = { 0 }; + enum bp_result result; + + /* Enable the PHY */ + + cntl.action = TRANSMITTER_CONTROL_ENABLE; + cntl.engine_id = enc->preferred_engine; + cntl.transmitter = enc10->base.transmitter; + cntl.pll_id = clock_source; + cntl.signal = signal; + if (cntl.signal == SIGNAL_TYPE_DVI_DUAL_LINK) + cntl.lanes_number = 8; + else + cntl.lanes_number = 4; + + cntl.hpd_sel = enc10->base.hpd_source; + + cntl.pixel_clock = pixel_clock; + cntl.color_depth = color_depth; + + result = link_transmitter_control(enc10, &cntl); + + if (result != BP_RESULT_OK) { + DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n", + __func__); + BREAK_TO_DEBUGGER(); + } +} + +/* enables DP PHY output */ +void dcn10_link_encoder_enable_dp_output( + struct link_encoder *enc, + const struct dc_link_settings *link_settings, + enum clock_source_id clock_source) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + struct bp_transmitter_control cntl = { 0 }; + enum bp_result result; + + /* Enable the PHY */ + + /* number_of_lanes is used for pixel clock adjust, + * but it's not passed to asic_control. + * We need to set number of lanes manually. + */ + configure_encoder(enc10, link_settings); + + cntl.action = TRANSMITTER_CONTROL_ENABLE; + cntl.engine_id = enc->preferred_engine; + cntl.transmitter = enc10->base.transmitter; + cntl.pll_id = clock_source; + cntl.signal = SIGNAL_TYPE_DISPLAY_PORT; + cntl.lanes_number = link_settings->lane_count; + cntl.hpd_sel = enc10->base.hpd_source; + cntl.pixel_clock = link_settings->link_rate + * LINK_RATE_REF_FREQ_IN_KHZ; + /* TODO: check if undefined works */ + cntl.color_depth = COLOR_DEPTH_UNDEFINED; + + result = link_transmitter_control(enc10, &cntl); + + if (result != BP_RESULT_OK) { + DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n", + __func__); + BREAK_TO_DEBUGGER(); + } +} + +/* enables DP PHY output in MST mode */ +void dcn10_link_encoder_enable_dp_mst_output( + struct link_encoder *enc, + const struct dc_link_settings *link_settings, + enum clock_source_id clock_source) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + struct bp_transmitter_control cntl = { 0 }; + enum bp_result result; + + /* Enable the PHY */ + + /* number_of_lanes is used for pixel clock adjust, + * but it's not passed to asic_control. + * We need to set number of lanes manually. + */ + configure_encoder(enc10, link_settings); + + cntl.action = TRANSMITTER_CONTROL_ENABLE; + cntl.engine_id = ENGINE_ID_UNKNOWN; + cntl.transmitter = enc10->base.transmitter; + cntl.pll_id = clock_source; + cntl.signal = SIGNAL_TYPE_DISPLAY_PORT_MST; + cntl.lanes_number = link_settings->lane_count; + cntl.hpd_sel = enc10->base.hpd_source; + cntl.pixel_clock = link_settings->link_rate + * LINK_RATE_REF_FREQ_IN_KHZ; + /* TODO: check if undefined works */ + cntl.color_depth = COLOR_DEPTH_UNDEFINED; + + result = link_transmitter_control(enc10, &cntl); + + if (result != BP_RESULT_OK) { + DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n", + __func__); + BREAK_TO_DEBUGGER(); + } +} +/* + * @brief + * Disable transmitter and its encoder + */ +void dcn10_link_encoder_disable_output( + struct link_encoder *enc, + enum signal_type signal) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + struct bp_transmitter_control cntl = { 0 }; + enum bp_result result; + + if (!dcn10_is_dig_enabled(enc)) { + /* OF_SKIP_POWER_DOWN_INACTIVE_ENCODER */ + return; + } + /* Power-down RX and disable GPU PHY should be paired. + * Disabling PHY without powering down RX may cause + * symbol lock loss, on which we will get DP Sink interrupt. + */ + + /* There is a case for the DP active dongles + * where we want to disable the PHY but keep RX powered, + * for those we need to ignore DP Sink interrupt + * by checking lane count that has been set + * on the last do_enable_output(). + */ + + /* disable transmitter */ + cntl.action = TRANSMITTER_CONTROL_DISABLE; + cntl.transmitter = enc10->base.transmitter; + cntl.hpd_sel = enc10->base.hpd_source; + cntl.signal = signal; + cntl.connector_obj_id = enc10->base.connector; + + result = link_transmitter_control(enc10, &cntl); + + if (result != BP_RESULT_OK) { + DC_LOG_ERROR("%s: Failed to execute VBIOS command table!\n", + __func__); + BREAK_TO_DEBUGGER(); + return; + } + + /* disable encoder */ + if (dc_is_dp_signal(signal)) + link_encoder_disable(enc10); +} + +void dcn10_link_encoder_dp_set_lane_settings( + struct link_encoder *enc, + const struct link_training_settings *link_settings) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + union dpcd_training_lane_set training_lane_set = { { 0 } }; + int32_t lane = 0; + struct bp_transmitter_control cntl = { 0 }; + + if (!link_settings) { + BREAK_TO_DEBUGGER(); + return; + } + + cntl.action = TRANSMITTER_CONTROL_SET_VOLTAGE_AND_PREEMPASIS; + cntl.transmitter = enc10->base.transmitter; + cntl.connector_obj_id = enc10->base.connector; + cntl.lanes_number = link_settings->link_settings.lane_count; + cntl.hpd_sel = enc10->base.hpd_source; + cntl.pixel_clock = link_settings->link_settings.link_rate * + LINK_RATE_REF_FREQ_IN_KHZ; + + for (lane = 0; lane < link_settings->link_settings.lane_count; lane++) { + /* translate lane settings */ + + training_lane_set.bits.VOLTAGE_SWING_SET = + link_settings->lane_settings[lane].VOLTAGE_SWING; + training_lane_set.bits.PRE_EMPHASIS_SET = + link_settings->lane_settings[lane].PRE_EMPHASIS; + + /* post cursor 2 setting only applies to HBR2 link rate */ + if (link_settings->link_settings.link_rate == LINK_RATE_HIGH2) { + /* this is passed to VBIOS + * to program post cursor 2 level + */ + training_lane_set.bits.POST_CURSOR2_SET = + link_settings->lane_settings[lane].POST_CURSOR2; + } + + cntl.lane_select = lane; + cntl.lane_settings = training_lane_set.raw; + + /* call VBIOS table to set voltage swing and pre-emphasis */ + link_transmitter_control(enc10, &cntl); + } +} + +/* set DP PHY test and training patterns */ +void dcn10_link_encoder_dp_set_phy_pattern( + struct link_encoder *enc, + const struct encoder_set_dp_phy_pattern_param *param) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + + switch (param->dp_phy_pattern) { + case DP_TEST_PATTERN_TRAINING_PATTERN1: + dcn10_link_encoder_set_dp_phy_pattern_training_pattern(enc, 0); + break; + case DP_TEST_PATTERN_TRAINING_PATTERN2: + dcn10_link_encoder_set_dp_phy_pattern_training_pattern(enc, 1); + break; + case DP_TEST_PATTERN_TRAINING_PATTERN3: + dcn10_link_encoder_set_dp_phy_pattern_training_pattern(enc, 2); + break; + case DP_TEST_PATTERN_TRAINING_PATTERN4: + dcn10_link_encoder_set_dp_phy_pattern_training_pattern(enc, 3); + break; + case DP_TEST_PATTERN_D102: + set_dp_phy_pattern_d102(enc10); + break; + case DP_TEST_PATTERN_SYMBOL_ERROR: + set_dp_phy_pattern_symbol_error(enc10); + break; + case DP_TEST_PATTERN_PRBS7: + set_dp_phy_pattern_prbs7(enc10); + break; + case DP_TEST_PATTERN_80BIT_CUSTOM: + set_dp_phy_pattern_80bit_custom( + enc10, param->custom_pattern); + break; + case DP_TEST_PATTERN_CP2520_1: + set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc10, 1); + break; + case DP_TEST_PATTERN_CP2520_2: + set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc10, 2); + break; + case DP_TEST_PATTERN_CP2520_3: + set_dp_phy_pattern_hbr2_compliance_cp2520_2(enc10, 3); + break; + case DP_TEST_PATTERN_VIDEO_MODE: { + set_dp_phy_pattern_passthrough_mode( + enc10, param->dp_panel_mode); + break; + } + + default: + /* invalid phy pattern */ + ASSERT_CRITICAL(false); + break; + } +} + +static void fill_stream_allocation_row_info( + const struct link_mst_stream_allocation *stream_allocation, + uint32_t *src, + uint32_t *slots) +{ + const struct stream_encoder *stream_enc = stream_allocation->stream_enc; + + if (stream_enc) { + *src = stream_enc->id; + *slots = stream_allocation->slot_count; + } else { + *src = 0; + *slots = 0; + } +} + +/* programs DP MST VC payload allocation */ +void dcn10_link_encoder_update_mst_stream_allocation_table( + struct link_encoder *enc, + const struct link_mst_stream_allocation_table *table) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + uint32_t value0 = 0; + uint32_t value1 = 0; + uint32_t value2 = 0; + uint32_t slots = 0; + uint32_t src = 0; + uint32_t retries = 0; + + /* For CZ, there are only 3 pipes. So Virtual channel is up 3.*/ + + /* --- Set MSE Stream Attribute - + * Setup VC Payload Table on Tx Side, + * Issue allocation change trigger + * to commit payload on both tx and rx side + */ + + /* we should clean-up table each time */ + + if (table->stream_count >= 1) { + fill_stream_allocation_row_info( + &table->stream_allocations[0], + &src, + &slots); + } else { + src = 0; + slots = 0; + } + + REG_UPDATE_2(DP_MSE_SAT0, + DP_MSE_SAT_SRC0, src, + DP_MSE_SAT_SLOT_COUNT0, slots); + + if (table->stream_count >= 2) { + fill_stream_allocation_row_info( + &table->stream_allocations[1], + &src, + &slots); + } else { + src = 0; + slots = 0; + } + + REG_UPDATE_2(DP_MSE_SAT0, + DP_MSE_SAT_SRC1, src, + DP_MSE_SAT_SLOT_COUNT1, slots); + + if (table->stream_count >= 3) { + fill_stream_allocation_row_info( + &table->stream_allocations[2], + &src, + &slots); + } else { + src = 0; + slots = 0; + } + + REG_UPDATE_2(DP_MSE_SAT1, + DP_MSE_SAT_SRC2, src, + DP_MSE_SAT_SLOT_COUNT2, slots); + + if (table->stream_count >= 4) { + fill_stream_allocation_row_info( + &table->stream_allocations[3], + &src, + &slots); + } else { + src = 0; + slots = 0; + } + + REG_UPDATE_2(DP_MSE_SAT1, + DP_MSE_SAT_SRC3, src, + DP_MSE_SAT_SLOT_COUNT3, slots); + + /* --- wait for transaction finish */ + + /* send allocation change trigger (ACT) ? + * this step first sends the ACT, + * then double buffers the SAT into the hardware + * making the new allocation active on the DP MST mode link + */ + + /* DP_MSE_SAT_UPDATE: + * 0 - No Action + * 1 - Update SAT with trigger + * 2 - Update SAT without trigger + */ + REG_UPDATE(DP_MSE_SAT_UPDATE, + DP_MSE_SAT_UPDATE, 1); + + /* wait for update to complete + * (i.e. DP_MSE_SAT_UPDATE field is reset to 0) + * then wait for the transmission + * of at least 16 MTP headers on immediate local link. + * i.e. DP_MSE_16_MTP_KEEPOUT field (read only) is reset to 0 + * a value of 1 indicates that DP MST mode + * is in the 16 MTP keepout region after a VC has been added. + * MST stream bandwidth (VC rate) can be configured + * after this bit is cleared + */ + do { + udelay(10); + + value0 = REG_READ(DP_MSE_SAT_UPDATE); + + REG_GET(DP_MSE_SAT_UPDATE, + DP_MSE_SAT_UPDATE, &value1); + + REG_GET(DP_MSE_SAT_UPDATE, + DP_MSE_16_MTP_KEEPOUT, &value2); + + /* bit field DP_MSE_SAT_UPDATE is set to 1 already */ + if (!value1 && !value2) + break; + ++retries; + } while (retries < DP_MST_UPDATE_MAX_RETRY); +} + +void dcn10_link_encoder_connect_dig_be_to_fe( + struct link_encoder *enc, + enum engine_id engine, + bool connect) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + uint32_t field; + + if (engine != ENGINE_ID_UNKNOWN) { + + REG_GET(DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, &field); + + if (connect) + field |= get_frontend_source(engine); + else + field &= ~get_frontend_source(engine); + + REG_UPDATE(DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, field); + } +} + + +#define HPD_REG(reg)\ + (enc10->hpd_regs->reg) + +#define HPD_REG_READ(reg_name) \ + dm_read_reg(CTX, HPD_REG(reg_name)) + +#define HPD_REG_UPDATE_N(reg_name, n, ...) \ + generic_reg_update_ex(CTX, \ + HPD_REG(reg_name), \ + HPD_REG_READ(reg_name), \ + n, __VA_ARGS__) + +#define HPD_REG_UPDATE(reg_name, field, val) \ + HPD_REG_UPDATE_N(reg_name, 1, \ + FN(reg_name, field), val) + +void dcn10_link_encoder_enable_hpd(struct link_encoder *enc) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + + HPD_REG_UPDATE(DC_HPD_CONTROL, + DC_HPD_EN, 1); +} + +void dcn10_link_encoder_disable_hpd(struct link_encoder *enc) +{ + struct dcn10_link_encoder *enc10 = TO_DCN10_LINK_ENC(enc); + + HPD_REG_UPDATE(DC_HPD_CONTROL, + DC_HPD_EN, 0); +} + + +#define AUX_REG(reg)\ + (enc10->aux_regs->reg) + +#define AUX_REG_READ(reg_name) \ + dm_read_reg(CTX, AUX_REG(reg_name)) + +#define AUX_REG_UPDATE_N(reg_name, n, ...) \ + generic_reg_update_ex(CTX, \ + AUX_REG(reg_name), \ + AUX_REG_READ(reg_name), \ + n, __VA_ARGS__) + +#define AUX_REG_UPDATE(reg_name, field, val) \ + AUX_REG_UPDATE_N(reg_name, 1, \ + FN(reg_name, field), val) + +#define AUX_REG_UPDATE_2(reg, f1, v1, f2, v2) \ + AUX_REG_UPDATE_N(reg, 2,\ + FN(reg, f1), v1,\ + FN(reg, f2), v2) + +static void aux_initialize( + struct dcn10_link_encoder *enc10) +{ + enum hpd_source_id hpd_source = enc10->base.hpd_source; + + AUX_REG_UPDATE_2(AUX_CONTROL, + AUX_HPD_SEL, hpd_source, + AUX_LS_READ_EN, 0); + + /* 1/4 window (the maximum allowed) */ + AUX_REG_UPDATE(AUX_DPHY_RX_CONTROL0, + AUX_RX_RECEIVE_WINDOW, 1); +} diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.h new file mode 100644 index 000000000000..2a97cdb2cfbb --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.h @@ -0,0 +1,330 @@ +/* + * Copyright 2012-15 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DC_LINK_ENCODER__DCN10_H__ +#define __DC_LINK_ENCODER__DCN10_H__ + +#include "link_encoder.h" + +#define TO_DCN10_LINK_ENC(link_encoder)\ + container_of(link_encoder, struct dcn10_link_encoder, base) + + +#define AUX_REG_LIST(id)\ + SRI(AUX_CONTROL, DP_AUX, id), \ + SRI(AUX_DPHY_RX_CONTROL0, DP_AUX, id) + +#define HPD_REG_LIST(id)\ + SRI(DC_HPD_CONTROL, HPD, id) + +#define LE_DCN_COMMON_REG_LIST(id) \ + SRI(DIG_BE_CNTL, DIG, id), \ + SRI(DIG_BE_EN_CNTL, DIG, id), \ + SRI(DP_CONFIG, DP, id), \ + SRI(DP_DPHY_CNTL, DP, id), \ + SRI(DP_DPHY_PRBS_CNTL, DP, id), \ + SRI(DP_DPHY_SCRAM_CNTL, DP, id),\ + SRI(DP_DPHY_SYM0, DP, id), \ + SRI(DP_DPHY_SYM1, DP, id), \ + SRI(DP_DPHY_SYM2, DP, id), \ + SRI(DP_DPHY_TRAINING_PATTERN_SEL, DP, id), \ + SRI(DP_LINK_CNTL, DP, id), \ + SRI(DP_LINK_FRAMING_CNTL, DP, id), \ + SRI(DP_MSE_SAT0, DP, id), \ + SRI(DP_MSE_SAT1, DP, id), \ + SRI(DP_MSE_SAT2, DP, id), \ + SRI(DP_MSE_SAT_UPDATE, DP, id), \ + SRI(DP_SEC_CNTL, DP, id), \ + SRI(DP_VID_STREAM_CNTL, DP, id), \ + SRI(DP_DPHY_FAST_TRAINING, DP, id), \ + SRI(DP_SEC_CNTL1, DP, id), \ + SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \ + SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \ + SRI(DP_DPHY_HBR2_PATTERN_CONTROL, DP, id) + +#define LE_DCN10_REG_LIST(id)\ + LE_DCN_COMMON_REG_LIST(id) + +struct dcn10_link_enc_aux_registers { + uint32_t AUX_CONTROL; + uint32_t AUX_DPHY_RX_CONTROL0; +}; + +struct dcn10_link_enc_hpd_registers { + uint32_t DC_HPD_CONTROL; +}; + +struct dcn10_link_enc_registers { + uint32_t DIG_BE_CNTL; + uint32_t DIG_BE_EN_CNTL; + uint32_t DP_CONFIG; + uint32_t DP_DPHY_CNTL; + uint32_t DP_DPHY_INTERNAL_CTRL; + uint32_t DP_DPHY_PRBS_CNTL; + uint32_t DP_DPHY_SCRAM_CNTL; + uint32_t DP_DPHY_SYM0; + uint32_t DP_DPHY_SYM1; + uint32_t DP_DPHY_SYM2; + uint32_t DP_DPHY_TRAINING_PATTERN_SEL; + uint32_t DP_LINK_CNTL; + uint32_t DP_LINK_FRAMING_CNTL; + uint32_t DP_MSE_SAT0; + uint32_t DP_MSE_SAT1; + uint32_t DP_MSE_SAT2; + uint32_t DP_MSE_SAT_UPDATE; + uint32_t DP_SEC_CNTL; + uint32_t DP_VID_STREAM_CNTL; + uint32_t DP_DPHY_FAST_TRAINING; + uint32_t DP_DPHY_BS_SR_SWAP_CNTL; + uint32_t DP_DPHY_HBR2_PATTERN_CONTROL; + uint32_t DP_SEC_CNTL1; +}; + +#define LE_SF(reg_name, field_name, post_fix)\ + .field_name = reg_name ## __ ## field_name ## post_fix + +#define LINK_ENCODER_MASK_SH_LIST_DCN10(mask_sh)\ + LE_SF(DIG0_DIG_BE_EN_CNTL, DIG_ENABLE, mask_sh),\ + LE_SF(DIG0_DIG_BE_CNTL, DIG_HPD_SELECT, mask_sh),\ + LE_SF(DIG0_DIG_BE_CNTL, DIG_MODE, mask_sh),\ + LE_SF(DIG0_DIG_BE_CNTL, DIG_FE_SOURCE_SELECT, mask_sh),\ + LE_SF(DP0_DP_DPHY_CNTL, DPHY_BYPASS, mask_sh),\ + LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE0, mask_sh),\ + LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE1, mask_sh),\ + LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE2, mask_sh),\ + LE_SF(DP0_DP_DPHY_CNTL, DPHY_ATEST_SEL_LANE3, mask_sh),\ + LE_SF(DP0_DP_DPHY_PRBS_CNTL, DPHY_PRBS_EN, mask_sh),\ + LE_SF(DP0_DP_DPHY_PRBS_CNTL, DPHY_PRBS_SEL, mask_sh),\ + LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM1, mask_sh),\ + LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM2, mask_sh),\ + LE_SF(DP0_DP_DPHY_SYM0, DPHY_SYM3, mask_sh),\ + LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM4, mask_sh),\ + LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM5, mask_sh),\ + LE_SF(DP0_DP_DPHY_SYM1, DPHY_SYM6, mask_sh),\ + LE_SF(DP0_DP_DPHY_SYM2, DPHY_SYM7, mask_sh),\ + LE_SF(DP0_DP_DPHY_SYM2, DPHY_SYM8, mask_sh),\ + LE_SF(DP0_DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_BS_COUNT, mask_sh),\ + LE_SF(DP0_DP_DPHY_SCRAM_CNTL, DPHY_SCRAMBLER_ADVANCE, mask_sh),\ + LE_SF(DP0_DP_DPHY_FAST_TRAINING, DPHY_RX_FAST_TRAINING_CAPABLE, mask_sh),\ + LE_SF(DP0_DP_DPHY_BS_SR_SWAP_CNTL, DPHY_LOAD_BS_COUNT, mask_sh),\ + LE_SF(DP0_DP_DPHY_TRAINING_PATTERN_SEL, DPHY_TRAINING_PATTERN_SEL, mask_sh),\ + LE_SF(DP0_DP_DPHY_HBR2_PATTERN_CONTROL, DP_DPHY_HBR2_PATTERN_CONTROL, mask_sh),\ + LE_SF(DP0_DP_LINK_CNTL, DP_LINK_TRAINING_COMPLETE, mask_sh),\ + LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_IDLE_BS_INTERVAL, mask_sh),\ + LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_VBID_DISABLE, mask_sh),\ + LE_SF(DP0_DP_LINK_FRAMING_CNTL, DP_VID_ENHANCED_FRAME_MODE, mask_sh),\ + LE_SF(DP0_DP_VID_STREAM_CNTL, DP_VID_STREAM_ENABLE, mask_sh),\ + LE_SF(DP0_DP_CONFIG, DP_UDI_LANES, mask_sh),\ + LE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP0_LINE_NUM, mask_sh),\ + LE_SF(DP0_DP_SEC_CNTL1, DP_SEC_GSP0_PRIORITY, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SRC0, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SRC1, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SLOT_COUNT0, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT0, DP_MSE_SAT_SLOT_COUNT1, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SRC2, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SRC3, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SLOT_COUNT2, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT1, DP_MSE_SAT_SLOT_COUNT3, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT_UPDATE, DP_MSE_SAT_UPDATE, mask_sh),\ + LE_SF(DP0_DP_MSE_SAT_UPDATE, DP_MSE_16_MTP_KEEPOUT, mask_sh),\ + LE_SF(DP_AUX0_AUX_CONTROL, AUX_HPD_SEL, mask_sh),\ + LE_SF(DP_AUX0_AUX_CONTROL, AUX_LS_READ_EN, mask_sh),\ + LE_SF(DP_AUX0_AUX_DPHY_RX_CONTROL0, AUX_RX_RECEIVE_WINDOW, mask_sh),\ + LE_SF(HPD0_DC_HPD_CONTROL, DC_HPD_EN, mask_sh) + +#define DCN_LINK_ENCODER_REG_FIELD_LIST(type) \ + type DIG_ENABLE;\ + type DIG_HPD_SELECT;\ + type DIG_MODE;\ + type DIG_FE_SOURCE_SELECT;\ + type DPHY_BYPASS;\ + type DPHY_ATEST_SEL_LANE0;\ + type DPHY_ATEST_SEL_LANE1;\ + type DPHY_ATEST_SEL_LANE2;\ + type DPHY_ATEST_SEL_LANE3;\ + type DPHY_PRBS_EN;\ + type DPHY_PRBS_SEL;\ + type DPHY_SYM1;\ + type DPHY_SYM2;\ + type DPHY_SYM3;\ + type DPHY_SYM4;\ + type DPHY_SYM5;\ + type DPHY_SYM6;\ + type DPHY_SYM7;\ + type DPHY_SYM8;\ + type DPHY_SCRAMBLER_BS_COUNT;\ + type DPHY_SCRAMBLER_ADVANCE;\ + type DPHY_RX_FAST_TRAINING_CAPABLE;\ + type DPHY_LOAD_BS_COUNT;\ + type DPHY_TRAINING_PATTERN_SEL;\ + type DP_DPHY_HBR2_PATTERN_CONTROL;\ + type DP_LINK_TRAINING_COMPLETE;\ + type DP_IDLE_BS_INTERVAL;\ + type DP_VBID_DISABLE;\ + type DP_VID_ENHANCED_FRAME_MODE;\ + type DP_VID_STREAM_ENABLE;\ + type DP_UDI_LANES;\ + type DP_SEC_GSP0_LINE_NUM;\ + type DP_SEC_GSP0_PRIORITY;\ + type DP_MSE_SAT_SRC0;\ + type DP_MSE_SAT_SRC1;\ + type DP_MSE_SAT_SRC2;\ + type DP_MSE_SAT_SRC3;\ + type DP_MSE_SAT_SLOT_COUNT0;\ + type DP_MSE_SAT_SLOT_COUNT1;\ + type DP_MSE_SAT_SLOT_COUNT2;\ + type DP_MSE_SAT_SLOT_COUNT3;\ + type DP_MSE_SAT_UPDATE;\ + type DP_MSE_16_MTP_KEEPOUT;\ + type AUX_HPD_SEL;\ + type AUX_LS_READ_EN;\ + type AUX_RX_RECEIVE_WINDOW;\ + type DC_HPD_EN + +struct dcn10_link_enc_shift { + DCN_LINK_ENCODER_REG_FIELD_LIST(uint8_t); +}; + +struct dcn10_link_enc_mask { + DCN_LINK_ENCODER_REG_FIELD_LIST(uint32_t); +}; + +struct dcn10_link_encoder { + struct link_encoder base; + const struct dcn10_link_enc_registers *link_regs; + const struct dcn10_link_enc_aux_registers *aux_regs; + const struct dcn10_link_enc_hpd_registers *hpd_regs; + const struct dcn10_link_enc_shift *link_shift; + const struct dcn10_link_enc_mask *link_mask; +}; + + +void dcn10_link_encoder_construct( + struct dcn10_link_encoder *enc10, + const struct encoder_init_data *init_data, + const struct encoder_feature_support *enc_features, + const struct dcn10_link_enc_registers *link_regs, + const struct dcn10_link_enc_aux_registers *aux_regs, + const struct dcn10_link_enc_hpd_registers *hpd_regs, + const struct dcn10_link_enc_shift *link_shift, + const struct dcn10_link_enc_mask *link_mask); + +bool dcn10_link_encoder_validate_dvi_output( + const struct dcn10_link_encoder *enc10, + enum signal_type connector_signal, + enum signal_type signal, + const struct dc_crtc_timing *crtc_timing); + +bool dcn10_link_encoder_validate_rgb_output( + const struct dcn10_link_encoder *enc10, + const struct dc_crtc_timing *crtc_timing); + +bool dcn10_link_encoder_validate_dp_output( + const struct dcn10_link_encoder *enc10, + const struct dc_crtc_timing *crtc_timing); + +bool dcn10_link_encoder_validate_wireless_output( + const struct dcn10_link_encoder *enc10, + const struct dc_crtc_timing *crtc_timing); + +bool dcn10_link_encoder_validate_output_with_stream( + struct link_encoder *enc, + const struct dc_stream_state *stream); + +/****************** HW programming ************************/ + +/* initialize HW */ /* why do we initialze aux in here? */ +void dcn10_link_encoder_hw_init(struct link_encoder *enc); + +void dcn10_link_encoder_destroy(struct link_encoder **enc); + +/* program DIG_MODE in DIG_BE */ +/* TODO can this be combined with enable_output? */ +void dcn10_link_encoder_setup( + struct link_encoder *enc, + enum signal_type signal); + +/* enables TMDS PHY output */ +/* TODO: still need depth or just pass in adjusted pixel clock? */ +void dcn10_link_encoder_enable_tmds_output( + struct link_encoder *enc, + enum clock_source_id clock_source, + enum dc_color_depth color_depth, + enum signal_type signal, + uint32_t pixel_clock); + +/* enables DP PHY output */ +void dcn10_link_encoder_enable_dp_output( + struct link_encoder *enc, + const struct dc_link_settings *link_settings, + enum clock_source_id clock_source); + +/* enables DP PHY output in MST mode */ +void dcn10_link_encoder_enable_dp_mst_output( + struct link_encoder *enc, + const struct dc_link_settings *link_settings, + enum clock_source_id clock_source); + +/* disable PHY output */ +void dcn10_link_encoder_disable_output( + struct link_encoder *enc, + enum signal_type signal); + +/* set DP lane settings */ +void dcn10_link_encoder_dp_set_lane_settings( + struct link_encoder *enc, + const struct link_training_settings *link_settings); + +void dcn10_link_encoder_dp_set_phy_pattern( + struct link_encoder *enc, + const struct encoder_set_dp_phy_pattern_param *param); + +/* programs DP MST VC payload allocation */ +void dcn10_link_encoder_update_mst_stream_allocation_table( + struct link_encoder *enc, + const struct link_mst_stream_allocation_table *table); + +void dcn10_link_encoder_connect_dig_be_to_fe( + struct link_encoder *enc, + enum engine_id engine, + bool connect); + +void dcn10_link_encoder_set_dp_phy_pattern_training_pattern( + struct link_encoder *enc, + uint32_t index); + +void dcn10_link_encoder_enable_hpd(struct link_encoder *enc); + +void dcn10_link_encoder_disable_hpd(struct link_encoder *enc); + +void dcn10_psr_program_dp_dphy_fast_training(struct link_encoder *enc, + bool exit_link_training_required); + +void dcn10_psr_program_secondary_packet(struct link_encoder *enc, + unsigned int sdp_transmit_line_num_deadline); + +bool dcn10_is_dig_enabled(struct link_encoder *enc); + +#endif /* __DC_LINK_ENCODER__DCN10_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c index c734b7fa5835..f2fbce0e3fc5 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c @@ -360,7 +360,7 @@ void optc1_program_timing( } -static void optc1_set_blank_data_double_buffer(struct timing_generator *optc, bool enable) +void optc1_set_blank_data_double_buffer(struct timing_generator *optc, bool enable) { struct optc *optc1 = DCN10TG_FROM_TG(optc); @@ -1257,20 +1257,20 @@ void optc1_read_otg_state(struct optc *optc1, OPTC_UNDERFLOW_OCCURRED_STATUS, &s->underflow_occurred_status); } -static void optc1_clear_optc_underflow(struct timing_generator *optc) +void optc1_clear_optc_underflow(struct timing_generator *optc) { struct optc *optc1 = DCN10TG_FROM_TG(optc); REG_UPDATE(OPTC_INPUT_GLOBAL_CONTROL, OPTC_UNDERFLOW_CLEAR, 1); } -static void optc1_tg_init(struct timing_generator *optc) +void optc1_tg_init(struct timing_generator *optc) { optc1_set_blank_data_double_buffer(optc, true); optc1_clear_optc_underflow(optc); } -static bool optc1_is_tg_enabled(struct timing_generator *optc) +bool optc1_is_tg_enabled(struct timing_generator *optc) { struct optc *optc1 = DCN10TG_FROM_TG(optc); uint32_t otg_enabled = 0; @@ -1281,7 +1281,7 @@ static bool optc1_is_tg_enabled(struct timing_generator *optc) } -static bool optc1_is_optc_underflow_occurred(struct timing_generator *optc) +bool optc1_is_optc_underflow_occurred(struct timing_generator *optc) { struct optc *optc1 = DCN10TG_FROM_TG(optc); uint32_t underflow_occurred = 0; diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.h index 89e09e5327a2..c62052f46460 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.h +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.h @@ -497,4 +497,14 @@ void optc1_program_stereo(struct timing_generator *optc, bool optc1_is_stereo_left_eye(struct timing_generator *optc); +void optc1_clear_optc_underflow(struct timing_generator *optc); + +void optc1_tg_init(struct timing_generator *optc); + +bool optc1_is_tg_enabled(struct timing_generator *optc); + +bool optc1_is_optc_underflow_occurred(struct timing_generator *optc); + +void optc1_set_blank_data_double_buffer(struct timing_generator *optc, bool enable); + #endif /* __DC_TIMING_GENERATOR_DCN10_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c index 2c0a3150bf2d..df5cb2d1d164 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c @@ -38,7 +38,7 @@ #include "dcn10/dcn10_hw_sequencer.h" #include "dce110/dce110_hw_sequencer.h" #include "dcn10/dcn10_opp.h" -#include "dce/dce_link_encoder.h" +#include "dcn10/dcn10_link_encoder.h" #include "dcn10/dcn10_stream_encoder.h" #include "dce/dce_clocks.h" #include "dce/dce_clock_source.h" @@ -214,13 +214,11 @@ static const struct dce_aduio_mask audio_mask = { AUX_REG_LIST(id)\ } -static const struct dce110_link_enc_aux_registers link_enc_aux_regs[] = { +static const struct dcn10_link_enc_aux_registers link_enc_aux_regs[] = { aux_regs(0), aux_regs(1), aux_regs(2), - aux_regs(3), - aux_regs(4), - aux_regs(5) + aux_regs(3) }; #define hpd_regs(id)\ @@ -228,13 +226,11 @@ static const struct dce110_link_enc_aux_registers link_enc_aux_regs[] = { HPD_REG_LIST(id)\ } -static const struct dce110_link_enc_hpd_registers link_enc_hpd_regs[] = { +static const struct dcn10_link_enc_hpd_registers link_enc_hpd_regs[] = { hpd_regs(0), hpd_regs(1), hpd_regs(2), - hpd_regs(3), - hpd_regs(4), - hpd_regs(5) + hpd_regs(3) }; #define link_regs(id)\ @@ -243,14 +239,19 @@ static const struct dce110_link_enc_hpd_registers link_enc_hpd_regs[] = { SRI(DP_DPHY_INTERNAL_CTRL, DP, id) \ } -static const struct dce110_link_enc_registers link_enc_regs[] = { +static const struct dcn10_link_enc_registers link_enc_regs[] = { link_regs(0), link_regs(1), link_regs(2), - link_regs(3), - link_regs(4), - link_regs(5), - link_regs(6), + link_regs(3) +}; + +static const struct dcn10_link_enc_shift le_shift = { + LINK_ENCODER_MASK_SH_LIST_DCN10(__SHIFT) +}; + +static const struct dcn10_link_enc_mask le_mask = { + LINK_ENCODER_MASK_SH_LIST_DCN10(_MASK) }; #define ipp_regs(id)\ @@ -446,6 +447,8 @@ static const struct dc_debug debug_defaults_drv = { .vsr_support = true, .performance_trace = false, .az_endpoint_mute_only = true, + .recovery_enabled = false, /*enable this by default after testing.*/ + .max_downscale_src_width = 3840, }; static const struct dc_debug debug_defaults_diags = { @@ -581,20 +584,22 @@ static const struct encoder_feature_support link_enc_feature = { struct link_encoder *dcn10_link_encoder_create( const struct encoder_init_data *enc_init_data) { - struct dce110_link_encoder *enc110 = - kzalloc(sizeof(struct dce110_link_encoder), GFP_KERNEL); + struct dcn10_link_encoder *enc10 = + kzalloc(sizeof(struct dcn10_link_encoder), GFP_KERNEL); - if (!enc110) + if (!enc10) return NULL; - dce110_link_encoder_construct(enc110, + dcn10_link_encoder_construct(enc10, enc_init_data, &link_enc_feature, &link_enc_regs[enc_init_data->transmitter], &link_enc_aux_regs[enc_init_data->channel - 1], - &link_enc_hpd_regs[enc_init_data->hpd_source]); + &link_enc_hpd_regs[enc_init_data->hpd_source], + &le_shift, + &le_mask); - return &enc110->base; + return &enc10->base; } struct clock_source *dcn10_clock_source_create( @@ -1021,6 +1026,7 @@ static bool construct( dc->caps.max_cursor_size = 256; dc->caps.max_slave_planes = 1; dc->caps.is_apu = true; + dc->caps.post_blend_color_processing = false; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_stream_encoder.c index befd8639ad55..653b7b2efe2e 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_stream_encoder.c @@ -603,11 +603,11 @@ void enc1_stream_encoder_set_mst_bandwidth( struct fixed31_32 avg_time_slots_per_mtp) { struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc); - uint32_t x = dal_fixed31_32_floor( + uint32_t x = dc_fixpt_floor( avg_time_slots_per_mtp); - uint32_t y = dal_fixed31_32_ceil( - dal_fixed31_32_shl( - dal_fixed31_32_sub_int( + uint32_t y = dc_fixpt_ceil( + dc_fixpt_shl( + dc_fixpt_sub_int( avg_time_slots_per_mtp, x), 26)); diff --git a/drivers/gpu/drm/amd/display/dc/dm_services.h b/drivers/gpu/drm/amd/display/dc/dm_services.h index 8eafe1af8a5e..4ff9b2bba178 100644 --- a/drivers/gpu/drm/amd/display/dc/dm_services.h +++ b/drivers/gpu/drm/amd/display/dc/dm_services.h @@ -355,10 +355,6 @@ void dm_perf_trace_timestamp(const char *func_name, unsigned int line); /* * Debug and verification hooks */ -bool dm_helpers_dc_conn_log( - struct dc_context *ctx, - struct log_entry *entry, - enum dc_log_type event); void dm_dtn_log_begin(struct dc_context *ctx); void dm_dtn_log_append_v(struct dc_context *ctx, const char *msg, ...); diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h b/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h index b1ad3553f900..47c19f8fe7d1 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h +++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h @@ -108,4 +108,17 @@ enum output_standard { dm_std_uninitialized = 0, dm_std_cvtr2, dm_std_cvt }; +enum mpc_combine_affinity { + dm_mpc_always_when_possible, + dm_mpc_reduce_voltage, + dm_mpc_reduce_voltage_and_clocks +}; + +enum self_refresh_affinity { + dm_try_to_allow_self_refresh_and_mclk_switch, + dm_allow_self_refresh_and_mclk_switch, + dm_allow_self_refresh, + dm_neither_self_refresh_nor_mclk_switch +}; + #endif diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h index ce750edc1e5f..7fa0375939ae 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h +++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h @@ -25,39 +25,39 @@ #ifndef __DISPLAY_MODE_STRUCTS_H__ #define __DISPLAY_MODE_STRUCTS_H__ -typedef struct _vcs_dpi_voltage_scaling_st voltage_scaling_st; -typedef struct _vcs_dpi_soc_bounding_box_st soc_bounding_box_st; -typedef struct _vcs_dpi_ip_params_st ip_params_st; -typedef struct _vcs_dpi_display_pipe_source_params_st display_pipe_source_params_st; -typedef struct _vcs_dpi_display_output_params_st display_output_params_st; -typedef struct _vcs_dpi_display_bandwidth_st display_bandwidth_st; -typedef struct _vcs_dpi_scaler_ratio_depth_st scaler_ratio_depth_st; -typedef struct _vcs_dpi_scaler_taps_st scaler_taps_st; -typedef struct _vcs_dpi_display_pipe_dest_params_st display_pipe_dest_params_st; -typedef struct _vcs_dpi_display_pipe_params_st display_pipe_params_st; -typedef struct _vcs_dpi_display_clocks_and_cfg_st display_clocks_and_cfg_st; -typedef struct _vcs_dpi_display_e2e_pipe_params_st display_e2e_pipe_params_st; -typedef struct _vcs_dpi_dchub_buffer_sizing_st dchub_buffer_sizing_st; -typedef struct _vcs_dpi_watermarks_perf_st watermarks_perf_st; -typedef struct _vcs_dpi_cstate_pstate_watermarks_st cstate_pstate_watermarks_st; -typedef struct _vcs_dpi_wm_calc_pipe_params_st wm_calc_pipe_params_st; -typedef struct _vcs_dpi_vratio_pre_st vratio_pre_st; -typedef struct _vcs_dpi_display_data_rq_misc_params_st display_data_rq_misc_params_st; -typedef struct _vcs_dpi_display_data_rq_sizing_params_st display_data_rq_sizing_params_st; -typedef struct _vcs_dpi_display_data_rq_dlg_params_st display_data_rq_dlg_params_st; -typedef struct _vcs_dpi_display_cur_rq_dlg_params_st display_cur_rq_dlg_params_st; -typedef struct _vcs_dpi_display_rq_dlg_params_st display_rq_dlg_params_st; -typedef struct _vcs_dpi_display_rq_sizing_params_st display_rq_sizing_params_st; -typedef struct _vcs_dpi_display_rq_misc_params_st display_rq_misc_params_st; -typedef struct _vcs_dpi_display_rq_params_st display_rq_params_st; -typedef struct _vcs_dpi_display_dlg_regs_st display_dlg_regs_st; -typedef struct _vcs_dpi_display_ttu_regs_st display_ttu_regs_st; -typedef struct _vcs_dpi_display_data_rq_regs_st display_data_rq_regs_st; -typedef struct _vcs_dpi_display_rq_regs_st display_rq_regs_st; -typedef struct _vcs_dpi_display_dlg_sys_params_st display_dlg_sys_params_st; -typedef struct _vcs_dpi_display_dlg_prefetch_param_st display_dlg_prefetch_param_st; -typedef struct _vcs_dpi_display_pipe_clock_st display_pipe_clock_st; -typedef struct _vcs_dpi_display_arb_params_st display_arb_params_st; +typedef struct _vcs_dpi_voltage_scaling_st voltage_scaling_st; +typedef struct _vcs_dpi_soc_bounding_box_st soc_bounding_box_st; +typedef struct _vcs_dpi_ip_params_st ip_params_st; +typedef struct _vcs_dpi_display_pipe_source_params_st display_pipe_source_params_st; +typedef struct _vcs_dpi_display_output_params_st display_output_params_st; +typedef struct _vcs_dpi_display_bandwidth_st display_bandwidth_st; +typedef struct _vcs_dpi_scaler_ratio_depth_st scaler_ratio_depth_st; +typedef struct _vcs_dpi_scaler_taps_st scaler_taps_st; +typedef struct _vcs_dpi_display_pipe_dest_params_st display_pipe_dest_params_st; +typedef struct _vcs_dpi_display_pipe_params_st display_pipe_params_st; +typedef struct _vcs_dpi_display_clocks_and_cfg_st display_clocks_and_cfg_st; +typedef struct _vcs_dpi_display_e2e_pipe_params_st display_e2e_pipe_params_st; +typedef struct _vcs_dpi_dchub_buffer_sizing_st dchub_buffer_sizing_st; +typedef struct _vcs_dpi_watermarks_perf_st watermarks_perf_st; +typedef struct _vcs_dpi_cstate_pstate_watermarks_st cstate_pstate_watermarks_st; +typedef struct _vcs_dpi_wm_calc_pipe_params_st wm_calc_pipe_params_st; +typedef struct _vcs_dpi_vratio_pre_st vratio_pre_st; +typedef struct _vcs_dpi_display_data_rq_misc_params_st display_data_rq_misc_params_st; +typedef struct _vcs_dpi_display_data_rq_sizing_params_st display_data_rq_sizing_params_st; +typedef struct _vcs_dpi_display_data_rq_dlg_params_st display_data_rq_dlg_params_st; +typedef struct _vcs_dpi_display_cur_rq_dlg_params_st display_cur_rq_dlg_params_st; +typedef struct _vcs_dpi_display_rq_dlg_params_st display_rq_dlg_params_st; +typedef struct _vcs_dpi_display_rq_sizing_params_st display_rq_sizing_params_st; +typedef struct _vcs_dpi_display_rq_misc_params_st display_rq_misc_params_st; +typedef struct _vcs_dpi_display_rq_params_st display_rq_params_st; +typedef struct _vcs_dpi_display_dlg_regs_st display_dlg_regs_st; +typedef struct _vcs_dpi_display_ttu_regs_st display_ttu_regs_st; +typedef struct _vcs_dpi_display_data_rq_regs_st display_data_rq_regs_st; +typedef struct _vcs_dpi_display_rq_regs_st display_rq_regs_st; +typedef struct _vcs_dpi_display_dlg_sys_params_st display_dlg_sys_params_st; +typedef struct _vcs_dpi_display_dlg_prefetch_param_st display_dlg_prefetch_param_st; +typedef struct _vcs_dpi_display_pipe_clock_st display_pipe_clock_st; +typedef struct _vcs_dpi_display_arb_params_st display_arb_params_st; struct _vcs_dpi_voltage_scaling_st { int state; @@ -72,89 +72,107 @@ struct _vcs_dpi_voltage_scaling_st { double dppclk_mhz; }; -struct _vcs_dpi_soc_bounding_box_st { - double sr_exit_time_us; - double sr_enter_plus_exit_time_us; - double urgent_latency_us; - double writeback_latency_us; - double ideal_dram_bw_after_urgent_percent; - unsigned int max_request_size_bytes; - double downspread_percent; - double dram_page_open_time_ns; - double dram_rw_turnaround_time_ns; - double dram_return_buffer_per_channel_bytes; - double dram_channel_width_bytes; +struct _vcs_dpi_soc_bounding_box_st { + double sr_exit_time_us; + double sr_enter_plus_exit_time_us; + double urgent_latency_us; + double urgent_latency_pixel_data_only_us; + double urgent_latency_pixel_mixed_with_vm_data_us; + double urgent_latency_vm_data_only_us; + double writeback_latency_us; + double ideal_dram_bw_after_urgent_percent; + double pct_ideal_dram_sdp_bw_after_urgent_pixel_only; // PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly + double pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm; + double pct_ideal_dram_sdp_bw_after_urgent_vm_only; + double max_avg_sdp_bw_use_normal_percent; + double max_avg_dram_bw_use_normal_percent; + unsigned int max_request_size_bytes; + double downspread_percent; + double dram_page_open_time_ns; + double dram_rw_turnaround_time_ns; + double dram_return_buffer_per_channel_bytes; + double dram_channel_width_bytes; double fabric_datapath_to_dcn_data_return_bytes; double dcn_downspread_percent; double dispclk_dppclk_vco_speed_mhz; double dfs_vco_period_ps; - unsigned int round_trip_ping_latency_dcfclk_cycles; - unsigned int urgent_out_of_order_return_per_channel_bytes; - unsigned int channel_interleave_bytes; - unsigned int num_banks; - unsigned int num_chans; - unsigned int vmm_page_size_bytes; - double dram_clock_change_latency_us; - double writeback_dram_clock_change_latency_us; - unsigned int return_bus_width_bytes; - unsigned int voltage_override; - double xfc_bus_transport_time_us; - double xfc_xbuf_latency_tolerance_us; + unsigned int urgent_out_of_order_return_per_channel_pixel_only_bytes; + unsigned int urgent_out_of_order_return_per_channel_pixel_and_vm_bytes; + unsigned int urgent_out_of_order_return_per_channel_vm_only_bytes; + unsigned int round_trip_ping_latency_dcfclk_cycles; + unsigned int urgent_out_of_order_return_per_channel_bytes; + unsigned int channel_interleave_bytes; + unsigned int num_banks; + unsigned int num_chans; + unsigned int vmm_page_size_bytes; + double dram_clock_change_latency_us; + double writeback_dram_clock_change_latency_us; + unsigned int return_bus_width_bytes; + unsigned int voltage_override; + double xfc_bus_transport_time_us; + double xfc_xbuf_latency_tolerance_us; + int use_urgent_burst_bw; struct _vcs_dpi_voltage_scaling_st clock_limits[7]; }; -struct _vcs_dpi_ip_params_st { - unsigned int max_inter_dcn_tile_repeaters; - unsigned int num_dsc; - unsigned int odm_capable; - unsigned int rob_buffer_size_kbytes; - unsigned int det_buffer_size_kbytes; - unsigned int dpte_buffer_size_in_pte_reqs; - unsigned int pde_proc_buffer_size_64k_reqs; - unsigned int dpp_output_buffer_pixels; - unsigned int opp_output_buffer_lines; - unsigned int pixel_chunk_size_kbytes; - unsigned char pte_enable; - unsigned int pte_chunk_size_kbytes; - unsigned int meta_chunk_size_kbytes; - unsigned int writeback_chunk_size_kbytes; - unsigned int line_buffer_size_bits; - unsigned int max_line_buffer_lines; - unsigned int writeback_luma_buffer_size_kbytes; - unsigned int writeback_chroma_buffer_size_kbytes; - unsigned int writeback_chroma_line_buffer_width_pixels; - unsigned int max_page_table_levels; - unsigned int max_num_dpp; - unsigned int max_num_otg; - unsigned int cursor_chunk_size; - unsigned int cursor_buffer_size; - unsigned int max_num_wb; - unsigned int max_dchub_pscl_bw_pix_per_clk; - unsigned int max_pscl_lb_bw_pix_per_clk; - unsigned int max_lb_vscl_bw_pix_per_clk; - unsigned int max_vscl_hscl_bw_pix_per_clk; - double max_hscl_ratio; - double max_vscl_ratio; - unsigned int hscl_mults; - unsigned int vscl_mults; - unsigned int max_hscl_taps; - unsigned int max_vscl_taps; - unsigned int xfc_supported; - unsigned int xfc_fill_constant_bytes; - double dispclk_ramp_margin_percent; - double xfc_fill_bw_overhead_percent; - double underscan_factor; - unsigned int min_vblank_lines; - unsigned int dppclk_delay_subtotal; - unsigned int dispclk_delay_subtotal; - unsigned int dcfclk_cstate_latency; - unsigned int dppclk_delay_scl; - unsigned int dppclk_delay_scl_lb_only; - unsigned int dppclk_delay_cnvc_formatter; - unsigned int dppclk_delay_cnvc_cursor; - unsigned int is_line_buffer_bpp_fixed; - unsigned int line_buffer_fixed_bpp; - unsigned int dcc_supported; +struct _vcs_dpi_ip_params_st { + bool gpuvm_enable; + bool hostvm_enable; + unsigned int gpuvm_max_page_table_levels; + unsigned int hostvm_max_page_table_levels; + unsigned int hostvm_cached_page_table_levels; + unsigned int pte_group_size_bytes; + unsigned int max_inter_dcn_tile_repeaters; + unsigned int num_dsc; + unsigned int odm_capable; + unsigned int rob_buffer_size_kbytes; + unsigned int det_buffer_size_kbytes; + unsigned int dpte_buffer_size_in_pte_reqs; + unsigned int pde_proc_buffer_size_64k_reqs; + unsigned int dpp_output_buffer_pixels; + unsigned int opp_output_buffer_lines; + unsigned int pixel_chunk_size_kbytes; + unsigned char pte_enable; + unsigned int pte_chunk_size_kbytes; + unsigned int meta_chunk_size_kbytes; + unsigned int writeback_chunk_size_kbytes; + unsigned int line_buffer_size_bits; + unsigned int max_line_buffer_lines; + unsigned int writeback_luma_buffer_size_kbytes; + unsigned int writeback_chroma_buffer_size_kbytes; + unsigned int writeback_chroma_line_buffer_width_pixels; + unsigned int max_page_table_levels; + unsigned int max_num_dpp; + unsigned int max_num_otg; + unsigned int cursor_chunk_size; + unsigned int cursor_buffer_size; + unsigned int max_num_wb; + unsigned int max_dchub_pscl_bw_pix_per_clk; + unsigned int max_pscl_lb_bw_pix_per_clk; + unsigned int max_lb_vscl_bw_pix_per_clk; + unsigned int max_vscl_hscl_bw_pix_per_clk; + double max_hscl_ratio; + double max_vscl_ratio; + unsigned int hscl_mults; + unsigned int vscl_mults; + unsigned int max_hscl_taps; + unsigned int max_vscl_taps; + unsigned int xfc_supported; + unsigned int xfc_fill_constant_bytes; + double dispclk_ramp_margin_percent; + double xfc_fill_bw_overhead_percent; + double underscan_factor; + unsigned int min_vblank_lines; + unsigned int dppclk_delay_subtotal; + unsigned int dispclk_delay_subtotal; + unsigned int dcfclk_cstate_latency; + unsigned int dppclk_delay_scl; + unsigned int dppclk_delay_scl_lb_only; + unsigned int dppclk_delay_cnvc_formatter; + unsigned int dppclk_delay_cnvc_cursor; + unsigned int is_line_buffer_bpp_fixed; + unsigned int line_buffer_fixed_bpp; + unsigned int dcc_supported; unsigned int IsLineBufferBppFixed; unsigned int LineBufferFixedBpp; @@ -169,41 +187,45 @@ struct _vcs_dpi_display_xfc_params_st { int xfc_slv_chunk_size_bytes; }; -struct _vcs_dpi_display_pipe_source_params_st { - int source_format; - unsigned char dcc; - unsigned int dcc_override; - unsigned int dcc_rate; - unsigned char dcc_use_global; - unsigned char vm; - unsigned char vm_levels_force_en; - unsigned int vm_levels_force; - int source_scan; - int sw_mode; - int macro_tile_size; - unsigned char is_display_sw; - unsigned int viewport_width; - unsigned int viewport_height; - unsigned int viewport_y_y; - unsigned int viewport_y_c; - unsigned int viewport_width_c; - unsigned int viewport_height_c; - unsigned int data_pitch; - unsigned int data_pitch_c; - unsigned int meta_pitch; - unsigned int meta_pitch_c; - unsigned int cur0_src_width; - int cur0_bpp; - unsigned int cur1_src_width; - int cur1_bpp; - int num_cursors; - unsigned char is_hsplit; - unsigned char dynamic_metadata_enable; - unsigned int dynamic_metadata_lines_before_active; - unsigned int dynamic_metadata_xmit_bytes; - unsigned int hsplit_grp; - unsigned char xfc_enable; - unsigned char xfc_slave; +struct _vcs_dpi_display_pipe_source_params_st { + int source_format; + unsigned char dcc; + unsigned int dcc_override; + unsigned int dcc_rate; + unsigned char dcc_use_global; + unsigned char vm; + bool gpuvm; // gpuvm enabled + bool hostvm; // hostvm enabled + bool gpuvm_levels_force_en; + unsigned int gpuvm_levels_force; + bool hostvm_levels_force_en; + unsigned int hostvm_levels_force; + int source_scan; + int sw_mode; + int macro_tile_size; + unsigned char is_display_sw; + unsigned int viewport_width; + unsigned int viewport_height; + unsigned int viewport_y_y; + unsigned int viewport_y_c; + unsigned int viewport_width_c; + unsigned int viewport_height_c; + unsigned int data_pitch; + unsigned int data_pitch_c; + unsigned int meta_pitch; + unsigned int meta_pitch_c; + unsigned int cur0_src_width; + int cur0_bpp; + unsigned int cur1_src_width; + int cur1_bpp; + int num_cursors; + unsigned char is_hsplit; + unsigned char dynamic_metadata_enable; + unsigned int dynamic_metadata_lines_before_active; + unsigned int dynamic_metadata_xmit_bytes; + unsigned int hsplit_grp; + unsigned char xfc_enable; + unsigned char xfc_slave; struct _vcs_dpi_display_xfc_params_st xfc_params; }; struct writeback_st { @@ -219,335 +241,335 @@ struct writeback_st { double wb_vratio; }; -struct _vcs_dpi_display_output_params_st { - int dp_lanes; - int output_bpp; - int dsc_enable; - int wb_enable; - int num_active_wb; - int opp_input_bpc; - int output_type; - int output_format; - int output_standard; - int dsc_slices; +struct _vcs_dpi_display_output_params_st { + int dp_lanes; + int output_bpp; + int dsc_enable; + int wb_enable; + int num_active_wb; + int output_bpc; + int output_type; + int output_format; + int output_standard; + int dsc_slices; struct writeback_st wb; }; -struct _vcs_dpi_display_bandwidth_st { - double total_bw_consumed_gbps; - double guaranteed_urgent_return_bw_gbps; -}; - -struct _vcs_dpi_scaler_ratio_depth_st { - double hscl_ratio; - double vscl_ratio; - double hscl_ratio_c; - double vscl_ratio_c; - double vinit; - double vinit_c; - double vinit_bot; - double vinit_bot_c; - int lb_depth; - int scl_enable; -}; - -struct _vcs_dpi_scaler_taps_st { - unsigned int htaps; - unsigned int vtaps; - unsigned int htaps_c; - unsigned int vtaps_c; -}; - -struct _vcs_dpi_display_pipe_dest_params_st { - unsigned int recout_width; - unsigned int recout_height; - unsigned int full_recout_width; - unsigned int full_recout_height; - unsigned int hblank_start; - unsigned int hblank_end; - unsigned int vblank_start; - unsigned int vblank_end; - unsigned int htotal; - unsigned int vtotal; - unsigned int vactive; - unsigned int hactive; - unsigned int vstartup_start; - unsigned int vupdate_offset; - unsigned int vupdate_width; - unsigned int vready_offset; - unsigned char interlaced; - unsigned char underscan; - double pixel_rate_mhz; - unsigned char synchronized_vblank_all_planes; - unsigned char otg_inst; - unsigned char odm_split_cnt; - unsigned char odm_combine; -}; - -struct _vcs_dpi_display_pipe_params_st { - display_pipe_source_params_st src; - display_pipe_dest_params_st dest; - scaler_ratio_depth_st scale_ratio_depth; - scaler_taps_st scale_taps; -}; - -struct _vcs_dpi_display_clocks_and_cfg_st { - int voltage; - double dppclk_mhz; - double refclk_mhz; - double dispclk_mhz; - double dcfclk_mhz; - double socclk_mhz; -}; - -struct _vcs_dpi_display_e2e_pipe_params_st { - display_pipe_params_st pipe; - display_output_params_st dout; - display_clocks_and_cfg_st clks_cfg; -}; - -struct _vcs_dpi_dchub_buffer_sizing_st { - unsigned int swath_width_y; - unsigned int swath_height_y; - unsigned int swath_height_c; - unsigned int detail_buffer_size_y; -}; - -struct _vcs_dpi_watermarks_perf_st { - double stutter_eff_in_active_region_percent; - double urgent_latency_supported_us; - double non_urgent_latency_supported_us; - double dram_clock_change_margin_us; - double dram_access_eff_percent; -}; - -struct _vcs_dpi_cstate_pstate_watermarks_st { - double cstate_exit_us; - double cstate_enter_plus_exit_us; - double pstate_change_us; -}; - -struct _vcs_dpi_wm_calc_pipe_params_st { - unsigned int num_dpp; - int voltage; - int output_type; - double dcfclk_mhz; - double socclk_mhz; - double dppclk_mhz; - double pixclk_mhz; - unsigned char interlace_en; - unsigned char pte_enable; - unsigned char dcc_enable; - double dcc_rate; - double bytes_per_pixel_c; - double bytes_per_pixel_y; - unsigned int swath_width_y; - unsigned int swath_height_y; - unsigned int swath_height_c; - unsigned int det_buffer_size_y; - double h_ratio; - double v_ratio; - unsigned int h_taps; - unsigned int h_total; - unsigned int v_total; - unsigned int v_active; - unsigned int e2e_index; - double display_pipe_line_delivery_time; - double read_bw; - unsigned int lines_in_det_y; - unsigned int lines_in_det_y_rounded_down_to_swath; - double full_det_buffering_time; - double dcfclk_deepsleep_mhz_per_plane; -}; - -struct _vcs_dpi_vratio_pre_st { - double vratio_pre_l; - double vratio_pre_c; -}; - -struct _vcs_dpi_display_data_rq_misc_params_st { - unsigned int full_swath_bytes; - unsigned int stored_swath_bytes; - unsigned int blk256_height; - unsigned int blk256_width; - unsigned int req_height; - unsigned int req_width; -}; - -struct _vcs_dpi_display_data_rq_sizing_params_st { - unsigned int chunk_bytes; - unsigned int min_chunk_bytes; - unsigned int meta_chunk_bytes; - unsigned int min_meta_chunk_bytes; - unsigned int mpte_group_bytes; - unsigned int dpte_group_bytes; -}; - -struct _vcs_dpi_display_data_rq_dlg_params_st { - unsigned int swath_width_ub; - unsigned int swath_height; - unsigned int req_per_swath_ub; - unsigned int meta_pte_bytes_per_frame_ub; - unsigned int dpte_req_per_row_ub; - unsigned int dpte_groups_per_row_ub; - unsigned int dpte_row_height; - unsigned int dpte_bytes_per_row_ub; - unsigned int meta_chunks_per_row_ub; - unsigned int meta_req_per_row_ub; - unsigned int meta_row_height; - unsigned int meta_bytes_per_row_ub; -}; - -struct _vcs_dpi_display_cur_rq_dlg_params_st { - unsigned char enable; - unsigned int swath_height; - unsigned int req_per_line; -}; - -struct _vcs_dpi_display_rq_dlg_params_st { - display_data_rq_dlg_params_st rq_l; - display_data_rq_dlg_params_st rq_c; - display_cur_rq_dlg_params_st rq_cur0; -}; - -struct _vcs_dpi_display_rq_sizing_params_st { - display_data_rq_sizing_params_st rq_l; - display_data_rq_sizing_params_st rq_c; -}; - -struct _vcs_dpi_display_rq_misc_params_st { - display_data_rq_misc_params_st rq_l; - display_data_rq_misc_params_st rq_c; -}; - -struct _vcs_dpi_display_rq_params_st { - unsigned char yuv420; - unsigned char yuv420_10bpc; - display_rq_misc_params_st misc; - display_rq_sizing_params_st sizing; - display_rq_dlg_params_st dlg; -}; - -struct _vcs_dpi_display_dlg_regs_st { - unsigned int refcyc_h_blank_end; - unsigned int dlg_vblank_end; - unsigned int min_dst_y_next_start; - unsigned int refcyc_per_htotal; - unsigned int refcyc_x_after_scaler; - unsigned int dst_y_after_scaler; - unsigned int dst_y_prefetch; - unsigned int dst_y_per_vm_vblank; - unsigned int dst_y_per_row_vblank; - unsigned int dst_y_per_vm_flip; - unsigned int dst_y_per_row_flip; - unsigned int ref_freq_to_pix_freq; - unsigned int vratio_prefetch; - unsigned int vratio_prefetch_c; - unsigned int refcyc_per_pte_group_vblank_l; - unsigned int refcyc_per_pte_group_vblank_c; - unsigned int refcyc_per_meta_chunk_vblank_l; - unsigned int refcyc_per_meta_chunk_vblank_c; - unsigned int refcyc_per_pte_group_flip_l; - unsigned int refcyc_per_pte_group_flip_c; - unsigned int refcyc_per_meta_chunk_flip_l; - unsigned int refcyc_per_meta_chunk_flip_c; - unsigned int dst_y_per_pte_row_nom_l; - unsigned int dst_y_per_pte_row_nom_c; - unsigned int refcyc_per_pte_group_nom_l; - unsigned int refcyc_per_pte_group_nom_c; - unsigned int dst_y_per_meta_row_nom_l; - unsigned int dst_y_per_meta_row_nom_c; - unsigned int refcyc_per_meta_chunk_nom_l; - unsigned int refcyc_per_meta_chunk_nom_c; - unsigned int refcyc_per_line_delivery_pre_l; - unsigned int refcyc_per_line_delivery_pre_c; - unsigned int refcyc_per_line_delivery_l; - unsigned int refcyc_per_line_delivery_c; - unsigned int chunk_hdl_adjust_cur0; - unsigned int chunk_hdl_adjust_cur1; - unsigned int vready_after_vcount0; - unsigned int dst_y_offset_cur0; - unsigned int dst_y_offset_cur1; - unsigned int xfc_reg_transfer_delay; - unsigned int xfc_reg_precharge_delay; - unsigned int xfc_reg_remote_surface_flip_latency; - unsigned int xfc_reg_prefetch_margin; - unsigned int dst_y_delta_drq_limit; -}; - -struct _vcs_dpi_display_ttu_regs_st { - unsigned int qos_level_low_wm; - unsigned int qos_level_high_wm; - unsigned int min_ttu_vblank; - unsigned int qos_level_flip; - unsigned int refcyc_per_req_delivery_l; - unsigned int refcyc_per_req_delivery_c; - unsigned int refcyc_per_req_delivery_cur0; - unsigned int refcyc_per_req_delivery_cur1; - unsigned int refcyc_per_req_delivery_pre_l; - unsigned int refcyc_per_req_delivery_pre_c; - unsigned int refcyc_per_req_delivery_pre_cur0; - unsigned int refcyc_per_req_delivery_pre_cur1; - unsigned int qos_level_fixed_l; - unsigned int qos_level_fixed_c; - unsigned int qos_level_fixed_cur0; - unsigned int qos_level_fixed_cur1; - unsigned int qos_ramp_disable_l; - unsigned int qos_ramp_disable_c; - unsigned int qos_ramp_disable_cur0; - unsigned int qos_ramp_disable_cur1; -}; - -struct _vcs_dpi_display_data_rq_regs_st { - unsigned int chunk_size; - unsigned int min_chunk_size; - unsigned int meta_chunk_size; - unsigned int min_meta_chunk_size; - unsigned int dpte_group_size; - unsigned int mpte_group_size; - unsigned int swath_height; - unsigned int pte_row_height_linear; -}; - -struct _vcs_dpi_display_rq_regs_st { - display_data_rq_regs_st rq_regs_l; - display_data_rq_regs_st rq_regs_c; - unsigned int drq_expansion_mode; - unsigned int prq_expansion_mode; - unsigned int mrq_expansion_mode; - unsigned int crq_expansion_mode; - unsigned int plane1_base_address; -}; - -struct _vcs_dpi_display_dlg_sys_params_st { - double t_mclk_wm_us; - double t_urg_wm_us; - double t_sr_wm_us; - double t_extra_us; - double mem_trip_us; - double t_srx_delay_us; - double deepsleep_dcfclk_mhz; - double total_flip_bw; - unsigned int total_flip_bytes; -}; - -struct _vcs_dpi_display_dlg_prefetch_param_st { - double prefetch_bw; - unsigned int flip_bytes; -}; - -struct _vcs_dpi_display_pipe_clock_st { - double dcfclk_mhz; - double dispclk_mhz; - double socclk_mhz; - double dscclk_mhz[6]; - double dppclk_mhz[6]; -}; - -struct _vcs_dpi_display_arb_params_st { - int max_req_outstanding; - int min_req_outstanding; - int sat_level_us; +struct _vcs_dpi_display_bandwidth_st { + double total_bw_consumed_gbps; + double guaranteed_urgent_return_bw_gbps; +}; + +struct _vcs_dpi_scaler_ratio_depth_st { + double hscl_ratio; + double vscl_ratio; + double hscl_ratio_c; + double vscl_ratio_c; + double vinit; + double vinit_c; + double vinit_bot; + double vinit_bot_c; + int lb_depth; + int scl_enable; +}; + +struct _vcs_dpi_scaler_taps_st { + unsigned int htaps; + unsigned int vtaps; + unsigned int htaps_c; + unsigned int vtaps_c; +}; + +struct _vcs_dpi_display_pipe_dest_params_st { + unsigned int recout_width; + unsigned int recout_height; + unsigned int full_recout_width; + unsigned int full_recout_height; + unsigned int hblank_start; + unsigned int hblank_end; + unsigned int vblank_start; + unsigned int vblank_end; + unsigned int htotal; + unsigned int vtotal; + unsigned int vactive; + unsigned int hactive; + unsigned int vstartup_start; + unsigned int vupdate_offset; + unsigned int vupdate_width; + unsigned int vready_offset; + unsigned char interlaced; + unsigned char underscan; + double pixel_rate_mhz; + unsigned char synchronized_vblank_all_planes; + unsigned char otg_inst; + unsigned char odm_split_cnt; + unsigned char odm_combine; +}; + +struct _vcs_dpi_display_pipe_params_st { + display_pipe_source_params_st src; + display_pipe_dest_params_st dest; + scaler_ratio_depth_st scale_ratio_depth; + scaler_taps_st scale_taps; +}; + +struct _vcs_dpi_display_clocks_and_cfg_st { + int voltage; + double dppclk_mhz; + double refclk_mhz; + double dispclk_mhz; + double dcfclk_mhz; + double socclk_mhz; +}; + +struct _vcs_dpi_display_e2e_pipe_params_st { + display_pipe_params_st pipe; + display_output_params_st dout; + display_clocks_and_cfg_st clks_cfg; +}; + +struct _vcs_dpi_dchub_buffer_sizing_st { + unsigned int swath_width_y; + unsigned int swath_height_y; + unsigned int swath_height_c; + unsigned int detail_buffer_size_y; +}; + +struct _vcs_dpi_watermarks_perf_st { + double stutter_eff_in_active_region_percent; + double urgent_latency_supported_us; + double non_urgent_latency_supported_us; + double dram_clock_change_margin_us; + double dram_access_eff_percent; +}; + +struct _vcs_dpi_cstate_pstate_watermarks_st { + double cstate_exit_us; + double cstate_enter_plus_exit_us; + double pstate_change_us; +}; + +struct _vcs_dpi_wm_calc_pipe_params_st { + unsigned int num_dpp; + int voltage; + int output_type; + double dcfclk_mhz; + double socclk_mhz; + double dppclk_mhz; + double pixclk_mhz; + unsigned char interlace_en; + unsigned char pte_enable; + unsigned char dcc_enable; + double dcc_rate; + double bytes_per_pixel_c; + double bytes_per_pixel_y; + unsigned int swath_width_y; + unsigned int swath_height_y; + unsigned int swath_height_c; + unsigned int det_buffer_size_y; + double h_ratio; + double v_ratio; + unsigned int h_taps; + unsigned int h_total; + unsigned int v_total; + unsigned int v_active; + unsigned int e2e_index; + double display_pipe_line_delivery_time; + double read_bw; + unsigned int lines_in_det_y; + unsigned int lines_in_det_y_rounded_down_to_swath; + double full_det_buffering_time; + double dcfclk_deepsleep_mhz_per_plane; +}; + +struct _vcs_dpi_vratio_pre_st { + double vratio_pre_l; + double vratio_pre_c; +}; + +struct _vcs_dpi_display_data_rq_misc_params_st { + unsigned int full_swath_bytes; + unsigned int stored_swath_bytes; + unsigned int blk256_height; + unsigned int blk256_width; + unsigned int req_height; + unsigned int req_width; +}; + +struct _vcs_dpi_display_data_rq_sizing_params_st { + unsigned int chunk_bytes; + unsigned int min_chunk_bytes; + unsigned int meta_chunk_bytes; + unsigned int min_meta_chunk_bytes; + unsigned int mpte_group_bytes; + unsigned int dpte_group_bytes; +}; + +struct _vcs_dpi_display_data_rq_dlg_params_st { + unsigned int swath_width_ub; + unsigned int swath_height; + unsigned int req_per_swath_ub; + unsigned int meta_pte_bytes_per_frame_ub; + unsigned int dpte_req_per_row_ub; + unsigned int dpte_groups_per_row_ub; + unsigned int dpte_row_height; + unsigned int dpte_bytes_per_row_ub; + unsigned int meta_chunks_per_row_ub; + unsigned int meta_req_per_row_ub; + unsigned int meta_row_height; + unsigned int meta_bytes_per_row_ub; +}; + +struct _vcs_dpi_display_cur_rq_dlg_params_st { + unsigned char enable; + unsigned int swath_height; + unsigned int req_per_line; +}; + +struct _vcs_dpi_display_rq_dlg_params_st { + display_data_rq_dlg_params_st rq_l; + display_data_rq_dlg_params_st rq_c; + display_cur_rq_dlg_params_st rq_cur0; +}; + +struct _vcs_dpi_display_rq_sizing_params_st { + display_data_rq_sizing_params_st rq_l; + display_data_rq_sizing_params_st rq_c; +}; + +struct _vcs_dpi_display_rq_misc_params_st { + display_data_rq_misc_params_st rq_l; + display_data_rq_misc_params_st rq_c; +}; + +struct _vcs_dpi_display_rq_params_st { + unsigned char yuv420; + unsigned char yuv420_10bpc; + display_rq_misc_params_st misc; + display_rq_sizing_params_st sizing; + display_rq_dlg_params_st dlg; +}; + +struct _vcs_dpi_display_dlg_regs_st { + unsigned int refcyc_h_blank_end; + unsigned int dlg_vblank_end; + unsigned int min_dst_y_next_start; + unsigned int refcyc_per_htotal; + unsigned int refcyc_x_after_scaler; + unsigned int dst_y_after_scaler; + unsigned int dst_y_prefetch; + unsigned int dst_y_per_vm_vblank; + unsigned int dst_y_per_row_vblank; + unsigned int dst_y_per_vm_flip; + unsigned int dst_y_per_row_flip; + unsigned int ref_freq_to_pix_freq; + unsigned int vratio_prefetch; + unsigned int vratio_prefetch_c; + unsigned int refcyc_per_pte_group_vblank_l; + unsigned int refcyc_per_pte_group_vblank_c; + unsigned int refcyc_per_meta_chunk_vblank_l; + unsigned int refcyc_per_meta_chunk_vblank_c; + unsigned int refcyc_per_pte_group_flip_l; + unsigned int refcyc_per_pte_group_flip_c; + unsigned int refcyc_per_meta_chunk_flip_l; + unsigned int refcyc_per_meta_chunk_flip_c; + unsigned int dst_y_per_pte_row_nom_l; + unsigned int dst_y_per_pte_row_nom_c; + unsigned int refcyc_per_pte_group_nom_l; + unsigned int refcyc_per_pte_group_nom_c; + unsigned int dst_y_per_meta_row_nom_l; + unsigned int dst_y_per_meta_row_nom_c; + unsigned int refcyc_per_meta_chunk_nom_l; + unsigned int refcyc_per_meta_chunk_nom_c; + unsigned int refcyc_per_line_delivery_pre_l; + unsigned int refcyc_per_line_delivery_pre_c; + unsigned int refcyc_per_line_delivery_l; + unsigned int refcyc_per_line_delivery_c; + unsigned int chunk_hdl_adjust_cur0; + unsigned int chunk_hdl_adjust_cur1; + unsigned int vready_after_vcount0; + unsigned int dst_y_offset_cur0; + unsigned int dst_y_offset_cur1; + unsigned int xfc_reg_transfer_delay; + unsigned int xfc_reg_precharge_delay; + unsigned int xfc_reg_remote_surface_flip_latency; + unsigned int xfc_reg_prefetch_margin; + unsigned int dst_y_delta_drq_limit; +}; + +struct _vcs_dpi_display_ttu_regs_st { + unsigned int qos_level_low_wm; + unsigned int qos_level_high_wm; + unsigned int min_ttu_vblank; + unsigned int qos_level_flip; + unsigned int refcyc_per_req_delivery_l; + unsigned int refcyc_per_req_delivery_c; + unsigned int refcyc_per_req_delivery_cur0; + unsigned int refcyc_per_req_delivery_cur1; + unsigned int refcyc_per_req_delivery_pre_l; + unsigned int refcyc_per_req_delivery_pre_c; + unsigned int refcyc_per_req_delivery_pre_cur0; + unsigned int refcyc_per_req_delivery_pre_cur1; + unsigned int qos_level_fixed_l; + unsigned int qos_level_fixed_c; + unsigned int qos_level_fixed_cur0; + unsigned int qos_level_fixed_cur1; + unsigned int qos_ramp_disable_l; + unsigned int qos_ramp_disable_c; + unsigned int qos_ramp_disable_cur0; + unsigned int qos_ramp_disable_cur1; +}; + +struct _vcs_dpi_display_data_rq_regs_st { + unsigned int chunk_size; + unsigned int min_chunk_size; + unsigned int meta_chunk_size; + unsigned int min_meta_chunk_size; + unsigned int dpte_group_size; + unsigned int mpte_group_size; + unsigned int swath_height; + unsigned int pte_row_height_linear; +}; + +struct _vcs_dpi_display_rq_regs_st { + display_data_rq_regs_st rq_regs_l; + display_data_rq_regs_st rq_regs_c; + unsigned int drq_expansion_mode; + unsigned int prq_expansion_mode; + unsigned int mrq_expansion_mode; + unsigned int crq_expansion_mode; + unsigned int plane1_base_address; +}; + +struct _vcs_dpi_display_dlg_sys_params_st { + double t_mclk_wm_us; + double t_urg_wm_us; + double t_sr_wm_us; + double t_extra_us; + double mem_trip_us; + double t_srx_delay_us; + double deepsleep_dcfclk_mhz; + double total_flip_bw; + unsigned int total_flip_bytes; +}; + +struct _vcs_dpi_display_dlg_prefetch_param_st { + double prefetch_bw; + unsigned int flip_bytes; +}; + +struct _vcs_dpi_display_pipe_clock_st { + double dcfclk_mhz; + double dispclk_mhz; + double socclk_mhz; + double dscclk_mhz[6]; + double dppclk_mhz[6]; +}; + +struct _vcs_dpi_display_arb_params_st { + int max_req_outstanding; + int min_req_outstanding; + int sat_level_us; }; #endif /*__DISPLAY_MODE_STRUCTS_H__*/ diff --git a/drivers/gpu/drm/amd/display/dc/dml/dml_inline_defs.h b/drivers/gpu/drm/amd/display/dc/dml/dml_inline_defs.h index f9cf08357989..e8ce08567cd8 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dml_inline_defs.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dml_inline_defs.h @@ -35,6 +35,16 @@ static inline double dml_min(double a, double b) return (double) dcn_bw_min2(a, b); } +static inline double dml_min3(double a, double b, double c) +{ + return dml_min(dml_min(a, b), c); +} + +static inline double dml_min4(double a, double b, double c, double d) +{ + return dml_min(dml_min(a, b), dml_min(c, d)); +} + static inline double dml_max(double a, double b) { return (double) dcn_bw_max2(a, b); diff --git a/drivers/gpu/drm/amd/display/dc/gpio/hw_factory.c b/drivers/gpu/drm/amd/display/dc/gpio/hw_factory.c index 61fe484da1a0..0caee3523017 100644 --- a/drivers/gpu/drm/amd/display/dc/gpio/hw_factory.c +++ b/drivers/gpu/drm/amd/display/dc/gpio/hw_factory.c @@ -75,9 +75,7 @@ bool dal_hw_factory_init( return true; case DCE_VERSION_11_0: case DCE_VERSION_11_2: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case DCE_VERSION_11_22: -#endif dal_hw_factory_dce110_init(factory); return true; case DCE_VERSION_12_0: diff --git a/drivers/gpu/drm/amd/display/dc/gpio/hw_translate.c b/drivers/gpu/drm/amd/display/dc/gpio/hw_translate.c index 910ae2b7bf64..55c707488541 100644 --- a/drivers/gpu/drm/amd/display/dc/gpio/hw_translate.c +++ b/drivers/gpu/drm/amd/display/dc/gpio/hw_translate.c @@ -72,9 +72,7 @@ bool dal_hw_translate_init( case DCE_VERSION_10_0: case DCE_VERSION_11_0: case DCE_VERSION_11_2: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case DCE_VERSION_11_22: -#endif dal_hw_translate_dce110_init(translate); return true; case DCE_VERSION_12_0: diff --git a/drivers/gpu/drm/amd/display/dc/i2caux/i2caux.c b/drivers/gpu/drm/amd/display/dc/i2caux/i2caux.c index c3d7c320fdba..14dc8c94d862 100644 --- a/drivers/gpu/drm/amd/display/dc/i2caux/i2caux.c +++ b/drivers/gpu/drm/amd/display/dc/i2caux/i2caux.c @@ -83,9 +83,7 @@ struct i2caux *dal_i2caux_create( case DCE_VERSION_8_3: return dal_i2caux_dce80_create(ctx); case DCE_VERSION_11_2: -#if defined(CONFIG_DRM_AMD_DC_VEGAM) case DCE_VERSION_11_22: -#endif return dal_i2caux_dce112_create(ctx); case DCE_VERSION_11_0: return dal_i2caux_dce110_create(ctx); diff --git a/drivers/gpu/drm/amd/display/dc/inc/dce_calcs.h b/drivers/gpu/drm/amd/display/dc/inc/dce_calcs.h index 933ea7a1e18b..eece165206f9 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/dce_calcs.h +++ b/drivers/gpu/drm/amd/display/dc/inc/dce_calcs.h @@ -43,9 +43,7 @@ enum bw_calcs_version { BW_CALCS_VERSION_POLARIS10, BW_CALCS_VERSION_POLARIS11, BW_CALCS_VERSION_POLARIS12, -#if defined(CONFIG_DRM_AMD_DC_VEGAM) BW_CALCS_VERSION_VEGAM, -#endif BW_CALCS_VERSION_STONEY, BW_CALCS_VERSION_VEGA10 }; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h index 331f8ff57ed7..97df82cddf82 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h @@ -121,6 +121,8 @@ struct hubp_funcs { void (*hubp_clk_cntl)(struct hubp *hubp, bool enable); void (*hubp_vtg_sel)(struct hubp *hubp, uint32_t otg_inst); void (*hubp_read_state)(struct hubp *hubp); + void (*hubp_disable_control)(struct hubp *hubp, bool disable_hubp); + unsigned int (*hubp_get_underflow_status)(struct hubp *hubp); }; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h index 29abf3ecb39c..63fc6c499789 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h @@ -83,11 +83,6 @@ struct hw_sequencer_funcs { int num_planes, struct dc_state *context); - void (*set_plane_config)( - const struct dc *dc, - struct pipe_ctx *pipe_ctx, - struct resource_context *res_ctx); - void (*program_gamut_remap)( struct pipe_ctx *pipe_ctx); diff --git a/drivers/gpu/drm/amd/display/dc/irq_types.h b/drivers/gpu/drm/amd/display/dc/irq_types.h index cc3b1bc6cedd..0b5f3a278c22 100644 --- a/drivers/gpu/drm/amd/display/dc/irq_types.h +++ b/drivers/gpu/drm/amd/display/dc/irq_types.h @@ -26,6 +26,8 @@ #ifndef __DAL_IRQ_TYPES_H__ #define __DAL_IRQ_TYPES_H__ +#include "os_types.h" + struct dc_context; typedef void (*interrupt_handler)(void *); diff --git a/drivers/gpu/drm/amd/display/include/dal_asic_id.h b/drivers/gpu/drm/amd/display/include/dal_asic_id.h index 1b987b6a347d..25029ed42d89 100644 --- a/drivers/gpu/drm/amd/display/include/dal_asic_id.h +++ b/drivers/gpu/drm/amd/display/include/dal_asic_id.h @@ -86,6 +86,7 @@ #define VI_POLARIS10_P_A0 80 #define VI_POLARIS11_M_A0 90 #define VI_POLARIS12_V_A0 100 +#define VI_VEGAM_A0 110 #define VI_UNKNOWN 0xFF @@ -98,14 +99,9 @@ (eChipRev < VI_POLARIS11_M_A0)) #define ASIC_REV_IS_POLARIS11_M(eChipRev) ((eChipRev >= VI_POLARIS11_M_A0) && \ (eChipRev < VI_POLARIS12_V_A0)) -#if defined(CONFIG_DRM_AMD_DC_VEGAM) -#define VI_VEGAM_A0 110 #define ASIC_REV_IS_POLARIS12_V(eChipRev) ((eChipRev >= VI_POLARIS12_V_A0) && \ (eChipRev < VI_VEGAM_A0)) #define ASIC_REV_IS_VEGAM(eChipRev) (eChipRev >= VI_VEGAM_A0) -#else -#define ASIC_REV_IS_POLARIS12_V(eChipRev) (eChipRev >= VI_POLARIS12_V_A0) -#endif /* DCE11 */ #define CZ_CARRIZO_A0 0x01 @@ -117,17 +113,19 @@ ((rev >= STONEY_A0) && (rev < CZ_UNKNOWN)) /* DCE12 */ +#define AI_UNKNOWN 0xFF #define AI_GREENLAND_P_A0 1 #define AI_GREENLAND_P_A1 2 #define AI_UNKNOWN 0xFF #define AI_VEGA12_P_A0 20 +#define AI_VEGA20_P_A0 40 #define ASICREV_IS_GREENLAND_M(eChipRev) (eChipRev < AI_VEGA12_P_A0) #define ASICREV_IS_GREENLAND_P(eChipRev) (eChipRev < AI_VEGA12_P_A0) -#define ASICREV_IS_VEGA12_P(eChipRev) ((eChipRev >= AI_VEGA12_P_A0) && (eChipRev < AI_UNKNOWN)) -#define ASICREV_IS_VEGA12_p(eChipRev) ((eChipRev >= AI_VEGA12_P_A0) && (eChipRev < AI_UNKNOWN)) +#define ASICREV_IS_VEGA12_P(eChipRev) ((eChipRev >= AI_VEGA12_P_A0) && (eChipRev < AI_VEGA20_P_A0)) +#define ASICREV_IS_VEGA20_P(eChipRev) ((eChipRev >= AI_VEGA20_P_A0) && (eChipRev < AI_UNKNOWN)) /* DCN1_0 */ #define INTERNAL_REV_RAVEN_A0 0x00 /* First spin of Raven */ diff --git a/drivers/gpu/drm/amd/display/include/dal_types.h b/drivers/gpu/drm/amd/display/include/dal_types.h index 5b1f8cef0c22..840142b65f8b 100644 --- a/drivers/gpu/drm/amd/display/include/dal_types.h +++ b/drivers/gpu/drm/amd/display/include/dal_types.h @@ -40,9 +40,7 @@ enum dce_version { DCE_VERSION_10_0, DCE_VERSION_11_0, DCE_VERSION_11_2, -#if defined(CONFIG_DRM_AMD_DC_VEGAM) DCE_VERSION_11_22, -#endif DCE_VERSION_12_0, DCE_VERSION_MAX, DCN_VERSION_1_0, diff --git a/drivers/gpu/drm/amd/display/include/fixed31_32.h b/drivers/gpu/drm/amd/display/include/fixed31_32.h index 16cbdb43d856..bb0d4ebba9f0 100644 --- a/drivers/gpu/drm/amd/display/include/fixed31_32.h +++ b/drivers/gpu/drm/amd/display/include/fixed31_32.h @@ -27,6 +27,12 @@ #define __DAL_FIXED31_32_H__ #define FIXED31_32_BITS_PER_FRACTIONAL_PART 32 +#ifndef LLONG_MIN +#define LLONG_MIN (1LL<<63) +#endif +#ifndef LLONG_MAX +#define LLONG_MAX (-1LL>>1) +#endif /* * @brief @@ -45,21 +51,22 @@ struct fixed31_32 { long long value; }; + /* * @brief * Useful constants */ -static const struct fixed31_32 dal_fixed31_32_zero = { 0 }; -static const struct fixed31_32 dal_fixed31_32_epsilon = { 1LL }; -static const struct fixed31_32 dal_fixed31_32_half = { 0x80000000LL }; -static const struct fixed31_32 dal_fixed31_32_one = { 0x100000000LL }; +static const struct fixed31_32 dc_fixpt_zero = { 0 }; +static const struct fixed31_32 dc_fixpt_epsilon = { 1LL }; +static const struct fixed31_32 dc_fixpt_half = { 0x80000000LL }; +static const struct fixed31_32 dc_fixpt_one = { 0x100000000LL }; -static const struct fixed31_32 dal_fixed31_32_pi = { 13493037705LL }; -static const struct fixed31_32 dal_fixed31_32_two_pi = { 26986075409LL }; -static const struct fixed31_32 dal_fixed31_32_e = { 11674931555LL }; -static const struct fixed31_32 dal_fixed31_32_ln2 = { 2977044471LL }; -static const struct fixed31_32 dal_fixed31_32_ln2_div_2 = { 1488522236LL }; +static const struct fixed31_32 dc_fixpt_pi = { 13493037705LL }; +static const struct fixed31_32 dc_fixpt_two_pi = { 26986075409LL }; +static const struct fixed31_32 dc_fixpt_e = { 11674931555LL }; +static const struct fixed31_32 dc_fixpt_ln2 = { 2977044471LL }; +static const struct fixed31_32 dc_fixpt_ln2_div_2 = { 1488522236LL }; /* * @brief @@ -70,24 +77,19 @@ static const struct fixed31_32 dal_fixed31_32_ln2_div_2 = { 1488522236LL }; * @brief * result = numerator / denominator */ -struct fixed31_32 dal_fixed31_32_from_fraction( - long long numerator, - long long denominator); +struct fixed31_32 dc_fixpt_from_fraction(long long numerator, long long denominator); /* * @brief * result = arg */ -struct fixed31_32 dal_fixed31_32_from_int_nonconst(long long arg); -static inline struct fixed31_32 dal_fixed31_32_from_int(long long arg) +static inline struct fixed31_32 dc_fixpt_from_int(int arg) { - if (__builtin_constant_p(arg)) { - struct fixed31_32 res; - BUILD_BUG_ON((LONG_MIN > arg) || (arg > LONG_MAX)); - res.value = arg << FIXED31_32_BITS_PER_FRACTIONAL_PART; - return res; - } else - return dal_fixed31_32_from_int_nonconst(arg); + struct fixed31_32 res; + + res.value = (long long) arg << FIXED31_32_BITS_PER_FRACTIONAL_PART; + + return res; } /* @@ -99,7 +101,7 @@ static inline struct fixed31_32 dal_fixed31_32_from_int(long long arg) * @brief * result = -arg */ -static inline struct fixed31_32 dal_fixed31_32_neg(struct fixed31_32 arg) +static inline struct fixed31_32 dc_fixpt_neg(struct fixed31_32 arg) { struct fixed31_32 res; @@ -112,10 +114,10 @@ static inline struct fixed31_32 dal_fixed31_32_neg(struct fixed31_32 arg) * @brief * result = abs(arg) := (arg >= 0) ? arg : -arg */ -static inline struct fixed31_32 dal_fixed31_32_abs(struct fixed31_32 arg) +static inline struct fixed31_32 dc_fixpt_abs(struct fixed31_32 arg) { if (arg.value < 0) - return dal_fixed31_32_neg(arg); + return dc_fixpt_neg(arg); else return arg; } @@ -129,8 +131,7 @@ static inline struct fixed31_32 dal_fixed31_32_abs(struct fixed31_32 arg) * @brief * result = arg1 < arg2 */ -static inline bool dal_fixed31_32_lt(struct fixed31_32 arg1, - struct fixed31_32 arg2) +static inline bool dc_fixpt_lt(struct fixed31_32 arg1, struct fixed31_32 arg2) { return arg1.value < arg2.value; } @@ -139,8 +140,7 @@ static inline bool dal_fixed31_32_lt(struct fixed31_32 arg1, * @brief * result = arg1 <= arg2 */ -static inline bool dal_fixed31_32_le(struct fixed31_32 arg1, - struct fixed31_32 arg2) +static inline bool dc_fixpt_le(struct fixed31_32 arg1, struct fixed31_32 arg2) { return arg1.value <= arg2.value; } @@ -149,8 +149,7 @@ static inline bool dal_fixed31_32_le(struct fixed31_32 arg1, * @brief * result = arg1 == arg2 */ -static inline bool dal_fixed31_32_eq(struct fixed31_32 arg1, - struct fixed31_32 arg2) +static inline bool dc_fixpt_eq(struct fixed31_32 arg1, struct fixed31_32 arg2) { return arg1.value == arg2.value; } @@ -159,8 +158,7 @@ static inline bool dal_fixed31_32_eq(struct fixed31_32 arg1, * @brief * result = min(arg1, arg2) := (arg1 <= arg2) ? arg1 : arg2 */ -static inline struct fixed31_32 dal_fixed31_32_min(struct fixed31_32 arg1, - struct fixed31_32 arg2) +static inline struct fixed31_32 dc_fixpt_min(struct fixed31_32 arg1, struct fixed31_32 arg2) { if (arg1.value <= arg2.value) return arg1; @@ -172,8 +170,7 @@ static inline struct fixed31_32 dal_fixed31_32_min(struct fixed31_32 arg1, * @brief * result = max(arg1, arg2) := (arg1 <= arg2) ? arg2 : arg1 */ -static inline struct fixed31_32 dal_fixed31_32_max(struct fixed31_32 arg1, - struct fixed31_32 arg2) +static inline struct fixed31_32 dc_fixpt_max(struct fixed31_32 arg1, struct fixed31_32 arg2) { if (arg1.value <= arg2.value) return arg2; @@ -187,14 +184,14 @@ static inline struct fixed31_32 dal_fixed31_32_max(struct fixed31_32 arg1, * result = | arg, when min_value < arg < max_value * | max_value, when arg >= max_value */ -static inline struct fixed31_32 dal_fixed31_32_clamp( +static inline struct fixed31_32 dc_fixpt_clamp( struct fixed31_32 arg, struct fixed31_32 min_value, struct fixed31_32 max_value) { - if (dal_fixed31_32_le(arg, min_value)) + if (dc_fixpt_le(arg, min_value)) return min_value; - else if (dal_fixed31_32_le(max_value, arg)) + else if (dc_fixpt_le(max_value, arg)) return max_value; else return arg; @@ -209,21 +206,30 @@ static inline struct fixed31_32 dal_fixed31_32_clamp( * @brief * result = arg << shift */ -struct fixed31_32 dal_fixed31_32_shl( - struct fixed31_32 arg, - unsigned char shift); +static inline struct fixed31_32 dc_fixpt_shl(struct fixed31_32 arg, unsigned char shift) +{ + ASSERT(((arg.value >= 0) && (arg.value <= LLONG_MAX >> shift)) || + ((arg.value < 0) && (arg.value >= ~(LLONG_MAX >> shift)))); + + arg.value = arg.value << shift; + + return arg; +} /* * @brief * result = arg >> shift */ -static inline struct fixed31_32 dal_fixed31_32_shr( - struct fixed31_32 arg, - unsigned char shift) +static inline struct fixed31_32 dc_fixpt_shr(struct fixed31_32 arg, unsigned char shift) { - struct fixed31_32 res; - res.value = arg.value >> shift; - return res; + bool negative = arg.value < 0; + + if (negative) + arg.value = -arg.value; + arg.value = arg.value >> shift; + if (negative) + arg.value = -arg.value; + return arg; } /* @@ -235,38 +241,50 @@ static inline struct fixed31_32 dal_fixed31_32_shr( * @brief * result = arg1 + arg2 */ -struct fixed31_32 dal_fixed31_32_add( - struct fixed31_32 arg1, - struct fixed31_32 arg2); +static inline struct fixed31_32 dc_fixpt_add(struct fixed31_32 arg1, struct fixed31_32 arg2) +{ + struct fixed31_32 res; + + ASSERT(((arg1.value >= 0) && (LLONG_MAX - arg1.value >= arg2.value)) || + ((arg1.value < 0) && (LLONG_MIN - arg1.value <= arg2.value))); + + res.value = arg1.value + arg2.value; + + return res; +} /* * @brief * result = arg1 + arg2 */ -static inline struct fixed31_32 dal_fixed31_32_add_int(struct fixed31_32 arg1, - int arg2) +static inline struct fixed31_32 dc_fixpt_add_int(struct fixed31_32 arg1, int arg2) { - return dal_fixed31_32_add(arg1, - dal_fixed31_32_from_int(arg2)); + return dc_fixpt_add(arg1, dc_fixpt_from_int(arg2)); } /* * @brief * result = arg1 - arg2 */ -struct fixed31_32 dal_fixed31_32_sub( - struct fixed31_32 arg1, - struct fixed31_32 arg2); +static inline struct fixed31_32 dc_fixpt_sub(struct fixed31_32 arg1, struct fixed31_32 arg2) +{ + struct fixed31_32 res; + + ASSERT(((arg2.value >= 0) && (LLONG_MIN + arg2.value <= arg1.value)) || + ((arg2.value < 0) && (LLONG_MAX + arg2.value >= arg1.value))); + + res.value = arg1.value - arg2.value; + + return res; +} /* * @brief * result = arg1 - arg2 */ -static inline struct fixed31_32 dal_fixed31_32_sub_int(struct fixed31_32 arg1, - int arg2) +static inline struct fixed31_32 dc_fixpt_sub_int(struct fixed31_32 arg1, int arg2) { - return dal_fixed31_32_sub(arg1, - dal_fixed31_32_from_int(arg2)); + return dc_fixpt_sub(arg1, dc_fixpt_from_int(arg2)); } @@ -279,49 +297,40 @@ static inline struct fixed31_32 dal_fixed31_32_sub_int(struct fixed31_32 arg1, * @brief * result = arg1 * arg2 */ -struct fixed31_32 dal_fixed31_32_mul( - struct fixed31_32 arg1, - struct fixed31_32 arg2); +struct fixed31_32 dc_fixpt_mul(struct fixed31_32 arg1, struct fixed31_32 arg2); /* * @brief * result = arg1 * arg2 */ -static inline struct fixed31_32 dal_fixed31_32_mul_int(struct fixed31_32 arg1, - int arg2) +static inline struct fixed31_32 dc_fixpt_mul_int(struct fixed31_32 arg1, int arg2) { - return dal_fixed31_32_mul(arg1, - dal_fixed31_32_from_int(arg2)); + return dc_fixpt_mul(arg1, dc_fixpt_from_int(arg2)); } /* * @brief * result = square(arg) := arg * arg */ -struct fixed31_32 dal_fixed31_32_sqr( - struct fixed31_32 arg); +struct fixed31_32 dc_fixpt_sqr(struct fixed31_32 arg); /* * @brief * result = arg1 / arg2 */ -static inline struct fixed31_32 dal_fixed31_32_div_int(struct fixed31_32 arg1, - long long arg2) +static inline struct fixed31_32 dc_fixpt_div_int(struct fixed31_32 arg1, long long arg2) { - return dal_fixed31_32_from_fraction(arg1.value, - dal_fixed31_32_from_int(arg2).value); + return dc_fixpt_from_fraction(arg1.value, dc_fixpt_from_int(arg2).value); } /* * @brief * result = arg1 / arg2 */ -static inline struct fixed31_32 dal_fixed31_32_div(struct fixed31_32 arg1, - struct fixed31_32 arg2) +static inline struct fixed31_32 dc_fixpt_div(struct fixed31_32 arg1, struct fixed31_32 arg2) { - return dal_fixed31_32_from_fraction(arg1.value, - arg2.value); + return dc_fixpt_from_fraction(arg1.value, arg2.value); } /* @@ -336,8 +345,7 @@ static inline struct fixed31_32 dal_fixed31_32_div(struct fixed31_32 arg1, * @note * No special actions taken in case argument is zero. */ -struct fixed31_32 dal_fixed31_32_recip( - struct fixed31_32 arg); +struct fixed31_32 dc_fixpt_recip(struct fixed31_32 arg); /* * @brief @@ -352,8 +360,7 @@ struct fixed31_32 dal_fixed31_32_recip( * Argument specified in radians, * internally it's normalized to [-2pi...2pi] range. */ -struct fixed31_32 dal_fixed31_32_sinc( - struct fixed31_32 arg); +struct fixed31_32 dc_fixpt_sinc(struct fixed31_32 arg); /* * @brief @@ -363,8 +370,7 @@ struct fixed31_32 dal_fixed31_32_sinc( * Argument specified in radians, * internally it's normalized to [-2pi...2pi] range. */ -struct fixed31_32 dal_fixed31_32_sin( - struct fixed31_32 arg); +struct fixed31_32 dc_fixpt_sin(struct fixed31_32 arg); /* * @brief @@ -376,8 +382,7 @@ struct fixed31_32 dal_fixed31_32_sin( * passing arguments outside that range * will cause incorrect result! */ -struct fixed31_32 dal_fixed31_32_cos( - struct fixed31_32 arg); +struct fixed31_32 dc_fixpt_cos(struct fixed31_32 arg); /* * @brief @@ -391,8 +396,7 @@ struct fixed31_32 dal_fixed31_32_cos( * @note * Currently, function is verified for abs(arg) <= 1. */ -struct fixed31_32 dal_fixed31_32_exp( - struct fixed31_32 arg); +struct fixed31_32 dc_fixpt_exp(struct fixed31_32 arg); /* * @brief @@ -404,8 +408,7 @@ struct fixed31_32 dal_fixed31_32_exp( * Currently, no special actions taken * in case of invalid argument(s). Take care! */ -struct fixed31_32 dal_fixed31_32_log( - struct fixed31_32 arg); +struct fixed31_32 dc_fixpt_log(struct fixed31_32 arg); /* * @brief @@ -419,9 +422,13 @@ struct fixed31_32 dal_fixed31_32_log( * @note * Currently, abs(arg1) should be less than 1. Take care! */ -struct fixed31_32 dal_fixed31_32_pow( - struct fixed31_32 arg1, - struct fixed31_32 arg2); +static inline struct fixed31_32 dc_fixpt_pow(struct fixed31_32 arg1, struct fixed31_32 arg2) +{ + return dc_fixpt_exp( + dc_fixpt_mul( + dc_fixpt_log(arg1), + arg2)); +} /* * @brief @@ -432,22 +439,56 @@ struct fixed31_32 dal_fixed31_32_pow( * @brief * result = floor(arg) := greatest integer lower than or equal to arg */ -int dal_fixed31_32_floor( - struct fixed31_32 arg); +static inline int dc_fixpt_floor(struct fixed31_32 arg) +{ + unsigned long long arg_value = arg.value > 0 ? arg.value : -arg.value; + + if (arg.value >= 0) + return (int)(arg_value >> FIXED31_32_BITS_PER_FRACTIONAL_PART); + else + return -(int)(arg_value >> FIXED31_32_BITS_PER_FRACTIONAL_PART); +} /* * @brief * result = round(arg) := integer nearest to arg */ -int dal_fixed31_32_round( - struct fixed31_32 arg); +static inline int dc_fixpt_round(struct fixed31_32 arg) +{ + unsigned long long arg_value = arg.value > 0 ? arg.value : -arg.value; + + const long long summand = dc_fixpt_half.value; + + ASSERT(LLONG_MAX - (long long)arg_value >= summand); + + arg_value += summand; + + if (arg.value >= 0) + return (int)(arg_value >> FIXED31_32_BITS_PER_FRACTIONAL_PART); + else + return -(int)(arg_value >> FIXED31_32_BITS_PER_FRACTIONAL_PART); +} /* * @brief * result = ceil(arg) := lowest integer greater than or equal to arg */ -int dal_fixed31_32_ceil( - struct fixed31_32 arg); +static inline int dc_fixpt_ceil(struct fixed31_32 arg) +{ + unsigned long long arg_value = arg.value > 0 ? arg.value : -arg.value; + + const long long summand = dc_fixpt_one.value - + dc_fixpt_epsilon.value; + + ASSERT(LLONG_MAX - (long long)arg_value >= summand); + + arg_value += summand; + + if (arg.value >= 0) + return (int)(arg_value >> FIXED31_32_BITS_PER_FRACTIONAL_PART); + else + return -(int)(arg_value >> FIXED31_32_BITS_PER_FRACTIONAL_PART); +} /* the following two function are used in scaler hw programming to convert fixed * point value to format 2 bits from integer part and 19 bits from fractional @@ -455,20 +496,31 @@ int dal_fixed31_32_ceil( * fractional */ -unsigned int dal_fixed31_32_u2d19( - struct fixed31_32 arg); +unsigned int dc_fixpt_u2d19(struct fixed31_32 arg); + +unsigned int dc_fixpt_u0d19(struct fixed31_32 arg); -unsigned int dal_fixed31_32_u0d19( - struct fixed31_32 arg); +unsigned int dc_fixpt_clamp_u0d14(struct fixed31_32 arg); +unsigned int dc_fixpt_clamp_u0d10(struct fixed31_32 arg); -unsigned int dal_fixed31_32_clamp_u0d14( - struct fixed31_32 arg); +int dc_fixpt_s4d19(struct fixed31_32 arg); -unsigned int dal_fixed31_32_clamp_u0d10( - struct fixed31_32 arg); +static inline struct fixed31_32 dc_fixpt_truncate(struct fixed31_32 arg, unsigned int frac_bits) +{ + bool negative = arg.value < 0; -int dal_fixed31_32_s4d19( - struct fixed31_32 arg); + if (frac_bits >= FIXED31_32_BITS_PER_FRACTIONAL_PART) { + ASSERT(frac_bits == FIXED31_32_BITS_PER_FRACTIONAL_PART); + return arg; + } + + if (negative) + arg.value = -arg.value; + arg.value &= (~0LL) << (FIXED31_32_BITS_PER_FRACTIONAL_PART - frac_bits); + if (negative) + arg.value = -arg.value; + return arg; +} #endif diff --git a/drivers/gpu/drm/amd/display/include/fixed32_32.h b/drivers/gpu/drm/amd/display/include/fixed32_32.h deleted file mode 100644 index 9c70341fe026..000000000000 --- a/drivers/gpu/drm/amd/display/include/fixed32_32.h +++ /dev/null @@ -1,129 +0,0 @@ -/* - * Copyright 2012-15 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - - -#ifndef __DAL_FIXED32_32_H__ -#define __DAL_FIXED32_32_H__ - -#include "os_types.h" - -struct fixed32_32 { - uint64_t value; -}; - -static const struct fixed32_32 dal_fixed32_32_zero = { 0 }; -static const struct fixed32_32 dal_fixed32_32_one = { 0x100000000LL }; -static const struct fixed32_32 dal_fixed32_32_half = { 0x80000000LL }; - -struct fixed32_32 dal_fixed32_32_from_fraction(uint32_t n, uint32_t d); -static inline struct fixed32_32 dal_fixed32_32_from_int(uint32_t value) -{ - struct fixed32_32 fx; - - fx.value = (uint64_t)value<<32; - return fx; -} - -struct fixed32_32 dal_fixed32_32_add( - struct fixed32_32 lhs, - struct fixed32_32 rhs); -struct fixed32_32 dal_fixed32_32_add_int( - struct fixed32_32 lhs, - uint32_t rhs); -struct fixed32_32 dal_fixed32_32_sub( - struct fixed32_32 lhs, - struct fixed32_32 rhs); -struct fixed32_32 dal_fixed32_32_sub_int( - struct fixed32_32 lhs, - uint32_t rhs); -struct fixed32_32 dal_fixed32_32_mul( - struct fixed32_32 lhs, - struct fixed32_32 rhs); -struct fixed32_32 dal_fixed32_32_mul_int( - struct fixed32_32 lhs, - uint32_t rhs); -struct fixed32_32 dal_fixed32_32_div( - struct fixed32_32 lhs, - struct fixed32_32 rhs); -struct fixed32_32 dal_fixed32_32_div_int( - struct fixed32_32 lhs, - uint32_t rhs); - -static inline struct fixed32_32 dal_fixed32_32_min(struct fixed32_32 lhs, - struct fixed32_32 rhs) -{ - return (lhs.value < rhs.value) ? lhs : rhs; -} - -static inline struct fixed32_32 dal_fixed32_32_max(struct fixed32_32 lhs, - struct fixed32_32 rhs) -{ - return (lhs.value > rhs.value) ? lhs : rhs; -} - -static inline bool dal_fixed32_32_gt(struct fixed32_32 lhs, struct fixed32_32 rhs) -{ - return lhs.value > rhs.value; -} - -static inline bool dal_fixed32_32_gt_int(struct fixed32_32 lhs, uint32_t rhs) -{ - return lhs.value > ((uint64_t)rhs<<32); -} - -static inline bool dal_fixed32_32_lt(struct fixed32_32 lhs, struct fixed32_32 rhs) -{ - return lhs.value < rhs.value; -} - -static inline bool dal_fixed32_32_lt_int(struct fixed32_32 lhs, uint32_t rhs) -{ - return lhs.value < ((uint64_t)rhs<<32); -} - -static inline bool dal_fixed32_32_le(struct fixed32_32 lhs, struct fixed32_32 rhs) -{ - return lhs.value <= rhs.value; -} - -static inline bool dal_fixed32_32_le_int(struct fixed32_32 lhs, uint32_t rhs) -{ - return lhs.value <= ((uint64_t)rhs<<32); -} - -static inline bool dal_fixed32_32_eq(struct fixed32_32 lhs, struct fixed32_32 rhs) -{ - return lhs.value == rhs.value; -} - -uint32_t dal_fixed32_32_ceil(struct fixed32_32 value); -static inline uint32_t dal_fixed32_32_floor(struct fixed32_32 value) -{ - return value.value>>32; -} - -uint32_t dal_fixed32_32_round(struct fixed32_32 value); - -#endif diff --git a/drivers/gpu/drm/amd/display/include/logger_interface.h b/drivers/gpu/drm/amd/display/include/logger_interface.h index 28dee960d509..dc98d6d4b2bd 100644 --- a/drivers/gpu/drm/amd/display/include/logger_interface.h +++ b/drivers/gpu/drm/amd/display/include/logger_interface.h @@ -190,4 +190,13 @@ void context_clock_trace( } \ } while (0) +#define DISPLAY_STATS_BEGIN(entry) \ + dm_logger_open(dc->ctx->logger, &entry, LOG_DISPLAYSTATS) + +#define DISPLAY_STATS(msg, ...) \ + dm_logger_append(&log_entry, msg, ##__VA_ARGS__) + +#define DISPLAY_STATS_END(entry) \ + dm_logger_close(&entry) + #endif /* __DAL_LOGGER_INTERFACE_H__ */ diff --git a/drivers/gpu/drm/amd/display/include/logger_types.h b/drivers/gpu/drm/amd/display/include/logger_types.h index b608a0830801..0a540b9897a6 100644 --- a/drivers/gpu/drm/amd/display/include/logger_types.h +++ b/drivers/gpu/drm/amd/display/include/logger_types.h @@ -98,7 +98,7 @@ enum dc_log_type { LOG_EVENT_UNDERFLOW, LOG_IF_TRACE, LOG_PERF_TRACE, - LOG_PROFILING, + LOG_DISPLAYSTATS, LOG_SECTION_TOTAL_COUNT }; diff --git a/drivers/gpu/drm/amd/display/modules/color/color_gamma.c b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c index 8446e0876ac8..0cd111d59018 100644 --- a/drivers/gpu/drm/amd/display/modules/color/color_gamma.c +++ b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c @@ -43,7 +43,7 @@ static bool de_pq_initialized; /* = false; */ /* one-time setup of X points */ void setup_x_points_distribution(void) { - struct fixed31_32 region_size = dal_fixed31_32_from_int(128); + struct fixed31_32 region_size = dc_fixpt_from_int(128); int32_t segment; uint32_t seg_offset; uint32_t index; @@ -53,8 +53,8 @@ void setup_x_points_distribution(void) coordinates_x[MAX_HW_POINTS + 1].x = region_size; for (segment = 6; segment > (6 - NUM_REGIONS); segment--) { - region_size = dal_fixed31_32_div_int(region_size, 2); - increment = dal_fixed31_32_div_int(region_size, + region_size = dc_fixpt_div_int(region_size, 2); + increment = dc_fixpt_div_int(region_size, NUM_PTS_IN_REGION); seg_offset = (segment + (NUM_REGIONS - 7)) * NUM_PTS_IN_REGION; coordinates_x[seg_offset].x = region_size; @@ -62,7 +62,7 @@ void setup_x_points_distribution(void) for (index = seg_offset + 1; index < seg_offset + NUM_PTS_IN_REGION; index++) { - coordinates_x[index].x = dal_fixed31_32_add + coordinates_x[index].x = dc_fixpt_add (coordinates_x[index-1].x, increment); } } @@ -72,63 +72,63 @@ static void compute_pq(struct fixed31_32 in_x, struct fixed31_32 *out_y) { /* consts for PQ gamma formula. */ const struct fixed31_32 m1 = - dal_fixed31_32_from_fraction(159301758, 1000000000); + dc_fixpt_from_fraction(159301758, 1000000000); const struct fixed31_32 m2 = - dal_fixed31_32_from_fraction(7884375, 100000); + dc_fixpt_from_fraction(7884375, 100000); const struct fixed31_32 c1 = - dal_fixed31_32_from_fraction(8359375, 10000000); + dc_fixpt_from_fraction(8359375, 10000000); const struct fixed31_32 c2 = - dal_fixed31_32_from_fraction(188515625, 10000000); + dc_fixpt_from_fraction(188515625, 10000000); const struct fixed31_32 c3 = - dal_fixed31_32_from_fraction(186875, 10000); + dc_fixpt_from_fraction(186875, 10000); struct fixed31_32 l_pow_m1; struct fixed31_32 base; - if (dal_fixed31_32_lt(in_x, dal_fixed31_32_zero)) - in_x = dal_fixed31_32_zero; + if (dc_fixpt_lt(in_x, dc_fixpt_zero)) + in_x = dc_fixpt_zero; - l_pow_m1 = dal_fixed31_32_pow(in_x, m1); - base = dal_fixed31_32_div( - dal_fixed31_32_add(c1, - (dal_fixed31_32_mul(c2, l_pow_m1))), - dal_fixed31_32_add(dal_fixed31_32_one, - (dal_fixed31_32_mul(c3, l_pow_m1)))); - *out_y = dal_fixed31_32_pow(base, m2); + l_pow_m1 = dc_fixpt_pow(in_x, m1); + base = dc_fixpt_div( + dc_fixpt_add(c1, + (dc_fixpt_mul(c2, l_pow_m1))), + dc_fixpt_add(dc_fixpt_one, + (dc_fixpt_mul(c3, l_pow_m1)))); + *out_y = dc_fixpt_pow(base, m2); } static void compute_de_pq(struct fixed31_32 in_x, struct fixed31_32 *out_y) { /* consts for dePQ gamma formula. */ const struct fixed31_32 m1 = - dal_fixed31_32_from_fraction(159301758, 1000000000); + dc_fixpt_from_fraction(159301758, 1000000000); const struct fixed31_32 m2 = - dal_fixed31_32_from_fraction(7884375, 100000); + dc_fixpt_from_fraction(7884375, 100000); const struct fixed31_32 c1 = - dal_fixed31_32_from_fraction(8359375, 10000000); + dc_fixpt_from_fraction(8359375, 10000000); const struct fixed31_32 c2 = - dal_fixed31_32_from_fraction(188515625, 10000000); + dc_fixpt_from_fraction(188515625, 10000000); const struct fixed31_32 c3 = - dal_fixed31_32_from_fraction(186875, 10000); + dc_fixpt_from_fraction(186875, 10000); struct fixed31_32 l_pow_m1; struct fixed31_32 base, div; - if (dal_fixed31_32_lt(in_x, dal_fixed31_32_zero)) - in_x = dal_fixed31_32_zero; + if (dc_fixpt_lt(in_x, dc_fixpt_zero)) + in_x = dc_fixpt_zero; - l_pow_m1 = dal_fixed31_32_pow(in_x, - dal_fixed31_32_div(dal_fixed31_32_one, m2)); - base = dal_fixed31_32_sub(l_pow_m1, c1); + l_pow_m1 = dc_fixpt_pow(in_x, + dc_fixpt_div(dc_fixpt_one, m2)); + base = dc_fixpt_sub(l_pow_m1, c1); - if (dal_fixed31_32_lt(base, dal_fixed31_32_zero)) - base = dal_fixed31_32_zero; + if (dc_fixpt_lt(base, dc_fixpt_zero)) + base = dc_fixpt_zero; - div = dal_fixed31_32_sub(c2, dal_fixed31_32_mul(c3, l_pow_m1)); + div = dc_fixpt_sub(c2, dc_fixpt_mul(c3, l_pow_m1)); - *out_y = dal_fixed31_32_pow(dal_fixed31_32_div(base, div), - dal_fixed31_32_div(dal_fixed31_32_one, m1)); + *out_y = dc_fixpt_pow(dc_fixpt_div(base, div), + dc_fixpt_div(dc_fixpt_one, m1)); } /* one-time pre-compute PQ values - only for sdr_white_level 80 */ @@ -138,14 +138,14 @@ void precompute_pq(void) struct fixed31_32 x; const struct hw_x_point *coord_x = coordinates_x + 32; struct fixed31_32 scaling_factor = - dal_fixed31_32_from_fraction(80, 10000); + dc_fixpt_from_fraction(80, 10000); /* pow function has problems with arguments too small */ for (i = 0; i < 32; i++) - pq_table[i] = dal_fixed31_32_zero; + pq_table[i] = dc_fixpt_zero; for (i = 32; i <= MAX_HW_POINTS; i++) { - x = dal_fixed31_32_mul(coord_x->x, scaling_factor); + x = dc_fixpt_mul(coord_x->x, scaling_factor); compute_pq(x, &pq_table[i]); ++coord_x; } @@ -158,7 +158,7 @@ void precompute_de_pq(void) struct fixed31_32 y; uint32_t begin_index, end_index; - struct fixed31_32 scaling_factor = dal_fixed31_32_from_int(125); + struct fixed31_32 scaling_factor = dc_fixpt_from_int(125); /* X points is 2^-25 to 2^7 * De-gamma X is 2^-12 to 2^0 – we are skipping first -12-(-25) = 13 regions @@ -167,11 +167,11 @@ void precompute_de_pq(void) end_index = begin_index + 12 * NUM_PTS_IN_REGION; for (i = 0; i <= begin_index; i++) - de_pq_table[i] = dal_fixed31_32_zero; + de_pq_table[i] = dc_fixpt_zero; for (; i <= end_index; i++) { compute_de_pq(coordinates_x[i].x, &y); - de_pq_table[i] = dal_fixed31_32_mul(y, scaling_factor); + de_pq_table[i] = dc_fixpt_mul(y, scaling_factor); } for (; i <= MAX_HW_POINTS; i++) @@ -195,15 +195,15 @@ static void build_coefficients(struct gamma_coefficients *coefficients, bool is_ uint32_t index = is_2_4 == true ? 0:1; do { - coefficients->a0[i] = dal_fixed31_32_from_fraction( + coefficients->a0[i] = dc_fixpt_from_fraction( numerator01[index], 10000000); - coefficients->a1[i] = dal_fixed31_32_from_fraction( + coefficients->a1[i] = dc_fixpt_from_fraction( numerator02[index], 1000); - coefficients->a2[i] = dal_fixed31_32_from_fraction( + coefficients->a2[i] = dc_fixpt_from_fraction( numerator03[index], 1000); - coefficients->a3[i] = dal_fixed31_32_from_fraction( + coefficients->a3[i] = dc_fixpt_from_fraction( numerator04[index], 1000); - coefficients->user_gamma[i] = dal_fixed31_32_from_fraction( + coefficients->user_gamma[i] = dc_fixpt_from_fraction( numerator05[index], 1000); ++i; @@ -218,33 +218,33 @@ static struct fixed31_32 translate_from_linear_space( struct fixed31_32 a3, struct fixed31_32 gamma) { - const struct fixed31_32 one = dal_fixed31_32_from_int(1); + const struct fixed31_32 one = dc_fixpt_from_int(1); - if (dal_fixed31_32_lt(one, arg)) + if (dc_fixpt_lt(one, arg)) return one; - if (dal_fixed31_32_le(arg, dal_fixed31_32_neg(a0))) - return dal_fixed31_32_sub( + if (dc_fixpt_le(arg, dc_fixpt_neg(a0))) + return dc_fixpt_sub( a2, - dal_fixed31_32_mul( - dal_fixed31_32_add( + dc_fixpt_mul( + dc_fixpt_add( one, a3), - dal_fixed31_32_pow( - dal_fixed31_32_neg(arg), - dal_fixed31_32_recip(gamma)))); - else if (dal_fixed31_32_le(a0, arg)) - return dal_fixed31_32_sub( - dal_fixed31_32_mul( - dal_fixed31_32_add( + dc_fixpt_pow( + dc_fixpt_neg(arg), + dc_fixpt_recip(gamma)))); + else if (dc_fixpt_le(a0, arg)) + return dc_fixpt_sub( + dc_fixpt_mul( + dc_fixpt_add( one, a3), - dal_fixed31_32_pow( + dc_fixpt_pow( arg, - dal_fixed31_32_recip(gamma))), + dc_fixpt_recip(gamma))), a2); else - return dal_fixed31_32_mul( + return dc_fixpt_mul( arg, a1); } @@ -259,25 +259,25 @@ static struct fixed31_32 translate_to_linear_space( { struct fixed31_32 linear; - a0 = dal_fixed31_32_mul(a0, a1); - if (dal_fixed31_32_le(arg, dal_fixed31_32_neg(a0))) + a0 = dc_fixpt_mul(a0, a1); + if (dc_fixpt_le(arg, dc_fixpt_neg(a0))) - linear = dal_fixed31_32_neg( - dal_fixed31_32_pow( - dal_fixed31_32_div( - dal_fixed31_32_sub(a2, arg), - dal_fixed31_32_add( - dal_fixed31_32_one, a3)), gamma)); + linear = dc_fixpt_neg( + dc_fixpt_pow( + dc_fixpt_div( + dc_fixpt_sub(a2, arg), + dc_fixpt_add( + dc_fixpt_one, a3)), gamma)); - else if (dal_fixed31_32_le(dal_fixed31_32_neg(a0), arg) && - dal_fixed31_32_le(arg, a0)) - linear = dal_fixed31_32_div(arg, a1); + else if (dc_fixpt_le(dc_fixpt_neg(a0), arg) && + dc_fixpt_le(arg, a0)) + linear = dc_fixpt_div(arg, a1); else - linear = dal_fixed31_32_pow( - dal_fixed31_32_div( - dal_fixed31_32_add(a2, arg), - dal_fixed31_32_add( - dal_fixed31_32_one, a3)), gamma); + linear = dc_fixpt_pow( + dc_fixpt_div( + dc_fixpt_add(a2, arg), + dc_fixpt_add( + dc_fixpt_one, a3)), gamma); return linear; } @@ -352,8 +352,8 @@ static bool find_software_points( right = axis_x[max_number - 1].b; } - if (dal_fixed31_32_le(left, hw_point) && - dal_fixed31_32_le(hw_point, right)) { + if (dc_fixpt_le(left, hw_point) && + dc_fixpt_le(hw_point, right)) { *index_to_start = i; *index_left = i; @@ -366,7 +366,7 @@ static bool find_software_points( return true; } else if ((i == *index_to_start) && - dal_fixed31_32_le(hw_point, left)) { + dc_fixpt_le(hw_point, left)) { *index_to_start = i; *index_left = i; *index_right = i; @@ -375,7 +375,7 @@ static bool find_software_points( return true; } else if ((i == max_number - 1) && - dal_fixed31_32_le(right, hw_point)) { + dc_fixpt_le(right, hw_point)) { *index_to_start = i; *index_left = i; *index_right = i; @@ -457,17 +457,17 @@ static bool build_custom_gamma_mapping_coefficients_worker( } if (hw_pos == HW_POINT_POSITION_MIDDLE) - point->coeff = dal_fixed31_32_div( - dal_fixed31_32_sub( + point->coeff = dc_fixpt_div( + dc_fixpt_sub( coord_x, left_pos), - dal_fixed31_32_sub( + dc_fixpt_sub( right_pos, left_pos)); else if (hw_pos == HW_POINT_POSITION_LEFT) - point->coeff = dal_fixed31_32_zero; + point->coeff = dc_fixpt_zero; else if (hw_pos == HW_POINT_POSITION_RIGHT) - point->coeff = dal_fixed31_32_from_int(2); + point->coeff = dc_fixpt_from_int(2); else { BREAK_TO_DEBUGGER(); return false; @@ -502,45 +502,45 @@ static struct fixed31_32 calculate_mapped_value( if ((point->left_index < 0) || (point->left_index > max_index)) { BREAK_TO_DEBUGGER(); - return dal_fixed31_32_zero; + return dc_fixpt_zero; } if ((point->right_index < 0) || (point->right_index > max_index)) { BREAK_TO_DEBUGGER(); - return dal_fixed31_32_zero; + return dc_fixpt_zero; } if (point->pos == HW_POINT_POSITION_MIDDLE) if (channel == CHANNEL_NAME_RED) - result = dal_fixed31_32_add( - dal_fixed31_32_mul( + result = dc_fixpt_add( + dc_fixpt_mul( point->coeff, - dal_fixed31_32_sub( + dc_fixpt_sub( rgb[point->right_index].r, rgb[point->left_index].r)), rgb[point->left_index].r); else if (channel == CHANNEL_NAME_GREEN) - result = dal_fixed31_32_add( - dal_fixed31_32_mul( + result = dc_fixpt_add( + dc_fixpt_mul( point->coeff, - dal_fixed31_32_sub( + dc_fixpt_sub( rgb[point->right_index].g, rgb[point->left_index].g)), rgb[point->left_index].g); else - result = dal_fixed31_32_add( - dal_fixed31_32_mul( + result = dc_fixpt_add( + dc_fixpt_mul( point->coeff, - dal_fixed31_32_sub( + dc_fixpt_sub( rgb[point->right_index].b, rgb[point->left_index].b)), rgb[point->left_index].b); else if (point->pos == HW_POINT_POSITION_LEFT) { BREAK_TO_DEBUGGER(); - result = dal_fixed31_32_zero; + result = dc_fixpt_zero; } else { BREAK_TO_DEBUGGER(); - result = dal_fixed31_32_one; + result = dc_fixpt_one; } return result; @@ -558,7 +558,7 @@ static void build_pq(struct pwl_float_data_ex *rgb_regamma, struct fixed31_32 x; struct fixed31_32 output; struct fixed31_32 scaling_factor = - dal_fixed31_32_from_fraction(sdr_white_level, 10000); + dc_fixpt_from_fraction(sdr_white_level, 10000); if (!pq_initialized && sdr_white_level == 80) { precompute_pq(); @@ -579,15 +579,15 @@ static void build_pq(struct pwl_float_data_ex *rgb_regamma, if (sdr_white_level == 80) { output = pq_table[i]; } else { - x = dal_fixed31_32_mul(coord_x->x, scaling_factor); + x = dc_fixpt_mul(coord_x->x, scaling_factor); compute_pq(x, &output); } /* should really not happen? */ - if (dal_fixed31_32_lt(output, dal_fixed31_32_zero)) - output = dal_fixed31_32_zero; - else if (dal_fixed31_32_lt(dal_fixed31_32_one, output)) - output = dal_fixed31_32_one; + if (dc_fixpt_lt(output, dc_fixpt_zero)) + output = dc_fixpt_zero; + else if (dc_fixpt_lt(dc_fixpt_one, output)) + output = dc_fixpt_one; rgb->r = output; rgb->g = output; @@ -605,7 +605,7 @@ static void build_de_pq(struct pwl_float_data_ex *de_pq, uint32_t i; struct fixed31_32 output; - struct fixed31_32 scaling_factor = dal_fixed31_32_from_int(125); + struct fixed31_32 scaling_factor = dc_fixpt_from_int(125); if (!de_pq_initialized) { precompute_de_pq(); @@ -616,9 +616,9 @@ static void build_de_pq(struct pwl_float_data_ex *de_pq, for (i = 0; i <= hw_points_num; i++) { output = de_pq_table[i]; /* should really not happen? */ - if (dal_fixed31_32_lt(output, dal_fixed31_32_zero)) - output = dal_fixed31_32_zero; - else if (dal_fixed31_32_lt(scaling_factor, output)) + if (dc_fixpt_lt(output, dc_fixpt_zero)) + output = dc_fixpt_zero; + else if (dc_fixpt_lt(scaling_factor, output)) output = scaling_factor; de_pq[i].r = output; de_pq[i].g = output; @@ -670,9 +670,9 @@ static void build_degamma(struct pwl_float_data_ex *curve, end_index = begin_index + 12 * NUM_PTS_IN_REGION; while (i != begin_index) { - curve[i].r = dal_fixed31_32_zero; - curve[i].g = dal_fixed31_32_zero; - curve[i].b = dal_fixed31_32_zero; + curve[i].r = dc_fixpt_zero; + curve[i].g = dc_fixpt_zero; + curve[i].b = dc_fixpt_zero; i++; } @@ -684,9 +684,9 @@ static void build_degamma(struct pwl_float_data_ex *curve, i++; } while (i != hw_points_num + 1) { - curve[i].r = dal_fixed31_32_one; - curve[i].g = dal_fixed31_32_one; - curve[i].b = dal_fixed31_32_one; + curve[i].r = dc_fixpt_one; + curve[i].g = dc_fixpt_one; + curve[i].b = dc_fixpt_one; i++; } } @@ -695,8 +695,8 @@ static void scale_gamma(struct pwl_float_data *pwl_rgb, const struct dc_gamma *ramp, struct dividers dividers) { - const struct fixed31_32 max_driver = dal_fixed31_32_from_int(0xFFFF); - const struct fixed31_32 max_os = dal_fixed31_32_from_int(0xFF00); + const struct fixed31_32 max_driver = dc_fixpt_from_int(0xFFFF); + const struct fixed31_32 max_os = dc_fixpt_from_int(0xFF00); struct fixed31_32 scaler = max_os; uint32_t i; struct pwl_float_data *rgb = pwl_rgb; @@ -705,9 +705,9 @@ static void scale_gamma(struct pwl_float_data *pwl_rgb, i = 0; do { - if (dal_fixed31_32_lt(max_os, ramp->entries.red[i]) || - dal_fixed31_32_lt(max_os, ramp->entries.green[i]) || - dal_fixed31_32_lt(max_os, ramp->entries.blue[i])) { + if (dc_fixpt_lt(max_os, ramp->entries.red[i]) || + dc_fixpt_lt(max_os, ramp->entries.green[i]) || + dc_fixpt_lt(max_os, ramp->entries.blue[i])) { scaler = max_driver; break; } @@ -717,40 +717,40 @@ static void scale_gamma(struct pwl_float_data *pwl_rgb, i = 0; do { - rgb->r = dal_fixed31_32_div( + rgb->r = dc_fixpt_div( ramp->entries.red[i], scaler); - rgb->g = dal_fixed31_32_div( + rgb->g = dc_fixpt_div( ramp->entries.green[i], scaler); - rgb->b = dal_fixed31_32_div( + rgb->b = dc_fixpt_div( ramp->entries.blue[i], scaler); ++rgb; ++i; } while (i != ramp->num_entries); - rgb->r = dal_fixed31_32_mul(rgb_last->r, + rgb->r = dc_fixpt_mul(rgb_last->r, dividers.divider1); - rgb->g = dal_fixed31_32_mul(rgb_last->g, + rgb->g = dc_fixpt_mul(rgb_last->g, dividers.divider1); - rgb->b = dal_fixed31_32_mul(rgb_last->b, + rgb->b = dc_fixpt_mul(rgb_last->b, dividers.divider1); ++rgb; - rgb->r = dal_fixed31_32_mul(rgb_last->r, + rgb->r = dc_fixpt_mul(rgb_last->r, dividers.divider2); - rgb->g = dal_fixed31_32_mul(rgb_last->g, + rgb->g = dc_fixpt_mul(rgb_last->g, dividers.divider2); - rgb->b = dal_fixed31_32_mul(rgb_last->b, + rgb->b = dc_fixpt_mul(rgb_last->b, dividers.divider2); ++rgb; - rgb->r = dal_fixed31_32_mul(rgb_last->r, + rgb->r = dc_fixpt_mul(rgb_last->r, dividers.divider3); - rgb->g = dal_fixed31_32_mul(rgb_last->g, + rgb->g = dc_fixpt_mul(rgb_last->g, dividers.divider3); - rgb->b = dal_fixed31_32_mul(rgb_last->b, + rgb->b = dc_fixpt_mul(rgb_last->b, dividers.divider3); } @@ -759,62 +759,62 @@ static void scale_gamma_dx(struct pwl_float_data *pwl_rgb, struct dividers dividers) { uint32_t i; - struct fixed31_32 min = dal_fixed31_32_zero; - struct fixed31_32 max = dal_fixed31_32_one; + struct fixed31_32 min = dc_fixpt_zero; + struct fixed31_32 max = dc_fixpt_one; - struct fixed31_32 delta = dal_fixed31_32_zero; - struct fixed31_32 offset = dal_fixed31_32_zero; + struct fixed31_32 delta = dc_fixpt_zero; + struct fixed31_32 offset = dc_fixpt_zero; for (i = 0 ; i < ramp->num_entries; i++) { - if (dal_fixed31_32_lt(ramp->entries.red[i], min)) + if (dc_fixpt_lt(ramp->entries.red[i], min)) min = ramp->entries.red[i]; - if (dal_fixed31_32_lt(ramp->entries.green[i], min)) + if (dc_fixpt_lt(ramp->entries.green[i], min)) min = ramp->entries.green[i]; - if (dal_fixed31_32_lt(ramp->entries.blue[i], min)) + if (dc_fixpt_lt(ramp->entries.blue[i], min)) min = ramp->entries.blue[i]; - if (dal_fixed31_32_lt(max, ramp->entries.red[i])) + if (dc_fixpt_lt(max, ramp->entries.red[i])) max = ramp->entries.red[i]; - if (dal_fixed31_32_lt(max, ramp->entries.green[i])) + if (dc_fixpt_lt(max, ramp->entries.green[i])) max = ramp->entries.green[i]; - if (dal_fixed31_32_lt(max, ramp->entries.blue[i])) + if (dc_fixpt_lt(max, ramp->entries.blue[i])) max = ramp->entries.blue[i]; } - if (dal_fixed31_32_lt(min, dal_fixed31_32_zero)) - delta = dal_fixed31_32_neg(min); + if (dc_fixpt_lt(min, dc_fixpt_zero)) + delta = dc_fixpt_neg(min); - offset = dal_fixed31_32_add(min, max); + offset = dc_fixpt_add(min, max); for (i = 0 ; i < ramp->num_entries; i++) { - pwl_rgb[i].r = dal_fixed31_32_div( - dal_fixed31_32_add( + pwl_rgb[i].r = dc_fixpt_div( + dc_fixpt_add( ramp->entries.red[i], delta), offset); - pwl_rgb[i].g = dal_fixed31_32_div( - dal_fixed31_32_add( + pwl_rgb[i].g = dc_fixpt_div( + dc_fixpt_add( ramp->entries.green[i], delta), offset); - pwl_rgb[i].b = dal_fixed31_32_div( - dal_fixed31_32_add( + pwl_rgb[i].b = dc_fixpt_div( + dc_fixpt_add( ramp->entries.blue[i], delta), offset); } - pwl_rgb[i].r = dal_fixed31_32_sub(dal_fixed31_32_mul_int( + pwl_rgb[i].r = dc_fixpt_sub(dc_fixpt_mul_int( pwl_rgb[i-1].r, 2), pwl_rgb[i-2].r); - pwl_rgb[i].g = dal_fixed31_32_sub(dal_fixed31_32_mul_int( + pwl_rgb[i].g = dc_fixpt_sub(dc_fixpt_mul_int( pwl_rgb[i-1].g, 2), pwl_rgb[i-2].g); - pwl_rgb[i].b = dal_fixed31_32_sub(dal_fixed31_32_mul_int( + pwl_rgb[i].b = dc_fixpt_sub(dc_fixpt_mul_int( pwl_rgb[i-1].b, 2), pwl_rgb[i-2].b); ++i; - pwl_rgb[i].r = dal_fixed31_32_sub(dal_fixed31_32_mul_int( + pwl_rgb[i].r = dc_fixpt_sub(dc_fixpt_mul_int( pwl_rgb[i-1].r, 2), pwl_rgb[i-2].r); - pwl_rgb[i].g = dal_fixed31_32_sub(dal_fixed31_32_mul_int( + pwl_rgb[i].g = dc_fixpt_sub(dc_fixpt_mul_int( pwl_rgb[i-1].g, 2), pwl_rgb[i-2].g); - pwl_rgb[i].b = dal_fixed31_32_sub(dal_fixed31_32_mul_int( + pwl_rgb[i].b = dc_fixpt_sub(dc_fixpt_mul_int( pwl_rgb[i-1].b, 2), pwl_rgb[i-2].b); } @@ -846,40 +846,40 @@ static void scale_user_regamma_ramp(struct pwl_float_data *pwl_rgb, i = 0; do { - rgb->r = dal_fixed31_32_from_fraction( + rgb->r = dc_fixpt_from_fraction( ramp->gamma[i], scaler); - rgb->g = dal_fixed31_32_from_fraction( + rgb->g = dc_fixpt_from_fraction( ramp->gamma[i + 256], scaler); - rgb->b = dal_fixed31_32_from_fraction( + rgb->b = dc_fixpt_from_fraction( ramp->gamma[i + 512], scaler); ++rgb; ++i; } while (i != GAMMA_RGB_256_ENTRIES); - rgb->r = dal_fixed31_32_mul(rgb_last->r, + rgb->r = dc_fixpt_mul(rgb_last->r, dividers.divider1); - rgb->g = dal_fixed31_32_mul(rgb_last->g, + rgb->g = dc_fixpt_mul(rgb_last->g, dividers.divider1); - rgb->b = dal_fixed31_32_mul(rgb_last->b, + rgb->b = dc_fixpt_mul(rgb_last->b, dividers.divider1); ++rgb; - rgb->r = dal_fixed31_32_mul(rgb_last->r, + rgb->r = dc_fixpt_mul(rgb_last->r, dividers.divider2); - rgb->g = dal_fixed31_32_mul(rgb_last->g, + rgb->g = dc_fixpt_mul(rgb_last->g, dividers.divider2); - rgb->b = dal_fixed31_32_mul(rgb_last->b, + rgb->b = dc_fixpt_mul(rgb_last->b, dividers.divider2); ++rgb; - rgb->r = dal_fixed31_32_mul(rgb_last->r, + rgb->r = dc_fixpt_mul(rgb_last->r, dividers.divider3); - rgb->g = dal_fixed31_32_mul(rgb_last->g, + rgb->g = dc_fixpt_mul(rgb_last->g, dividers.divider3); - rgb->b = dal_fixed31_32_mul(rgb_last->b, + rgb->b = dc_fixpt_mul(rgb_last->b, dividers.divider3); } @@ -913,7 +913,7 @@ static void apply_lut_1d( struct fixed31_32 lut2; const int max_lut_index = 4095; const struct fixed31_32 max_lut_index_f = - dal_fixed31_32_from_int_nonconst(max_lut_index); + dc_fixpt_from_int(max_lut_index); int32_t index = 0, index_next = 0; struct fixed31_32 index_f; struct fixed31_32 delta_lut; @@ -931,10 +931,10 @@ static void apply_lut_1d( else regamma_y = &tf_pts->blue[i]; - norm_y = dal_fixed31_32_mul(max_lut_index_f, + norm_y = dc_fixpt_mul(max_lut_index_f, *regamma_y); - index = dal_fixed31_32_floor(norm_y); - index_f = dal_fixed31_32_from_int_nonconst(index); + index = dc_fixpt_floor(norm_y); + index_f = dc_fixpt_from_int(index); if (index < 0 || index > max_lut_index) continue; @@ -953,11 +953,11 @@ static void apply_lut_1d( } // we have everything now, so interpolate - delta_lut = dal_fixed31_32_sub(lut2, lut1); - delta_index = dal_fixed31_32_sub(norm_y, index_f); + delta_lut = dc_fixpt_sub(lut2, lut1); + delta_index = dc_fixpt_sub(norm_y, index_f); - *regamma_y = dal_fixed31_32_add(lut1, - dal_fixed31_32_mul(delta_index, delta_lut)); + *regamma_y = dc_fixpt_add(lut1, + dc_fixpt_mul(delta_index, delta_lut)); } } } @@ -973,7 +973,7 @@ static void build_evenly_distributed_points( uint32_t i = 0; do { - struct fixed31_32 value = dal_fixed31_32_from_fraction(i, + struct fixed31_32 value = dc_fixpt_from_fraction(i, numberof_points - 1); p->r = value; @@ -984,21 +984,21 @@ static void build_evenly_distributed_points( ++i; } while (i != numberof_points); - p->r = dal_fixed31_32_div(p_last->r, dividers.divider1); - p->g = dal_fixed31_32_div(p_last->g, dividers.divider1); - p->b = dal_fixed31_32_div(p_last->b, dividers.divider1); + p->r = dc_fixpt_div(p_last->r, dividers.divider1); + p->g = dc_fixpt_div(p_last->g, dividers.divider1); + p->b = dc_fixpt_div(p_last->b, dividers.divider1); ++p; - p->r = dal_fixed31_32_div(p_last->r, dividers.divider2); - p->g = dal_fixed31_32_div(p_last->g, dividers.divider2); - p->b = dal_fixed31_32_div(p_last->b, dividers.divider2); + p->r = dc_fixpt_div(p_last->r, dividers.divider2); + p->g = dc_fixpt_div(p_last->g, dividers.divider2); + p->b = dc_fixpt_div(p_last->b, dividers.divider2); ++p; - p->r = dal_fixed31_32_div(p_last->r, dividers.divider3); - p->g = dal_fixed31_32_div(p_last->g, dividers.divider3); - p->b = dal_fixed31_32_div(p_last->b, dividers.divider3); + p->r = dc_fixpt_div(p_last->r, dividers.divider3); + p->g = dc_fixpt_div(p_last->g, dividers.divider3); + p->b = dc_fixpt_div(p_last->b, dividers.divider3); } static inline void copy_rgb_regamma_to_coordinates_x( @@ -1094,7 +1094,7 @@ static void interpolate_user_regamma(uint32_t hw_points_num, struct fixed31_32 *tf_point; struct fixed31_32 hw_x; struct fixed31_32 norm_factor = - dal_fixed31_32_from_int_nonconst(255); + dc_fixpt_from_int(255); struct fixed31_32 norm_x; struct fixed31_32 index_f; struct fixed31_32 lut1; @@ -1105,9 +1105,9 @@ static void interpolate_user_regamma(uint32_t hw_points_num, i = 0; /* fixed_pt library has problems handling too small values */ while (i != 32) { - tf_pts->red[i] = dal_fixed31_32_zero; - tf_pts->green[i] = dal_fixed31_32_zero; - tf_pts->blue[i] = dal_fixed31_32_zero; + tf_pts->red[i] = dc_fixpt_zero; + tf_pts->green[i] = dc_fixpt_zero; + tf_pts->blue[i] = dc_fixpt_zero; ++i; } while (i <= hw_points_num + 1) { @@ -1129,12 +1129,12 @@ static void interpolate_user_regamma(uint32_t hw_points_num, } else hw_x = coordinates_x[i].x; - norm_x = dal_fixed31_32_mul(norm_factor, hw_x); - index = dal_fixed31_32_floor(norm_x); + norm_x = dc_fixpt_mul(norm_factor, hw_x); + index = dc_fixpt_floor(norm_x); if (index < 0 || index > 255) continue; - index_f = dal_fixed31_32_from_int_nonconst(index); + index_f = dc_fixpt_from_int(index); index_next = (index == 255) ? index : index + 1; if (color == 0) { @@ -1149,11 +1149,11 @@ static void interpolate_user_regamma(uint32_t hw_points_num, } // we have everything now, so interpolate - delta_lut = dal_fixed31_32_sub(lut2, lut1); - delta_index = dal_fixed31_32_sub(norm_x, index_f); + delta_lut = dc_fixpt_sub(lut2, lut1); + delta_index = dc_fixpt_sub(norm_x, index_f); - *tf_point = dal_fixed31_32_add(lut1, - dal_fixed31_32_mul(delta_index, delta_lut)); + *tf_point = dc_fixpt_add(lut1, + dc_fixpt_mul(delta_index, delta_lut)); } ++i; } @@ -1168,15 +1168,15 @@ static void build_new_custom_resulted_curve( i = 0; while (i != hw_points_num + 1) { - tf_pts->red[i] = dal_fixed31_32_clamp( - tf_pts->red[i], dal_fixed31_32_zero, - dal_fixed31_32_one); - tf_pts->green[i] = dal_fixed31_32_clamp( - tf_pts->green[i], dal_fixed31_32_zero, - dal_fixed31_32_one); - tf_pts->blue[i] = dal_fixed31_32_clamp( - tf_pts->blue[i], dal_fixed31_32_zero, - dal_fixed31_32_one); + tf_pts->red[i] = dc_fixpt_clamp( + tf_pts->red[i], dc_fixpt_zero, + dc_fixpt_one); + tf_pts->green[i] = dc_fixpt_clamp( + tf_pts->green[i], dc_fixpt_zero, + dc_fixpt_one); + tf_pts->blue[i] = dc_fixpt_clamp( + tf_pts->blue[i], dc_fixpt_zero, + dc_fixpt_one); ++i; } @@ -1290,9 +1290,9 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf, if (!coeff) goto coeff_alloc_fail; - dividers.divider1 = dal_fixed31_32_from_fraction(3, 2); - dividers.divider2 = dal_fixed31_32_from_int(2); - dividers.divider3 = dal_fixed31_32_from_fraction(5, 2); + dividers.divider1 = dc_fixpt_from_fraction(3, 2); + dividers.divider2 = dc_fixpt_from_int(2); + dividers.divider3 = dc_fixpt_from_fraction(5, 2); tf = output_tf->tf; @@ -1357,15 +1357,15 @@ bool calculate_user_regamma_coeff(struct dc_transfer_func *output_tf, uint32_t i = 0; do { - coeff.a0[i] = dal_fixed31_32_from_fraction( + coeff.a0[i] = dc_fixpt_from_fraction( regamma->coeff.A0[i], 10000000); - coeff.a1[i] = dal_fixed31_32_from_fraction( + coeff.a1[i] = dc_fixpt_from_fraction( regamma->coeff.A1[i], 1000); - coeff.a2[i] = dal_fixed31_32_from_fraction( + coeff.a2[i] = dc_fixpt_from_fraction( regamma->coeff.A2[i], 1000); - coeff.a3[i] = dal_fixed31_32_from_fraction( + coeff.a3[i] = dc_fixpt_from_fraction( regamma->coeff.A3[i], 1000); - coeff.user_gamma[i] = dal_fixed31_32_from_fraction( + coeff.user_gamma[i] = dc_fixpt_from_fraction( regamma->coeff.gamma[i], 1000); ++i; @@ -1374,9 +1374,9 @@ bool calculate_user_regamma_coeff(struct dc_transfer_func *output_tf, i = 0; /* fixed_pt library has problems handling too small values */ while (i != 32) { - output_tf->tf_pts.red[i] = dal_fixed31_32_zero; - output_tf->tf_pts.green[i] = dal_fixed31_32_zero; - output_tf->tf_pts.blue[i] = dal_fixed31_32_zero; + output_tf->tf_pts.red[i] = dc_fixpt_zero; + output_tf->tf_pts.green[i] = dc_fixpt_zero; + output_tf->tf_pts.blue[i] = dc_fixpt_zero; ++coord_x; ++i; } @@ -1423,9 +1423,9 @@ bool calculate_user_regamma_ramp(struct dc_transfer_func *output_tf, if (!rgb_regamma) goto rgb_regamma_alloc_fail; - dividers.divider1 = dal_fixed31_32_from_fraction(3, 2); - dividers.divider2 = dal_fixed31_32_from_int(2); - dividers.divider3 = dal_fixed31_32_from_fraction(5, 2); + dividers.divider1 = dc_fixpt_from_fraction(3, 2); + dividers.divider2 = dc_fixpt_from_int(2); + dividers.divider3 = dc_fixpt_from_fraction(5, 2); scale_user_regamma_ramp(rgb_user, ®amma->ramp, dividers); @@ -1496,9 +1496,9 @@ bool mod_color_calculate_degamma_params(struct dc_transfer_func *input_tf, if (!coeff) goto coeff_alloc_fail; - dividers.divider1 = dal_fixed31_32_from_fraction(3, 2); - dividers.divider2 = dal_fixed31_32_from_int(2); - dividers.divider3 = dal_fixed31_32_from_fraction(5, 2); + dividers.divider1 = dc_fixpt_from_fraction(3, 2); + dividers.divider2 = dc_fixpt_from_int(2); + dividers.divider3 = dc_fixpt_from_fraction(5, 2); tf = input_tf->tf; diff --git a/drivers/gpu/drm/amd/display/modules/inc/mod_stats.h b/drivers/gpu/drm/amd/display/modules/inc/mod_stats.h index 3230e2adb870..3812094b52e8 100644 --- a/drivers/gpu/drm/amd/display/modules/inc/mod_stats.h +++ b/drivers/gpu/drm/amd/display/modules/inc/mod_stats.h @@ -46,6 +46,10 @@ void mod_stats_dump(struct mod_stats *mod_stats); void mod_stats_reset_data(struct mod_stats *mod_stats); +void mod_stats_update_event(struct mod_stats *mod_stats, + char *event_string, + unsigned int length); + void mod_stats_update_flip(struct mod_stats *mod_stats, unsigned long timestamp_in_ns); diff --git a/drivers/gpu/drm/amd/display/modules/stats/stats.c b/drivers/gpu/drm/amd/display/modules/stats/stats.c index 48e02197919f..3f7d47fdc367 100644 --- a/drivers/gpu/drm/amd/display/modules/stats/stats.c +++ b/drivers/gpu/drm/amd/display/modules/stats/stats.c @@ -36,9 +36,14 @@ #define DAL_STATS_ENTRIES_REGKEY_DEFAULT 0x00350000 #define DAL_STATS_ENTRIES_REGKEY_MAX 0x01000000 +#define DAL_STATS_EVENT_ENTRIES_DEFAULT 0x00000100 + #define MOD_STATS_NUM_VSYNCS 5 +#define MOD_STATS_EVENT_STRING_MAX 512 struct stats_time_cache { + unsigned int entry_id; + unsigned long flip_timestamp_in_ns; unsigned long vupdate_timestamp_in_ns; @@ -63,15 +68,26 @@ struct stats_time_cache { unsigned int flags; }; +struct stats_event_cache { + unsigned int entry_id; + char event_string[MOD_STATS_EVENT_STRING_MAX]; +}; + struct core_stats { struct mod_stats public; struct dc *dc; + bool enabled; + unsigned int entries; + unsigned int event_entries; + unsigned int entry_id; + struct stats_time_cache *time; unsigned int index; - bool enabled; - unsigned int entries; + struct stats_event_cache *events; + unsigned int event_index; + }; #define MOD_STATS_TO_CORE(mod_stats)\ @@ -99,12 +115,12 @@ struct mod_stats *mod_stats_create(struct dc *dc) unsigned int reg_data; int i = 0; + if (dc == NULL) + goto fail_construct; + core_stats = kzalloc(sizeof(struct core_stats), GFP_KERNEL); if (core_stats == NULL) - goto fail_alloc_context; - - if (dc == NULL) goto fail_construct; core_stats->dc = dc; @@ -125,27 +141,45 @@ struct mod_stats *mod_stats_create(struct dc *dc) else core_stats->entries = reg_data; } + core_stats->time = kzalloc( + sizeof(struct stats_time_cache) * + core_stats->entries, + GFP_KERNEL); + + if (core_stats->time == NULL) + goto fail_construct_time; - core_stats->time = kzalloc(sizeof(struct stats_time_cache) * core_stats->entries, + core_stats->event_entries = DAL_STATS_EVENT_ENTRIES_DEFAULT; + core_stats->events = kzalloc( + sizeof(struct stats_event_cache) * + core_stats->event_entries, GFP_KERNEL); + + if (core_stats->events == NULL) + goto fail_construct_events; + } else { core_stats->entries = 0; } - if (core_stats->time == NULL) - goto fail_construct; - /* Purposely leave index 0 unused so we don't need special logic to * handle calculation cases that depend on previous flip data. */ core_stats->index = 1; + core_stats->event_index = 0; + + // Keeps track of ordering within the different stats structures + core_stats->entry_id = 0; return &core_stats->public; -fail_construct: +fail_construct_events: + kfree(core_stats->time); + +fail_construct_time: kfree(core_stats); -fail_alloc_context: +fail_construct: return NULL; } @@ -157,6 +191,9 @@ void mod_stats_destroy(struct mod_stats *mod_stats) if (core_stats->time != NULL) kfree(core_stats->time); + if (core_stats->events != NULL) + kfree(core_stats->events); + kfree(core_stats); } } @@ -167,7 +204,11 @@ void mod_stats_dump(struct mod_stats *mod_stats) struct dal_logger *logger = NULL; struct core_stats *core_stats = NULL; struct stats_time_cache *time = NULL; + struct stats_event_cache *events = NULL; + unsigned int time_index = 1; + unsigned int event_index = 0; unsigned int index = 0; + struct log_entry log_entry; if (mod_stats == NULL) return; @@ -176,45 +217,62 @@ void mod_stats_dump(struct mod_stats *mod_stats) dc = core_stats->dc; logger = dc->ctx->logger; time = core_stats->time; - - //LogEntry* pLog = GetLog()->Open(LogMajor_ISR, LogMinor_ISR_FreeSyncSW); - - //if (!pLog->IsDummyEntry()) - { - dm_logger_write(logger, LOG_PROFILING, "==Display Caps==\n"); - dm_logger_write(logger, LOG_PROFILING, "\n"); - dm_logger_write(logger, LOG_PROFILING, "\n"); - - dm_logger_write(logger, LOG_PROFILING, "==Stats==\n"); - dm_logger_write(logger, LOG_PROFILING, - "render avgRender minWindow midPoint maxWindow vsyncToFlip flipToVsync #vsyncBetweenFlip #frame insertDuration vTotalMin vTotalMax eventTrigs vSyncTime1 vSyncTime2 vSyncTime3 vSyncTime4 vSyncTime5 flags\n"); - - for (int i = 0; i < core_stats->index && i < core_stats->entries; i++) { - dm_logger_write(logger, LOG_PROFILING, - "%u %u %u %u %u %u %u %u %u %u %u %u %u %u %u %u %u %u %u", - time[i].render_time_in_us, - time[i].avg_render_time_in_us_last_ten, - time[i].min_window, - time[i].lfc_mid_point_in_us, - time[i].max_window, - time[i].vsync_to_flip_time_in_us, - time[i].flip_to_vsync_time_in_us, - time[i].num_vsync_between_flips, - time[i].num_frames_inserted, - time[i].inserted_duration_in_us, - time[i].v_total_min, - time[i].v_total_max, - time[i].event_triggers, - time[i].v_sync_time_in_us[0], - time[i].v_sync_time_in_us[1], - time[i].v_sync_time_in_us[2], - time[i].v_sync_time_in_us[3], - time[i].v_sync_time_in_us[4], - time[i].flags); + events = core_stats->events; + + DISPLAY_STATS_BEGIN(log_entry); + + DISPLAY_STATS("==Display Caps==\n"); + + DISPLAY_STATS("==Display Stats==\n"); + + DISPLAY_STATS("%10s %10s %10s %10s %10s" + " %11s %11s %17s %10s %14s" + " %10s %10s %10s %10s %10s" + " %10s %10s %10s %10s\n", + "render", "avgRender", + "minWindow", "midPoint", "maxWindow", + "vsyncToFlip", "flipToVsync", "vsyncsBetweenFlip", + "numFrame", "insertDuration", + "vTotalMin", "vTotalMax", "eventTrigs", + "vSyncTime1", "vSyncTime2", "vSyncTime3", + "vSyncTime4", "vSyncTime5", "flags"); + + for (int i = 0; i < core_stats->entry_id; i++) { + if (event_index < core_stats->event_index && + i == events[event_index].entry_id) { + DISPLAY_STATS("%s\n", events[event_index].event_string); + event_index++; + } else if (time_index < core_stats->index && + i == time[time_index].entry_id) { + DISPLAY_STATS("%10u %10u %10u %10u %10u" + " %11u %11u %17u %10u %14u" + " %10u %10u %10u %10u %10u" + " %10u %10u %10u %10u\n", + time[time_index].render_time_in_us, + time[time_index].avg_render_time_in_us_last_ten, + time[time_index].min_window, + time[time_index].lfc_mid_point_in_us, + time[time_index].max_window, + time[time_index].vsync_to_flip_time_in_us, + time[time_index].flip_to_vsync_time_in_us, + time[time_index].num_vsync_between_flips, + time[time_index].num_frames_inserted, + time[time_index].inserted_duration_in_us, + time[time_index].v_total_min, + time[time_index].v_total_max, + time[time_index].event_triggers, + time[time_index].v_sync_time_in_us[0], + time[time_index].v_sync_time_in_us[1], + time[time_index].v_sync_time_in_us[2], + time[time_index].v_sync_time_in_us[3], + time[time_index].v_sync_time_in_us[4], + time[time_index].flags); + + time_index++; } } - //GetLog()->Close(pLog); - //GetLog()->UnSetLogMask(LogMajor_ISR, LogMinor_ISR_FreeSyncSW); + + DISPLAY_STATS_END(log_entry); } void mod_stats_reset_data(struct mod_stats *mod_stats) @@ -231,7 +289,46 @@ void mod_stats_reset_data(struct mod_stats *mod_stats) memset(core_stats->time, 0, sizeof(struct stats_time_cache) * core_stats->entries); + memset(core_stats->events, 0, + sizeof(struct stats_event_cache) * core_stats->event_entries); + core_stats->index = 1; + core_stats->event_index = 0; + + // Keeps track of ordering within the different stats structures + core_stats->entry_id = 0; +} + +void mod_stats_update_event(struct mod_stats *mod_stats, + char *event_string, + unsigned int length) +{ + struct core_stats *core_stats = NULL; + struct stats_event_cache *events = NULL; + unsigned int index = 0; + unsigned int copy_length = 0; + + if (mod_stats == NULL) + return; + + core_stats = MOD_STATS_TO_CORE(mod_stats); + + if (core_stats->event_index >= core_stats->event_entries) + return; + + events = core_stats->events; + index = core_stats->event_index; + + copy_length = length; + if (length > MOD_STATS_EVENT_STRING_MAX) + copy_length = MOD_STATS_EVENT_STRING_MAX; + + memcpy(&events[index].event_string, event_string, copy_length); + events[index].event_string[copy_length - 1] = '\0'; + + events[index].entry_id = core_stats->entry_id; + core_stats->event_index++; + core_stats->entry_id++; } void mod_stats_update_flip(struct mod_stats *mod_stats, @@ -272,7 +369,9 @@ void mod_stats_update_flip(struct mod_stats *mod_stats, (timestamp_in_ns - time[index - 1].vupdate_timestamp_in_ns) / 1000; + time[index].entry_id = core_stats->entry_id; core_stats->index++; + core_stats->entry_id++; } void mod_stats_update_vupdate(struct mod_stats *mod_stats, diff --git a/drivers/gpu/drm/amd/include/amd_shared.h b/drivers/gpu/drm/amd/include/amd_shared.h index 33de33016bda..b178176b72ac 100644 --- a/drivers/gpu/drm/amd/include/amd_shared.h +++ b/drivers/gpu/drm/amd/include/amd_shared.h @@ -92,7 +92,7 @@ enum amd_powergating_state { #define AMD_CG_SUPPORT_GFX_3D_CGLS (1 << 21) #define AMD_CG_SUPPORT_DRM_MGCG (1 << 22) #define AMD_CG_SUPPORT_DF_MGCG (1 << 23) - +#define AMD_CG_SUPPORT_VCN_MGCG (1 << 24) /* PG flags */ #define AMD_PG_SUPPORT_GFX_PG (1 << 0) #define AMD_PG_SUPPORT_GFX_SMG (1 << 1) @@ -108,6 +108,7 @@ enum amd_powergating_state { #define AMD_PG_SUPPORT_GFX_QUICK_MG (1 << 11) #define AMD_PG_SUPPORT_GFX_PIPELINE (1 << 12) #define AMD_PG_SUPPORT_MMHUB (1 << 13) +#define AMD_PG_SUPPORT_VCN (1 << 14) enum PP_FEATURE_MASK { PP_SCLK_DPM_MASK = 0x1, diff --git a/drivers/gpu/drm/amd/include/asic_reg/dce/dce_12_0_offset.h b/drivers/gpu/drm/amd/include/asic_reg/dce/dce_12_0_offset.h index f730d0629020..b6f74bf4af02 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/dce/dce_12_0_offset.h +++ b/drivers/gpu/drm/amd/include/asic_reg/dce/dce_12_0_offset.h @@ -2095,6 +2095,18 @@ #define mmDC_GPIO_AUX_CTRL_2_BASE_IDX 2 #define mmDC_GPIO_RXEN 0x212f #define mmDC_GPIO_RXEN_BASE_IDX 2 +#define mmDC_GPIO_AUX_CTRL_3 0x2130 +#define mmDC_GPIO_AUX_CTRL_3_BASE_IDX 2 +#define mmDC_GPIO_AUX_CTRL_4 0x2131 +#define mmDC_GPIO_AUX_CTRL_4_BASE_IDX 2 +#define mmDC_GPIO_AUX_CTRL_5 0x2132 +#define mmDC_GPIO_AUX_CTRL_5_BASE_IDX 2 +#define mmAUXI2C_PAD_ALL_PWR_OK 0x2133 +#define mmAUXI2C_PAD_ALL_PWR_OK_BASE_IDX 2 +#define mmDC_GPIO_PULLUPEN 0x2134 +#define mmDC_GPIO_PULLUPEN_BASE_IDX 2 +#define mmDC_GPIO_AUX_CTRL_6 0x2135 +#define mmDC_GPIO_AUX_CTRL_6_BASE_IDX 2 #define mmBPHYC_DAC_MACRO_CNTL 0x2136 #define mmBPHYC_DAC_MACRO_CNTL_BASE_IDX 2 #define mmDAC_MACRO_CNTL_RESERVED0 0x2136 diff --git a/drivers/gpu/drm/amd/include/asic_reg/dce/dce_12_0_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/dce/dce_12_0_sh_mask.h index 6d3162c42957..bcd190a3fcdd 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/dce/dce_12_0_sh_mask.h +++ b/drivers/gpu/drm/amd/include/asic_reg/dce/dce_12_0_sh_mask.h @@ -10971,6 +10971,158 @@ #define DC_GPIO_RXEN__DC_GPIO_BLON_RXEN_MASK 0x00100000L #define DC_GPIO_RXEN__DC_GPIO_DIGON_RXEN_MASK 0x00200000L #define DC_GPIO_RXEN__DC_GPIO_ENA_BL_RXEN_MASK 0x00400000L +//DC_GPIO_AUX_CTRL_3 +#define DC_GPIO_AUX_CTRL_3__AUX1_NEN_RTERM__SHIFT 0x0 +#define DC_GPIO_AUX_CTRL_3__AUX2_NEN_RTERM__SHIFT 0x1 +#define DC_GPIO_AUX_CTRL_3__AUX3_NEN_RTERM__SHIFT 0x2 +#define DC_GPIO_AUX_CTRL_3__AUX4_NEN_RTERM__SHIFT 0x3 +#define DC_GPIO_AUX_CTRL_3__AUX5_NEN_RTERM__SHIFT 0x4 +#define DC_GPIO_AUX_CTRL_3__AUX6_NEN_RTERM__SHIFT 0x5 +#define DC_GPIO_AUX_CTRL_3__AUX1_DP_DN_SWAP__SHIFT 0x8 +#define DC_GPIO_AUX_CTRL_3__AUX2_DP_DN_SWAP__SHIFT 0x9 +#define DC_GPIO_AUX_CTRL_3__AUX3_DP_DN_SWAP__SHIFT 0xa +#define DC_GPIO_AUX_CTRL_3__AUX4_DP_DN_SWAP__SHIFT 0xb +#define DC_GPIO_AUX_CTRL_3__AUX5_DP_DN_SWAP__SHIFT 0xc +#define DC_GPIO_AUX_CTRL_3__AUX6_DP_DN_SWAP__SHIFT 0xd +#define DC_GPIO_AUX_CTRL_3__AUX1_HYS_TUNE__SHIFT 0x10 +#define DC_GPIO_AUX_CTRL_3__AUX2_HYS_TUNE__SHIFT 0x12 +#define DC_GPIO_AUX_CTRL_3__AUX3_HYS_TUNE__SHIFT 0x14 +#define DC_GPIO_AUX_CTRL_3__AUX4_HYS_TUNE__SHIFT 0x16 +#define DC_GPIO_AUX_CTRL_3__AUX5_HYS_TUNE__SHIFT 0x18 +#define DC_GPIO_AUX_CTRL_3__AUX6_HYS_TUNE__SHIFT 0x1a +#define DC_GPIO_AUX_CTRL_3__AUX1_NEN_RTERM_MASK 0x00000001L +#define DC_GPIO_AUX_CTRL_3__AUX2_NEN_RTERM_MASK 0x00000002L +#define DC_GPIO_AUX_CTRL_3__AUX3_NEN_RTERM_MASK 0x00000004L +#define DC_GPIO_AUX_CTRL_3__AUX4_NEN_RTERM_MASK 0x00000008L +#define DC_GPIO_AUX_CTRL_3__AUX5_NEN_RTERM_MASK 0x00000010L +#define DC_GPIO_AUX_CTRL_3__AUX6_NEN_RTERM_MASK 0x00000020L +#define DC_GPIO_AUX_CTRL_3__AUX1_DP_DN_SWAP_MASK 0x00000100L +#define DC_GPIO_AUX_CTRL_3__AUX2_DP_DN_SWAP_MASK 0x00000200L +#define DC_GPIO_AUX_CTRL_3__AUX3_DP_DN_SWAP_MASK 0x00000400L +#define DC_GPIO_AUX_CTRL_3__AUX4_DP_DN_SWAP_MASK 0x00000800L +#define DC_GPIO_AUX_CTRL_3__AUX5_DP_DN_SWAP_MASK 0x00001000L +#define DC_GPIO_AUX_CTRL_3__AUX6_DP_DN_SWAP_MASK 0x00002000L +#define DC_GPIO_AUX_CTRL_3__AUX1_HYS_TUNE_MASK 0x00030000L +#define DC_GPIO_AUX_CTRL_3__AUX2_HYS_TUNE_MASK 0x000C0000L +#define DC_GPIO_AUX_CTRL_3__AUX3_HYS_TUNE_MASK 0x00300000L +#define DC_GPIO_AUX_CTRL_3__AUX4_HYS_TUNE_MASK 0x00C00000L +#define DC_GPIO_AUX_CTRL_3__AUX5_HYS_TUNE_MASK 0x03000000L +#define DC_GPIO_AUX_CTRL_3__AUX6_HYS_TUNE_MASK 0x0C000000L +//DC_GPIO_AUX_CTRL_4 +#define DC_GPIO_AUX_CTRL_4__AUX1_AUX_CTRL__SHIFT 0x0 +#define DC_GPIO_AUX_CTRL_4__AUX2_AUX_CTRL__SHIFT 0x4 +#define DC_GPIO_AUX_CTRL_4__AUX3_AUX_CTRL__SHIFT 0x8 +#define DC_GPIO_AUX_CTRL_4__AUX4_AUX_CTRL__SHIFT 0xc +#define DC_GPIO_AUX_CTRL_4__AUX5_AUX_CTRL__SHIFT 0x10 +#define DC_GPIO_AUX_CTRL_4__AUX6_AUX_CTRL__SHIFT 0x14 +#define DC_GPIO_AUX_CTRL_4__AUX1_AUX_CTRL_MASK 0x0000000FL +#define DC_GPIO_AUX_CTRL_4__AUX2_AUX_CTRL_MASK 0x000000F0L +#define DC_GPIO_AUX_CTRL_4__AUX3_AUX_CTRL_MASK 0x00000F00L +#define DC_GPIO_AUX_CTRL_4__AUX4_AUX_CTRL_MASK 0x0000F000L +#define DC_GPIO_AUX_CTRL_4__AUX5_AUX_CTRL_MASK 0x000F0000L +#define DC_GPIO_AUX_CTRL_4__AUX6_AUX_CTRL_MASK 0x00F00000L +//DC_GPIO_AUX_CTRL_5 +#define DC_GPIO_AUX_CTRL_5__AUX1_VOD_TUNE__SHIFT 0x0 +#define DC_GPIO_AUX_CTRL_5__AUX2_VOD_TUNE__SHIFT 0x2 +#define DC_GPIO_AUX_CTRL_5__AUX3_VOD_TUNE__SHIFT 0x4 +#define DC_GPIO_AUX_CTRL_5__AUX4_VOD_TUNE__SHIFT 0x6 +#define DC_GPIO_AUX_CTRL_5__AUX5_VOD_TUNE__SHIFT 0x8 +#define DC_GPIO_AUX_CTRL_5__AUX6_VOD_TUNE__SHIFT 0xa +#define DC_GPIO_AUX_CTRL_5__DDC_PAD1_I2CMODE__SHIFT 0xc +#define DC_GPIO_AUX_CTRL_5__DDC_PAD2_I2CMODE__SHIFT 0xd +#define DC_GPIO_AUX_CTRL_5__DDC_PAD3_I2CMODE__SHIFT 0xe +#define DC_GPIO_AUX_CTRL_5__DDC_PAD4_I2CMODE__SHIFT 0xf +#define DC_GPIO_AUX_CTRL_5__DDC_PAD5_I2CMODE__SHIFT 0x10 +#define DC_GPIO_AUX_CTRL_5__DDC_PAD6_I2CMODE__SHIFT 0x11 +#define DC_GPIO_AUX_CTRL_5__DDC1_I2C_VPH_1V2_EN__SHIFT 0x12 +#define DC_GPIO_AUX_CTRL_5__DDC2_I2C_VPH_1V2_EN__SHIFT 0x13 +#define DC_GPIO_AUX_CTRL_5__DDC3_I2C_VPH_1V2_EN__SHIFT 0x14 +#define DC_GPIO_AUX_CTRL_5__DDC4_I2C_VPH_1V2_EN__SHIFT 0x15 +#define DC_GPIO_AUX_CTRL_5__DDC5_I2C_VPH_1V2_EN__SHIFT 0x16 +#define DC_GPIO_AUX_CTRL_5__DDC6_I2C_VPH_1V2_EN__SHIFT 0x17 +#define DC_GPIO_AUX_CTRL_5__DDC1_PAD_I2C_CTRL__SHIFT 0x18 +#define DC_GPIO_AUX_CTRL_5__DDC2_PAD_I2C_CTRL__SHIFT 0x19 +#define DC_GPIO_AUX_CTRL_5__DDC3_PAD_I2C_CTRL__SHIFT 0x1a +#define DC_GPIO_AUX_CTRL_5__DDC4_PAD_I2C_CTRL__SHIFT 0x1b +#define DC_GPIO_AUX_CTRL_5__DDC5_PAD_I2C_CTRL__SHIFT 0x1c +#define DC_GPIO_AUX_CTRL_5__DDC6_PAD_I2C_CTRL__SHIFT 0x1d +#define DC_GPIO_AUX_CTRL_5__AUX1_VOD_TUNE_MASK 0x00000003L +#define DC_GPIO_AUX_CTRL_5__AUX2_VOD_TUNE_MASK 0x0000000CL +#define DC_GPIO_AUX_CTRL_5__AUX3_VOD_TUNE_MASK 0x00000030L +#define DC_GPIO_AUX_CTRL_5__AUX4_VOD_TUNE_MASK 0x000000C0L +#define DC_GPIO_AUX_CTRL_5__AUX5_VOD_TUNE_MASK 0x00000300L +#define DC_GPIO_AUX_CTRL_5__AUX6_VOD_TUNE_MASK 0x00000C00L +#define DC_GPIO_AUX_CTRL_5__DDC_PAD1_I2CMODE_MASK 0x00001000L +#define DC_GPIO_AUX_CTRL_5__DDC_PAD2_I2CMODE_MASK 0x00002000L +#define DC_GPIO_AUX_CTRL_5__DDC_PAD3_I2CMODE_MASK 0x00004000L +#define DC_GPIO_AUX_CTRL_5__DDC_PAD4_I2CMODE_MASK 0x00008000L +#define DC_GPIO_AUX_CTRL_5__DDC_PAD5_I2CMODE_MASK 0x00010000L +#define DC_GPIO_AUX_CTRL_5__DDC_PAD6_I2CMODE_MASK 0x00020000L +#define DC_GPIO_AUX_CTRL_5__DDC1_I2C_VPH_1V2_EN_MASK 0x00040000L +#define DC_GPIO_AUX_CTRL_5__DDC2_I2C_VPH_1V2_EN_MASK 0x00080000L +#define DC_GPIO_AUX_CTRL_5__DDC3_I2C_VPH_1V2_EN_MASK 0x00100000L +#define DC_GPIO_AUX_CTRL_5__DDC4_I2C_VPH_1V2_EN_MASK 0x00200000L +#define DC_GPIO_AUX_CTRL_5__DDC5_I2C_VPH_1V2_EN_MASK 0x00400000L +#define DC_GPIO_AUX_CTRL_5__DDC6_I2C_VPH_1V2_EN_MASK 0x00800000L +#define DC_GPIO_AUX_CTRL_5__DDC1_PAD_I2C_CTRL_MASK 0x01000000L +#define DC_GPIO_AUX_CTRL_5__DDC2_PAD_I2C_CTRL_MASK 0x02000000L +#define DC_GPIO_AUX_CTRL_5__DDC3_PAD_I2C_CTRL_MASK 0x04000000L +#define DC_GPIO_AUX_CTRL_5__DDC4_PAD_I2C_CTRL_MASK 0x08000000L +#define DC_GPIO_AUX_CTRL_5__DDC5_PAD_I2C_CTRL_MASK 0x10000000L +#define DC_GPIO_AUX_CTRL_5__DDC6_PAD_I2C_CTRL_MASK 0x20000000L +//AUXI2C_PAD_ALL_PWR_OK +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY1_ALL_PWR_OK__SHIFT 0x0 +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY2_ALL_PWR_OK__SHIFT 0x1 +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY3_ALL_PWR_OK__SHIFT 0x2 +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY4_ALL_PWR_OK__SHIFT 0x3 +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY5_ALL_PWR_OK__SHIFT 0x4 +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY6_ALL_PWR_OK__SHIFT 0x5 +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY1_ALL_PWR_OK_MASK 0x00000001L +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY2_ALL_PWR_OK_MASK 0x00000002L +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY3_ALL_PWR_OK_MASK 0x00000004L +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY4_ALL_PWR_OK_MASK 0x00000008L +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY5_ALL_PWR_OK_MASK 0x00000010L +#define AUXI2C_PAD_ALL_PWR_OK__AUXI2C_PHY6_ALL_PWR_OK_MASK 0x00000020L +//DC_GPIO_PULLUPEN +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICA_PU_EN__SHIFT 0x0 +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICB_PU_EN__SHIFT 0x1 +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICC_PU_EN__SHIFT 0x2 +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICD_PU_EN__SHIFT 0x3 +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICE_PU_EN__SHIFT 0x4 +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICF_PU_EN__SHIFT 0x5 +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICG_PU_EN__SHIFT 0x6 +#define DC_GPIO_PULLUPEN__DC_GPIO_HSYNCA_PU_EN__SHIFT 0x8 +#define DC_GPIO_PULLUPEN__DC_GPIO_VSYNCA_PU_EN__SHIFT 0x9 +#define DC_GPIO_PULLUPEN__DC_GPIO_HPD1_PU_EN__SHIFT 0xe +#define DC_GPIO_PULLUPEN__DC_GPIO_BLON_PU_EN__SHIFT 0x14 +#define DC_GPIO_PULLUPEN__DC_GPIO_DIGON_PU_EN__SHIFT 0x15 +#define DC_GPIO_PULLUPEN__DC_GPIO_ENA_BL_PU_EN__SHIFT 0x16 +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICA_PU_EN_MASK 0x00000001L +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICB_PU_EN_MASK 0x00000002L +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICC_PU_EN_MASK 0x00000004L +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICD_PU_EN_MASK 0x00000008L +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICE_PU_EN_MASK 0x00000010L +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICF_PU_EN_MASK 0x00000020L +#define DC_GPIO_PULLUPEN__DC_GPIO_GENERICG_PU_EN_MASK 0x00000040L +#define DC_GPIO_PULLUPEN__DC_GPIO_HSYNCA_PU_EN_MASK 0x00000100L +#define DC_GPIO_PULLUPEN__DC_GPIO_VSYNCA_PU_EN_MASK 0x00000200L +#define DC_GPIO_PULLUPEN__DC_GPIO_HPD1_PU_EN_MASK 0x00004000L +#define DC_GPIO_PULLUPEN__DC_GPIO_BLON_PU_EN_MASK 0x00100000L +#define DC_GPIO_PULLUPEN__DC_GPIO_DIGON_PU_EN_MASK 0x00200000L +#define DC_GPIO_PULLUPEN__DC_GPIO_ENA_BL_PU_EN_MASK 0x00400000L +//DC_GPIO_AUX_CTRL_6 +#define DC_GPIO_AUX_CTRL_6__AUX1_PAD_RXSEL__SHIFT 0x0 +#define DC_GPIO_AUX_CTRL_6__AUX2_PAD_RXSEL__SHIFT 0x2 +#define DC_GPIO_AUX_CTRL_6__AUX3_PAD_RXSEL__SHIFT 0x4 +#define DC_GPIO_AUX_CTRL_6__AUX4_PAD_RXSEL__SHIFT 0x6 +#define DC_GPIO_AUX_CTRL_6__AUX5_PAD_RXSEL__SHIFT 0x8 +#define DC_GPIO_AUX_CTRL_6__AUX6_PAD_RXSEL__SHIFT 0xa +#define DC_GPIO_AUX_CTRL_6__AUX1_PAD_RXSEL_MASK 0x00000003L +#define DC_GPIO_AUX_CTRL_6__AUX2_PAD_RXSEL_MASK 0x0000000CL +#define DC_GPIO_AUX_CTRL_6__AUX3_PAD_RXSEL_MASK 0x00000030L +#define DC_GPIO_AUX_CTRL_6__AUX4_PAD_RXSEL_MASK 0x000000C0L +#define DC_GPIO_AUX_CTRL_6__AUX5_PAD_RXSEL_MASK 0x00000300L +#define DC_GPIO_AUX_CTRL_6__AUX6_PAD_RXSEL_MASK 0x00000C00L //BPHYC_DAC_MACRO_CNTL #define BPHYC_DAC_MACRO_CNTL__BPHYC_DAC_WHITE_LEVEL__SHIFT 0x0 #define BPHYC_DAC_MACRO_CNTL__BPHYC_DAC_WHITE_FINE_CONTROL__SHIFT 0x8 diff --git a/drivers/gpu/drm/amd/include/asic_reg/df/df_1_7_offset.h b/drivers/gpu/drm/amd/include/asic_reg/df/df_1_7_offset.h index 2b305dd021e8..e6044e27a913 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/df/df_1_7_offset.h +++ b/drivers/gpu/drm/amd/include/asic_reg/df/df_1_7_offset.h @@ -30,4 +30,8 @@ #define mmDF_CS_AON0_DramBaseAddress0 0x0044 #define mmDF_CS_AON0_DramBaseAddress0_BASE_IDX 0 +#define mmDF_CS_AON0_CoherentSlaveModeCtrlA0 0x0214 +#define mmDF_CS_AON0_CoherentSlaveModeCtrlA0_BASE_IDX 0 + + #endif diff --git a/drivers/gpu/drm/amd/include/asic_reg/df/df_1_7_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/df/df_1_7_sh_mask.h index 2ba849798924..a78c99480e2d 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/df/df_1_7_sh_mask.h +++ b/drivers/gpu/drm/amd/include/asic_reg/df/df_1_7_sh_mask.h @@ -45,4 +45,8 @@ #define DF_CS_AON0_DramBaseAddress0__IntLvAddrSel_MASK 0x00000700L #define DF_CS_AON0_DramBaseAddress0__DramBaseAddr_MASK 0xFFFFF000L +//DF_CS_AON0_CoherentSlaveModeCtrlA0 +#define DF_CS_AON0_CoherentSlaveModeCtrlA0__ForceParWrRMW__SHIFT 0x3 +#define DF_CS_AON0_CoherentSlaveModeCtrlA0__ForceParWrRMW_MASK 0x00000008L + #endif diff --git a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_default.h b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_default.h new file mode 100644 index 000000000000..e58c207ac980 --- /dev/null +++ b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_default.h @@ -0,0 +1,26 @@ +/* + * Copyright (C) 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ +#ifndef _df_3_6_DEFAULT_HEADER +#define _df_3_6_DEFAULT_HEADER + +#define mmFabricConfigAccessControl_DEFAULT 0x00000000 + +#endif diff --git a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h new file mode 100644 index 000000000000..a9575db8d7aa --- /dev/null +++ b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_offset.h @@ -0,0 +1,33 @@ +/* + * Copyright (C) 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ +#ifndef _df_3_6_OFFSET_HEADER +#define _df_3_6_OFFSET_HEADER + +#define mmFabricConfigAccessControl 0x0410 +#define mmFabricConfigAccessControl_BASE_IDX 0 + +#define mmDF_PIE_AON0_DfGlobalClkGater 0x00fc +#define mmDF_PIE_AON0_DfGlobalClkGater_BASE_IDX 0 + +#define mmDF_CS_UMC_AON0_DramBaseAddress0 0x0044 +#define mmDF_CS_UMC_AON0_DramBaseAddress0_BASE_IDX 0 + +#endif diff --git a/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h new file mode 100644 index 000000000000..88f7c69df6b9 --- /dev/null +++ b/drivers/gpu/drm/amd/include/asic_reg/df/df_3_6_sh_mask.h @@ -0,0 +1,48 @@ +/* + * Copyright (C) 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ +#ifndef _df_3_6_SH_MASK_HEADER +#define _df_3_6_SH_MASK_HEADER + +/* FabricConfigAccessControl */ +#define FabricConfigAccessControl__CfgRegInstAccEn__SHIFT 0x0 +#define FabricConfigAccessControl__CfgRegInstAccRegLock__SHIFT 0x1 +#define FabricConfigAccessControl__CfgRegInstID__SHIFT 0x10 +#define FabricConfigAccessControl__CfgRegInstAccEn_MASK 0x00000001L +#define FabricConfigAccessControl__CfgRegInstAccRegLock_MASK 0x00000002L +#define FabricConfigAccessControl__CfgRegInstID_MASK 0x00FF0000L + +/* DF_PIE_AON0_DfGlobalClkGater */ +#define DF_PIE_AON0_DfGlobalClkGater__MGCGMode__SHIFT 0x0 +#define DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK 0x0000000FL + +/* DF_CS_AON0_DramBaseAddress0 */ +#define DF_CS_UMC_AON0_DramBaseAddress0__AddrRngVal__SHIFT 0x0 +#define DF_CS_UMC_AON0_DramBaseAddress0__LgcyMmioHoleEn__SHIFT 0x1 +#define DF_CS_UMC_AON0_DramBaseAddress0__IntLvNumChan__SHIFT 0x4 +#define DF_CS_UMC_AON0_DramBaseAddress0__IntLvAddrSel__SHIFT 0x8 +#define DF_CS_UMC_AON0_DramBaseAddress0__DramBaseAddr__SHIFT 0xc +#define DF_CS_UMC_AON0_DramBaseAddress0__AddrRngVal_MASK 0x00000001L +#define DF_CS_UMC_AON0_DramBaseAddress0__LgcyMmioHoleEn_MASK 0x00000002L +#define DF_CS_UMC_AON0_DramBaseAddress0__IntLvNumChan_MASK 0x000000F0L +#define DF_CS_UMC_AON0_DramBaseAddress0__IntLvAddrSel_MASK 0x00000700L +#define DF_CS_UMC_AON0_DramBaseAddress0__DramBaseAddr_MASK 0xFFFFF000L + +#endif diff --git a/drivers/gpu/drm/amd/include/atomfirmware.h b/drivers/gpu/drm/amd/include/atomfirmware.h index de177ce8ca80..c6c1666ac120 100644 --- a/drivers/gpu/drm/amd/include/atomfirmware.h +++ b/drivers/gpu/drm/amd/include/atomfirmware.h @@ -1219,6 +1219,41 @@ struct atom_gfx_info_v2_3 { uint32_t rm21_sram_vmin_value; }; +struct atom_gfx_info_v2_4 { + struct atom_common_table_header table_header; + uint8_t gfxip_min_ver; + uint8_t gfxip_max_ver; + uint8_t gc_num_se; + uint8_t max_tile_pipes; + uint8_t gc_num_cu_per_sh; + uint8_t gc_num_sh_per_se; + uint8_t gc_num_rb_per_se; + uint8_t gc_num_tccs; + uint32_t regaddr_cp_dma_src_addr; + uint32_t regaddr_cp_dma_src_addr_hi; + uint32_t regaddr_cp_dma_dst_addr; + uint32_t regaddr_cp_dma_dst_addr_hi; + uint32_t regaddr_cp_dma_command; + uint32_t regaddr_cp_status; + uint32_t regaddr_rlc_gpu_clock_32; + uint32_t rlc_gpu_timer_refclk; + uint8_t active_cu_per_sh; + uint8_t active_rb_per_se; + uint16_t gcgoldenoffset; + uint16_t gc_num_gprs; + uint16_t gc_gsprim_buff_depth; + uint16_t gc_parameter_cache_depth; + uint16_t gc_wave_size; + uint16_t gc_max_waves_per_simd; + uint16_t gc_lds_size; + uint8_t gc_num_max_gs_thds; + uint8_t gc_gs_table_depth; + uint8_t gc_double_offchip_lds_buffer; + uint8_t gc_max_scratch_slots_per_cu; + uint32_t sram_rm_fuses_val; + uint32_t sram_custom_rm_fuses_val; +}; + /* *************************************************************************** Data Table smu_info structure diff --git a/drivers/gpu/drm/amd/include/soc15_ih_clientid.h b/drivers/gpu/drm/amd/include/soc15_ih_clientid.h index a12d4f27cfa4..12e196c15bbe 100644 --- a/drivers/gpu/drm/amd/include/soc15_ih_clientid.h +++ b/drivers/gpu/drm/amd/include/soc15_ih_clientid.h @@ -43,6 +43,7 @@ enum soc15_ih_clientid { SOC15_IH_CLIENTID_SE2SH = 0x0c, SOC15_IH_CLIENTID_SE3SH = 0x0d, SOC15_IH_CLIENTID_SYSHUB = 0x0e, + SOC15_IH_CLIENTID_UVD1 = 0x0e, SOC15_IH_CLIENTID_THM = 0x0f, SOC15_IH_CLIENTID_UVD = 0x10, SOC15_IH_CLIENTID_VCE0 = 0x11, diff --git a/drivers/gpu/drm/amd/include/vega20_ip_offset.h b/drivers/gpu/drm/amd/include/vega20_ip_offset.h new file mode 100644 index 000000000000..2a2a9cc8bedb --- /dev/null +++ b/drivers/gpu/drm/amd/include/vega20_ip_offset.h @@ -0,0 +1,1051 @@ +/* + * Copyright (C) 2018 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ +#ifndef _vega20_ip_offset_HEADER +#define _vega20_ip_offset_HEADER + +#define MAX_INSTANCE 6 +#define MAX_SEGMENT 6 + + +struct IP_BASE_INSTANCE +{ + unsigned int segment[MAX_SEGMENT]; +}; + +struct IP_BASE +{ + struct IP_BASE_INSTANCE instance[MAX_INSTANCE]; +}; + + +static const struct IP_BASE ATHUB_BASE ={ { { { 0x00000C20, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE CLK_BASE ={ { { { 0x00016C00, 0x00016E00, 0x00017000, 0x00017200, 0x0001B000, 0x0001B200 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE DCE_BASE ={ { { { 0x00000012, 0x000000C0, 0x000034C0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE DF_BASE ={ { { { 0x00007000, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE FUSE_BASE ={ { { { 0x00017400, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE GC_BASE ={ { { { 0x00002000, 0x0000A000, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE HDP_BASE ={ { { { 0x00000F20, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE MMHUB_BASE ={ { { { 0x0001A000, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE MP0_BASE ={ { { { 0x00016000, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE MP1_BASE ={ { { { 0x00016000, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE NBIO_BASE ={ { { { 0x00000000, 0x00000014, 0x00000D20, 0x00010400, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE OSSSYS_BASE ={ { { { 0x000010A0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE SDMA0_BASE ={ { { { 0x00001260, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE SDMA1_BASE ={ { { { 0x00001860, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE SMUIO_BASE ={ { { { 0x00016800, 0x00016A00, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE THM_BASE ={ { { { 0x00016600, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE UMC_BASE ={ { { { 0x00014000, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE UVD_BASE ={ { { { 0x00007800, 0x00007E00, 0, 0, 0, 0 } }, + { { 0, 0x00009000, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +/* Adjust VCE_BASE to make vce_4_1 use vce_4_0 offset header files*/ +static const struct IP_BASE VCE_BASE ={ { { { 0x00007E00/* 0x00008800 */, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE XDMA_BASE ={ { { { 0x00003400, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; +static const struct IP_BASE RSMU_BASE ={ { { { 0x00012000, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } }, + { { 0, 0, 0, 0, 0, 0 } } } }; + + +#define ATHUB_BASE__INST0_SEG0 0x00000C20 +#define ATHUB_BASE__INST0_SEG1 0 +#define ATHUB_BASE__INST0_SEG2 0 +#define ATHUB_BASE__INST0_SEG3 0 +#define ATHUB_BASE__INST0_SEG4 0 +#define ATHUB_BASE__INST0_SEG5 0 + +#define ATHUB_BASE__INST1_SEG0 0 +#define ATHUB_BASE__INST1_SEG1 0 +#define ATHUB_BASE__INST1_SEG2 0 +#define ATHUB_BASE__INST1_SEG3 0 +#define ATHUB_BASE__INST1_SEG4 0 +#define ATHUB_BASE__INST1_SEG5 0 + +#define ATHUB_BASE__INST2_SEG0 0 +#define ATHUB_BASE__INST2_SEG1 0 +#define ATHUB_BASE__INST2_SEG2 0 +#define ATHUB_BASE__INST2_SEG3 0 +#define ATHUB_BASE__INST2_SEG4 0 +#define ATHUB_BASE__INST2_SEG5 0 + +#define ATHUB_BASE__INST3_SEG0 0 +#define ATHUB_BASE__INST3_SEG1 0 +#define ATHUB_BASE__INST3_SEG2 0 +#define ATHUB_BASE__INST3_SEG3 0 +#define ATHUB_BASE__INST3_SEG4 0 +#define ATHUB_BASE__INST3_SEG5 0 + +#define ATHUB_BASE__INST4_SEG0 0 +#define ATHUB_BASE__INST4_SEG1 0 +#define ATHUB_BASE__INST4_SEG2 0 +#define ATHUB_BASE__INST4_SEG3 0 +#define ATHUB_BASE__INST4_SEG4 0 +#define ATHUB_BASE__INST4_SEG5 0 + +#define ATHUB_BASE__INST5_SEG0 0 +#define ATHUB_BASE__INST5_SEG1 0 +#define ATHUB_BASE__INST5_SEG2 0 +#define ATHUB_BASE__INST5_SEG3 0 +#define ATHUB_BASE__INST5_SEG4 0 +#define ATHUB_BASE__INST5_SEG5 0 + +#define CLK_BASE__INST0_SEG0 0x00016C00 +#define CLK_BASE__INST0_SEG1 0x00016E00 +#define CLK_BASE__INST0_SEG2 0x00017000 +#define CLK_BASE__INST0_SEG3 0x00017200 +#define CLK_BASE__INST0_SEG4 0x0001B000 +#define CLK_BASE__INST0_SEG5 0x0001B200 + +#define CLK_BASE__INST1_SEG0 0 +#define CLK_BASE__INST1_SEG1 0 +#define CLK_BASE__INST1_SEG2 0 +#define CLK_BASE__INST1_SEG3 0 +#define CLK_BASE__INST1_SEG4 0 +#define CLK_BASE__INST1_SEG5 0 + +#define CLK_BASE__INST2_SEG0 0 +#define CLK_BASE__INST2_SEG1 0 +#define CLK_BASE__INST2_SEG2 0 +#define CLK_BASE__INST2_SEG3 0 +#define CLK_BASE__INST2_SEG4 0 +#define CLK_BASE__INST2_SEG5 0 + +#define CLK_BASE__INST3_SEG0 0 +#define CLK_BASE__INST3_SEG1 0 +#define CLK_BASE__INST3_SEG2 0 +#define CLK_BASE__INST3_SEG3 0 +#define CLK_BASE__INST3_SEG4 0 +#define CLK_BASE__INST3_SEG5 0 + +#define CLK_BASE__INST4_SEG0 0 +#define CLK_BASE__INST4_SEG1 0 +#define CLK_BASE__INST4_SEG2 0 +#define CLK_BASE__INST4_SEG3 0 +#define CLK_BASE__INST4_SEG4 0 +#define CLK_BASE__INST4_SEG5 0 + +#define CLK_BASE__INST5_SEG0 0 +#define CLK_BASE__INST5_SEG1 0 +#define CLK_BASE__INST5_SEG2 0 +#define CLK_BASE__INST5_SEG3 0 +#define CLK_BASE__INST5_SEG4 0 +#define CLK_BASE__INST5_SEG5 0 + +#define DCE_BASE__INST0_SEG0 0x00000012 +#define DCE_BASE__INST0_SEG1 0x000000C0 +#define DCE_BASE__INST0_SEG2 0x000034C0 +#define DCE_BASE__INST0_SEG3 0 +#define DCE_BASE__INST0_SEG4 0 +#define DCE_BASE__INST0_SEG5 0 + +#define DCE_BASE__INST1_SEG0 0 +#define DCE_BASE__INST1_SEG1 0 +#define DCE_BASE__INST1_SEG2 0 +#define DCE_BASE__INST1_SEG3 0 +#define DCE_BASE__INST1_SEG4 0 +#define DCE_BASE__INST1_SEG5 0 + +#define DCE_BASE__INST2_SEG0 0 +#define DCE_BASE__INST2_SEG1 0 +#define DCE_BASE__INST2_SEG2 0 +#define DCE_BASE__INST2_SEG3 0 +#define DCE_BASE__INST2_SEG4 0 +#define DCE_BASE__INST2_SEG5 0 + +#define DCE_BASE__INST3_SEG0 0 +#define DCE_BASE__INST3_SEG1 0 +#define DCE_BASE__INST3_SEG2 0 +#define DCE_BASE__INST3_SEG3 0 +#define DCE_BASE__INST3_SEG4 0 +#define DCE_BASE__INST3_SEG5 0 + +#define DCE_BASE__INST4_SEG0 0 +#define DCE_BASE__INST4_SEG1 0 +#define DCE_BASE__INST4_SEG2 0 +#define DCE_BASE__INST4_SEG3 0 +#define DCE_BASE__INST4_SEG4 0 +#define DCE_BASE__INST4_SEG5 0 + +#define DCE_BASE__INST5_SEG0 0 +#define DCE_BASE__INST5_SEG1 0 +#define DCE_BASE__INST5_SEG2 0 +#define DCE_BASE__INST5_SEG3 0 +#define DCE_BASE__INST5_SEG4 0 +#define DCE_BASE__INST5_SEG5 0 + +#define DF_BASE__INST0_SEG0 0x00007000 +#define DF_BASE__INST0_SEG1 0 +#define DF_BASE__INST0_SEG2 0 +#define DF_BASE__INST0_SEG3 0 +#define DF_BASE__INST0_SEG4 0 +#define DF_BASE__INST0_SEG5 0 + +#define DF_BASE__INST1_SEG0 0 +#define DF_BASE__INST1_SEG1 0 +#define DF_BASE__INST1_SEG2 0 +#define DF_BASE__INST1_SEG3 0 +#define DF_BASE__INST1_SEG4 0 +#define DF_BASE__INST1_SEG5 0 + +#define DF_BASE__INST2_SEG0 0 +#define DF_BASE__INST2_SEG1 0 +#define DF_BASE__INST2_SEG2 0 +#define DF_BASE__INST2_SEG3 0 +#define DF_BASE__INST2_SEG4 0 +#define DF_BASE__INST2_SEG5 0 + +#define DF_BASE__INST3_SEG0 0 +#define DF_BASE__INST3_SEG1 0 +#define DF_BASE__INST3_SEG2 0 +#define DF_BASE__INST3_SEG3 0 +#define DF_BASE__INST3_SEG4 0 +#define DF_BASE__INST3_SEG5 0 + +#define DF_BASE__INST4_SEG0 0 +#define DF_BASE__INST4_SEG1 0 +#define DF_BASE__INST4_SEG2 0 +#define DF_BASE__INST4_SEG3 0 +#define DF_BASE__INST4_SEG4 0 +#define DF_BASE__INST4_SEG5 0 + +#define DF_BASE__INST5_SEG0 0 +#define DF_BASE__INST5_SEG1 0 +#define DF_BASE__INST5_SEG2 0 +#define DF_BASE__INST5_SEG3 0 +#define DF_BASE__INST5_SEG4 0 +#define DF_BASE__INST5_SEG5 0 + +#define FUSE_BASE__INST0_SEG0 0x00017400 +#define FUSE_BASE__INST0_SEG1 0 +#define FUSE_BASE__INST0_SEG2 0 +#define FUSE_BASE__INST0_SEG3 0 +#define FUSE_BASE__INST0_SEG4 0 +#define FUSE_BASE__INST0_SEG5 0 + +#define FUSE_BASE__INST1_SEG0 0 +#define FUSE_BASE__INST1_SEG1 0 +#define FUSE_BASE__INST1_SEG2 0 +#define FUSE_BASE__INST1_SEG3 0 +#define FUSE_BASE__INST1_SEG4 0 +#define FUSE_BASE__INST1_SEG5 0 + +#define FUSE_BASE__INST2_SEG0 0 +#define FUSE_BASE__INST2_SEG1 0 +#define FUSE_BASE__INST2_SEG2 0 +#define FUSE_BASE__INST2_SEG3 0 +#define FUSE_BASE__INST2_SEG4 0 +#define FUSE_BASE__INST2_SEG5 0 + +#define FUSE_BASE__INST3_SEG0 0 +#define FUSE_BASE__INST3_SEG1 0 +#define FUSE_BASE__INST3_SEG2 0 +#define FUSE_BASE__INST3_SEG3 0 +#define FUSE_BASE__INST3_SEG4 0 +#define FUSE_BASE__INST3_SEG5 0 + +#define FUSE_BASE__INST4_SEG0 0 +#define FUSE_BASE__INST4_SEG1 0 +#define FUSE_BASE__INST4_SEG2 0 +#define FUSE_BASE__INST4_SEG3 0 +#define FUSE_BASE__INST4_SEG4 0 +#define FUSE_BASE__INST4_SEG5 0 + +#define FUSE_BASE__INST5_SEG0 0 +#define FUSE_BASE__INST5_SEG1 0 +#define FUSE_BASE__INST5_SEG2 0 +#define FUSE_BASE__INST5_SEG3 0 +#define FUSE_BASE__INST5_SEG4 0 +#define FUSE_BASE__INST5_SEG5 0 + +#define GC_BASE__INST0_SEG0 0x00002000 +#define GC_BASE__INST0_SEG1 0x0000A000 +#define GC_BASE__INST0_SEG2 0 +#define GC_BASE__INST0_SEG3 0 +#define GC_BASE__INST0_SEG4 0 +#define GC_BASE__INST0_SEG5 0 + +#define GC_BASE__INST1_SEG0 0 +#define GC_BASE__INST1_SEG1 0 +#define GC_BASE__INST1_SEG2 0 +#define GC_BASE__INST1_SEG3 0 +#define GC_BASE__INST1_SEG4 0 +#define GC_BASE__INST1_SEG5 0 + +#define GC_BASE__INST2_SEG0 0 +#define GC_BASE__INST2_SEG1 0 +#define GC_BASE__INST2_SEG2 0 +#define GC_BASE__INST2_SEG3 0 +#define GC_BASE__INST2_SEG4 0 +#define GC_BASE__INST2_SEG5 0 + +#define GC_BASE__INST3_SEG0 0 +#define GC_BASE__INST3_SEG1 0 +#define GC_BASE__INST3_SEG2 0 +#define GC_BASE__INST3_SEG3 0 +#define GC_BASE__INST3_SEG4 0 +#define GC_BASE__INST3_SEG5 0 + +#define GC_BASE__INST4_SEG0 0 +#define GC_BASE__INST4_SEG1 0 +#define GC_BASE__INST4_SEG2 0 +#define GC_BASE__INST4_SEG3 0 +#define GC_BASE__INST4_SEG4 0 +#define GC_BASE__INST4_SEG5 0 + +#define GC_BASE__INST5_SEG0 0 +#define GC_BASE__INST5_SEG1 0 +#define GC_BASE__INST5_SEG2 0 +#define GC_BASE__INST5_SEG3 0 +#define GC_BASE__INST5_SEG4 0 +#define GC_BASE__INST5_SEG5 0 + +#define HDP_BASE__INST0_SEG0 0x00000F20 +#define HDP_BASE__INST0_SEG1 0 +#define HDP_BASE__INST0_SEG2 0 +#define HDP_BASE__INST0_SEG3 0 +#define HDP_BASE__INST0_SEG4 0 +#define HDP_BASE__INST0_SEG5 0 + +#define HDP_BASE__INST1_SEG0 0 +#define HDP_BASE__INST1_SEG1 0 +#define HDP_BASE__INST1_SEG2 0 +#define HDP_BASE__INST1_SEG3 0 +#define HDP_BASE__INST1_SEG4 0 +#define HDP_BASE__INST1_SEG5 0 + +#define HDP_BASE__INST2_SEG0 0 +#define HDP_BASE__INST2_SEG1 0 +#define HDP_BASE__INST2_SEG2 0 +#define HDP_BASE__INST2_SEG3 0 +#define HDP_BASE__INST2_SEG4 0 +#define HDP_BASE__INST2_SEG5 0 + +#define HDP_BASE__INST3_SEG0 0 +#define HDP_BASE__INST3_SEG1 0 +#define HDP_BASE__INST3_SEG2 0 +#define HDP_BASE__INST3_SEG3 0 +#define HDP_BASE__INST3_SEG4 0 +#define HDP_BASE__INST3_SEG5 0 + +#define HDP_BASE__INST4_SEG0 0 +#define HDP_BASE__INST4_SEG1 0 +#define HDP_BASE__INST4_SEG2 0 +#define HDP_BASE__INST4_SEG3 0 +#define HDP_BASE__INST4_SEG4 0 +#define HDP_BASE__INST4_SEG5 0 + +#define HDP_BASE__INST5_SEG0 0 +#define HDP_BASE__INST5_SEG1 0 +#define HDP_BASE__INST5_SEG2 0 +#define HDP_BASE__INST5_SEG3 0 +#define HDP_BASE__INST5_SEG4 0 +#define HDP_BASE__INST5_SEG5 0 + +#define MMHUB_BASE__INST0_SEG0 0x0001A000 +#define MMHUB_BASE__INST0_SEG1 0 +#define MMHUB_BASE__INST0_SEG2 0 +#define MMHUB_BASE__INST0_SEG3 0 +#define MMHUB_BASE__INST0_SEG4 0 +#define MMHUB_BASE__INST0_SEG5 0 + +#define MMHUB_BASE__INST1_SEG0 0 +#define MMHUB_BASE__INST1_SEG1 0 +#define MMHUB_BASE__INST1_SEG2 0 +#define MMHUB_BASE__INST1_SEG3 0 +#define MMHUB_BASE__INST1_SEG4 0 +#define MMHUB_BASE__INST1_SEG5 0 + +#define MMHUB_BASE__INST2_SEG0 0 +#define MMHUB_BASE__INST2_SEG1 0 +#define MMHUB_BASE__INST2_SEG2 0 +#define MMHUB_BASE__INST2_SEG3 0 +#define MMHUB_BASE__INST2_SEG4 0 +#define MMHUB_BASE__INST2_SEG5 0 + +#define MMHUB_BASE__INST3_SEG0 0 +#define MMHUB_BASE__INST3_SEG1 0 +#define MMHUB_BASE__INST3_SEG2 0 +#define MMHUB_BASE__INST3_SEG3 0 +#define MMHUB_BASE__INST3_SEG4 0 +#define MMHUB_BASE__INST3_SEG5 0 + +#define MMHUB_BASE__INST4_SEG0 0 +#define MMHUB_BASE__INST4_SEG1 0 +#define MMHUB_BASE__INST4_SEG2 0 +#define MMHUB_BASE__INST4_SEG3 0 +#define MMHUB_BASE__INST4_SEG4 0 +#define MMHUB_BASE__INST4_SEG5 0 + +#define MMHUB_BASE__INST5_SEG0 0 +#define MMHUB_BASE__INST5_SEG1 0 +#define MMHUB_BASE__INST5_SEG2 0 +#define MMHUB_BASE__INST5_SEG3 0 +#define MMHUB_BASE__INST5_SEG4 0 +#define MMHUB_BASE__INST5_SEG5 0 + +#define MP0_BASE__INST0_SEG0 0x00016000 +#define MP0_BASE__INST0_SEG1 0 +#define MP0_BASE__INST0_SEG2 0 +#define MP0_BASE__INST0_SEG3 0 +#define MP0_BASE__INST0_SEG4 0 +#define MP0_BASE__INST0_SEG5 0 + +#define MP0_BASE__INST1_SEG0 0 +#define MP0_BASE__INST1_SEG1 0 +#define MP0_BASE__INST1_SEG2 0 +#define MP0_BASE__INST1_SEG3 0 +#define MP0_BASE__INST1_SEG4 0 +#define MP0_BASE__INST1_SEG5 0 + +#define MP0_BASE__INST2_SEG0 0 +#define MP0_BASE__INST2_SEG1 0 +#define MP0_BASE__INST2_SEG2 0 +#define MP0_BASE__INST2_SEG3 0 +#define MP0_BASE__INST2_SEG4 0 +#define MP0_BASE__INST2_SEG5 0 + +#define MP0_BASE__INST3_SEG0 0 +#define MP0_BASE__INST3_SEG1 0 +#define MP0_BASE__INST3_SEG2 0 +#define MP0_BASE__INST3_SEG3 0 +#define MP0_BASE__INST3_SEG4 0 +#define MP0_BASE__INST3_SEG5 0 + +#define MP0_BASE__INST4_SEG0 0 +#define MP0_BASE__INST4_SEG1 0 +#define MP0_BASE__INST4_SEG2 0 +#define MP0_BASE__INST4_SEG3 0 +#define MP0_BASE__INST4_SEG4 0 +#define MP0_BASE__INST4_SEG5 0 + +#define MP0_BASE__INST5_SEG0 0 +#define MP0_BASE__INST5_SEG1 0 +#define MP0_BASE__INST5_SEG2 0 +#define MP0_BASE__INST5_SEG3 0 +#define MP0_BASE__INST5_SEG4 0 +#define MP0_BASE__INST5_SEG5 0 + +#define MP1_BASE__INST0_SEG0 0x00016000 +#define MP1_BASE__INST0_SEG1 0 +#define MP1_BASE__INST0_SEG2 0 +#define MP1_BASE__INST0_SEG3 0 +#define MP1_BASE__INST0_SEG4 0 +#define MP1_BASE__INST0_SEG5 0 + +#define MP1_BASE__INST1_SEG0 0 +#define MP1_BASE__INST1_SEG1 0 +#define MP1_BASE__INST1_SEG2 0 +#define MP1_BASE__INST1_SEG3 0 +#define MP1_BASE__INST1_SEG4 0 +#define MP1_BASE__INST1_SEG5 0 + +#define MP1_BASE__INST2_SEG0 0 +#define MP1_BASE__INST2_SEG1 0 +#define MP1_BASE__INST2_SEG2 0 +#define MP1_BASE__INST2_SEG3 0 +#define MP1_BASE__INST2_SEG4 0 +#define MP1_BASE__INST2_SEG5 0 + +#define MP1_BASE__INST3_SEG0 0 +#define MP1_BASE__INST3_SEG1 0 +#define MP1_BASE__INST3_SEG2 0 +#define MP1_BASE__INST3_SEG3 0 +#define MP1_BASE__INST3_SEG4 0 +#define MP1_BASE__INST3_SEG5 0 + +#define MP1_BASE__INST4_SEG0 0 +#define MP1_BASE__INST4_SEG1 0 +#define MP1_BASE__INST4_SEG2 0 +#define MP1_BASE__INST4_SEG3 0 +#define MP1_BASE__INST4_SEG4 0 +#define MP1_BASE__INST4_SEG5 0 + +#define MP1_BASE__INST5_SEG0 0 +#define MP1_BASE__INST5_SEG1 0 +#define MP1_BASE__INST5_SEG2 0 +#define MP1_BASE__INST5_SEG3 0 +#define MP1_BASE__INST5_SEG4 0 +#define MP1_BASE__INST5_SEG5 0 + +#define NBIO_BASE__INST0_SEG0 0x00000000 +#define NBIO_BASE__INST0_SEG1 0x00000014 +#define NBIO_BASE__INST0_SEG2 0x00000D20 +#define NBIO_BASE__INST0_SEG3 0x00010400 +#define NBIO_BASE__INST0_SEG4 0 +#define NBIO_BASE__INST0_SEG5 0 + +#define NBIO_BASE__INST1_SEG0 0 +#define NBIO_BASE__INST1_SEG1 0 +#define NBIO_BASE__INST1_SEG2 0 +#define NBIO_BASE__INST1_SEG3 0 +#define NBIO_BASE__INST1_SEG4 0 +#define NBIO_BASE__INST1_SEG5 0 + +#define NBIO_BASE__INST2_SEG0 0 +#define NBIO_BASE__INST2_SEG1 0 +#define NBIO_BASE__INST2_SEG2 0 +#define NBIO_BASE__INST2_SEG3 0 +#define NBIO_BASE__INST2_SEG4 0 +#define NBIO_BASE__INST2_SEG5 0 + +#define NBIO_BASE__INST3_SEG0 0 +#define NBIO_BASE__INST3_SEG1 0 +#define NBIO_BASE__INST3_SEG2 0 +#define NBIO_BASE__INST3_SEG3 0 +#define NBIO_BASE__INST3_SEG4 0 +#define NBIO_BASE__INST3_SEG5 0 + +#define NBIO_BASE__INST4_SEG0 0 +#define NBIO_BASE__INST4_SEG1 0 +#define NBIO_BASE__INST4_SEG2 0 +#define NBIO_BASE__INST4_SEG3 0 +#define NBIO_BASE__INST4_SEG4 0 +#define NBIO_BASE__INST4_SEG5 0 + +#define NBIO_BASE__INST5_SEG0 0 +#define NBIO_BASE__INST5_SEG1 0 +#define NBIO_BASE__INST5_SEG2 0 +#define NBIO_BASE__INST5_SEG3 0 +#define NBIO_BASE__INST5_SEG4 0 +#define NBIO_BASE__INST5_SEG5 0 + +#define OSSSYS_BASE__INST0_SEG0 0x000010A0 +#define OSSSYS_BASE__INST0_SEG1 0 +#define OSSSYS_BASE__INST0_SEG2 0 +#define OSSSYS_BASE__INST0_SEG3 0 +#define OSSSYS_BASE__INST0_SEG4 0 +#define OSSSYS_BASE__INST0_SEG5 0 + +#define OSSSYS_BASE__INST1_SEG0 0 +#define OSSSYS_BASE__INST1_SEG1 0 +#define OSSSYS_BASE__INST1_SEG2 0 +#define OSSSYS_BASE__INST1_SEG3 0 +#define OSSSYS_BASE__INST1_SEG4 0 +#define OSSSYS_BASE__INST1_SEG5 0 + +#define OSSSYS_BASE__INST2_SEG0 0 +#define OSSSYS_BASE__INST2_SEG1 0 +#define OSSSYS_BASE__INST2_SEG2 0 +#define OSSSYS_BASE__INST2_SEG3 0 +#define OSSSYS_BASE__INST2_SEG4 0 +#define OSSSYS_BASE__INST2_SEG5 0 + +#define OSSSYS_BASE__INST3_SEG0 0 +#define OSSSYS_BASE__INST3_SEG1 0 +#define OSSSYS_BASE__INST3_SEG2 0 +#define OSSSYS_BASE__INST3_SEG3 0 +#define OSSSYS_BASE__INST3_SEG4 0 +#define OSSSYS_BASE__INST3_SEG5 0 + +#define OSSSYS_BASE__INST4_SEG0 0 +#define OSSSYS_BASE__INST4_SEG1 0 +#define OSSSYS_BASE__INST4_SEG2 0 +#define OSSSYS_BASE__INST4_SEG3 0 +#define OSSSYS_BASE__INST4_SEG4 0 +#define OSSSYS_BASE__INST4_SEG5 0 + +#define OSSSYS_BASE__INST5_SEG0 0 +#define OSSSYS_BASE__INST5_SEG1 0 +#define OSSSYS_BASE__INST5_SEG2 0 +#define OSSSYS_BASE__INST5_SEG3 0 +#define OSSSYS_BASE__INST5_SEG4 0 +#define OSSSYS_BASE__INST5_SEG5 0 + +#define SDMA0_BASE__INST0_SEG0 0x00001260 +#define SDMA0_BASE__INST0_SEG1 0 +#define SDMA0_BASE__INST0_SEG2 0 +#define SDMA0_BASE__INST0_SEG3 0 +#define SDMA0_BASE__INST0_SEG4 0 +#define SDMA0_BASE__INST0_SEG5 0 + +#define SDMA0_BASE__INST1_SEG0 0 +#define SDMA0_BASE__INST1_SEG1 0 +#define SDMA0_BASE__INST1_SEG2 0 +#define SDMA0_BASE__INST1_SEG3 0 +#define SDMA0_BASE__INST1_SEG4 0 +#define SDMA0_BASE__INST1_SEG5 0 + +#define SDMA0_BASE__INST2_SEG0 0 +#define SDMA0_BASE__INST2_SEG1 0 +#define SDMA0_BASE__INST2_SEG2 0 +#define SDMA0_BASE__INST2_SEG3 0 +#define SDMA0_BASE__INST2_SEG4 0 +#define SDMA0_BASE__INST2_SEG5 0 + +#define SDMA0_BASE__INST3_SEG0 0 +#define SDMA0_BASE__INST3_SEG1 0 +#define SDMA0_BASE__INST3_SEG2 0 +#define SDMA0_BASE__INST3_SEG3 0 +#define SDMA0_BASE__INST3_SEG4 0 +#define SDMA0_BASE__INST3_SEG5 0 + +#define SDMA0_BASE__INST4_SEG0 0 +#define SDMA0_BASE__INST4_SEG1 0 +#define SDMA0_BASE__INST4_SEG2 0 +#define SDMA0_BASE__INST4_SEG3 0 +#define SDMA0_BASE__INST4_SEG4 0 +#define SDMA0_BASE__INST4_SEG5 0 + +#define SDMA0_BASE__INST5_SEG0 0 +#define SDMA0_BASE__INST5_SEG1 0 +#define SDMA0_BASE__INST5_SEG2 0 +#define SDMA0_BASE__INST5_SEG3 0 +#define SDMA0_BASE__INST5_SEG4 0 +#define SDMA0_BASE__INST5_SEG5 0 + +#define SDMA1_BASE__INST0_SEG0 0x00001860 +#define SDMA1_BASE__INST0_SEG1 0 +#define SDMA1_BASE__INST0_SEG2 0 +#define SDMA1_BASE__INST0_SEG3 0 +#define SDMA1_BASE__INST0_SEG4 0 +#define SDMA1_BASE__INST0_SEG5 0 + +#define SDMA1_BASE__INST1_SEG0 0 +#define SDMA1_BASE__INST1_SEG1 0 +#define SDMA1_BASE__INST1_SEG2 0 +#define SDMA1_BASE__INST1_SEG3 0 +#define SDMA1_BASE__INST1_SEG4 0 +#define SDMA1_BASE__INST1_SEG5 0 + +#define SDMA1_BASE__INST2_SEG0 0 +#define SDMA1_BASE__INST2_SEG1 0 +#define SDMA1_BASE__INST2_SEG2 0 +#define SDMA1_BASE__INST2_SEG3 0 +#define SDMA1_BASE__INST2_SEG4 0 +#define SDMA1_BASE__INST2_SEG5 0 + +#define SDMA1_BASE__INST3_SEG0 0 +#define SDMA1_BASE__INST3_SEG1 0 +#define SDMA1_BASE__INST3_SEG2 0 +#define SDMA1_BASE__INST3_SEG3 0 +#define SDMA1_BASE__INST3_SEG4 0 +#define SDMA1_BASE__INST3_SEG5 0 + +#define SDMA1_BASE__INST4_SEG0 0 +#define SDMA1_BASE__INST4_SEG1 0 +#define SDMA1_BASE__INST4_SEG2 0 +#define SDMA1_BASE__INST4_SEG3 0 +#define SDMA1_BASE__INST4_SEG4 0 +#define SDMA1_BASE__INST4_SEG5 0 + +#define SDMA1_BASE__INST5_SEG0 0 +#define SDMA1_BASE__INST5_SEG1 0 +#define SDMA1_BASE__INST5_SEG2 0 +#define SDMA1_BASE__INST5_SEG3 0 +#define SDMA1_BASE__INST5_SEG4 0 +#define SDMA1_BASE__INST5_SEG5 0 + +#define SMUIO_BASE__INST0_SEG0 0x00016800 +#define SMUIO_BASE__INST0_SEG1 0x00016A00 +#define SMUIO_BASE__INST0_SEG2 0 +#define SMUIO_BASE__INST0_SEG3 0 +#define SMUIO_BASE__INST0_SEG4 0 +#define SMUIO_BASE__INST0_SEG5 0 + +#define SMUIO_BASE__INST1_SEG0 0 +#define SMUIO_BASE__INST1_SEG1 0 +#define SMUIO_BASE__INST1_SEG2 0 +#define SMUIO_BASE__INST1_SEG3 0 +#define SMUIO_BASE__INST1_SEG4 0 +#define SMUIO_BASE__INST1_SEG5 0 + +#define SMUIO_BASE__INST2_SEG0 0 +#define SMUIO_BASE__INST2_SEG1 0 +#define SMUIO_BASE__INST2_SEG2 0 +#define SMUIO_BASE__INST2_SEG3 0 +#define SMUIO_BASE__INST2_SEG4 0 +#define SMUIO_BASE__INST2_SEG5 0 + +#define SMUIO_BASE__INST3_SEG0 0 +#define SMUIO_BASE__INST3_SEG1 0 +#define SMUIO_BASE__INST3_SEG2 0 +#define SMUIO_BASE__INST3_SEG3 0 +#define SMUIO_BASE__INST3_SEG4 0 +#define SMUIO_BASE__INST3_SEG5 0 + +#define SMUIO_BASE__INST4_SEG0 0 +#define SMUIO_BASE__INST4_SEG1 0 +#define SMUIO_BASE__INST4_SEG2 0 +#define SMUIO_BASE__INST4_SEG3 0 +#define SMUIO_BASE__INST4_SEG4 0 +#define SMUIO_BASE__INST4_SEG5 0 + +#define SMUIO_BASE__INST5_SEG0 0 +#define SMUIO_BASE__INST5_SEG1 0 +#define SMUIO_BASE__INST5_SEG2 0 +#define SMUIO_BASE__INST5_SEG3 0 +#define SMUIO_BASE__INST5_SEG4 0 +#define SMUIO_BASE__INST5_SEG5 0 + +#define THM_BASE__INST0_SEG0 0x00016600 +#define THM_BASE__INST0_SEG1 0 +#define THM_BASE__INST0_SEG2 0 +#define THM_BASE__INST0_SEG3 0 +#define THM_BASE__INST0_SEG4 0 +#define THM_BASE__INST0_SEG5 0 + +#define THM_BASE__INST1_SEG0 0 +#define THM_BASE__INST1_SEG1 0 +#define THM_BASE__INST1_SEG2 0 +#define THM_BASE__INST1_SEG3 0 +#define THM_BASE__INST1_SEG4 0 +#define THM_BASE__INST1_SEG5 0 + +#define THM_BASE__INST2_SEG0 0 +#define THM_BASE__INST2_SEG1 0 +#define THM_BASE__INST2_SEG2 0 +#define THM_BASE__INST2_SEG3 0 +#define THM_BASE__INST2_SEG4 0 +#define THM_BASE__INST2_SEG5 0 + +#define THM_BASE__INST3_SEG0 0 +#define THM_BASE__INST3_SEG1 0 +#define THM_BASE__INST3_SEG2 0 +#define THM_BASE__INST3_SEG3 0 +#define THM_BASE__INST3_SEG4 0 +#define THM_BASE__INST3_SEG5 0 + +#define THM_BASE__INST4_SEG0 0 +#define THM_BASE__INST4_SEG1 0 +#define THM_BASE__INST4_SEG2 0 +#define THM_BASE__INST4_SEG3 0 +#define THM_BASE__INST4_SEG4 0 +#define THM_BASE__INST4_SEG5 0 + +#define THM_BASE__INST5_SEG0 0 +#define THM_BASE__INST5_SEG1 0 +#define THM_BASE__INST5_SEG2 0 +#define THM_BASE__INST5_SEG3 0 +#define THM_BASE__INST5_SEG4 0 +#define THM_BASE__INST5_SEG5 0 + +#define UMC_BASE__INST0_SEG0 0x00014000 +#define UMC_BASE__INST0_SEG1 0 +#define UMC_BASE__INST0_SEG2 0 +#define UMC_BASE__INST0_SEG3 0 +#define UMC_BASE__INST0_SEG4 0 +#define UMC_BASE__INST0_SEG5 0 + +#define UMC_BASE__INST1_SEG0 0 +#define UMC_BASE__INST1_SEG1 0 +#define UMC_BASE__INST1_SEG2 0 +#define UMC_BASE__INST1_SEG3 0 +#define UMC_BASE__INST1_SEG4 0 +#define UMC_BASE__INST1_SEG5 0 + +#define UMC_BASE__INST2_SEG0 0 +#define UMC_BASE__INST2_SEG1 0 +#define UMC_BASE__INST2_SEG2 0 +#define UMC_BASE__INST2_SEG3 0 +#define UMC_BASE__INST2_SEG4 0 +#define UMC_BASE__INST2_SEG5 0 + +#define UMC_BASE__INST3_SEG0 0 +#define UMC_BASE__INST3_SEG1 0 +#define UMC_BASE__INST3_SEG2 0 +#define UMC_BASE__INST3_SEG3 0 +#define UMC_BASE__INST3_SEG4 0 +#define UMC_BASE__INST3_SEG5 0 + +#define UMC_BASE__INST4_SEG0 0 +#define UMC_BASE__INST4_SEG1 0 +#define UMC_BASE__INST4_SEG2 0 +#define UMC_BASE__INST4_SEG3 0 +#define UMC_BASE__INST4_SEG4 0 +#define UMC_BASE__INST4_SEG5 0 + +#define UMC_BASE__INST5_SEG0 0 +#define UMC_BASE__INST5_SEG1 0 +#define UMC_BASE__INST5_SEG2 0 +#define UMC_BASE__INST5_SEG3 0 +#define UMC_BASE__INST5_SEG4 0 +#define UMC_BASE__INST5_SEG5 0 + +#define UVD_BASE__INST0_SEG0 0x00007800 +#define UVD_BASE__INST0_SEG1 0x00007E00 +#define UVD_BASE__INST0_SEG2 0 +#define UVD_BASE__INST0_SEG3 0 +#define UVD_BASE__INST0_SEG4 0 +#define UVD_BASE__INST0_SEG5 0 + +#define UVD_BASE__INST1_SEG0 0 +#define UVD_BASE__INST1_SEG1 0x00009000 +#define UVD_BASE__INST1_SEG2 0 +#define UVD_BASE__INST1_SEG3 0 +#define UVD_BASE__INST1_SEG4 0 +#define UVD_BASE__INST1_SEG5 0 + +#define UVD_BASE__INST2_SEG0 0 +#define UVD_BASE__INST2_SEG1 0 +#define UVD_BASE__INST2_SEG2 0 +#define UVD_BASE__INST2_SEG3 0 +#define UVD_BASE__INST2_SEG4 0 +#define UVD_BASE__INST2_SEG5 0 + +#define UVD_BASE__INST3_SEG0 0 +#define UVD_BASE__INST3_SEG1 0 +#define UVD_BASE__INST3_SEG2 0 +#define UVD_BASE__INST3_SEG3 0 +#define UVD_BASE__INST3_SEG4 0 +#define UVD_BASE__INST3_SEG5 0 + +#define UVD_BASE__INST4_SEG0 0 +#define UVD_BASE__INST4_SEG1 0 +#define UVD_BASE__INST4_SEG2 0 +#define UVD_BASE__INST4_SEG3 0 +#define UVD_BASE__INST4_SEG4 0 +#define UVD_BASE__INST4_SEG5 0 + +#define UVD_BASE__INST5_SEG0 0 +#define UVD_BASE__INST5_SEG1 0 +#define UVD_BASE__INST5_SEG2 0 +#define UVD_BASE__INST5_SEG3 0 +#define UVD_BASE__INST5_SEG4 0 +#define UVD_BASE__INST5_SEG5 0 + +#define VCE_BASE__INST0_SEG0 0x00008800 +#define VCE_BASE__INST0_SEG1 0 +#define VCE_BASE__INST0_SEG2 0 +#define VCE_BASE__INST0_SEG3 0 +#define VCE_BASE__INST0_SEG4 0 +#define VCE_BASE__INST0_SEG5 0 + +#define VCE_BASE__INST1_SEG0 0 +#define VCE_BASE__INST1_SEG1 0 +#define VCE_BASE__INST1_SEG2 0 +#define VCE_BASE__INST1_SEG3 0 +#define VCE_BASE__INST1_SEG4 0 +#define VCE_BASE__INST1_SEG5 0 + +#define VCE_BASE__INST2_SEG0 0 +#define VCE_BASE__INST2_SEG1 0 +#define VCE_BASE__INST2_SEG2 0 +#define VCE_BASE__INST2_SEG3 0 +#define VCE_BASE__INST2_SEG4 0 +#define VCE_BASE__INST2_SEG5 0 + +#define VCE_BASE__INST3_SEG0 0 +#define VCE_BASE__INST3_SEG1 0 +#define VCE_BASE__INST3_SEG2 0 +#define VCE_BASE__INST3_SEG3 0 +#define VCE_BASE__INST3_SEG4 0 +#define VCE_BASE__INST3_SEG5 0 + +#define VCE_BASE__INST4_SEG0 0 +#define VCE_BASE__INST4_SEG1 0 +#define VCE_BASE__INST4_SEG2 0 +#define VCE_BASE__INST4_SEG3 0 +#define VCE_BASE__INST4_SEG4 0 +#define VCE_BASE__INST4_SEG5 0 + +#define VCE_BASE__INST5_SEG0 0 +#define VCE_BASE__INST5_SEG1 0 +#define VCE_BASE__INST5_SEG2 0 +#define VCE_BASE__INST5_SEG3 0 +#define VCE_BASE__INST5_SEG4 0 +#define VCE_BASE__INST5_SEG5 0 + +#define XDMA_BASE__INST0_SEG0 0x00003400 +#define XDMA_BASE__INST0_SEG1 0 +#define XDMA_BASE__INST0_SEG2 0 +#define XDMA_BASE__INST0_SEG3 0 +#define XDMA_BASE__INST0_SEG4 0 +#define XDMA_BASE__INST0_SEG5 0 + +#define XDMA_BASE__INST1_SEG0 0 +#define XDMA_BASE__INST1_SEG1 0 +#define XDMA_BASE__INST1_SEG2 0 +#define XDMA_BASE__INST1_SEG3 0 +#define XDMA_BASE__INST1_SEG4 0 +#define XDMA_BASE__INST1_SEG5 0 + +#define XDMA_BASE__INST2_SEG0 0 +#define XDMA_BASE__INST2_SEG1 0 +#define XDMA_BASE__INST2_SEG2 0 +#define XDMA_BASE__INST2_SEG3 0 +#define XDMA_BASE__INST2_SEG4 0 +#define XDMA_BASE__INST2_SEG5 0 + +#define XDMA_BASE__INST3_SEG0 0 +#define XDMA_BASE__INST3_SEG1 0 +#define XDMA_BASE__INST3_SEG2 0 +#define XDMA_BASE__INST3_SEG3 0 +#define XDMA_BASE__INST3_SEG4 0 +#define XDMA_BASE__INST3_SEG5 0 + +#define XDMA_BASE__INST4_SEG0 0 +#define XDMA_BASE__INST4_SEG1 0 +#define XDMA_BASE__INST4_SEG2 0 +#define XDMA_BASE__INST4_SEG3 0 +#define XDMA_BASE__INST4_SEG4 0 +#define XDMA_BASE__INST4_SEG5 0 + +#define XDMA_BASE__INST5_SEG0 0 +#define XDMA_BASE__INST5_SEG1 0 +#define XDMA_BASE__INST5_SEG2 0 +#define XDMA_BASE__INST5_SEG3 0 +#define XDMA_BASE__INST5_SEG4 0 +#define XDMA_BASE__INST5_SEG5 0 + +#define RSMU_BASE__INST0_SEG0 0x00012000 +#define RSMU_BASE__INST0_SEG1 0 +#define RSMU_BASE__INST0_SEG2 0 +#define RSMU_BASE__INST0_SEG3 0 +#define RSMU_BASE__INST0_SEG4 0 +#define RSMU_BASE__INST0_SEG5 0 + +#define RSMU_BASE__INST1_SEG0 0 +#define RSMU_BASE__INST1_SEG1 0 +#define RSMU_BASE__INST1_SEG2 0 +#define RSMU_BASE__INST1_SEG3 0 +#define RSMU_BASE__INST1_SEG4 0 +#define RSMU_BASE__INST1_SEG5 0 + +#define RSMU_BASE__INST2_SEG0 0 +#define RSMU_BASE__INST2_SEG1 0 +#define RSMU_BASE__INST2_SEG2 0 +#define RSMU_BASE__INST2_SEG3 0 +#define RSMU_BASE__INST2_SEG4 0 +#define RSMU_BASE__INST2_SEG5 0 + +#define RSMU_BASE__INST3_SEG0 0 +#define RSMU_BASE__INST3_SEG1 0 +#define RSMU_BASE__INST3_SEG2 0 +#define RSMU_BASE__INST3_SEG3 0 +#define RSMU_BASE__INST3_SEG4 0 +#define RSMU_BASE__INST3_SEG5 0 + +#define RSMU_BASE__INST4_SEG0 0 +#define RSMU_BASE__INST4_SEG1 0 +#define RSMU_BASE__INST4_SEG2 0 +#define RSMU_BASE__INST4_SEG3 0 +#define RSMU_BASE__INST4_SEG4 0 +#define RSMU_BASE__INST4_SEG5 0 + +#define RSMU_BASE__INST5_SEG0 0 +#define RSMU_BASE__INST5_SEG1 0 +#define RSMU_BASE__INST5_SEG2 0 +#define RSMU_BASE__INST5_SEG3 0 +#define RSMU_BASE__INST5_SEG4 0 +#define RSMU_BASE__INST5_SEG5 0 + +#endif + diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c b/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c index e411012b3dcb..a0bb921fac22 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c @@ -132,6 +132,15 @@ int phm_apply_state_adjust_rules(struct pp_hwmgr *hwmgr, return 0; } +int phm_apply_clock_adjust_rules(struct pp_hwmgr *hwmgr) +{ + PHM_FUNC_CHECK(hwmgr); + + if (hwmgr->hwmgr_func->apply_clocks_adjust_rules != NULL) + return hwmgr->hwmgr_func->apply_clocks_adjust_rules(hwmgr); + return 0; +} + int phm_powerdown_uvd(struct pp_hwmgr *hwmgr) { PHM_FUNC_CHECK(hwmgr); @@ -161,6 +170,16 @@ int phm_disable_clock_power_gatings(struct pp_hwmgr *hwmgr) return 0; } +int phm_pre_display_configuration_changed(struct pp_hwmgr *hwmgr) +{ + PHM_FUNC_CHECK(hwmgr); + + if (NULL != hwmgr->hwmgr_func->pre_display_config_changed) + hwmgr->hwmgr_func->pre_display_config_changed(hwmgr); + + return 0; + +} int phm_display_configuration_changed(struct pp_hwmgr *hwmgr) { diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c index 71b42331f185..e63bc47dc715 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c @@ -151,6 +151,7 @@ int hwmgr_early_init(struct pp_hwmgr *hwmgr) hwmgr->feature_mask &= ~PP_GFXOFF_MASK; switch (hwmgr->chip_id) { case CHIP_VEGA10: + case CHIP_VEGA20: hwmgr->smumgr_funcs = &vega10_smu_funcs; vega10_hwmgr_init(hwmgr); break; diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c b/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c index 308bff2b5d1d..0af13c154328 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/pp_psm.c @@ -265,6 +265,15 @@ int psm_adjust_power_state_dynamic(struct pp_hwmgr *hwmgr, bool skip, if (skip) return 0; + if (!hwmgr->ps) + /* + * for vega12/vega20 which does not support power state manager + * DAL clock limits should also be honoured + */ + phm_apply_clock_adjust_rules(hwmgr); + + phm_pre_display_configuration_changed(hwmgr); + phm_display_configuration_changed(hwmgr); if (hwmgr->ps) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomctrl.c b/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomctrl.c index cf99c5eaf080..7047e29755c3 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomctrl.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/ppatomctrl.c @@ -320,7 +320,7 @@ int atomctrl_get_memory_pll_dividers_ai(struct pp_hwmgr *hwmgr, pp_atomctrl_memory_clock_param_ai *mpll_param) { struct amdgpu_device *adev = hwmgr->adev; - COMPUTE_MEMORY_CLOCK_PARAM_PARAMETERS_V2_3 mpll_parameters = {0}; + COMPUTE_MEMORY_CLOCK_PARAM_PARAMETERS_V2_3 mpll_parameters = {{0}, 0, 0}; int result; mpll_parameters.ulClock.ulClock = cpu_to_le32(clock_value); @@ -1104,10 +1104,8 @@ int atomctrl_get_voltage_evv_on_sclk( GetIndexIntoMasterTable(COMMAND, GetVoltageInfo), (uint32_t *)&get_voltage_info_param_space); - if (0 != result) - return result; - - *voltage = le16_to_cpu(((GET_EVV_VOLTAGE_INFO_OUTPUT_PARAMETER_V1_2 *) + *voltage = result ? 0 : + le16_to_cpu(((GET_EVV_VOLTAGE_INFO_OUTPUT_PARAMETER_V1_2 *) (&get_voltage_info_param_space))->usVoltageLevel); return result; @@ -1312,8 +1310,7 @@ int atomctrl_read_efuse(struct pp_hwmgr *hwmgr, uint16_t start_index, result = amdgpu_atom_execute_table(adev->mode_info.atom_context, GetIndexIntoMasterTable(COMMAND, ReadEfuseValue), (uint32_t *)&efuse_param); - if (!result) - *efuse = le32_to_cpu(efuse_param.ulEfuseValue) & mask; + *efuse = result ? 0 : le32_to_cpu(efuse_param.ulEfuseValue) & mask; return result; } @@ -1354,11 +1351,8 @@ int atomctrl_get_voltage_evv_on_sclk_ai(struct pp_hwmgr *hwmgr, uint8_t voltage_ GetIndexIntoMasterTable(COMMAND, GetVoltageInfo), (uint32_t *)&get_voltage_info_param_space); - if (0 != result) - return result; - - *voltage = le32_to_cpu(((GET_EVV_VOLTAGE_INFO_OUTPUT_PARAMETER_V1_3 *) - (&get_voltage_info_param_space))->ulVoltageLevel); + *voltage = result ? 0 : + le32_to_cpu(((GET_EVV_VOLTAGE_INFO_OUTPUT_PARAMETER_V1_3 *)(&get_voltage_info_param_space))->ulVoltageLevel); return result; } @@ -1552,15 +1546,17 @@ void atomctrl_get_voltage_range(struct pp_hwmgr *hwmgr, uint32_t *max_vddc, case CHIP_FIJI: *max_vddc = le32_to_cpu(((ATOM_ASIC_PROFILING_INFO_V3_3 *)profile)->ulMaxVddc/4); *min_vddc = le32_to_cpu(((ATOM_ASIC_PROFILING_INFO_V3_3 *)profile)->ulMinVddc/4); - break; + return; case CHIP_POLARIS11: case CHIP_POLARIS10: case CHIP_POLARIS12: *max_vddc = le32_to_cpu(((ATOM_ASIC_PROFILING_INFO_V3_6 *)profile)->ulMaxVddc/100); *min_vddc = le32_to_cpu(((ATOM_ASIC_PROFILING_INFO_V3_6 *)profile)->ulMinVddc/100); - break; - default: return; + default: + break; } } + *max_vddc = 0; + *min_vddc = 0; } diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c index 2f69bfa478a7..85f84f4d8be5 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu10_hwmgr.c @@ -600,7 +600,10 @@ static int smu10_dpm_force_dpm_level(struct pp_hwmgr *hwmgr, data->gfx_min_freq_limit/100); smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetHardMinFclkByFreq, + hwmgr->display_config->num_display > 3 ? + SMU10_UMD_PSTATE_PEAK_FCLK : SMU10_UMD_PSTATE_MIN_FCLK); + smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetHardMinSocclkByFreq, SMU10_UMD_PSTATE_MIN_SOCCLK); @@ -1125,6 +1128,23 @@ static int smu10_set_mmhub_powergating_by_smu(struct pp_hwmgr *hwmgr) return smum_send_msg_to_smc(hwmgr, PPSMC_MSG_PowerGateMmHub); } +static void smu10_powergate_vcn(struct pp_hwmgr *hwmgr, bool bgate) +{ + if (bgate) { + amdgpu_device_ip_set_powergating_state(hwmgr->adev, + AMD_IP_BLOCK_TYPE_VCN, + AMD_PG_STATE_GATE); + smum_send_msg_to_smc_with_parameter(hwmgr, + PPSMC_MSG_PowerDownVcn, 0); + } else { + smum_send_msg_to_smc_with_parameter(hwmgr, + PPSMC_MSG_PowerUpVcn, 0); + amdgpu_device_ip_set_powergating_state(hwmgr->adev, + AMD_IP_BLOCK_TYPE_VCN, + AMD_PG_STATE_UNGATE); + } +} + static const struct pp_hwmgr_func smu10_hwmgr_funcs = { .backend_init = smu10_hwmgr_backend_init, .backend_fini = smu10_hwmgr_backend_fini, @@ -1133,7 +1153,7 @@ static const struct pp_hwmgr_func smu10_hwmgr_funcs = { .force_dpm_level = smu10_dpm_force_dpm_level, .get_power_state_size = smu10_get_power_state_size, .powerdown_uvd = NULL, - .powergate_uvd = NULL, + .powergate_uvd = smu10_powergate_vcn, .powergate_vce = NULL, .get_mclk = smu10_dpm_get_mclk, .get_sclk = smu10_dpm_get_sclk, diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c index 8eb3f5176646..45e9b8cb169d 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c @@ -860,7 +860,8 @@ static void smu7_setup_voltage_range_from_vbios(struct pp_hwmgr *hwmgr) struct phm_ppt_v1_clock_voltage_dependency_table *dep_sclk_table; struct phm_ppt_v1_information *table_info = (struct phm_ppt_v1_information *)(hwmgr->pptable); - uint32_t min_vddc, max_vddc; + uint32_t min_vddc = 0; + uint32_t max_vddc = 0; if (!table_info) return; @@ -1018,7 +1019,7 @@ static int smu7_enable_sclk_mclk_dpm(struct pp_hwmgr *hwmgr) struct smu7_hwmgr *data = (struct smu7_hwmgr *)(hwmgr->backend); /* enable SCLK dpm */ - if (!data->sclk_dpm_key_disabled) + if (!data->sclk_dpm_key_disabled) { if (hwmgr->chip_id == CHIP_VEGAM) smu7_disable_sclk_vce_handshake(hwmgr); @@ -1026,6 +1027,7 @@ static int smu7_enable_sclk_mclk_dpm(struct pp_hwmgr *hwmgr) (0 == smum_send_msg_to_smc(hwmgr, PPSMC_MSG_DPM_Enable)), "Failed to enable SCLK DPM during DPM Start Function!", return -EINVAL); + } /* enable MCLK dpm */ if (0 == data->mclk_dpm_key_disabled) { diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_powertune.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_powertune.c index 99b29ff45d91..c952845833d7 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_powertune.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_powertune.c @@ -936,45 +936,49 @@ int smu7_enable_didt_config(struct pp_hwmgr *hwmgr) if (hwmgr->chip_id == CHIP_POLARIS10) { result = smu7_program_pt_config_registers(hwmgr, GCCACConfig_Polaris10); - PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", return result); + PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", goto error); result = smu7_program_pt_config_registers(hwmgr, DIDTConfig_Polaris10); - PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", return result); + PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", goto error); } else if (hwmgr->chip_id == CHIP_POLARIS11) { result = smu7_program_pt_config_registers(hwmgr, GCCACConfig_Polaris11); - PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", return result); + PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", goto error); if (hwmgr->is_kicker) result = smu7_program_pt_config_registers(hwmgr, DIDTConfig_Polaris11_Kicker); else result = smu7_program_pt_config_registers(hwmgr, DIDTConfig_Polaris11); - PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", return result); + PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", goto error); } else if (hwmgr->chip_id == CHIP_POLARIS12) { result = smu7_program_pt_config_registers(hwmgr, GCCACConfig_Polaris11); - PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", return result); + PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", goto error); result = smu7_program_pt_config_registers(hwmgr, DIDTConfig_Polaris12); - PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", return result); + PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", goto error); } else if (hwmgr->chip_id == CHIP_VEGAM) { result = smu7_program_pt_config_registers(hwmgr, GCCACConfig_VegaM); - PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", return result); + PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", goto error); result = smu7_program_pt_config_registers(hwmgr, DIDTConfig_VegaM); - PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", return result); + PP_ASSERT_WITH_CODE((result == 0), "DIDT Config failed.", goto error); } } cgs_write_register(hwmgr->device, mmGRBM_GFX_INDEX, value2); result = smu7_enable_didt(hwmgr, true); - PP_ASSERT_WITH_CODE((result == 0), "EnableDiDt failed.", return result); + PP_ASSERT_WITH_CODE((result == 0), "EnableDiDt failed.", goto error); if (hwmgr->chip_id == CHIP_POLARIS11) { result = smum_send_msg_to_smc(hwmgr, (uint16_t)(PPSMC_MSG_EnableDpmDidt)); PP_ASSERT_WITH_CODE((0 == result), - "Failed to enable DPM DIDT.", return result); + "Failed to enable DPM DIDT.", goto error); } mutex_unlock(&adev->grbm_idx_mutex); adev->gfx.rlc.funcs->exit_safe_mode(adev); } return 0; +error: + mutex_unlock(&adev->grbm_idx_mutex); + adev->gfx.rlc.funcs->exit_safe_mode(adev); + return result; } int smu7_disable_didt_config(struct pp_hwmgr *hwmgr) @@ -992,17 +996,20 @@ int smu7_disable_didt_config(struct pp_hwmgr *hwmgr) result = smu7_enable_didt(hwmgr, false); PP_ASSERT_WITH_CODE((result == 0), "Post DIDT enable clock gating failed.", - return result); + goto error); if (hwmgr->chip_id == CHIP_POLARIS11) { result = smum_send_msg_to_smc(hwmgr, (uint16_t)(PPSMC_MSG_DisableDpmDidt)); PP_ASSERT_WITH_CODE((0 == result), - "Failed to disable DPM DIDT.", return result); + "Failed to disable DPM DIDT.", goto error); } adev->gfx.rlc.funcs->exit_safe_mode(adev); } return 0; +error: + adev->gfx.rlc.funcs->exit_safe_mode(adev); + return result; } int smu7_enable_smc_cac(struct pp_hwmgr *hwmgr) diff --git a/drivers/gpu/drm/amd/powerplay/inc/hardwaremanager.h b/drivers/gpu/drm/amd/powerplay/inc/hardwaremanager.h index 9bb87857a20f..a202247c9894 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/hardwaremanager.h +++ b/drivers/gpu/drm/amd/powerplay/inc/hardwaremanager.h @@ -410,7 +410,10 @@ extern int phm_apply_state_adjust_rules(struct pp_hwmgr *hwmgr, struct pp_power_state *adjusted_ps, const struct pp_power_state *current_ps); +extern int phm_apply_clock_adjust_rules(struct pp_hwmgr *hwmgr); + extern int phm_force_dpm_levels(struct pp_hwmgr *hwmgr, enum amd_dpm_forced_level level); +extern int phm_pre_display_configuration_changed(struct pp_hwmgr *hwmgr); extern int phm_display_configuration_changed(struct pp_hwmgr *hwmgr); extern int phm_notify_smc_display_config_after_ps_adjustment(struct pp_hwmgr *hwmgr); extern int phm_register_irq_handlers(struct pp_hwmgr *hwmgr); diff --git a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h index 3c321c7d9626..b99fb8ac822c 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h +++ b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h @@ -229,6 +229,8 @@ struct pp_hwmgr_func { struct pp_power_state *prequest_ps, const struct pp_power_state *pcurrent_ps); + int (*apply_clocks_adjust_rules)(struct pp_hwmgr *hwmgr); + int (*force_dpm_level)(struct pp_hwmgr *hw_mgr, enum amd_dpm_forced_level level); @@ -252,6 +254,7 @@ struct pp_hwmgr_func { const void *state); int (*enable_clock_power_gating)(struct pp_hwmgr *hwmgr); int (*notify_smc_display_config_after_ps_adjustment)(struct pp_hwmgr *hwmgr); + int (*pre_display_config_changed)(struct pp_hwmgr *hwmgr); int (*display_config_changed)(struct pp_hwmgr *hwmgr); int (*disable_clock_power_gating)(struct pp_hwmgr *hwmgr); int (*update_clock_gatings)(struct pp_hwmgr *hwmgr, diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smu7_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smu7_smumgr.c index 64d33b775906..d644a9bb9078 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/smu7_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/smu7_smumgr.c @@ -283,11 +283,9 @@ int smu7_read_smc_sram_dword(struct pp_hwmgr *hwmgr, uint32_t smc_addr, uint32_t result = smu7_set_smc_sram_address(hwmgr, smc_addr, limit); - if (result) - return result; + *value = result ? 0 : cgs_read_register(hwmgr->device, mmSMC_IND_DATA_11); - *value = cgs_read_register(hwmgr->device, mmSMC_IND_DATA_11); - return 0; + return result; } int smu7_write_smc_sram_dword(struct pp_hwmgr *hwmgr, uint32_t smc_addr, uint32_t value, uint32_t limit) diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smumgr.c index ee236dfbf1d6..c9837935f0f5 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/smumgr.c @@ -45,6 +45,7 @@ MODULE_FIRMWARE("amdgpu/vegam_smc.bin"); MODULE_FIRMWARE("amdgpu/vega10_smc.bin"); MODULE_FIRMWARE("amdgpu/vega10_acg_smc.bin"); MODULE_FIRMWARE("amdgpu/vega12_smc.bin"); +MODULE_FIRMWARE("amdgpu/vega20_smc.bin"); int smum_thermal_avfs_enable(struct pp_hwmgr *hwmgr) { diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c index c9a563399330..2de48959ac93 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c @@ -1366,10 +1366,12 @@ static int vegam_program_memory_timing_parameters(struct pp_hwmgr *hwmgr) { struct smu7_hwmgr *hw_data = (struct smu7_hwmgr *)(hwmgr->backend); struct vegam_smumgr *smu_data = (struct vegam_smumgr *)(hwmgr->smu_backend); - struct SMU75_Discrete_MCArbDramTimingTable arb_regs = {0}; + struct SMU75_Discrete_MCArbDramTimingTable arb_regs; uint32_t i, j; int result = 0; + memset(&arb_regs, 0, sizeof(SMU75_Discrete_MCArbDramTimingTable)); + for (i = 0; i < hw_data->dpm_table.sclk_table.count; i++) { for (j = 0; j < hw_data->dpm_table.mclk_table.count; j++) { result = vegam_populate_memory_timing_parameters(hwmgr, @@ -2377,6 +2379,5 @@ const struct pp_smumgr_func vegam_smu_funcs = { .update_sclk_threshold = vegam_update_sclk_threshold, .is_hw_avfs_present = vegam_is_hw_avfs_present, .thermal_avfs_enable = vegam_thermal_avfs_enable, - .is_dpm_running = vegam_is_dpm_running, .thermal_setup_fan_table = vegam_thermal_setup_fan_table, }; diff --git a/drivers/gpu/drm/scheduler/gpu_scheduler.c b/drivers/gpu/drm/scheduler/gpu_scheduler.c index a364fc0b38c3..df1578d6f42e 100644 --- a/drivers/gpu/drm/scheduler/gpu_scheduler.c +++ b/drivers/gpu/drm/scheduler/gpu_scheduler.c @@ -139,7 +139,6 @@ int drm_sched_entity_init(struct drm_gpu_scheduler *sched, entity->last_scheduled = NULL; spin_lock_init(&entity->rq_lock); - spin_lock_init(&entity->queue_lock); spsc_queue_init(&entity->job_queue); atomic_set(&entity->fence_seq, 0); @@ -413,6 +412,10 @@ drm_sched_entity_pop_job(struct drm_sched_entity *entity) * * @sched_job The pointer to job required to submit * + * Note: To guarantee that the order of insertion to queue matches + * the job's fence sequence number this function should be + * called with drm_sched_job_init under common lock. + * * Returns 0 for success, negative error code otherwise. */ void drm_sched_entity_push_job(struct drm_sched_job *sched_job, @@ -423,11 +426,8 @@ void drm_sched_entity_push_job(struct drm_sched_job *sched_job, trace_drm_sched_job(sched_job, entity); - spin_lock(&entity->queue_lock); first = spsc_queue_push(&entity->job_queue, &sched_job->queue_node); - spin_unlock(&entity->queue_lock); - /* first job wakes up scheduler */ if (first) { /* Add the entity to the run queue */ @@ -593,7 +593,12 @@ void drm_sched_job_recovery(struct drm_gpu_scheduler *sched) } EXPORT_SYMBOL(drm_sched_job_recovery); -/* init a sched_job with basic field */ +/** + * Init a sched_job with basic field + * + * Note: Refer to drm_sched_entity_push_job documentation + * for locking considerations. + */ int drm_sched_job_init(struct drm_sched_job *job, struct drm_gpu_scheduler *sched, struct drm_sched_entity *entity, diff --git a/drivers/gpu/drm/scheduler/sched_fence.c b/drivers/gpu/drm/scheduler/sched_fence.c index 69aab086b913..df4461648e3f 100644 --- a/drivers/gpu/drm/scheduler/sched_fence.c +++ b/drivers/gpu/drm/scheduler/sched_fence.c @@ -87,7 +87,7 @@ static bool drm_sched_fence_enable_signaling(struct dma_fence *f) } /** - * amd_sched_fence_free - free up the fence memory + * drm_sched_fence_free - free up the fence memory * * @rcu: RCU callback head * @@ -98,12 +98,11 @@ static void drm_sched_fence_free(struct rcu_head *rcu) struct dma_fence *f = container_of(rcu, struct dma_fence, rcu); struct drm_sched_fence *fence = to_drm_sched_fence(f); - dma_fence_put(fence->parent); kmem_cache_free(sched_fence_slab, fence); } /** - * amd_sched_fence_release_scheduled - callback that fence can be freed + * drm_sched_fence_release_scheduled - callback that fence can be freed * * @fence: fence * @@ -114,11 +113,12 @@ static void drm_sched_fence_release_scheduled(struct dma_fence *f) { struct drm_sched_fence *fence = to_drm_sched_fence(f); + dma_fence_put(fence->parent); call_rcu(&fence->finished.rcu, drm_sched_fence_free); } /** - * amd_sched_fence_release_finished - drop extra reference + * drm_sched_fence_release_finished - drop extra reference * * @f: fence * |