From d025e9e2b890db679f1246037bf65bd4be512627 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Thu, 29 Nov 2012 10:35:41 -0500 Subject: drm/radeon: do not move bo to different placement at each cs The bo creation placement is where the bo will be. Instead of trying to move bo at each command stream let this work to another worker thread that will use more advance heuristic. agd5f: remove leftover unused variable Signed-off-by: Jerome Glisse Reviewed-by: Alex Deucher --- drivers/gpu/drm/radeon/radeon.h | 1 + drivers/gpu/drm/radeon/radeon_object.c | 18 ++++++++---------- 2 files changed, 9 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index 5d68346b2c01..1b9120a875ef 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -318,6 +318,7 @@ struct radeon_bo { struct list_head list; /* Protected by tbo.reserved */ u32 placements[3]; + u32 busy_placements[3]; struct ttm_placement placement; struct ttm_buffer_object tbo; struct ttm_bo_kmap_obj kmap; diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index bfb332e616dc..93d3445477be 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -84,7 +84,6 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) rbo->placement.fpfn = 0; rbo->placement.lpfn = 0; rbo->placement.placement = rbo->placements; - rbo->placement.busy_placement = rbo->placements; if (domain & RADEON_GEM_DOMAIN_VRAM) rbo->placements[c++] = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_VRAM; @@ -105,6 +104,14 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) if (!c) rbo->placements[c++] = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM; rbo->placement.num_placement = c; + + c = 0; + rbo->placement.busy_placement = rbo->busy_placements; + if (rbo->rdev->flags & RADEON_IS_AGP) { + rbo->busy_placements[c++] = TTM_PL_FLAG_WC | TTM_PL_FLAG_TT; + } else { + rbo->busy_placements[c++] = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_TT; + } rbo->placement.num_busy_placement = c; } @@ -350,7 +357,6 @@ int radeon_bo_list_validate(struct list_head *head) { struct radeon_bo_list *lobj; struct radeon_bo *bo; - u32 domain; int r; r = ttm_eu_reserve_buffers(head); @@ -360,17 +366,9 @@ int radeon_bo_list_validate(struct list_head *head) list_for_each_entry(lobj, head, tv.head) { bo = lobj->bo; if (!bo->pin_count) { - domain = lobj->wdomain ? lobj->wdomain : lobj->rdomain; - - retry: - radeon_ttm_placement_from_domain(bo, domain); r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false); if (unlikely(r)) { - if (r != -ERESTARTSYS && domain == RADEON_GEM_DOMAIN_VRAM) { - domain |= RADEON_GEM_DOMAIN_GTT; - goto retry; - } return r; } } -- cgit v1.2.3 From bd25f0783dc3fb72e1e2779c2b99b2d34b67fa8a Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Tue, 11 Dec 2012 11:56:52 -0500 Subject: drm/radeon: fix amd afusion gpu setup aka sumo v2 Set the proper number of tile pipe that should be a multiple of pipe depending on the number of se engine. Fix: https://bugs.freedesktop.org/show_bug.cgi?id=56405 https://bugs.freedesktop.org/show_bug.cgi?id=56720 v2: Don't change sumo2 Signed-off-by: Jerome Glisse Cc: stable@vger.kernel.org Reviewed-by: Alex Deucher --- drivers/gpu/drm/radeon/evergreen.c | 8 ++++---- drivers/gpu/drm/radeon/evergreend.h | 2 ++ 2 files changed, 6 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/radeon/evergreen.c b/drivers/gpu/drm/radeon/evergreen.c index c66251e4a9b9..8dbc69a6e5bd 100644 --- a/drivers/gpu/drm/radeon/evergreen.c +++ b/drivers/gpu/drm/radeon/evergreen.c @@ -1821,7 +1821,7 @@ static void evergreen_gpu_init(struct radeon_device *rdev) case CHIP_SUMO: rdev->config.evergreen.num_ses = 1; rdev->config.evergreen.max_pipes = 4; - rdev->config.evergreen.max_tile_pipes = 2; + rdev->config.evergreen.max_tile_pipes = 4; if (rdev->pdev->device == 0x9648) rdev->config.evergreen.max_simds = 3; else if ((rdev->pdev->device == 0x9647) || @@ -1844,7 +1844,7 @@ static void evergreen_gpu_init(struct radeon_device *rdev) rdev->config.evergreen.sc_prim_fifo_size = 0x40; rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; - gb_addr_config = REDWOOD_GB_ADDR_CONFIG_GOLDEN; + gb_addr_config = SUMO_GB_ADDR_CONFIG_GOLDEN; break; case CHIP_SUMO2: rdev->config.evergreen.num_ses = 1; @@ -1866,7 +1866,7 @@ static void evergreen_gpu_init(struct radeon_device *rdev) rdev->config.evergreen.sc_prim_fifo_size = 0x40; rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; - gb_addr_config = REDWOOD_GB_ADDR_CONFIG_GOLDEN; + gb_addr_config = SUMO2_GB_ADDR_CONFIG_GOLDEN; break; case CHIP_BARTS: rdev->config.evergreen.num_ses = 2; @@ -1914,7 +1914,7 @@ static void evergreen_gpu_init(struct radeon_device *rdev) break; case CHIP_CAICOS: rdev->config.evergreen.num_ses = 1; - rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_pipes = 2; rdev->config.evergreen.max_tile_pipes = 2; rdev->config.evergreen.max_simds = 2; rdev->config.evergreen.max_backends = 1 * rdev->config.evergreen.num_ses; diff --git a/drivers/gpu/drm/radeon/evergreend.h b/drivers/gpu/drm/radeon/evergreend.h index 7b4a650e33b2..a5f3d0d244a2 100644 --- a/drivers/gpu/drm/radeon/evergreend.h +++ b/drivers/gpu/drm/radeon/evergreend.h @@ -45,6 +45,8 @@ #define TURKS_GB_ADDR_CONFIG_GOLDEN 0x02010002 #define CEDAR_GB_ADDR_CONFIG_GOLDEN 0x02010001 #define CAICOS_GB_ADDR_CONFIG_GOLDEN 0x02010001 +#define SUMO_GB_ADDR_CONFIG_GOLDEN 0x02010002 +#define SUMO2_GB_ADDR_CONFIG_GOLDEN 0x02010002 /* Registers */ -- cgit v1.2.3 From 6253e4c75d96006c06b9ac8f417eba873de2497b Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 12 Dec 2012 14:30:32 -0500 Subject: drm/radeon: improve mc_stop/mc_resume on r5xx-r7xx Along the same lines of what was done for evergreen+ in the last kernel. Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/r600_reg.h | 9 +++ drivers/gpu/drm/radeon/radeon_asic.h | 1 + drivers/gpu/drm/radeon/rv515.c | 122 ++++++++++++++++++++++++++++++----- 3 files changed, 116 insertions(+), 16 deletions(-) diff --git a/drivers/gpu/drm/radeon/r600_reg.h b/drivers/gpu/drm/radeon/r600_reg.h index 2b960cb5c18a..909219b1bf80 100644 --- a/drivers/gpu/drm/radeon/r600_reg.h +++ b/drivers/gpu/drm/radeon/r600_reg.h @@ -96,6 +96,15 @@ #define R600_CONFIG_F0_BASE 0x542C #define R600_CONFIG_APER_SIZE 0x5430 +#define R600_BIF_FB_EN 0x5490 +#define R600_FB_READ_EN (1 << 0) +#define R600_FB_WRITE_EN (1 << 1) + +#define R600_CITF_CNTL 0x200c +#define R600_BLACKOUT_MASK 0x00000003 + +#define R700_MC_CITF_CNTL 0x25c0 + #define R600_ROM_CNTL 0x1600 # define R600_SCK_OVERWRITE (1 << 1) # define R600_SCK_PRESCALE_CRYSTAL_CLK_SHIFT 28 diff --git a/drivers/gpu/drm/radeon/radeon_asic.h b/drivers/gpu/drm/radeon/radeon_asic.h index ae56673d2410..c338931190a5 100644 --- a/drivers/gpu/drm/radeon/radeon_asic.h +++ b/drivers/gpu/drm/radeon/radeon_asic.h @@ -263,6 +263,7 @@ extern int rs690_mc_wait_for_idle(struct radeon_device *rdev); struct rv515_mc_save { u32 vga_render_control; u32 vga_hdp_control; + bool crtc_enabled[2]; }; int rv515_init(struct radeon_device *rdev); diff --git a/drivers/gpu/drm/radeon/rv515.c b/drivers/gpu/drm/radeon/rv515.c index 785d09590b24..2bb6d0e84b3d 100644 --- a/drivers/gpu/drm/radeon/rv515.c +++ b/drivers/gpu/drm/radeon/rv515.c @@ -40,6 +40,12 @@ static int rv515_debugfs_ga_info_init(struct radeon_device *rdev); static void rv515_gpu_init(struct radeon_device *rdev); int rv515_mc_wait_for_idle(struct radeon_device *rdev); +static const u32 crtc_offsets[2] = +{ + 0, + AVIVO_D2CRTC_H_TOTAL - AVIVO_D1CRTC_H_TOTAL +}; + void rv515_debugfs(struct radeon_device *rdev) { if (r100_debugfs_rbbm_init(rdev)) { @@ -281,30 +287,114 @@ static int rv515_debugfs_ga_info_init(struct radeon_device *rdev) void rv515_mc_stop(struct radeon_device *rdev, struct rv515_mc_save *save) { + u32 crtc_enabled, tmp, frame_count, blackout; + int i, j; + save->vga_render_control = RREG32(R_000300_VGA_RENDER_CONTROL); save->vga_hdp_control = RREG32(R_000328_VGA_HDP_CONTROL); - /* Stop all video */ - WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 0); + /* disable VGA render */ WREG32(R_000300_VGA_RENDER_CONTROL, 0); - WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 1); - WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 1); - WREG32(R_006080_D1CRTC_CONTROL, 0); - WREG32(R_006880_D2CRTC_CONTROL, 0); - WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 0); - WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 0); - WREG32(R_000330_D1VGA_CONTROL, 0); - WREG32(R_000338_D2VGA_CONTROL, 0); + /* blank the display controllers */ + for (i = 0; i < rdev->num_crtc; i++) { + crtc_enabled = RREG32(AVIVO_D1CRTC_CONTROL + crtc_offsets[i]) & AVIVO_CRTC_EN; + if (crtc_enabled) { + save->crtc_enabled[i] = true; + tmp = RREG32(AVIVO_D1CRTC_CONTROL + crtc_offsets[i]); + if (!(tmp & AVIVO_CRTC_DISP_READ_REQUEST_DISABLE)) { + radeon_wait_for_vblank(rdev, i); + tmp |= AVIVO_CRTC_DISP_READ_REQUEST_DISABLE; + WREG32(AVIVO_D1CRTC_CONTROL + crtc_offsets[i], tmp); + } + /* wait for the next frame */ + frame_count = radeon_get_vblank_counter(rdev, i); + for (j = 0; j < rdev->usec_timeout; j++) { + if (radeon_get_vblank_counter(rdev, i) != frame_count) + break; + udelay(1); + } + } else { + save->crtc_enabled[i] = false; + } + } + + radeon_mc_wait_for_idle(rdev); + + if (rdev->family >= CHIP_R600) { + if (rdev->family >= CHIP_RV770) + blackout = RREG32(R700_MC_CITF_CNTL); + else + blackout = RREG32(R600_CITF_CNTL); + if ((blackout & R600_BLACKOUT_MASK) != R600_BLACKOUT_MASK) { + /* Block CPU access */ + WREG32(R600_BIF_FB_EN, 0); + /* blackout the MC */ + blackout |= R600_BLACKOUT_MASK; + if (rdev->family >= CHIP_RV770) + WREG32(R700_MC_CITF_CNTL, blackout); + else + WREG32(R600_CITF_CNTL, blackout); + } + } } void rv515_mc_resume(struct radeon_device *rdev, struct rv515_mc_save *save) { - WREG32(R_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS, rdev->mc.vram_start); - WREG32(R_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS, rdev->mc.vram_start); - WREG32(R_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS, rdev->mc.vram_start); - WREG32(R_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS, rdev->mc.vram_start); - WREG32(R_000310_VGA_MEMORY_BASE_ADDRESS, rdev->mc.vram_start); - /* Unlock host access */ + u32 tmp, frame_count; + int i, j; + + /* update crtc base addresses */ + for (i = 0; i < rdev->num_crtc; i++) { + if (rdev->family >= CHIP_RV770) { + if (i == 1) { + WREG32(R700_D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, + upper_32_bits(rdev->mc.vram_start)); + WREG32(R700_D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, + upper_32_bits(rdev->mc.vram_start)); + } else { + WREG32(R700_D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, + upper_32_bits(rdev->mc.vram_start)); + WREG32(R700_D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, + upper_32_bits(rdev->mc.vram_start)); + } + } + WREG32(R_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS + crtc_offsets[i], + (u32)rdev->mc.vram_start); + WREG32(R_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS + crtc_offsets[i], + (u32)rdev->mc.vram_start); + } + WREG32(R_000310_VGA_MEMORY_BASE_ADDRESS, (u32)rdev->mc.vram_start); + + if (rdev->family >= CHIP_R600) { + /* unblackout the MC */ + if (rdev->family >= CHIP_RV770) + tmp = RREG32(R700_MC_CITF_CNTL); + else + tmp = RREG32(R600_CITF_CNTL); + tmp &= ~R600_BLACKOUT_MASK; + if (rdev->family >= CHIP_RV770) + WREG32(R700_MC_CITF_CNTL, tmp); + else + WREG32(R600_CITF_CNTL, tmp); + /* allow CPU access */ + WREG32(R600_BIF_FB_EN, R600_FB_READ_EN | R600_FB_WRITE_EN); + } + + for (i = 0; i < rdev->num_crtc; i++) { + if (save->crtc_enabled[i]) { + tmp = RREG32(AVIVO_D1CRTC_CONTROL + crtc_offsets[i]); + tmp &= ~AVIVO_CRTC_DISP_READ_REQUEST_DISABLE; + WREG32(AVIVO_D1CRTC_CONTROL + crtc_offsets[i], tmp); + /* wait for the next frame */ + frame_count = radeon_get_vblank_counter(rdev, i); + for (j = 0; j < rdev->usec_timeout; j++) { + if (radeon_get_vblank_counter(rdev, i) != frame_count) + break; + udelay(1); + } + } + } + /* Unlock vga access */ WREG32(R_000328_VGA_HDP_CONTROL, save->vga_hdp_control); mdelay(1); WREG32(R_000300_VGA_RENDER_CONTROL, save->vga_render_control); -- cgit v1.2.3 From b997a8ba26377895506a26a3f2b8f1e7abc4ed22 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Mon, 3 Dec 2012 18:07:25 -0500 Subject: drm/radeon: add register headers for CP DMA on r6xx-SI Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/evergreend.h | 47 ++++++++++++++++++++++++++++++++++++ drivers/gpu/drm/radeon/r600d.h | 32 +++++++++++++++++++++++++ drivers/gpu/drm/radeon/sid.h | 48 +++++++++++++++++++++++++++++++++++++ 3 files changed, 127 insertions(+) diff --git a/drivers/gpu/drm/radeon/evergreend.h b/drivers/gpu/drm/radeon/evergreend.h index a5f3d0d244a2..cb9baaac9e85 100644 --- a/drivers/gpu/drm/radeon/evergreend.h +++ b/drivers/gpu/drm/radeon/evergreend.h @@ -1035,6 +1035,53 @@ #define PACKET3_WAIT_REG_MEM 0x3C #define PACKET3_MEM_WRITE 0x3D #define PACKET3_INDIRECT_BUFFER 0x32 +#define PACKET3_CP_DMA 0x41 +/* 1. header + * 2. SRC_ADDR_LO or DATA [31:0] + * 3. CP_SYNC [31] | SRC_SEL [30:29] | ENGINE [27] | DST_SEL [21:20] | + * SRC_ADDR_HI [7:0] + * 4. DST_ADDR_LO [31:0] + * 5. DST_ADDR_HI [7:0] + * 6. COMMAND [29:22] | BYTE_COUNT [20:0] + */ +# define PACKET3_CP_DMA_DST_SEL(x) ((x) << 20) + /* 0 - SRC_ADDR + * 1 - GDS + */ +# define PACKET3_CP_DMA_ENGINE(x) ((x) << 27) + /* 0 - ME + * 1 - PFP + */ +# define PACKET3_CP_DMA_SRC_SEL(x) ((x) << 29) + /* 0 - SRC_ADDR + * 1 - GDS + * 2 - DATA + */ +# define PACKET3_CP_DMA_CP_SYNC (1 << 31) +/* COMMAND */ +# define PACKET3_CP_DMA_DIS_WC (1 << 21) +# define PACKET3_CP_DMA_CMD_SRC_SWAP(x) ((x) << 23) + /* 0 - none + * 1 - 8 in 16 + * 2 - 8 in 32 + * 3 - 8 in 64 + */ +# define PACKET3_CP_DMA_CMD_DST_SWAP(x) ((x) << 24) + /* 0 - none + * 1 - 8 in 16 + * 2 - 8 in 32 + * 3 - 8 in 64 + */ +# define PACKET3_CP_DMA_CMD_SAS (1 << 26) + /* 0 - memory + * 1 - register + */ +# define PACKET3_CP_DMA_CMD_DAS (1 << 27) + /* 0 - memory + * 1 - register + */ +# define PACKET3_CP_DMA_CMD_SAIC (1 << 28) +# define PACKET3_CP_DMA_CMD_DAIC (1 << 29) #define PACKET3_SURFACE_SYNC 0x43 # define PACKET3_CB0_DEST_BASE_ENA (1 << 6) # define PACKET3_CB1_DEST_BASE_ENA (1 << 7) diff --git a/drivers/gpu/drm/radeon/r600d.h b/drivers/gpu/drm/radeon/r600d.h index a596c554a3a0..4a53402b1852 100644 --- a/drivers/gpu/drm/radeon/r600d.h +++ b/drivers/gpu/drm/radeon/r600d.h @@ -1186,6 +1186,38 @@ #define PACKET3_WAIT_REG_MEM 0x3C #define PACKET3_MEM_WRITE 0x3D #define PACKET3_INDIRECT_BUFFER 0x32 +#define PACKET3_CP_DMA 0x41 +/* 1. header + * 2. SRC_ADDR_LO [31:0] + * 3. CP_SYNC [31] | SRC_ADDR_HI [7:0] + * 4. DST_ADDR_LO [31:0] + * 5. DST_ADDR_HI [7:0] + * 6. COMMAND [29:22] | BYTE_COUNT [20:0] + */ +# define PACKET3_CP_DMA_CP_SYNC (1 << 31) +/* COMMAND */ +# define PACKET3_CP_DMA_CMD_SRC_SWAP(x) ((x) << 23) + /* 0 - none + * 1 - 8 in 16 + * 2 - 8 in 32 + * 3 - 8 in 64 + */ +# define PACKET3_CP_DMA_CMD_DST_SWAP(x) ((x) << 24) + /* 0 - none + * 1 - 8 in 16 + * 2 - 8 in 32 + * 3 - 8 in 64 + */ +# define PACKET3_CP_DMA_CMD_SAS (1 << 26) + /* 0 - memory + * 1 - register + */ +# define PACKET3_CP_DMA_CMD_DAS (1 << 27) + /* 0 - memory + * 1 - register + */ +# define PACKET3_CP_DMA_CMD_SAIC (1 << 28) +# define PACKET3_CP_DMA_CMD_DAIC (1 << 29) #define PACKET3_SURFACE_SYNC 0x43 # define PACKET3_CB0_DEST_BASE_ENA (1 << 6) # define PACKET3_TC_ACTION_ENA (1 << 23) diff --git a/drivers/gpu/drm/radeon/sid.h b/drivers/gpu/drm/radeon/sid.h index e153c254fbfb..62b46215d423 100644 --- a/drivers/gpu/drm/radeon/sid.h +++ b/drivers/gpu/drm/radeon/sid.h @@ -849,6 +849,54 @@ #define PACKET3_WAIT_REG_MEM 0x3C #define PACKET3_MEM_WRITE 0x3D #define PACKET3_COPY_DATA 0x40 +#define PACKET3_CP_DMA 0x41 +/* 1. header + * 2. SRC_ADDR_LO or DATA [31:0] + * 3. CP_SYNC [31] | SRC_SEL [30:29] | ENGINE [27] | DST_SEL [21:20] | + * SRC_ADDR_HI [7:0] + * 4. DST_ADDR_LO [31:0] + * 5. DST_ADDR_HI [7:0] + * 6. COMMAND [30:21] | BYTE_COUNT [20:0] + */ +# define PACKET3_CP_DMA_DST_SEL(x) ((x) << 20) + /* 0 - SRC_ADDR + * 1 - GDS + */ +# define PACKET3_CP_DMA_ENGINE(x) ((x) << 27) + /* 0 - ME + * 1 - PFP + */ +# define PACKET3_CP_DMA_SRC_SEL(x) ((x) << 29) + /* 0 - SRC_ADDR + * 1 - GDS + * 2 - DATA + */ +# define PACKET3_CP_DMA_CP_SYNC (1 << 31) +/* COMMAND */ +# define PACKET3_CP_DMA_DIS_WC (1 << 21) +# define PACKET3_CP_DMA_CMD_SRC_SWAP(x) ((x) << 23) + /* 0 - none + * 1 - 8 in 16 + * 2 - 8 in 32 + * 3 - 8 in 64 + */ +# define PACKET3_CP_DMA_CMD_DST_SWAP(x) ((x) << 24) + /* 0 - none + * 1 - 8 in 16 + * 2 - 8 in 32 + * 3 - 8 in 64 + */ +# define PACKET3_CP_DMA_CMD_SAS (1 << 26) + /* 0 - memory + * 1 - register + */ +# define PACKET3_CP_DMA_CMD_DAS (1 << 27) + /* 0 - memory + * 1 - register + */ +# define PACKET3_CP_DMA_CMD_SAIC (1 << 28) +# define PACKET3_CP_DMA_CMD_DAIC (1 << 29) +# define PACKET3_CP_DMA_CMD_RAW_WAIT (1 << 30) #define PACKET3_PFP_SYNC_ME 0x42 #define PACKET3_SURFACE_SYNC 0x43 # define PACKET3_DEST_BASE_0_ENA (1 << 0) -- cgit v1.2.3 From 6830f585724e1db5609dab0b059ea2e338bd9cc2 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Mon, 3 Dec 2012 19:03:23 -0500 Subject: drm/radeon: add support for CP DMA packet to r6xx/r7xx CS checker Currently only memory to memory transfers are allowed. Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/r600_cs.c | 72 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 72 insertions(+) diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c index 211c40252fe0..5d6e7f959e75 100644 --- a/drivers/gpu/drm/radeon/r600_cs.c +++ b/drivers/gpu/drm/radeon/r600_cs.c @@ -1949,6 +1949,78 @@ static int r600_packet3_check(struct radeon_cs_parser *p, ib[idx+2] = upper_32_bits(offset) & 0xff; } break; + case PACKET3_CP_DMA: + { + u32 command, size; + u64 offset, tmp; + if (pkt->count != 4) { + DRM_ERROR("bad CP DMA\n"); + return -EINVAL; + } + command = radeon_get_ib_value(p, idx+4); + size = command & 0x1fffff; + if (command & PACKET3_CP_DMA_CMD_SAS) { + /* src address space is register */ + DRM_ERROR("CP DMA SAS not supported\n"); + return -EINVAL; + } else { + if (command & PACKET3_CP_DMA_CMD_SAIC) { + DRM_ERROR("CP DMA SAIC only supported for registers\n"); + return -EINVAL; + } + /* src address space is memory */ + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad CP DMA SRC\n"); + return -EINVAL; + } + + tmp = radeon_get_ib_value(p, idx) + + ((u64)(radeon_get_ib_value(p, idx+1) & 0xff) << 32); + + offset = reloc->lobj.gpu_offset + tmp; + + if ((tmp + size) > radeon_bo_size(reloc->robj)) { + dev_warn(p->dev, "CP DMA src buffer too small (%llu %lu)\n", + tmp + size, radeon_bo_size(reloc->robj)); + return -EINVAL; + } + + ib[idx] = offset; + ib[idx+1] = (ib[idx+1] & 0xffffff00) | (upper_32_bits(offset) & 0xff); + } + if (command & PACKET3_CP_DMA_CMD_DAS) { + /* dst address space is register */ + DRM_ERROR("CP DMA DAS not supported\n"); + return -EINVAL; + } else { + /* dst address space is memory */ + if (command & PACKET3_CP_DMA_CMD_DAIC) { + DRM_ERROR("CP DMA DAIC only supported for registers\n"); + return -EINVAL; + } + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad CP DMA DST\n"); + return -EINVAL; + } + + tmp = radeon_get_ib_value(p, idx+2) + + ((u64)(radeon_get_ib_value(p, idx+3) & 0xff) << 32); + + offset = reloc->lobj.gpu_offset + tmp; + + if ((tmp + size) > radeon_bo_size(reloc->robj)) { + dev_warn(p->dev, "CP DMA dst buffer too small (%llu %lu)\n", + tmp + size, radeon_bo_size(reloc->robj)); + return -EINVAL; + } + + ib[idx+2] = offset; + ib[idx+3] = upper_32_bits(offset) & 0xff; + } + break; + } case PACKET3_SURFACE_SYNC: if (pkt->count != 3) { DRM_ERROR("bad SURFACE_SYNC\n"); -- cgit v1.2.3 From 8770b86b3e02c3e30f2ffc42753ff9d62bc428bf Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Mon, 3 Dec 2012 19:18:30 -0500 Subject: drm/radeon: add support for CP DMA packet to evergreen CS checker Currently only memory and GDS transfers are allowed. Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/evergreen_cs.c | 89 +++++++++++++++++++++++++++++++++++ 1 file changed, 89 insertions(+) diff --git a/drivers/gpu/drm/radeon/evergreen_cs.c b/drivers/gpu/drm/radeon/evergreen_cs.c index c042e497e450..5435879ad535 100644 --- a/drivers/gpu/drm/radeon/evergreen_cs.c +++ b/drivers/gpu/drm/radeon/evergreen_cs.c @@ -2232,6 +2232,95 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, ib[idx+2] = upper_32_bits(offset) & 0xff; } break; + case PACKET3_CP_DMA: + { + u32 command, size, info; + u64 offset, tmp; + if (pkt->count != 4) { + DRM_ERROR("bad CP DMA\n"); + return -EINVAL; + } + command = radeon_get_ib_value(p, idx+4); + size = command & 0x1fffff; + info = radeon_get_ib_value(p, idx+1); + if (command & PACKET3_CP_DMA_CMD_SAS) { + /* src address space is register */ + /* GDS is ok */ + if (((info & 0x60000000) >> 29) != 1) { + DRM_ERROR("CP DMA SAS not supported\n"); + return -EINVAL; + } + } else { + if (command & PACKET3_CP_DMA_CMD_SAIC) { + DRM_ERROR("CP DMA SAIC only supported for registers\n"); + return -EINVAL; + } + /* src address space is memory */ + if (((info & 0x60000000) >> 29) == 0) { + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad CP DMA SRC\n"); + return -EINVAL; + } + + tmp = radeon_get_ib_value(p, idx) + + ((u64)(radeon_get_ib_value(p, idx+1) & 0xff) << 32); + + offset = reloc->lobj.gpu_offset + tmp; + + if ((tmp + size) > radeon_bo_size(reloc->robj)) { + dev_warn(p->dev, "CP DMA src buffer too small (%llu %lu)\n", + tmp + size, radeon_bo_size(reloc->robj)); + return -EINVAL; + } + + ib[idx] = offset; + ib[idx+1] = (ib[idx+1] & 0xffffff00) | (upper_32_bits(offset) & 0xff); + } else if (((info & 0x60000000) >> 29) != 2) { + DRM_ERROR("bad CP DMA SRC_SEL\n"); + return -EINVAL; + } + } + if (command & PACKET3_CP_DMA_CMD_DAS) { + /* dst address space is register */ + /* GDS is ok */ + if (((info & 0x00300000) >> 20) != 1) { + DRM_ERROR("CP DMA DAS not supported\n"); + return -EINVAL; + } + } else { + /* dst address space is memory */ + if (command & PACKET3_CP_DMA_CMD_DAIC) { + DRM_ERROR("CP DMA DAIC only supported for registers\n"); + return -EINVAL; + } + if (((info & 0x00300000) >> 20) == 0) { + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad CP DMA DST\n"); + return -EINVAL; + } + + tmp = radeon_get_ib_value(p, idx+2) + + ((u64)(radeon_get_ib_value(p, idx+3) & 0xff) << 32); + + offset = reloc->lobj.gpu_offset + tmp; + + if ((tmp + size) > radeon_bo_size(reloc->robj)) { + dev_warn(p->dev, "CP DMA dst buffer too small (%llu %lu)\n", + tmp + size, radeon_bo_size(reloc->robj)); + return -EINVAL; + } + + ib[idx+2] = offset; + ib[idx+3] = upper_32_bits(offset) & 0xff; + } else { + DRM_ERROR("bad CP DMA DST_SEL\n"); + return -EINVAL; + } + } + break; + } case PACKET3_SURFACE_SYNC: if (pkt->count != 3) { DRM_ERROR("bad SURFACE_SYNC\n"); -- cgit v1.2.3 From 94e014ee98e98dedb080ed1cdf510a583ed0514b Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Mon, 3 Dec 2012 19:32:54 -0500 Subject: drm/radeon/cayman: add VM CS checker support for CP DMA Need to verify for copies involving registers. Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/evergreen_cs.c | 47 +++++++++++++++++++++++++++++++++++ 1 file changed, 47 insertions(+) diff --git a/drivers/gpu/drm/radeon/evergreen_cs.c b/drivers/gpu/drm/radeon/evergreen_cs.c index 5435879ad535..62c227104781 100644 --- a/drivers/gpu/drm/radeon/evergreen_cs.c +++ b/drivers/gpu/drm/radeon/evergreen_cs.c @@ -2932,6 +2932,7 @@ static int evergreen_vm_packet3_check(struct radeon_device *rdev, u32 idx = pkt->idx + 1; u32 idx_value = ib[idx]; u32 start_reg, end_reg, reg, i; + u32 command, info; switch (pkt->opcode) { case PACKET3_NOP: @@ -3006,6 +3007,52 @@ static int evergreen_vm_packet3_check(struct radeon_device *rdev, return -EINVAL; } break; + case PACKET3_CP_DMA: + command = ib[idx + 4]; + info = ib[idx + 1]; + if (command & PACKET3_CP_DMA_CMD_SAS) { + /* src address space is register */ + if (((info & 0x60000000) >> 29) == 0) { + start_reg = idx_value << 2; + if (command & PACKET3_CP_DMA_CMD_SAIC) { + reg = start_reg; + if (!evergreen_vm_reg_valid(reg)) { + DRM_ERROR("CP DMA Bad SRC register\n"); + return -EINVAL; + } + } else { + for (i = 0; i < (command & 0x1fffff); i++) { + reg = start_reg + (4 * i); + if (!evergreen_vm_reg_valid(reg)) { + DRM_ERROR("CP DMA Bad SRC register\n"); + return -EINVAL; + } + } + } + } + } + if (command & PACKET3_CP_DMA_CMD_DAS) { + /* dst address space is register */ + if (((info & 0x00300000) >> 20) == 0) { + start_reg = ib[idx + 2]; + if (command & PACKET3_CP_DMA_CMD_DAIC) { + reg = start_reg; + if (!evergreen_vm_reg_valid(reg)) { + DRM_ERROR("CP DMA Bad DST register\n"); + return -EINVAL; + } + } else { + for (i = 0; i < (command & 0x1fffff); i++) { + reg = start_reg + (4 * i); + if (!evergreen_vm_reg_valid(reg)) { + DRM_ERROR("CP DMA Bad DST register\n"); + return -EINVAL; + } + } + } + } + } + break; default: return -EINVAL; } -- cgit v1.2.3 From 5aa709be7e60a8296859766935f92bce51465341 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Mon, 3 Dec 2012 19:42:37 -0500 Subject: drm/radeon/si: add VM CS checker support for CP DMA Need to verify for copies involving registers. Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/si.c | 47 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 47 insertions(+) diff --git a/drivers/gpu/drm/radeon/si.c b/drivers/gpu/drm/radeon/si.c index f6e7815e1860..7e835d94df4a 100644 --- a/drivers/gpu/drm/radeon/si.c +++ b/drivers/gpu/drm/radeon/si.c @@ -2550,6 +2550,7 @@ static int si_vm_packet3_gfx_check(struct radeon_device *rdev, u32 idx = pkt->idx + 1; u32 idx_value = ib[idx]; u32 start_reg, end_reg, reg, i; + u32 command, info; switch (pkt->opcode) { case PACKET3_NOP: @@ -2649,6 +2650,52 @@ static int si_vm_packet3_gfx_check(struct radeon_device *rdev, return -EINVAL; } break; + case PACKET3_CP_DMA: + command = ib[idx + 4]; + info = ib[idx + 1]; + if (command & PACKET3_CP_DMA_CMD_SAS) { + /* src address space is register */ + if (((info & 0x60000000) >> 29) == 0) { + start_reg = idx_value << 2; + if (command & PACKET3_CP_DMA_CMD_SAIC) { + reg = start_reg; + if (!si_vm_reg_valid(reg)) { + DRM_ERROR("CP DMA Bad SRC register\n"); + return -EINVAL; + } + } else { + for (i = 0; i < (command & 0x1fffff); i++) { + reg = start_reg + (4 * i); + if (!si_vm_reg_valid(reg)) { + DRM_ERROR("CP DMA Bad SRC register\n"); + return -EINVAL; + } + } + } + } + } + if (command & PACKET3_CP_DMA_CMD_DAS) { + /* dst address space is register */ + if (((info & 0x00300000) >> 20) == 0) { + start_reg = ib[idx + 2]; + if (command & PACKET3_CP_DMA_CMD_DAIC) { + reg = start_reg; + if (!si_vm_reg_valid(reg)) { + DRM_ERROR("CP DMA Bad DST register\n"); + return -EINVAL; + } + } else { + for (i = 0; i < (command & 0x1fffff); i++) { + reg = start_reg + (4 * i); + if (!si_vm_reg_valid(reg)) { + DRM_ERROR("CP DMA Bad DST register\n"); + return -EINVAL; + } + } + } + } + } + break; default: DRM_ERROR("Invalid GFX packet3: 0x%x\n", pkt->opcode); return -EINVAL; -- cgit v1.2.3 From 86a1881d08f65a42c17071a59c0088dbe2870246 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Wed, 12 Dec 2012 16:43:15 -0500 Subject: drm/radeon: fix fence driver for dma ring when wb is disabled The dma ring can't write to register thus have to write to memory its fence value. This ensure that it doesn't try to use scratch register for dma ring fence driver. Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=58166 Signed-off-by: Jerome Glisse Reviewed-by: Alex Deucher --- drivers/gpu/drm/radeon/r600.c | 3 ++- drivers/gpu/drm/radeon/radeon_fence.c | 2 +- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c index a76eca18f134..2aaf147969bd 100644 --- a/drivers/gpu/drm/radeon/r600.c +++ b/drivers/gpu/drm/radeon/r600.c @@ -2533,11 +2533,12 @@ void r600_dma_fence_ring_emit(struct radeon_device *rdev, { struct radeon_ring *ring = &rdev->ring[fence->ring]; u64 addr = rdev->fence_drv[fence->ring].gpu_addr; + /* write the fence */ radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_FENCE, 0, 0, 0)); radeon_ring_write(ring, addr & 0xfffffffc); radeon_ring_write(ring, (upper_32_bits(addr) & 0xff)); - radeon_ring_write(ring, fence->seq); + radeon_ring_write(ring, lower_32_bits(fence->seq)); /* generate an interrupt */ radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_TRAP, 0, 0, 0)); } diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c index 22bd6c2c2740..410a975a8eec 100644 --- a/drivers/gpu/drm/radeon/radeon_fence.c +++ b/drivers/gpu/drm/radeon/radeon_fence.c @@ -772,7 +772,7 @@ int radeon_fence_driver_start_ring(struct radeon_device *rdev, int ring) int r; radeon_scratch_free(rdev, rdev->fence_drv[ring].scratch_reg); - if (rdev->wb.use_event) { + if (rdev->wb.use_event || !radeon_ring_supports_scratch_reg(rdev, &rdev->ring[ring])) { rdev->fence_drv[ring].scratch_reg = 0; index = R600_WB_EVENT_OFFSET + ring * 4; } else { -- cgit v1.2.3