diff options
149 files changed, 5671 insertions, 3120 deletions
diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c index 2f9ec14be3b1..a4b5af03dcc1 100644 --- a/arch/x86/kernel/early-quirks.c +++ b/arch/x86/kernel/early-quirks.c @@ -550,6 +550,7 @@ static const struct pci_device_id intel_early_ids[] __initconst = { INTEL_ICL_11_IDS(&gen11_early_ops), INTEL_EHL_IDS(&gen11_early_ops), INTEL_TGL_12_IDS(&gen11_early_ops), + INTEL_RKL_IDS(&gen11_early_ops), }; struct resource intel_graphics_stolen_res __ro_after_init = DEFINE_RES_MEM(0, 0); diff --git a/drivers/char/agp/intel-gtt.c b/drivers/char/agp/intel-gtt.c index 4b34a5195c65..5bfdf222d5f9 100644 --- a/drivers/char/agp/intel-gtt.c +++ b/drivers/char/agp/intel-gtt.c @@ -304,8 +304,10 @@ static int intel_gtt_setup_scratch_page(void) if (intel_private.needs_dmar) { dma_addr = pci_map_page(intel_private.pcidev, page, 0, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); - if (pci_dma_mapping_error(intel_private.pcidev, dma_addr)) + if (pci_dma_mapping_error(intel_private.pcidev, dma_addr)) { + __free_page(page); return -EINVAL; + } intel_private.scratch_page_dma = dma_addr; } else diff --git a/drivers/dma-buf/selftests.h b/drivers/dma-buf/selftests.h index 55918ef9adab..bc8cea67bf1e 100644 --- a/drivers/dma-buf/selftests.h +++ b/drivers/dma-buf/selftests.h @@ -5,7 +5,7 @@ * a module parameter. It must be unique and legal for a C identifier. * * The function should be of type int function(void). It may be conditionally - * compiled using #if IS_ENABLED(DRM_I915_SELFTEST). + * compiled using #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST). * * Tests are executed in order by igt/dmabuf_selftest */ diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile index b0da6ea6e3f1..41a27fd5dbc7 100644 --- a/drivers/gpu/drm/i915/Makefile +++ b/drivers/gpu/drm/i915/Makefile @@ -78,6 +78,8 @@ gt-y += \ gt/debugfs_engines.o \ gt/debugfs_gt.o \ gt/debugfs_gt_pm.o \ + gt/gen2_engine_cs.o \ + gt/gen6_engine_cs.o \ gt/gen6_ppgtt.o \ gt/gen7_renderclear.o \ gt/gen8_ppgtt.o \ diff --git a/drivers/gpu/drm/i915/display/icl_dsi.c b/drivers/gpu/drm/i915/display/icl_dsi.c index 4fec5bd64920..8c55f5bee9ab 100644 --- a/drivers/gpu/drm/i915/display/icl_dsi.c +++ b/drivers/gpu/drm/i915/display/icl_dsi.c @@ -1469,8 +1469,7 @@ static void gen11_dsi_get_config(struct intel_encoder *encoder, pipe_config->pipe_bpp = bdw_get_pipemisc_bpp(crtc); if (gen11_dsi_is_periodic_cmd_mode(intel_dsi)) - pipe_config->hw.adjusted_mode.private_flags |= - I915_MODE_FLAG_DSI_PERIODIC_CMD_MODE; + pipe_config->mode_flags |= I915_MODE_FLAG_DSI_PERIODIC_CMD_MODE; } static int gen11_dsi_dsc_compute_config(struct intel_encoder *encoder, @@ -1558,10 +1557,6 @@ static int gen11_dsi_compute_config(struct intel_encoder *encoder, pipe_config->port_clock = afe_clk(encoder, pipe_config) / 5; - /* We would not operate in periodic command mode */ - pipe_config->hw.adjusted_mode.private_flags &= - ~I915_MODE_FLAG_DSI_PERIODIC_CMD_MODE; - /* * In case of TE GATE cmd mode, we * receive TE from the slave if @@ -1569,14 +1564,14 @@ static int gen11_dsi_compute_config(struct intel_encoder *encoder, */ if (is_cmd_mode(intel_dsi)) { if (intel_dsi->ports == (BIT(PORT_B) | BIT(PORT_A))) - pipe_config->hw.adjusted_mode.private_flags |= + pipe_config->mode_flags |= I915_MODE_FLAG_DSI_USE_TE1 | I915_MODE_FLAG_DSI_USE_TE0; else if (intel_dsi->ports == BIT(PORT_B)) - pipe_config->hw.adjusted_mode.private_flags |= + pipe_config->mode_flags |= I915_MODE_FLAG_DSI_USE_TE1; else - pipe_config->hw.adjusted_mode.private_flags |= + pipe_config->mode_flags |= I915_MODE_FLAG_DSI_USE_TE0; } @@ -1954,6 +1949,7 @@ void icl_dsi_init(struct drm_i915_private *dev_priv) return; err: + drm_connector_cleanup(connector); drm_encoder_cleanup(&encoder->base); kfree(intel_dsi); kfree(intel_connector); diff --git a/drivers/gpu/drm/i915/display/intel_atomic.c b/drivers/gpu/drm/i915/display/intel_atomic.c index d043057d2fa0..630f49b7aa01 100644 --- a/drivers/gpu/drm/i915/display/intel_atomic.c +++ b/drivers/gpu/drm/i915/display/intel_atomic.c @@ -249,9 +249,11 @@ intel_crtc_duplicate_state(struct drm_crtc *crtc) crtc_state->update_wm_post = false; crtc_state->fifo_changed = false; crtc_state->preload_luts = false; + crtc_state->inherited = false; crtc_state->wm.need_postvbl_update = false; crtc_state->fb_bits = 0; crtc_state->update_planes = 0; + crtc_state->dsb = NULL; return &crtc_state->uapi; } @@ -292,6 +294,8 @@ intel_crtc_destroy_state(struct drm_crtc *crtc, { struct intel_crtc_state *crtc_state = to_intel_crtc_state(state); + drm_WARN_ON(crtc->dev, crtc_state->dsb); + __drm_atomic_helper_crtc_destroy_state(&crtc_state->uapi); intel_crtc_free_hw_state(crtc_state); kfree(crtc_state); diff --git a/drivers/gpu/drm/i915/display/intel_bios.c b/drivers/gpu/drm/i915/display/intel_bios.c index 839124647202..6593e2c38043 100644 --- a/drivers/gpu/drm/i915/display/intel_bios.c +++ b/drivers/gpu/drm/i915/display/intel_bios.c @@ -479,7 +479,7 @@ parse_sdvo_panel_data(struct drm_i915_private *dev_priv, struct drm_display_mode *panel_fixed_mode; int index; - index = i915_modparams.vbt_sdvo_panel_type; + index = dev_priv->params.vbt_sdvo_panel_type; if (index == -2) { drm_dbg_kms(&dev_priv->drm, "Ignore SDVO panel mode from BIOS VBT tables.\n"); @@ -829,9 +829,9 @@ parse_edp(struct drm_i915_private *dev_priv, const struct bdb_header *bdb) u8 vswing; /* Don't read from VBT if module parameter has valid value*/ - if (i915_modparams.edp_vswing) { + if (dev_priv->params.edp_vswing) { dev_priv->vbt.edp.low_vswing = - i915_modparams.edp_vswing == 1; + dev_priv->params.edp_vswing == 1; } else { vswing = (edp->edp_vswing_preemph >> (panel_type * 4)) & 0xF; dev_priv->vbt.edp.low_vswing = vswing == 0; @@ -1619,30 +1619,18 @@ static u8 map_ddc_pin(struct drm_i915_private *dev_priv, u8 vbt_pin) return 0; } -static enum port dvo_port_to_port(u8 dvo_port) +static enum port __dvo_port_to_port(int n_ports, int n_dvo, + const int port_mapping[][3], u8 dvo_port) { - /* - * Each DDI port can have more than one value on the "DVO Port" field, - * so look for all the possible values for each port. - */ - static const int dvo_ports[][3] = { - [PORT_A] = { DVO_PORT_HDMIA, DVO_PORT_DPA, -1}, - [PORT_B] = { DVO_PORT_HDMIB, DVO_PORT_DPB, -1}, - [PORT_C] = { DVO_PORT_HDMIC, DVO_PORT_DPC, -1}, - [PORT_D] = { DVO_PORT_HDMID, DVO_PORT_DPD, -1}, - [PORT_E] = { DVO_PORT_CRT, DVO_PORT_HDMIE, DVO_PORT_DPE}, - [PORT_F] = { DVO_PORT_HDMIF, DVO_PORT_DPF, -1}, - [PORT_G] = { DVO_PORT_HDMIG, DVO_PORT_DPG, -1}, - }; enum port port; int i; - for (port = PORT_A; port < ARRAY_SIZE(dvo_ports); port++) { - for (i = 0; i < ARRAY_SIZE(dvo_ports[port]); i++) { - if (dvo_ports[port][i] == -1) + for (port = PORT_A; port < n_ports; port++) { + for (i = 0; i < n_dvo; i++) { + if (port_mapping[port][i] == -1) break; - if (dvo_port == dvo_ports[port][i]) + if (dvo_port == port_mapping[port][i]) return port; } } @@ -1650,6 +1638,48 @@ static enum port dvo_port_to_port(u8 dvo_port) return PORT_NONE; } +static enum port dvo_port_to_port(struct drm_i915_private *dev_priv, + u8 dvo_port) +{ + /* + * Each DDI port can have more than one value on the "DVO Port" field, + * so look for all the possible values for each port. + */ + static const int port_mapping[][3] = { + [PORT_A] = { DVO_PORT_HDMIA, DVO_PORT_DPA, -1 }, + [PORT_B] = { DVO_PORT_HDMIB, DVO_PORT_DPB, -1 }, + [PORT_C] = { DVO_PORT_HDMIC, DVO_PORT_DPC, -1 }, + [PORT_D] = { DVO_PORT_HDMID, DVO_PORT_DPD, -1 }, + [PORT_E] = { DVO_PORT_HDMIE, DVO_PORT_DPE, DVO_PORT_CRT }, + [PORT_F] = { DVO_PORT_HDMIF, DVO_PORT_DPF, -1 }, + [PORT_G] = { DVO_PORT_HDMIG, DVO_PORT_DPG, -1 }, + }; + /* + * Bspec lists the ports as A, B, C, D - however internally in our + * driver we keep them as PORT_A, PORT_B, PORT_D and PORT_E so the + * registers in Display Engine match the right offsets. Apply the + * mapping here to translate from VBT to internal convention. + */ + static const int rkl_port_mapping[][3] = { + [PORT_A] = { DVO_PORT_HDMIA, DVO_PORT_DPA, -1 }, + [PORT_B] = { DVO_PORT_HDMIB, DVO_PORT_DPB, -1 }, + [PORT_C] = { -1 }, + [PORT_D] = { DVO_PORT_HDMIC, DVO_PORT_DPC, -1 }, + [PORT_E] = { DVO_PORT_HDMID, DVO_PORT_DPD, -1 }, + }; + + if (IS_ROCKETLAKE(dev_priv)) + return __dvo_port_to_port(ARRAY_SIZE(rkl_port_mapping), + ARRAY_SIZE(rkl_port_mapping[0]), + rkl_port_mapping, + dvo_port); + else + return __dvo_port_to_port(ARRAY_SIZE(port_mapping), + ARRAY_SIZE(port_mapping[0]), + port_mapping, + dvo_port); +} + static void parse_ddi_port(struct drm_i915_private *dev_priv, struct display_device_data *devdata, u8 bdb_version) @@ -1659,7 +1689,7 @@ static void parse_ddi_port(struct drm_i915_private *dev_priv, bool is_dvi, is_hdmi, is_dp, is_edp, is_crt; enum port port; - port = dvo_port_to_port(child->dvo_port); + port = dvo_port_to_port(dev_priv, child->dvo_port); if (port == PORT_NONE) return; @@ -2603,10 +2633,10 @@ enum aux_ch intel_bios_port_aux_ch(struct drm_i915_private *dev_priv, aux_ch = AUX_CH_B; break; case DP_AUX_C: - aux_ch = AUX_CH_C; + aux_ch = IS_ROCKETLAKE(dev_priv) ? AUX_CH_D : AUX_CH_C; break; case DP_AUX_D: - aux_ch = AUX_CH_D; + aux_ch = IS_ROCKETLAKE(dev_priv) ? AUX_CH_E : AUX_CH_D; break; case DP_AUX_E: aux_ch = AUX_CH_E; diff --git a/drivers/gpu/drm/i915/display/intel_bw.c b/drivers/gpu/drm/i915/display/intel_bw.c index fef04e2d954e..bd060404d249 100644 --- a/drivers/gpu/drm/i915/display/intel_bw.c +++ b/drivers/gpu/drm/i915/display/intel_bw.c @@ -5,12 +5,12 @@ #include <drm/drm_atomic_state_helper.h> +#include "intel_atomic.h" #include "intel_bw.h" +#include "intel_cdclk.h" #include "intel_display_types.h" -#include "intel_sideband.h" -#include "intel_atomic.h" #include "intel_pm.h" - +#include "intel_sideband.h" /* Parameters for Qclk Geyserville (QGV) */ struct intel_qgv_point { @@ -199,6 +199,12 @@ static const struct intel_sa_info tgl_sa_info = { .displayrtids = 256, }; +static const struct intel_sa_info rkl_sa_info = { + .deburst = 16, + .deprogbwlimit = 20, /* GB/s */ + .displayrtids = 128, +}; + static int icl_get_bw_info(struct drm_i915_private *dev_priv, const struct intel_sa_info *sa) { struct intel_qgv_info qi = {}; @@ -309,7 +315,9 @@ void intel_bw_init_hw(struct drm_i915_private *dev_priv) if (!HAS_DISPLAY(dev_priv)) return; - if (IS_GEN(dev_priv, 12)) + if (IS_ROCKETLAKE(dev_priv)) + icl_get_bw_info(dev_priv, &rkl_sa_info); + else if (IS_GEN(dev_priv, 12)) icl_get_bw_info(dev_priv, &tgl_sa_info); else if (IS_GEN(dev_priv, 11)) icl_get_bw_info(dev_priv, &icl_sa_info); @@ -420,6 +428,141 @@ intel_atomic_get_bw_state(struct intel_atomic_state *state) return to_intel_bw_state(bw_state); } +int skl_bw_calc_min_cdclk(struct intel_atomic_state *state) +{ + struct drm_i915_private *dev_priv = to_i915(state->base.dev); + struct intel_bw_state *new_bw_state = NULL; + struct intel_bw_state *old_bw_state = NULL; + const struct intel_crtc_state *crtc_state; + struct intel_crtc *crtc; + int max_bw = 0; + int slice_id; + enum pipe pipe; + int i; + + for_each_new_intel_crtc_in_state(state, crtc, crtc_state, i) { + enum plane_id plane_id; + struct intel_dbuf_bw *crtc_bw; + + new_bw_state = intel_atomic_get_bw_state(state); + if (IS_ERR(new_bw_state)) + return PTR_ERR(new_bw_state); + + old_bw_state = intel_atomic_get_old_bw_state(state); + + crtc_bw = &new_bw_state->dbuf_bw[crtc->pipe]; + + memset(&crtc_bw->used_bw, 0, sizeof(crtc_bw->used_bw)); + + if (!crtc_state->hw.active) + continue; + + for_each_plane_id_on_crtc(crtc, plane_id) { + const struct skl_ddb_entry *plane_alloc = + &crtc_state->wm.skl.plane_ddb_y[plane_id]; + const struct skl_ddb_entry *uv_plane_alloc = + &crtc_state->wm.skl.plane_ddb_uv[plane_id]; + unsigned int data_rate = crtc_state->data_rate[plane_id]; + unsigned int dbuf_mask = 0; + + dbuf_mask |= skl_ddb_dbuf_slice_mask(dev_priv, plane_alloc); + dbuf_mask |= skl_ddb_dbuf_slice_mask(dev_priv, uv_plane_alloc); + + /* + * FIXME: To calculate that more properly we probably + * need to to split per plane data_rate into data_rate_y + * and data_rate_uv for multiplanar formats in order not + * to get accounted those twice if they happen to reside + * on different slices. + * However for pre-icl this would work anyway because + * we have only single slice and for icl+ uv plane has + * non-zero data rate. + * So in worst case those calculation are a bit + * pessimistic, which shouldn't pose any significant + * problem anyway. + */ + for_each_dbuf_slice_in_mask(slice_id, dbuf_mask) + crtc_bw->used_bw[slice_id] += data_rate; + } + } + + if (!old_bw_state) + return 0; + + for_each_pipe(dev_priv, pipe) { + struct intel_dbuf_bw *crtc_bw; + + crtc_bw = &new_bw_state->dbuf_bw[pipe]; + + for_each_dbuf_slice(slice_id) { + /* + * Current experimental observations show that contrary + * to BSpec we get underruns once we exceed 64 * CDCLK + * for slices in total. + * As a temporary measure in order not to keep CDCLK + * bumped up all the time we calculate CDCLK according + * to this formula for overall bw consumed by slices. + */ + max_bw += crtc_bw->used_bw[slice_id]; + } + } + + new_bw_state->min_cdclk = max_bw / 64; + + if (new_bw_state->min_cdclk != old_bw_state->min_cdclk) { + int ret = intel_atomic_lock_global_state(&new_bw_state->base); + + if (ret) + return ret; + } + + return 0; +} + +int intel_bw_calc_min_cdclk(struct intel_atomic_state *state) +{ + struct drm_i915_private *dev_priv = to_i915(state->base.dev); + struct intel_bw_state *new_bw_state = NULL; + struct intel_bw_state *old_bw_state = NULL; + const struct intel_crtc_state *crtc_state; + struct intel_crtc *crtc; + int min_cdclk = 0; + enum pipe pipe; + int i; + + for_each_new_intel_crtc_in_state(state, crtc, crtc_state, i) { + new_bw_state = intel_atomic_get_bw_state(state); + if (IS_ERR(new_bw_state)) + return PTR_ERR(new_bw_state); + + old_bw_state = intel_atomic_get_old_bw_state(state); + } + + if (!old_bw_state) + return 0; + + for_each_pipe(dev_priv, pipe) { + struct intel_cdclk_state *cdclk_state; + + cdclk_state = intel_atomic_get_new_cdclk_state(state); + if (!cdclk_state) + return 0; + + min_cdclk = max(cdclk_state->min_cdclk[pipe], min_cdclk); + } + + new_bw_state->min_cdclk = min_cdclk; + + if (new_bw_state->min_cdclk != old_bw_state->min_cdclk) { + int ret = intel_atomic_lock_global_state(&new_bw_state->base); + + if (ret) + return ret; + } + + return 0; +} + int intel_bw_atomic_check(struct intel_atomic_state *state) { struct drm_i915_private *dev_priv = to_i915(state->base.dev); diff --git a/drivers/gpu/drm/i915/display/intel_bw.h b/drivers/gpu/drm/i915/display/intel_bw.h index bbcaaa73ec1b..46c6eecbd917 100644 --- a/drivers/gpu/drm/i915/display/intel_bw.h +++ b/drivers/gpu/drm/i915/display/intel_bw.h @@ -9,14 +9,20 @@ #include <drm/drm_atomic.h> #include "intel_display.h" +#include "intel_display_power.h" #include "intel_global_state.h" struct drm_i915_private; struct intel_atomic_state; struct intel_crtc_state; +struct intel_dbuf_bw { + int used_bw[I915_MAX_DBUF_SLICES]; +}; + struct intel_bw_state { struct intel_global_state base; + struct intel_dbuf_bw dbuf_bw[I915_MAX_PIPES]; /* * Contains a bit mask, used to determine, whether correspondent @@ -36,6 +42,8 @@ struct intel_bw_state { /* bitmask of active pipes */ u8 active_pipes; + + int min_cdclk; }; #define to_intel_bw_state(x) container_of((x), struct intel_bw_state, base) @@ -56,5 +64,7 @@ void intel_bw_crtc_update(struct intel_bw_state *bw_state, const struct intel_crtc_state *crtc_state); int icl_pcode_restrict_qgv_points(struct drm_i915_private *dev_priv, u32 points_mask); +int intel_bw_calc_min_cdclk(struct intel_atomic_state *state); +int skl_bw_calc_min_cdclk(struct intel_atomic_state *state); #endif /* __INTEL_BW_H__ */ diff --git a/drivers/gpu/drm/i915/display/intel_cdclk.c b/drivers/gpu/drm/i915/display/intel_cdclk.c index 979a0241fdcb..45f7f33d1144 100644 --- a/drivers/gpu/drm/i915/display/intel_cdclk.c +++ b/drivers/gpu/drm/i915/display/intel_cdclk.c @@ -21,7 +21,10 @@ * DEALINGS IN THE SOFTWARE. */ +#include <linux/time.h> + #include "intel_atomic.h" +#include "intel_bw.h" #include "intel_cdclk.h" #include "intel_display_types.h" #include "intel_sideband.h" @@ -2094,6 +2097,7 @@ static int intel_compute_min_cdclk(struct intel_cdclk_state *cdclk_state) { struct intel_atomic_state *state = cdclk_state->base.state; struct drm_i915_private *dev_priv = to_i915(state->base.dev); + struct intel_bw_state *bw_state = NULL; struct intel_crtc *crtc; struct intel_crtc_state *crtc_state; int min_cdclk, i; @@ -2106,6 +2110,10 @@ static int intel_compute_min_cdclk(struct intel_cdclk_state *cdclk_state) if (min_cdclk < 0) return min_cdclk; + bw_state = intel_atomic_get_bw_state(state); + if (IS_ERR(bw_state)) + return PTR_ERR(bw_state); + if (cdclk_state->min_cdclk[i] == min_cdclk) continue; @@ -2117,9 +2125,15 @@ static int intel_compute_min_cdclk(struct intel_cdclk_state *cdclk_state) } min_cdclk = cdclk_state->force_min_cdclk; - for_each_pipe(dev_priv, pipe) + for_each_pipe(dev_priv, pipe) { min_cdclk = max(cdclk_state->min_cdclk[pipe], min_cdclk); + if (!bw_state) + continue; + + min_cdclk = max(bw_state->min_cdclk, min_cdclk); + } + return min_cdclk; } @@ -2700,29 +2714,59 @@ static int vlv_hrawclk(struct drm_i915_private *dev_priv) CCK_DISPLAY_REF_CLOCK_CONTROL); } -static int g4x_hrawclk(struct drm_i915_private *dev_priv) +static int i9xx_hrawclk(struct drm_i915_private *dev_priv) { u32 clkcfg; - /* hrawclock is 1/4 the FSB frequency */ - clkcfg = intel_de_read(dev_priv, CLKCFG); - switch (clkcfg & CLKCFG_FSB_MASK) { - case CLKCFG_FSB_400: - return 100000; - case CLKCFG_FSB_533: - return 133333; - case CLKCFG_FSB_667: - return 166667; - case CLKCFG_FSB_800: - return 200000; - case CLKCFG_FSB_1067: - case CLKCFG_FSB_1067_ALT: - return 266667; - case CLKCFG_FSB_1333: - case CLKCFG_FSB_1333_ALT: - return 333333; - default: - return 133333; + /* + * hrawclock is 1/4 the FSB frequency + * + * Note that this only reads the state of the FSB + * straps, not the actual FSB frequency. Some BIOSen + * let you configure each independently. Ideally we'd + * read out the actual FSB frequency but sadly we + * don't know which registers have that information, + * and all the relevant docs have gone to bit heaven :( + */ + clkcfg = intel_de_read(dev_priv, CLKCFG) & CLKCFG_FSB_MASK; + + if (IS_MOBILE(dev_priv)) { + switch (clkcfg) { + case CLKCFG_FSB_400: + return 100000; + case CLKCFG_FSB_533: + return 133333; + case CLKCFG_FSB_667: + return 166667; + case CLKCFG_FSB_800: + return 200000; + case CLKCFG_FSB_1067: + return 266667; + case CLKCFG_FSB_1333: + return 333333; + default: + MISSING_CASE(clkcfg); + return 133333; + } + } else { + switch (clkcfg) { + case CLKCFG_FSB_400_ALT: + return 100000; + case CLKCFG_FSB_533: + return 133333; + case CLKCFG_FSB_667: + return 166667; + case CLKCFG_FSB_800: + return 200000; + case CLKCFG_FSB_1067_ALT: + return 266667; + case CLKCFG_FSB_1333_ALT: + return 333333; + case CLKCFG_FSB_1600_ALT: + return 400000; + default: + return 133333; + } } } @@ -2743,8 +2787,8 @@ u32 intel_read_rawclk(struct drm_i915_private *dev_priv) freq = pch_rawclk(dev_priv); else if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) freq = vlv_hrawclk(dev_priv); - else if (IS_G4X(dev_priv) || IS_PINEVIEW(dev_priv)) - freq = g4x_hrawclk(dev_priv); + else if (INTEL_GEN(dev_priv) >= 3) + freq = i9xx_hrawclk(dev_priv); else /* no rawclk on other platforms, or no need to know it */ return 0; @@ -2760,25 +2804,30 @@ void intel_init_cdclk_hooks(struct drm_i915_private *dev_priv) { if (INTEL_GEN(dev_priv) >= 12) { dev_priv->display.set_cdclk = bxt_set_cdclk; + dev_priv->display.bw_calc_min_cdclk = skl_bw_calc_min_cdclk; dev_priv->display.modeset_calc_cdclk = bxt_modeset_calc_cdclk; dev_priv->display.calc_voltage_level = tgl_calc_voltage_level; dev_priv->cdclk.table = icl_cdclk_table; } else if (IS_ELKHARTLAKE(dev_priv)) { dev_priv->display.set_cdclk = bxt_set_cdclk; + dev_priv->display.bw_calc_min_cdclk = skl_bw_calc_min_cdclk; dev_priv->display.modeset_calc_cdclk = bxt_modeset_calc_cdclk; dev_priv->display.calc_voltage_level = ehl_calc_voltage_level; dev_priv->cdclk.table = icl_cdclk_table; } else if (INTEL_GEN(dev_priv) >= 11) { dev_priv->display.set_cdclk = bxt_set_cdclk; + dev_priv->display.bw_calc_min_cdclk = skl_bw_calc_min_cdclk; dev_priv->display.modeset_calc_cdclk = bxt_modeset_calc_cdclk; dev_priv->display.calc_voltage_level = icl_calc_voltage_level; dev_priv->cdclk.table = icl_cdclk_table; } else if (IS_CANNONLAKE(dev_priv)) { + dev_priv->display.bw_calc_min_cdclk = skl_bw_calc_min_cdclk; dev_priv->display.set_cdclk = bxt_set_cdclk; dev_priv->display.modeset_calc_cdclk = bxt_modeset_calc_cdclk; dev_priv->display.calc_voltage_level = cnl_calc_voltage_level; dev_priv->cdclk.table = cnl_cdclk_table; } else if (IS_GEN9_LP(dev_priv)) { + dev_priv->display.bw_calc_min_cdclk = skl_bw_calc_min_cdclk; dev_priv->display.set_cdclk = bxt_set_cdclk; dev_priv->display.modeset_calc_cdclk = bxt_modeset_calc_cdclk; dev_priv->display.calc_voltage_level = bxt_calc_voltage_level; @@ -2787,18 +2836,23 @@ void intel_init_cdclk_hooks(struct drm_i915_private *dev_priv) else dev_priv->cdclk.table = bxt_cdclk_table; } else if (IS_GEN9_BC(dev_priv)) { + dev_priv->display.bw_calc_min_cdclk = skl_bw_calc_min_cdclk; dev_priv->display.set_cdclk = skl_set_cdclk; dev_priv->display.modeset_calc_cdclk = skl_modeset_calc_cdclk; } else if (IS_BROADWELL(dev_priv)) { + dev_priv->display.bw_calc_min_cdclk = intel_bw_calc_min_cdclk; dev_priv->display.set_cdclk = bdw_set_cdclk; dev_priv->display.modeset_calc_cdclk = bdw_modeset_calc_cdclk; } else if (IS_CHERRYVIEW(dev_priv)) { + dev_priv->display.bw_calc_min_cdclk = intel_bw_calc_min_cdclk; dev_priv->display.set_cdclk = chv_set_cdclk; dev_priv->display.modeset_calc_cdclk = vlv_modeset_calc_cdclk; } else if (IS_VALLEYVIEW(dev_priv)) { + dev_priv->display.bw_calc_min_cdclk = intel_bw_calc_min_cdclk; dev_priv->display.set_cdclk = vlv_set_cdclk; dev_priv->display.modeset_calc_cdclk = vlv_modeset_calc_cdclk; } else { + dev_priv->display.bw_calc_min_cdclk = intel_bw_calc_min_cdclk; dev_priv->display.modeset_calc_cdclk = fixed_modeset_calc_cdclk; } diff --git a/drivers/gpu/drm/i915/display/intel_color.c b/drivers/gpu/drm/i915/display/intel_color.c index 98ece9cd7cdd..945bb03bdd4d 100644 --- a/drivers/gpu/drm/i915/display/intel_color.c +++ b/drivers/gpu/drm/i915/display/intel_color.c @@ -714,16 +714,16 @@ static void bdw_load_lut_10(struct intel_crtc *crtc, intel_de_write(dev_priv, PREC_PAL_INDEX(pipe), 0); } -static void ivb_load_lut_ext_max(struct intel_crtc *crtc) +static void ivb_load_lut_ext_max(const struct intel_crtc_state *crtc_state) { + struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - struct intel_dsb *dsb = intel_dsb_get(crtc); enum pipe pipe = crtc->pipe; /* Program the max register to clamp values > 1.0. */ - intel_dsb_reg_write(dsb, PREC_PAL_EXT_GC_MAX(pipe, 0), 1 << 16); - intel_dsb_reg_write(dsb, PREC_PAL_EXT_GC_MAX(pipe, 1), 1 << 16); - intel_dsb_reg_write(dsb, PREC_PAL_EXT_GC_MAX(pipe, 2), 1 << 16); + intel_dsb_reg_write(crtc_state, PREC_PAL_EXT_GC_MAX(pipe, 0), 1 << 16); + intel_dsb_reg_write(crtc_state, PREC_PAL_EXT_GC_MAX(pipe, 1), 1 << 16); + intel_dsb_reg_write(crtc_state, PREC_PAL_EXT_GC_MAX(pipe, 2), 1 << 16); /* * Program the gc max 2 register to clamp values > 1.0. @@ -731,15 +731,13 @@ static void ivb_load_lut_ext_max(struct intel_crtc *crtc) * from 3.0 to 7.0 */ if (INTEL_GEN(dev_priv) >= 10 || IS_GEMINILAKE(dev_priv)) { - intel_dsb_reg_write(dsb, PREC_PAL_EXT2_GC_MAX(pipe, 0), + intel_dsb_reg_write(crtc_state, PREC_PAL_EXT2_GC_MAX(pipe, 0), 1 << 16); - intel_dsb_reg_write(dsb, PREC_PAL_EXT2_GC_MAX(pipe, 1), + intel_dsb_reg_write(crtc_state, PREC_PAL_EXT2_GC_MAX(pipe, 1), 1 << 16); - intel_dsb_reg_write(dsb, PREC_PAL_EXT2_GC_MAX(pipe, 2), + intel_dsb_reg_write(crtc_state, PREC_PAL_EXT2_GC_MAX(pipe, 2), 1 << 16); } - - intel_dsb_put(dsb); } static void ivb_load_luts(const struct intel_crtc_state *crtc_state) @@ -753,7 +751,7 @@ static void ivb_load_luts(const struct intel_crtc_state *crtc_state) } else if (crtc_state->gamma_mode == GAMMA_MODE_MODE_SPLIT) { ivb_load_lut_10(crtc, degamma_lut, PAL_PREC_SPLIT_MODE | PAL_PREC_INDEX_VALUE(0)); - ivb_load_lut_ext_max(crtc); + ivb_load_lut_ext_max(crtc_state); ivb_load_lut_10(crtc, gamma_lut, PAL_PREC_SPLIT_MODE | PAL_PREC_INDEX_VALUE(512)); } else { @@ -761,7 +759,7 @@ static void ivb_load_luts(const struct intel_crtc_state *crtc_state) ivb_load_lut_10(crtc, blob, PAL_PREC_INDEX_VALUE(0)); - ivb_load_lut_ext_max(crtc); + ivb_load_lut_ext_max(crtc_state); } } @@ -776,7 +774,7 @@ static void bdw_load_luts(const struct intel_crtc_state *crtc_state) } else if (crtc_state->gamma_mode == GAMMA_MODE_MODE_SPLIT) { bdw_load_lut_10(crtc, degamma_lut, PAL_PREC_SPLIT_MODE | PAL_PREC_INDEX_VALUE(0)); - ivb_load_lut_ext_max(crtc); + ivb_load_lut_ext_max(crtc_state); bdw_load_lut_10(crtc, gamma_lut, PAL_PREC_SPLIT_MODE | PAL_PREC_INDEX_VALUE(512)); } else { @@ -784,7 +782,7 @@ static void bdw_load_luts(const struct intel_crtc_state *crtc_state) bdw_load_lut_10(crtc, blob, PAL_PREC_INDEX_VALUE(0)); - ivb_load_lut_ext_max(crtc); + ivb_load_lut_ext_max(crtc_state); } } @@ -877,7 +875,7 @@ static void glk_load_luts(const struct intel_crtc_state *crtc_state) ilk_load_lut_8(crtc, gamma_lut); } else { bdw_load_lut_10(crtc, gamma_lut, PAL_PREC_INDEX_VALUE(0)); - ivb_load_lut_ext_max(crtc); + ivb_load_lut_ext_max(crtc_state); } } @@ -900,14 +898,12 @@ icl_load_gcmax(const struct intel_crtc_state *crtc_state, const struct drm_color_lut *color) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct intel_dsb *dsb = intel_dsb_get(crtc); enum pipe pipe = crtc->pipe; /* FIXME LUT entries are 16 bit only, so we can prog 0xFFFF max */ - intel_dsb_reg_write(dsb, PREC_PAL_GC_MAX(pipe, 0), color->red); - intel_dsb_reg_write(dsb, PREC_PAL_GC_MAX(pipe, 1), color->green); - intel_dsb_reg_write(dsb, PREC_PAL_GC_MAX(pipe, 2), color->blue); - intel_dsb_put(dsb); + intel_dsb_reg_write(crtc_state, PREC_PAL_GC_MAX(pipe, 0), color->red); + intel_dsb_reg_write(crtc_state, PREC_PAL_GC_MAX(pipe, 1), color->green); + intel_dsb_reg_write(crtc_state, PREC_PAL_GC_MAX(pipe, 2), color->blue); } static void @@ -916,7 +912,6 @@ icl_program_gamma_superfine_segment(const struct intel_crtc_state *crtc_state) struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); const struct drm_property_blob *blob = crtc_state->hw.gamma_lut; const struct drm_color_lut *lut = blob->data; - struct intel_dsb *dsb = intel_dsb_get(crtc); enum pipe pipe = crtc->pipe; int i; @@ -927,19 +922,17 @@ icl_program_gamma_superfine_segment(const struct intel_crtc_state *crtc_state) * 9 entries, corresponding to values 0, 1/(8 * 128 * 256), * 2/(8 * 128 * 256) ... 8/(8 * 128 * 256). */ - intel_dsb_reg_write(dsb, PREC_PAL_MULTI_SEG_INDEX(pipe), + intel_dsb_reg_write(crtc_state, PREC_PAL_MULTI_SEG_INDEX(pipe), PAL_PREC_AUTO_INCREMENT); for (i = 0; i < 9; i++) { const struct drm_color_lut *entry = &lut[i]; - intel_dsb_indexed_reg_write(dsb, PREC_PAL_MULTI_SEG_DATA(pipe), + intel_dsb_indexed_reg_write(crtc_state, PREC_PAL_MULTI_SEG_DATA(pipe), ilk_lut_12p4_ldw(entry)); - intel_dsb_indexed_reg_write(dsb, PREC_PAL_MULTI_SEG_DATA(pipe), + intel_dsb_indexed_reg_write(crtc_state, PREC_PAL_MULTI_SEG_DATA(pipe), ilk_lut_12p4_udw(entry)); } - - intel_dsb_put(dsb); } static void @@ -949,7 +942,6 @@ icl_program_gamma_multi_segment(const struct intel_crtc_state *crtc_state) const struct drm_property_blob *blob = crtc_state->hw.gamma_lut; const struct drm_color_lut *lut = blob->data; const struct drm_color_lut *entry; - struct intel_dsb *dsb = intel_dsb_get(crtc); enum pipe pipe = crtc->pipe; int i; @@ -963,12 +955,13 @@ icl_program_gamma_multi_segment(const struct intel_crtc_state *crtc_state) * PAL_PREC_INDEX[0] and PAL_PREC_INDEX[1] map to seg2[1], * seg2[0] being unused by the hardware. */ - intel_dsb_reg_write(dsb, PREC_PAL_INDEX(pipe), PAL_PREC_AUTO_INCREMENT); + intel_dsb_reg_write(crtc_state, PREC_PAL_INDEX(pipe), + PAL_PREC_AUTO_INCREMENT); for (i = 1; i < 257; i++) { entry = &lut[i * 8]; - intel_dsb_indexed_reg_write(dsb, PREC_PAL_DATA(pipe), + intel_dsb_indexed_reg_write(crtc_state, PREC_PAL_DATA(pipe), ilk_lut_12p4_ldw(entry)); - intel_dsb_indexed_reg_write(dsb, PREC_PAL_DATA(pipe), + intel_dsb_indexed_reg_write(crtc_state, PREC_PAL_DATA(pipe), ilk_lut_12p4_udw(entry)); } @@ -986,24 +979,22 @@ icl_program_gamma_multi_segment(const struct intel_crtc_state *crtc_state) */ for (i = 0; i < 256; i++) { entry = &lut[i * 8 * 128]; - intel_dsb_indexed_reg_write(dsb, PREC_PAL_DATA(pipe), + intel_dsb_indexed_reg_write(crtc_state, PREC_PAL_DATA(pipe), ilk_lut_12p4_ldw(entry)); - intel_dsb_indexed_reg_write(dsb, PREC_PAL_DATA(pipe), + intel_dsb_indexed_reg_write(crtc_state, PREC_PAL_DATA(pipe), ilk_lut_12p4_udw(entry)); } /* The last entry in the LUT is to be programmed in GCMAX */ entry = &lut[256 * 8 * 128]; icl_load_gcmax(crtc_state, entry); - ivb_load_lut_ext_max(crtc); - intel_dsb_put(dsb); + ivb_load_lut_ext_max(crtc_state); } static void icl_load_luts(const struct intel_crtc_state *crtc_state) { const struct drm_property_blob *gamma_lut = crtc_state->hw.gamma_lut; struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct intel_dsb *dsb = intel_dsb_get(crtc); if (crtc_state->hw.degamma_lut) glk_load_degamma_lut(crtc_state); @@ -1018,11 +1009,10 @@ static void icl_load_luts(const struct intel_crtc_state *crtc_state) break; default: bdw_load_lut_10(crtc, gamma_lut, PAL_PREC_INDEX_VALUE(0)); - ivb_load_lut_ext_max(crtc); + ivb_load_lut_ext_max(crtc_state); } - intel_dsb_commit(dsb); - intel_dsb_put(dsb); + intel_dsb_commit(crtc_state); } static u32 chv_cgm_degamma_ldw(const struct drm_color_lut *color) diff --git a/drivers/gpu/drm/i915/display/intel_combo_phy.c b/drivers/gpu/drm/i915/display/intel_combo_phy.c index 9ff05ec12115..77b04bb3ec62 100644 --- a/drivers/gpu/drm/i915/display/intel_combo_phy.c +++ b/drivers/gpu/drm/i915/display/intel_combo_phy.c @@ -181,11 +181,25 @@ static void cnl_combo_phys_uninit(struct drm_i915_private *dev_priv) intel_de_write(dev_priv, CHICKEN_MISC_2, val); } +static bool has_phy_misc(struct drm_i915_private *i915, enum phy phy) +{ + /* + * Some platforms only expect PHY_MISC to be programmed for PHY-A and + * PHY-B and may not even have instances of the register for the + * other combo PHY's. + */ + if (IS_ELKHARTLAKE(i915) || + IS_ROCKETLAKE(i915)) + return phy < PHY_C; + + return true; +} + static bool icl_combo_phy_enabled(struct drm_i915_private *dev_priv, enum phy phy) { /* The PHY C added by EHL has no PHY_MISC register */ - if (IS_ELKHARTLAKE(dev_priv) && phy == PHY_C) + if (!has_phy_misc(dev_priv, phy)) return intel_de_read(dev_priv, ICL_PORT_COMP_DW0(phy)) & COMP_INIT; else return !(intel_de_read(dev_priv, ICL_PHY_MISC(phy)) & @@ -220,6 +234,27 @@ static bool ehl_vbt_ddi_d_present(struct drm_i915_private *i915) return false; } +static bool phy_is_master(struct drm_i915_private *dev_priv, enum phy phy) +{ + /* + * Certain PHYs are connected to compensation resistors and act + * as masters to other PHYs. + * + * ICL,TGL: + * A(master) -> B(slave), C(slave) + * RKL: + * A(master) -> B(slave) + * C(master) -> D(slave) + * + * We must set the IREFGEN bit for any PHY acting as a master + * to another PHY. + */ + if (IS_ROCKETLAKE(dev_priv) && phy == PHY_C) + return true; + + return phy == PHY_A; +} + static bool icl_combo_phy_verify_state(struct drm_i915_private *dev_priv, enum phy phy) { @@ -231,7 +266,7 @@ static bool icl_combo_phy_verify_state(struct drm_i915_private *dev_priv, ret = cnl_verify_procmon_ref_values(dev_priv, phy); - if (phy == PHY_A) { + if (phy_is_master(dev_priv, phy)) { ret &= check_phy_reg(dev_priv, phy, ICL_PORT_COMP_DW8(phy), IREFGEN, IREFGEN); @@ -317,12 +352,7 @@ static void icl_combo_phys_init(struct drm_i915_private *dev_priv) continue; } - /* - * Although EHL adds a combo PHY C, there's no PHY_MISC - * register for it and no need to program the - * DE_IO_COMP_PWR_DOWN setting on PHY C. - */ - if (IS_ELKHARTLAKE(dev_priv) && phy == PHY_C) + if (!has_phy_misc(dev_priv, phy)) goto skip_phy_misc; /* @@ -347,7 +377,7 @@ static void icl_combo_phys_init(struct drm_i915_private *dev_priv) skip_phy_misc: cnl_set_procmon_ref_values(dev_priv, phy); - if (phy == PHY_A) { + if (phy_is_master(dev_priv, phy)) { val = intel_de_read(dev_priv, ICL_PORT_COMP_DW8(phy)); val |= IREFGEN; intel_de_write(dev_priv, ICL_PORT_COMP_DW8(phy), val); @@ -376,12 +406,7 @@ static void icl_combo_phys_uninit(struct drm_i915_private *dev_priv) "Combo PHY %c HW state changed unexpectedly\n", phy_name(phy)); - /* - * Although EHL adds a combo PHY C, there's no PHY_MISC - * register for it and no need to program the - * DE_IO_COMP_PWR_DOWN setting on PHY C. - */ - if (IS_ELKHARTLAKE(dev_priv) && phy == PHY_C) + if (!has_phy_misc(dev_priv, phy)) goto skip_phy_misc; val = intel_de_read(dev_priv, ICL_PHY_MISC(phy)); diff --git a/drivers/gpu/drm/i915/display/intel_crt.c b/drivers/gpu/drm/i915/display/intel_crt.c index 2f5b9a4baafd..5b4510ce5693 100644 --- a/drivers/gpu/drm/i915/display/intel_crt.c +++ b/drivers/gpu/drm/i915/display/intel_crt.c @@ -833,7 +833,7 @@ intel_crt_detect(struct drm_connector *connector, connector->base.id, connector->name, force); - if (i915_modparams.load_detect_test) { + if (dev_priv->params.load_detect_test) { wakeref = intel_display_power_get(dev_priv, intel_encoder->power_domain); goto load_detect; @@ -889,7 +889,7 @@ load_detect: else if (INTEL_GEN(dev_priv) < 4) status = intel_crt_load_detect(crt, to_intel_crtc(connector->state->crtc)->pipe); - else if (i915_modparams.load_detect_test) + else if (dev_priv->params.load_detect_test) status = connector_status_disconnected; else status = connector_status_unknown; diff --git a/drivers/gpu/drm/i915/display/intel_csr.c b/drivers/gpu/drm/i915/display/intel_csr.c index 3112572cfb7d..f22a7645c249 100644 --- a/drivers/gpu/drm/i915/display/intel_csr.c +++ b/drivers/gpu/drm/i915/display/intel_csr.c @@ -40,6 +40,10 @@ #define GEN12_CSR_MAX_FW_SIZE ICL_CSR_MAX_FW_SIZE +#define RKL_CSR_PATH "i915/rkl_dmc_ver2_01.bin" +#define RKL_CSR_VERSION_REQUIRED CSR_VERSION(2, 1) +MODULE_FIRMWARE(RKL_CSR_PATH); + #define TGL_CSR_PATH "i915/tgl_dmc_ver2_06.bin" #define TGL_CSR_VERSION_REQUIRED CSR_VERSION(2, 6) #define TGL_CSR_MAX_FW_SIZE 0x6000 @@ -682,7 +686,11 @@ void intel_csr_ucode_init(struct drm_i915_private *dev_priv) */ intel_csr_runtime_pm_get(dev_priv); - if (INTEL_GEN(dev_priv) >= 12) { + if (IS_ROCKETLAKE(dev_priv)) { + csr->fw_path = RKL_CSR_PATH; + csr->required_version = RKL_CSR_VERSION_REQUIRED; + csr->max_fw_size = GEN12_CSR_MAX_FW_SIZE; + } else if (INTEL_GEN(dev_priv) >= 12) { csr->fw_path = TGL_CSR_PATH; csr->required_version = TGL_CSR_VERSION_REQUIRED; /* Allow to load fw via parameter using the last known size */ @@ -699,7 +707,9 @@ void intel_csr_ucode_init(struct drm_i915_private *dev_priv) csr->fw_path = GLK_CSR_PATH; csr->required_version = GLK_CSR_VERSION_REQUIRED; csr->max_fw_size = GLK_CSR_MAX_FW_SIZE; - } else if (IS_KABYLAKE(dev_priv) || IS_COFFEELAKE(dev_priv)) { + } else if (IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) { csr->fw_path = KBL_CSR_PATH; csr->required_version = KBL_CSR_VERSION_REQUIRED; csr->max_fw_size = KBL_CSR_MAX_FW_SIZE; @@ -713,15 +723,15 @@ void intel_csr_ucode_init(struct drm_i915_private *dev_priv) csr->max_fw_size = BXT_CSR_MAX_FW_SIZE; } - if (i915_modparams.dmc_firmware_path) { - if (strlen(i915_modparams.dmc_firmware_path) == 0) { + if (dev_priv->params.dmc_firmware_path) { + if (strlen(dev_priv->params.dmc_firmware_path) == 0) { csr->fw_path = NULL; drm_info(&dev_priv->drm, "Disabling CSR firmware and runtime PM\n"); return; } - csr->fw_path = i915_modparams.dmc_firmware_path; + csr->fw_path = dev_priv->params.dmc_firmware_path; /* Bypass version check for firmware override. */ csr->required_version = 0; } diff --git a/drivers/gpu/drm/i915/display/intel_ddi.c b/drivers/gpu/drm/i915/display/intel_ddi.c index 0575a1eea2a1..884b507c5f55 100644 --- a/drivers/gpu/drm/i915/display/intel_ddi.c +++ b/drivers/gpu/drm/i915/display/intel_ddi.c @@ -639,11 +639,25 @@ struct tgl_dkl_phy_ddi_buf_trans { static const struct tgl_dkl_phy_ddi_buf_trans tgl_dkl_phy_dp_ddi_trans[] = { /* VS pre-emp Non-trans mV Pre-emph dB */ { 0x7, 0x0, 0x00 }, /* 0 0 400mV 0 dB */ - { 0x5, 0x0, 0x03 }, /* 0 1 400mV 3.5 dB */ - { 0x2, 0x0, 0x0b }, /* 0 2 400mV 6 dB */ + { 0x5, 0x0, 0x05 }, /* 0 1 400mV 3.5 dB */ + { 0x2, 0x0, 0x0B }, /* 0 2 400mV 6 dB */ + { 0x0, 0x0, 0x18 }, /* 0 3 400mV 9.5 dB */ + { 0x5, 0x0, 0x00 }, /* 1 0 600mV 0 dB */ + { 0x2, 0x0, 0x08 }, /* 1 1 600mV 3.5 dB */ + { 0x0, 0x0, 0x14 }, /* 1 2 600mV 6 dB */ + { 0x2, 0x0, 0x00 }, /* 2 0 800mV 0 dB */ + { 0x0, 0x0, 0x0B }, /* 2 1 800mV 3.5 dB */ + { 0x0, 0x0, 0x00 }, /* 3 0 1200mV 0 dB HDMI default */ +}; + +static const struct tgl_dkl_phy_ddi_buf_trans tgl_dkl_phy_dp_ddi_trans_hbr2[] = { + /* VS pre-emp Non-trans mV Pre-emph dB */ + { 0x7, 0x0, 0x00 }, /* 0 0 400mV 0 dB */ + { 0x5, 0x0, 0x05 }, /* 0 1 400mV 3.5 dB */ + { 0x2, 0x0, 0x0B }, /* 0 2 400mV 6 dB */ { 0x0, 0x0, 0x19 }, /* 0 3 400mV 9.5 dB */ { 0x5, 0x0, 0x00 }, /* 1 0 600mV 0 dB */ - { 0x2, 0x0, 0x03 }, /* 1 1 600mV 3.5 dB */ + { 0x2, 0x0, 0x08 }, /* 1 1 600mV 3.5 dB */ { 0x0, 0x0, 0x14 }, /* 1 2 600mV 6 dB */ { 0x2, 0x0, 0x00 }, /* 2 0 800mV 0 dB */ { 0x0, 0x0, 0x0B }, /* 2 1 800mV 3.5 dB */ @@ -722,10 +736,14 @@ skl_get_buf_trans_dp(struct drm_i915_private *dev_priv, int *n_entries) static const struct ddi_buf_trans * kbl_get_buf_trans_dp(struct drm_i915_private *dev_priv, int *n_entries) { - if (IS_KBL_ULX(dev_priv) || IS_CFL_ULX(dev_priv)) { + if (IS_KBL_ULX(dev_priv) || + IS_CFL_ULX(dev_priv) || + IS_CML_ULX(dev_priv)) { *n_entries = ARRAY_SIZE(kbl_y_ddi_translations_dp); return kbl_y_ddi_translations_dp; - } else if (IS_KBL_ULT(dev_priv) || IS_CFL_ULT(dev_priv)) { + } else if (IS_KBL_ULT(dev_priv) || + IS_CFL_ULT(dev_priv) || + IS_CML_ULT(dev_priv)) { *n_entries = ARRAY_SIZE(kbl_u_ddi_translations_dp); return kbl_u_ddi_translations_dp; } else { @@ -738,12 +756,16 @@ static const struct ddi_buf_trans * skl_get_buf_trans_edp(struct drm_i915_private *dev_priv, int *n_entries) { if (dev_priv->vbt.edp.low_vswing) { - if (IS_SKL_ULX(dev_priv) || IS_KBL_ULX(dev_priv) || - IS_CFL_ULX(dev_priv)) { + if (IS_SKL_ULX(dev_priv) || + IS_KBL_ULX(dev_priv) || + IS_CFL_ULX(dev_priv) || + IS_CML_ULX(dev_priv)) { *n_entries = ARRAY_SIZE(skl_y_ddi_translations_edp); return skl_y_ddi_translations_edp; - } else if (IS_SKL_ULT(dev_priv) || IS_KBL_ULT(dev_priv) || - IS_CFL_ULT(dev_priv)) { + } else if (IS_SKL_ULT(dev_priv) || + IS_KBL_ULT(dev_priv) || + IS_CFL_ULT(dev_priv) || + IS_CML_ULT(dev_priv)) { *n_entries = ARRAY_SIZE(skl_u_ddi_translations_edp); return skl_u_ddi_translations_edp; } else { @@ -752,7 +774,9 @@ skl_get_buf_trans_edp(struct drm_i915_private *dev_priv, int *n_entries) } } - if (IS_KABYLAKE(dev_priv) || IS_COFFEELAKE(dev_priv)) + if (IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) return kbl_get_buf_trans_dp(dev_priv, n_entries); else return skl_get_buf_trans_dp(dev_priv, n_entries); @@ -761,8 +785,10 @@ skl_get_buf_trans_edp(struct drm_i915_private *dev_priv, int *n_entries) static const struct ddi_buf_trans * skl_get_buf_trans_hdmi(struct drm_i915_private *dev_priv, int *n_entries) { - if (IS_SKL_ULX(dev_priv) || IS_KBL_ULX(dev_priv) || - IS_CFL_ULX(dev_priv)) { + if (IS_SKL_ULX(dev_priv) || + IS_KBL_ULX(dev_priv) || + IS_CFL_ULX(dev_priv) || + IS_CML_ULX(dev_priv)) { *n_entries = ARRAY_SIZE(skl_y_ddi_translations_hdmi); return skl_y_ddi_translations_hdmi; } else { @@ -784,7 +810,9 @@ static const struct ddi_buf_trans * intel_ddi_get_buf_trans_dp(struct drm_i915_private *dev_priv, enum port port, int *n_entries) { - if (IS_KABYLAKE(dev_priv) || IS_COFFEELAKE(dev_priv)) { + if (IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) { const struct ddi_buf_trans *ddi_translations = kbl_get_buf_trans_dp(dev_priv, n_entries); *n_entries = skl_buf_trans_num_entries(port, *n_entries); @@ -1014,6 +1042,22 @@ tgl_get_combo_buf_trans(struct drm_i915_private *dev_priv, int type, int rate, return tgl_combo_phy_ddi_translations_dp_hbr; } +static const struct tgl_dkl_phy_ddi_buf_trans * +tgl_get_dkl_buf_trans(struct drm_i915_private *dev_priv, int type, int rate, + int *n_entries) +{ + if (type == INTEL_OUTPUT_HDMI) { + *n_entries = ARRAY_SIZE(tgl_dkl_phy_hdmi_ddi_trans); + return tgl_dkl_phy_hdmi_ddi_trans; + } else if (rate > 270000) { + *n_entries = ARRAY_SIZE(tgl_dkl_phy_dp_ddi_trans_hbr2); + return tgl_dkl_phy_dp_ddi_trans_hbr2; + } + + *n_entries = ARRAY_SIZE(tgl_dkl_phy_dp_ddi_trans); + return tgl_dkl_phy_dp_ddi_trans; +} + static int intel_ddi_hdmi_level(struct intel_encoder *encoder) { struct drm_i915_private *dev_priv = to_i915(encoder->base.dev); @@ -1025,7 +1069,8 @@ static int intel_ddi_hdmi_level(struct intel_encoder *encoder) tgl_get_combo_buf_trans(dev_priv, INTEL_OUTPUT_HDMI, 0, &n_entries); else - n_entries = ARRAY_SIZE(tgl_dkl_phy_hdmi_ddi_trans); + tgl_get_dkl_buf_trans(dev_priv, INTEL_OUTPUT_HDMI, 0, + &n_entries); default_entry = n_entries - 1; } else if (INTEL_GEN(dev_priv) == 11) { if (intel_phy_is_combo(dev_priv, phy)) @@ -1608,7 +1653,6 @@ void intel_ddi_enable_transcoder_func(struct intel_encoder *encoder, struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); enum transcoder cpu_transcoder = crtc_state->cpu_transcoder; - u32 ctl; if (INTEL_GEN(dev_priv) >= 11) { enum transcoder master_transcoder = crtc_state->master_transcoder; @@ -1626,10 +1670,9 @@ void intel_ddi_enable_transcoder_func(struct intel_encoder *encoder, TRANS_DDI_FUNC_CTL2(cpu_transcoder), ctl2); } - ctl = intel_ddi_transcoder_func_reg_val_get(encoder, crtc_state); - if (intel_crtc_has_type(crtc_state, INTEL_OUTPUT_DP_MST)) - ctl |= TRANS_DDI_DP_VC_PAYLOAD_ALLOC; - intel_de_write(dev_priv, TRANS_DDI_FUNC_CTL(cpu_transcoder), ctl); + intel_de_write(dev_priv, TRANS_DDI_FUNC_CTL(cpu_transcoder), + intel_ddi_transcoder_func_reg_val_get(encoder, + crtc_state)); } /* @@ -2095,10 +2138,10 @@ static void bxt_ddi_vswing_sequence(struct intel_encoder *encoder, ddi_translations[level].deemphasis); } -u8 intel_ddi_dp_voltage_max(struct intel_encoder *encoder) +static u8 intel_ddi_dp_voltage_max(struct intel_dp *intel_dp) { + struct intel_encoder *encoder = &dp_to_dig_port(intel_dp)->base; struct drm_i915_private *dev_priv = to_i915(encoder->base.dev); - struct intel_dp *intel_dp = enc_to_intel_dp(encoder); enum port port = encoder->port; enum phy phy = intel_port_to_phy(dev_priv, port); int n_entries; @@ -2108,7 +2151,8 @@ u8 intel_ddi_dp_voltage_max(struct intel_encoder *encoder) tgl_get_combo_buf_trans(dev_priv, encoder->type, intel_dp->link_rate, &n_entries); else - n_entries = ARRAY_SIZE(tgl_dkl_phy_dp_ddi_trans); + tgl_get_dkl_buf_trans(dev_priv, encoder->type, + intel_dp->link_rate, &n_entries); } else if (INTEL_GEN(dev_priv) == 11) { if (IS_ELKHARTLAKE(dev_priv)) ehl_get_combo_buf_trans(dev_priv, encoder->type, @@ -2151,19 +2195,9 @@ u8 intel_ddi_dp_voltage_max(struct intel_encoder *encoder) * used on all DDI platforms. Should that change we need to * rethink this code. */ -u8 intel_ddi_dp_pre_emphasis_max(struct intel_encoder *encoder, u8 voltage_swing) +static u8 intel_ddi_dp_preemph_max(struct intel_dp *intel_dp) { - switch (voltage_swing & DP_TRAIN_VOLTAGE_SWING_MASK) { - case DP_TRAIN_VOLTAGE_SWING_LEVEL_0: - return DP_TRAIN_PRE_EMPH_LEVEL_3; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_1: - return DP_TRAIN_PRE_EMPH_LEVEL_2; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_2: - return DP_TRAIN_PRE_EMPH_LEVEL_1; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_3: - default: - return DP_TRAIN_PRE_EMPH_LEVEL_0; - } + return DP_TRAIN_PRE_EMPH_LEVEL_3; } static void cnl_ddi_vswing_program(struct intel_encoder *encoder, @@ -2585,15 +2619,17 @@ tgl_dkl_phy_ddi_vswing_sequence(struct intel_encoder *encoder, int link_clock, enum tc_port tc_port = intel_port_to_tc(dev_priv, encoder->port); const struct tgl_dkl_phy_ddi_buf_trans *ddi_translations; u32 n_entries, val, ln, dpcnt_mask, dpcnt_val; + int rate = 0; if (type == INTEL_OUTPUT_HDMI) { - n_entries = ARRAY_SIZE(tgl_dkl_phy_hdmi_ddi_trans); - ddi_translations = tgl_dkl_phy_hdmi_ddi_trans; - } else { - n_entries = ARRAY_SIZE(tgl_dkl_phy_dp_ddi_trans); - ddi_translations = tgl_dkl_phy_dp_ddi_trans; + struct intel_dp *intel_dp = enc_to_intel_dp(encoder); + + rate = intel_dp->link_rate; } + ddi_translations = tgl_get_dkl_buf_trans(dev_priv, encoder->type, rate, + &n_entries); + if (level >= n_entries) level = n_entries - 1; @@ -4155,11 +4191,6 @@ void intel_ddi_get_config(struct intel_encoder *encoder, if (drm_WARN_ON(&dev_priv->drm, transcoder_is_dsi(cpu_transcoder))) return; - if (INTEL_GEN(dev_priv) >= 12) { - intel_dp->regs.dp_tp_ctl = TGL_DP_TP_CTL(cpu_transcoder); - intel_dp->regs.dp_tp_status = TGL_DP_TP_STATUS(cpu_transcoder); - } - intel_dsc_get_config(encoder, pipe_config); temp = intel_de_read(dev_priv, TRANS_DDI_FUNC_CTL(cpu_transcoder)); @@ -4261,6 +4292,16 @@ void intel_ddi_get_config(struct intel_encoder *encoder, break; } + if (INTEL_GEN(dev_priv) >= 12) { + enum transcoder transcoder = + intel_dp_mst_is_slave_trans(pipe_config) ? + pipe_config->mst_master_transcoder : + pipe_config->cpu_transcoder; + + intel_dp->regs.dp_tp_ctl = TGL_DP_TP_CTL(transcoder); + intel_dp->regs.dp_tp_status = TGL_DP_TP_STATUS(transcoder); + } + pipe_config->has_audio = intel_ddi_is_audio_enabled(dev_priv, cpu_transcoder); @@ -4523,6 +4564,9 @@ intel_ddi_init_dp_connector(struct intel_digital_port *intel_dig_port) else intel_dig_port->dp.set_signal_levels = hsw_set_signal_levels; + intel_dig_port->dp.voltage_max = intel_ddi_dp_voltage_max; + intel_dig_port->dp.preemph_max = intel_ddi_dp_preemph_max; + if (INTEL_GEN(dev_priv) < 12) { intel_dig_port->dp.regs.dp_tp_ctl = DP_TP_CTL(port); intel_dig_port->dp.regs.dp_tp_status = DP_TP_STATUS(port); diff --git a/drivers/gpu/drm/i915/display/intel_ddi.h b/drivers/gpu/drm/i915/display/intel_ddi.h index fbdf8ddde486..077e9dbbe367 100644 --- a/drivers/gpu/drm/i915/display/intel_ddi.h +++ b/drivers/gpu/drm/i915/display/intel_ddi.h @@ -42,9 +42,6 @@ void intel_ddi_compute_min_voltage_level(struct drm_i915_private *dev_priv, struct intel_crtc_state *crtc_state); u32 bxt_signal_levels(struct intel_dp *intel_dp); u32 ddi_signal_levels(struct intel_dp *intel_dp); -u8 intel_ddi_dp_voltage_max(struct intel_encoder *encoder); -u8 intel_ddi_dp_pre_emphasis_max(struct intel_encoder *encoder, - u8 voltage_swing); int intel_ddi_toggle_hdcp_signalling(struct intel_encoder *intel_encoder, bool enable); void icl_sanitize_encoder_pll_mapping(struct intel_encoder *encoder); diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c index c1836095ea38..a11bb675f9b3 100644 --- a/drivers/gpu/drm/i915/display/intel_display.c +++ b/drivers/gpu/drm/i915/display/intel_display.c @@ -4812,11 +4812,18 @@ u32 glk_plane_color_ctl(const struct intel_crtc_state *crtc_state, plane_color_ctl |= glk_plane_color_ctl_alpha(plane_state); if (fb->format->is_yuv && !icl_is_hdr_plane(dev_priv, plane->id)) { - if (plane_state->hw.color_encoding == DRM_COLOR_YCBCR_BT709) + switch (plane_state->hw.color_encoding) { + case DRM_COLOR_YCBCR_BT709: plane_color_ctl |= PLANE_COLOR_CSC_MODE_YUV709_TO_RGB709; - else - plane_color_ctl |= PLANE_COLOR_CSC_MODE_YUV601_TO_RGB709; - + break; + case DRM_COLOR_YCBCR_BT2020: + plane_color_ctl |= + PLANE_COLOR_CSC_MODE_YUV2020_TO_RGB2020; + break; + default: + plane_color_ctl |= + PLANE_COLOR_CSC_MODE_YUV601_TO_RGB601; + } if (plane_state->hw.color_range == DRM_COLOR_YCBCR_FULL_RANGE) plane_color_ctl |= PLANE_COLOR_YUV_RANGE_CORRECTION_DISABLE; } else if (fb->format->is_yuv) { @@ -4879,7 +4886,7 @@ void intel_prepare_reset(struct drm_i915_private *dev_priv) int ret; /* reset doesn't touch the display */ - if (!i915_modparams.force_reset_modeset_test && + if (!dev_priv->params.force_reset_modeset_test && !gpu_reset_clobbers_display(dev_priv)) return; @@ -6424,8 +6431,7 @@ static bool hsw_post_update_enable_ips(const struct intel_crtc_state *old_crtc_s * We can't read out IPS on broadwell, assume the worst and * forcibly enable IPS on the first fastset. */ - if (new_crtc_state->update_pipe && - old_crtc_state->hw.adjusted_mode.private_flags & I915_MODE_FLAG_INHERITED) + if (new_crtc_state->update_pipe && old_crtc_state->inherited) return true; return !old_crtc_state->ips_enabled; @@ -7212,30 +7218,33 @@ bool intel_phy_is_combo(struct drm_i915_private *dev_priv, enum phy phy) { if (phy == PHY_NONE) return false; - - if (IS_ELKHARTLAKE(dev_priv)) + else if (IS_ROCKETLAKE(dev_priv)) + return phy <= PHY_D; + else if (IS_ELKHARTLAKE(dev_priv)) return phy <= PHY_C; - - if (INTEL_GEN(dev_priv) >= 11) + else if (INTEL_GEN(dev_priv) >= 11) return phy <= PHY_B; - - return false; + else + return false; } bool intel_phy_is_tc(struct drm_i915_private *dev_priv, enum phy phy) { - if (INTEL_GEN(dev_priv) >= 12) + if (IS_ROCKETLAKE(dev_priv)) + return false; + else if (INTEL_GEN(dev_priv) >= 12) return phy >= PHY_D && phy <= PHY_I; - - if (INTEL_GEN(dev_priv) >= 11 && !IS_ELKHARTLAKE(dev_priv)) + else if (INTEL_GEN(dev_priv) >= 11 && !IS_ELKHARTLAKE(dev_priv)) return phy >= PHY_C && phy <= PHY_F; - - return false; + else + return false; } enum phy intel_port_to_phy(struct drm_i915_private *i915, enum port port) { - if (IS_ELKHARTLAKE(i915) && port == PORT_D) + if (IS_ROCKETLAKE(i915) && port >= PORT_D) + return (enum phy)port - 1; + else if (IS_ELKHARTLAKE(i915) && port == PORT_D) return PHY_A; return (enum phy)port; @@ -7579,6 +7588,8 @@ static void intel_crtc_disable_noatomic(struct intel_crtc *crtc, to_intel_bw_state(dev_priv->bw_obj.state); struct intel_cdclk_state *cdclk_state = to_intel_cdclk_state(dev_priv->cdclk.obj.state); + struct intel_dbuf_state *dbuf_state = + to_intel_dbuf_state(dev_priv->dbuf.obj.state); struct intel_crtc_state *crtc_state = to_intel_crtc_state(crtc->base.state); enum intel_display_power_domain domain; @@ -7652,6 +7663,8 @@ static void intel_crtc_disable_noatomic(struct intel_crtc *crtc, cdclk_state->min_voltage_level[pipe] = 0; cdclk_state->active_pipes &= ~BIT(pipe); + dbuf_state->active_pipes &= ~BIT(pipe); + bw_state->data_rate[pipe] = 0; bw_state->num_active_planes[pipe] = 0; } @@ -7869,7 +7882,7 @@ bool hsw_crtc_state_ips_capable(const struct intel_crtc_state *crtc_state) if (!hsw_crtc_supports_ips(crtc)) return false; - if (!i915_modparams.enable_ips) + if (!dev_priv->params.enable_ips) return false; if (crtc_state->pipe_bpp > 24) @@ -8140,8 +8153,8 @@ static void intel_panel_sanitize_ssc(struct drm_i915_private *dev_priv) static bool intel_panel_use_ssc(struct drm_i915_private *dev_priv) { - if (i915_modparams.panel_use_ssc >= 0) - return i915_modparams.panel_use_ssc != 0; + if (dev_priv->params.panel_use_ssc >= 0) + return dev_priv->params.panel_use_ssc != 0; return dev_priv->vbt.lvds_use_ssc && !(dev_priv->quirks & QUIRK_LVDS_SSC_DISABLE); } @@ -10882,7 +10895,7 @@ static bool hsw_get_transcoder_state(struct intel_crtc *crtc, struct drm_device *dev = crtc->base.dev; struct drm_i915_private *dev_priv = to_i915(dev); enum intel_display_power_domain power_domain; - unsigned long panel_transcoder_mask = 0; + unsigned long panel_transcoder_mask = BIT(TRANSCODER_EDP); unsigned long enabled_panel_transcoders = 0; enum transcoder panel_transcoder; intel_wakeref_t wf; @@ -10892,9 +10905,6 @@ static bool hsw_get_transcoder_state(struct intel_crtc *crtc, panel_transcoder_mask |= BIT(TRANSCODER_DSI_0) | BIT(TRANSCODER_DSI_1); - if (HAS_TRANSCODER(dev_priv, TRANSCODER_EDP)) - panel_transcoder_mask |= BIT(TRANSCODER_EDP); - /* * The pipe->transcoder mapping is fixed with the exception of the eDP * and DSI transcoders handled below. @@ -10905,9 +10915,8 @@ static bool hsw_get_transcoder_state(struct intel_crtc *crtc, * XXX: Do intel_display_power_get_if_enabled before reading this (for * consistency and less surprising code; it's in always on power). */ - for_each_set_bit(panel_transcoder, - &panel_transcoder_mask, - ARRAY_SIZE(INTEL_INFO(dev_priv)->trans_offsets)) { + for_each_cpu_transcoder_masked(dev_priv, panel_transcoder, + panel_transcoder_mask) { bool force_thru = false; enum pipe trans_pipe; @@ -12499,7 +12508,7 @@ static int icl_check_nv12_planes(struct intel_crtc_state *crtc_state) continue; for_each_intel_plane_on_crtc(&dev_priv->drm, crtc, linked) { - if (!icl_is_nv12_y_plane(linked->id)) + if (!icl_is_nv12_y_plane(dev_priv, linked->id)) continue; if (crtc_state->active_planes & BIT(linked->id)) @@ -12545,6 +12554,10 @@ static int icl_check_nv12_planes(struct intel_crtc_state *crtc_state) plane_state->cus_ctl |= PLANE_CUS_PLANE_7; else if (linked->id == PLANE_SPRITE4) plane_state->cus_ctl |= PLANE_CUS_PLANE_6; + else if (linked->id == PLANE_SPRITE3) + plane_state->cus_ctl |= PLANE_CUS_PLANE_5_RKL; + else if (linked->id == PLANE_SPRITE2) + plane_state->cus_ctl |= PLANE_CUS_PLANE_4_RKL; else MISSING_CASE(linked->id); } @@ -13570,8 +13583,8 @@ pipe_config_mismatch(bool fastset, const struct intel_crtc *crtc, static bool fastboot_enabled(struct drm_i915_private *dev_priv) { - if (i915_modparams.fastboot != -1) - return i915_modparams.fastboot; + if (dev_priv->params.fastboot != -1) + return dev_priv->params.fastboot; /* Enable fastboot by default on Skylake and newer */ if (INTEL_GEN(dev_priv) >= 9) @@ -13595,8 +13608,7 @@ intel_pipe_config_compare(const struct intel_crtc_state *current_config, bool ret = true; u32 bp_gamma = 0; bool fixup_inherited = fastset && - (current_config->hw.mode.private_flags & I915_MODE_FLAG_INHERITED) && - !(pipe_config->hw.mode.private_flags & I915_MODE_FLAG_INHERITED); + current_config->inherited && !pipe_config->inherited; if (fixup_inherited && !fastboot_enabled(dev_priv)) { drm_dbg_kms(&dev_priv->drm, @@ -14007,10 +14019,10 @@ static void verify_wm_state(struct intel_crtc *crtc, hw_enabled_slices = intel_enabled_dbuf_slices_mask(dev_priv); if (INTEL_GEN(dev_priv) >= 11 && - hw_enabled_slices != dev_priv->enabled_dbuf_slices_mask) + hw_enabled_slices != dev_priv->dbuf.enabled_slices) drm_err(&dev_priv->drm, "mismatch in DBUF Slices (expected 0x%x, got 0x%x)\n", - dev_priv->enabled_dbuf_slices_mask, + dev_priv->dbuf.enabled_slices, hw_enabled_slices); /* planes */ @@ -14404,6 +14416,8 @@ intel_crtc_update_active_timings(const struct intel_crtc_state *crtc_state) drm_calc_timestamping_constants(&crtc->base, adjusted_mode); + crtc->mode_flags = crtc_state->mode_flags; + /* * The scanline counter increments at the leading edge of hsync. * @@ -14551,20 +14565,12 @@ static int intel_modeset_checks(struct intel_atomic_state *state) state->modeset = true; state->active_pipes = intel_calc_active_pipes(state, dev_priv->active_pipes); - state->active_pipe_changes = state->active_pipes ^ dev_priv->active_pipes; - - if (state->active_pipe_changes) { + if (state->active_pipes != dev_priv->active_pipes) { ret = _intel_atomic_lock_global_state(state); if (ret) return ret; } - ret = intel_modeset_calc_cdclk(state); - if (ret) - return ret; - - intel_modeset_clear_plls(state); - if (IS_HASWELL(dev_priv)) return hsw_mode_set_planes_workaround(state); @@ -14641,11 +14647,10 @@ static bool active_planes_affects_min_cdclk(struct drm_i915_private *dev_priv) /* See {hsw,vlv,ivb}_plane_ratio() */ return IS_BROADWELL(dev_priv) || IS_HASWELL(dev_priv) || IS_CHERRYVIEW(dev_priv) || IS_VALLEYVIEW(dev_priv) || - IS_IVYBRIDGE(dev_priv); + IS_IVYBRIDGE(dev_priv) || (INTEL_GEN(dev_priv) >= 11); } -static int intel_atomic_check_planes(struct intel_atomic_state *state, - bool *need_cdclk_calc) +static int intel_atomic_check_planes(struct intel_atomic_state *state) { struct drm_i915_private *dev_priv = to_i915(state->base.dev); struct intel_crtc_state *old_crtc_state, *new_crtc_state; @@ -14687,7 +14692,13 @@ static int intel_atomic_check_planes(struct intel_atomic_state *state, old_active_planes = old_crtc_state->active_planes & ~BIT(PLANE_CURSOR); new_active_planes = new_crtc_state->active_planes & ~BIT(PLANE_CURSOR); - if (hweight8(old_active_planes) == hweight8(new_active_planes)) + /* + * Not only the number of planes, but if the plane configuration had + * changed might already mean we need to recompute min CDCLK, + * because different planes might consume different amount of Dbuf bandwidth + * according to formula: Bw per plane = Pixel rate * bpp * pipe/plane scale factor + */ + if (old_active_planes == new_active_planes) continue; ret = intel_crtc_add_planes_to_state(state, crtc, new_active_planes); @@ -14695,6 +14706,21 @@ static int intel_atomic_check_planes(struct intel_atomic_state *state, return ret; } + return 0; +} + +static int intel_atomic_check_cdclk(struct intel_atomic_state *state, + bool *need_cdclk_calc) +{ + struct drm_i915_private *dev_priv = to_i915(state->base.dev); + struct intel_cdclk_state *new_cdclk_state; + struct intel_plane_state *plane_state; + struct intel_bw_state *new_bw_state; + struct intel_plane *plane; + int min_cdclk = 0; + enum pipe pipe; + int ret; + int i; /* * active_planes bitmask has been updated, and potentially * affected planes are part of the state. We can now @@ -14706,6 +14732,30 @@ static int intel_atomic_check_planes(struct intel_atomic_state *state, return ret; } + new_cdclk_state = intel_atomic_get_new_cdclk_state(state); + + if (new_cdclk_state && new_cdclk_state->force_min_cdclk_changed) + *need_cdclk_calc = true; + + ret = dev_priv->display.bw_calc_min_cdclk(state); + if (ret) + return ret; + + new_bw_state = intel_atomic_get_new_bw_state(state); + + if (!new_cdclk_state || !new_bw_state) + return 0; + + for_each_pipe(dev_priv, pipe) { + min_cdclk = max(new_cdclk_state->min_cdclk[pipe], min_cdclk); + + /* + * Currently do this change only if we need to increase + */ + if (new_bw_state->min_cdclk > min_cdclk) + *need_cdclk_calc = true; + } + return 0; } @@ -14757,16 +14807,13 @@ static int intel_atomic_check(struct drm_device *dev, struct drm_i915_private *dev_priv = to_i915(dev); struct intel_atomic_state *state = to_intel_atomic_state(_state); struct intel_crtc_state *old_crtc_state, *new_crtc_state; - struct intel_cdclk_state *new_cdclk_state; struct intel_crtc *crtc; int ret, i; bool any_ms = false; - /* Catch I915_MODE_FLAG_INHERITED */ for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) { - if (new_crtc_state->uapi.mode.private_flags != - old_crtc_state->uapi.mode.private_flags) + if (new_crtc_state->inherited != old_crtc_state->inherited) new_crtc_state->uapi.mode_changed = true; } @@ -14868,14 +14915,10 @@ static int intel_atomic_check(struct drm_device *dev, if (ret) goto fail; - ret = intel_atomic_check_planes(state, &any_ms); + ret = intel_atomic_check_planes(state); if (ret) goto fail; - new_cdclk_state = intel_atomic_get_new_cdclk_state(state); - if (new_cdclk_state && new_cdclk_state->force_min_cdclk_changed) - any_ms = true; - /* * distrust_bios_wm will force a full dbuf recomputation * but the hardware state will only get updated accordingly @@ -14896,10 +14939,6 @@ static int intel_atomic_check(struct drm_device *dev, goto fail; } - ret = intel_atomic_check_crtcs(state); - if (ret) - goto fail; - intel_fbc_choose_crtc(dev_priv, state); ret = calc_watermark_data(state); if (ret) @@ -14909,6 +14948,22 @@ static int intel_atomic_check(struct drm_device *dev, if (ret) goto fail; + ret = intel_atomic_check_cdclk(state, &any_ms); + if (ret) + goto fail; + + if (any_ms) { + ret = intel_modeset_calc_cdclk(state); + if (ret) + return ret; + + intel_modeset_clear_plls(state); + } + + ret = intel_atomic_check_crtcs(state); + if (ret) + goto fail; + for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) { if (!needs_modeset(new_crtc_state) && @@ -14939,8 +14994,24 @@ static int intel_atomic_check(struct drm_device *dev, static int intel_atomic_prepare_commit(struct intel_atomic_state *state) { - return drm_atomic_helper_prepare_planes(state->base.dev, - &state->base); + struct intel_crtc_state *crtc_state; + struct intel_crtc *crtc; + int i, ret; + + ret = drm_atomic_helper_prepare_planes(state->base.dev, &state->base); + if (ret < 0) + return ret; + + for_each_new_intel_crtc_in_state(state, crtc, crtc_state, i) { + bool mode_changed = needs_modeset(crtc_state); + + if (mode_changed || crtc_state->update_pipe || + crtc_state->uapi.color_mgmt_changed) { + intel_dsb_prepare(crtc_state); + } + } + + return 0; } u32 intel_crtc_get_vblank_counter(struct intel_crtc *crtc) @@ -15112,7 +15183,7 @@ static void intel_update_crtc(struct intel_atomic_state *state, * of enabling them on the CRTC's first fastset. */ if (new_crtc_state->update_pipe && !modeset && - old_crtc_state->hw.mode.private_flags & I915_MODE_FLAG_INHERITED) + old_crtc_state->inherited) intel_crtc_arm_fifo_underrun(crtc, new_crtc_state); } @@ -15204,29 +15275,6 @@ static void intel_commit_modeset_enables(struct intel_atomic_state *state) } } -static void icl_dbuf_slice_pre_update(struct intel_atomic_state *state) -{ - struct drm_i915_private *dev_priv = to_i915(state->base.dev); - u8 hw_enabled_slices = dev_priv->enabled_dbuf_slices_mask; - u8 required_slices = state->enabled_dbuf_slices_mask; - u8 slices_union = hw_enabled_slices | required_slices; - - /* If 2nd DBuf slice required, enable it here */ - if (INTEL_GEN(dev_priv) >= 11 && slices_union != hw_enabled_slices) - icl_dbuf_slices_update(dev_priv, slices_union); -} - -static void icl_dbuf_slice_post_update(struct intel_atomic_state *state) -{ - struct drm_i915_private *dev_priv = to_i915(state->base.dev); - u8 hw_enabled_slices = dev_priv->enabled_dbuf_slices_mask; - u8 required_slices = state->enabled_dbuf_slices_mask; - - /* If 2nd DBuf slice is no more required disable it */ - if (INTEL_GEN(dev_priv) >= 11 && required_slices != hw_enabled_slices) - icl_dbuf_slices_update(dev_priv, required_slices); -} - static void skl_commit_modeset_enables(struct intel_atomic_state *state) { struct drm_i915_private *dev_priv = to_i915(state->base.dev); @@ -15393,15 +15441,27 @@ static void intel_atomic_commit_fence_wait(struct intel_atomic_state *intel_stat &wait_reset); } +static void intel_cleanup_dsbs(struct intel_atomic_state *state) +{ + struct intel_crtc_state *old_crtc_state, *new_crtc_state; + struct intel_crtc *crtc; + int i; + + for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state, + new_crtc_state, i) + intel_dsb_cleanup(old_crtc_state); +} + static void intel_atomic_cleanup_work(struct work_struct *work) { - struct drm_atomic_state *state = - container_of(work, struct drm_atomic_state, commit_work); - struct drm_i915_private *i915 = to_i915(state->dev); + struct intel_atomic_state *state = + container_of(work, struct intel_atomic_state, base.commit_work); + struct drm_i915_private *i915 = to_i915(state->base.dev); - drm_atomic_helper_cleanup_planes(&i915->drm, state); - drm_atomic_helper_commit_cleanup_done(state); - drm_atomic_state_put(state); + intel_cleanup_dsbs(state); + drm_atomic_helper_cleanup_planes(&i915->drm, &state->base); + drm_atomic_helper_commit_cleanup_done(&state->base); + drm_atomic_state_put(&state->base); intel_atomic_helper_free_state(i915); } @@ -15467,9 +15527,7 @@ static void intel_atomic_commit_tail(struct intel_atomic_state *state) if (state->modeset) intel_encoders_update_prepare(state); - /* Enable all new slices, we might need */ - if (state->modeset) - icl_dbuf_slice_pre_update(state); + intel_dbuf_pre_plane_update(state); /* Now enable the clocks, plane, pipe, and connectors that we set up. */ dev_priv->display.commit_modeset_enables(state); @@ -15524,9 +15582,7 @@ static void intel_atomic_commit_tail(struct intel_atomic_state *state) dev_priv->display.optimize_watermarks(state, crtc); } - /* Disable all slices, we don't need */ - if (state->modeset) - icl_dbuf_slice_post_update(state); + intel_dbuf_post_plane_update(state); for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) { intel_post_plane_update(state, crtc); @@ -15535,6 +15591,13 @@ static void intel_atomic_commit_tail(struct intel_atomic_state *state) modeset_put_power_domains(dev_priv, put_domains[i]); intel_modeset_verify_crtc(crtc, state, old_crtc_state, new_crtc_state); + + /* + * DSB cleanup is done in cleanup_work aligning with framebuffer + * cleanup. So copy and reset the dsb structure to sync with + * commit_done and later do dsb cleanup in cleanup_work. + */ + old_crtc_state->dsb = fetch_and_zero(&new_crtc_state->dsb); } /* Underruns don't always raise interrupts, so check manually */ @@ -15684,8 +15747,15 @@ static int intel_atomic_commit(struct drm_device *dev, intel_atomic_swap_global_state(state); if (ret) { + struct intel_crtc_state *new_crtc_state; + struct intel_crtc *crtc; + int i; + i915_sw_fence_commit(&state->commit_ready); + for_each_new_intel_crtc_in_state(state, crtc, new_crtc_state, i) + intel_dsb_cleanup(new_crtc_state); + drm_atomic_helper_cleanup_planes(dev, &state->base); intel_runtime_pm_put(&dev_priv->runtime_pm, state->wakeref); return ret; @@ -16745,7 +16815,12 @@ static void intel_setup_outputs(struct drm_i915_private *dev_priv) if (!HAS_DISPLAY(dev_priv) || !INTEL_DISPLAY_ENABLED(dev_priv)) return; - if (INTEL_GEN(dev_priv) >= 12) { + if (IS_ROCKETLAKE(dev_priv)) { + intel_ddi_init(dev_priv, PORT_A); + intel_ddi_init(dev_priv, PORT_B); + intel_ddi_init(dev_priv, PORT_D); /* DDI TC1 */ + intel_ddi_init(dev_priv, PORT_E); /* DDI TC2 */ + } else if (INTEL_GEN(dev_priv) >= 12) { intel_ddi_init(dev_priv, PORT_A); intel_ddi_init(dev_priv, PORT_B); intel_ddi_init(dev_priv, PORT_D); @@ -17420,23 +17495,35 @@ void intel_modeset_init_hw(struct drm_i915_private *i915) { struct intel_cdclk_state *cdclk_state = to_intel_cdclk_state(i915->cdclk.obj.state); + struct intel_dbuf_state *dbuf_state = + to_intel_dbuf_state(i915->dbuf.obj.state); intel_update_cdclk(i915); intel_dump_cdclk_config(&i915->cdclk.hw, "Current CDCLK"); cdclk_state->logical = cdclk_state->actual = i915->cdclk.hw; + + dbuf_state->enabled_slices = i915->dbuf.enabled_slices; } static int sanitize_watermarks_add_affected(struct drm_atomic_state *state) { struct drm_plane *plane; - struct drm_crtc *crtc; + struct intel_crtc *crtc; - drm_for_each_crtc(crtc, state->dev) { - struct drm_crtc_state *crtc_state; + for_each_intel_crtc(state->dev, crtc) { + struct intel_crtc_state *crtc_state; - crtc_state = drm_atomic_get_crtc_state(state, crtc); + crtc_state = intel_atomic_get_crtc_state(state, crtc); if (IS_ERR(crtc_state)) return PTR_ERR(crtc_state); + + if (crtc_state->hw.active) { + /* + * Preserve the inherited flag to avoid + * taking the full modeset path. + */ + crtc_state->inherited = true; + } } drm_for_each_plane(plane, state->dev) { @@ -17578,6 +17665,15 @@ retry: } if (crtc_state->hw.active) { + /* + * We've not yet detected sink capabilities + * (audio,infoframes,etc.) and thus we don't want to + * force a full state recomputation yet. We want that to + * happen only for the first real commit from userspace. + * So preserve the inherited flag for the time being. + */ + crtc_state->inherited = true; + ret = drm_atomic_add_affected_planes(state, &crtc->base); if (ret) goto out; @@ -17665,7 +17761,8 @@ static void intel_mode_config_init(struct drm_i915_private *i915) if (IS_I845G(i915) || IS_I865G(i915)) { mode_config->cursor_width = IS_I845G(i915) ? 64 : 512; mode_config->cursor_height = 1023; - } else if (IS_GEN(i915, 2)) { + } else if (IS_I830(i915) || IS_I85X(i915) || + IS_I915G(i915) || IS_I915GM(i915)) { mode_config->cursor_width = 64; mode_config->cursor_height = 64; } else { @@ -17711,6 +17808,10 @@ int intel_modeset_init_noirq(struct drm_i915_private *i915) if (ret) return ret; + ret = intel_dbuf_init(i915); + if (ret) + return ret; + ret = intel_bw_init(i915); if (ret) return ret; @@ -18227,6 +18328,8 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) struct drm_i915_private *dev_priv = to_i915(dev); struct intel_cdclk_state *cdclk_state = to_intel_cdclk_state(dev_priv->cdclk.obj.state); + struct intel_dbuf_state *dbuf_state = + to_intel_dbuf_state(dev_priv->dbuf.obj.state); enum pipe pipe; struct intel_crtc *crtc; struct intel_encoder *encoder; @@ -18257,7 +18360,8 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) enableddisabled(crtc_state->hw.active)); } - dev_priv->active_pipes = cdclk_state->active_pipes = active_pipes; + dev_priv->active_pipes = cdclk_state->active_pipes = + dbuf_state->active_pipes = active_pipes; readout_plane_state(dev_priv); @@ -18348,7 +18452,7 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) * set a flag to indicate that a full recalculation is * needed on the next commit. */ - mode->private_flags = I915_MODE_FLAG_INHERITED; + crtc_state->inherited = true; intel_crtc_compute_pixel_rate(crtc_state); diff --git a/drivers/gpu/drm/i915/display/intel_display.h b/drivers/gpu/drm/i915/display/intel_display.h index efb4da205ea2..b7a6d56bac5f 100644 --- a/drivers/gpu/drm/i915/display/intel_display.h +++ b/drivers/gpu/drm/i915/display/intel_display.h @@ -187,6 +187,13 @@ enum plane_id { for ((__p) = PLANE_PRIMARY; (__p) < I915_MAX_PLANES; (__p)++) \ for_each_if((__crtc)->plane_ids_mask & BIT(__p)) +#define for_each_dbuf_slice_in_mask(__slice, __mask) \ + for ((__slice) = DBUF_S1; (__slice) < I915_MAX_DBUF_SLICES; (__slice)++) \ + for_each_if((BIT(__slice)) & (__mask)) + +#define for_each_dbuf_slice(__slice) \ + for_each_dbuf_slice_in_mask(__slice, BIT(I915_MAX_DBUF_SLICES) - 1) + enum port { PORT_NONE = -1, diff --git a/drivers/gpu/drm/i915/display/intel_display_debugfs.c b/drivers/gpu/drm/i915/display/intel_display_debugfs.c index 2b640d8ab9d2..d1cb48b3f462 100644 --- a/drivers/gpu/drm/i915/display/intel_display_debugfs.c +++ b/drivers/gpu/drm/i915/display/intel_display_debugfs.c @@ -125,7 +125,7 @@ static int i915_ips_status(struct seq_file *m, void *unused) wakeref = intel_runtime_pm_get(&dev_priv->runtime_pm); seq_printf(m, "Enabled by kernel parameter: %s\n", - yesno(i915_modparams.enable_ips)); + yesno(dev_priv->params.enable_ips)); if (INTEL_GEN(dev_priv) >= 8) { seq_puts(m, "Currently: unknown\n"); @@ -2218,7 +2218,8 @@ int intel_connector_debugfs_add(struct drm_connector *connector) } if (INTEL_GEN(dev_priv) >= 10 && - (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort || + ((connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort && + !to_intel_connector(connector)->mst_port) || connector->connector_type == DRM_MODE_CONNECTOR_eDP)) debugfs_create_file("i915_dsc_fec_support", S_IRUGO, root, connector, &i915_dsc_fec_support_fops); diff --git a/drivers/gpu/drm/i915/display/intel_display_power.c b/drivers/gpu/drm/i915/display/intel_display_power.c index 49998906cc61..834162bc5a3f 100644 --- a/drivers/gpu/drm/i915/display/intel_display_power.c +++ b/drivers/gpu/drm/i915/display/intel_display_power.c @@ -1161,7 +1161,7 @@ static bool gen9_dc_off_power_well_enabled(struct drm_i915_private *dev_priv, static void gen9_assert_dbuf_enabled(struct drm_i915_private *dev_priv) { u8 hw_enabled_dbuf_slices = intel_enabled_dbuf_slices_mask(dev_priv); - u8 enabled_dbuf_slices = dev_priv->enabled_dbuf_slices_mask; + u8 enabled_dbuf_slices = dev_priv->dbuf.enabled_slices; drm_WARN(&dev_priv->drm, hw_enabled_dbuf_slices != enabled_dbuf_slices, @@ -1943,22 +1943,29 @@ static u64 __async_put_domains_mask(struct i915_power_domains *power_domains) static bool assert_async_put_domain_masks_disjoint(struct i915_power_domains *power_domains) { - return !WARN_ON(power_domains->async_put_domains[0] & - power_domains->async_put_domains[1]); + struct drm_i915_private *i915 = container_of(power_domains, + struct drm_i915_private, + power_domains); + return !drm_WARN_ON(&i915->drm, power_domains->async_put_domains[0] & + power_domains->async_put_domains[1]); } static bool __async_put_domains_state_ok(struct i915_power_domains *power_domains) { + struct drm_i915_private *i915 = container_of(power_domains, + struct drm_i915_private, + power_domains); enum intel_display_power_domain domain; bool err = false; err |= !assert_async_put_domain_masks_disjoint(power_domains); - err |= WARN_ON(!!power_domains->async_put_wakeref != - !!__async_put_domains_mask(power_domains)); + err |= drm_WARN_ON(&i915->drm, !!power_domains->async_put_wakeref != + !!__async_put_domains_mask(power_domains)); for_each_power_domain(domain, __async_put_domains_mask(power_domains)) - err |= WARN_ON(power_domains->domain_use_count[domain] != 1); + err |= drm_WARN_ON(&i915->drm, + power_domains->domain_use_count[domain] != 1); return !err; } @@ -2200,11 +2207,14 @@ static void queue_async_put_domains_work(struct i915_power_domains *power_domains, intel_wakeref_t wakeref) { - WARN_ON(power_domains->async_put_wakeref); + struct drm_i915_private *i915 = container_of(power_domains, + struct drm_i915_private, + power_domains); + drm_WARN_ON(&i915->drm, power_domains->async_put_wakeref); power_domains->async_put_wakeref = wakeref; - WARN_ON(!queue_delayed_work(system_unbound_wq, - &power_domains->async_put_work, - msecs_to_jiffies(100))); + drm_WARN_ON(&i915->drm, !queue_delayed_work(system_unbound_wq, + &power_domains->async_put_work, + msecs_to_jiffies(100))); } static void @@ -2913,6 +2923,53 @@ void intel_display_power_put(struct drm_i915_private *dev_priv, BIT_ULL(POWER_DOMAIN_AUX_I_TBT) | \ BIT_ULL(POWER_DOMAIN_TC_COLD_OFF)) +#define RKL_PW_4_POWER_DOMAINS ( \ + BIT_ULL(POWER_DOMAIN_PIPE_C) | \ + BIT_ULL(POWER_DOMAIN_PIPE_C_PANEL_FITTER) | \ + BIT_ULL(POWER_DOMAIN_TRANSCODER_C) | \ + BIT_ULL(POWER_DOMAIN_INIT)) + +#define RKL_PW_3_POWER_DOMAINS ( \ + RKL_PW_4_POWER_DOMAINS | \ + BIT_ULL(POWER_DOMAIN_PIPE_B) | \ + BIT_ULL(POWER_DOMAIN_PIPE_B_PANEL_FITTER) | \ + BIT_ULL(POWER_DOMAIN_AUDIO) | \ + BIT_ULL(POWER_DOMAIN_VGA) | \ + BIT_ULL(POWER_DOMAIN_TRANSCODER_B) | \ + BIT_ULL(POWER_DOMAIN_PORT_DDI_D_LANES) | \ + BIT_ULL(POWER_DOMAIN_PORT_DDI_E_LANES) | \ + BIT_ULL(POWER_DOMAIN_AUX_D) | \ + BIT_ULL(POWER_DOMAIN_AUX_E) | \ + BIT_ULL(POWER_DOMAIN_INIT)) + +/* + * There is no PW_2/PG_2 on RKL. + * + * RKL PW_1/PG_1 domains (under HW/DMC control): + * - DBUF function (note: registers are in PW0) + * - PIPE_A and its planes and VDSC/joining, except VGA + * - transcoder A + * - DDI_A and DDI_B + * - FBC + * + * RKL PW_0/PG_0 domains (under HW/DMC control): + * - PCI + * - clocks except port PLL + * - shared functions: + * * interrupts except pipe interrupts + * * MBus except PIPE_MBUS_DBOX_CTL + * * DBUF registers + * - central power except FBC + * - top-level GTC (DDI-level GTC is in the well associated with the DDI) + */ + +#define RKL_DISPLAY_DC_OFF_POWER_DOMAINS ( \ + RKL_PW_3_POWER_DOMAINS | \ + BIT_ULL(POWER_DOMAIN_MODESET) | \ + BIT_ULL(POWER_DOMAIN_AUX_A) | \ + BIT_ULL(POWER_DOMAIN_AUX_B) | \ + BIT_ULL(POWER_DOMAIN_INIT)) + static const struct i915_power_well_ops i9xx_always_on_power_well_ops = { .sync_hw = i9xx_power_well_sync_hw_noop, .enable = i9xx_always_on_power_well_noop, @@ -4283,6 +4340,140 @@ static const struct i915_power_well_desc tgl_power_wells[] = { }, }; +static const struct i915_power_well_desc rkl_power_wells[] = { + { + .name = "always-on", + .always_on = true, + .domains = POWER_DOMAIN_MASK, + .ops = &i9xx_always_on_power_well_ops, + .id = DISP_PW_ID_NONE, + }, + { + .name = "power well 1", + /* Handled by the DMC firmware */ + .always_on = true, + .domains = 0, + .ops = &hsw_power_well_ops, + .id = SKL_DISP_PW_1, + { + .hsw.regs = &hsw_power_well_regs, + .hsw.idx = ICL_PW_CTL_IDX_PW_1, + .hsw.has_fuses = true, + }, + }, + { + .name = "DC off", + .domains = RKL_DISPLAY_DC_OFF_POWER_DOMAINS, + .ops = &gen9_dc_off_power_well_ops, + .id = SKL_DISP_DC_OFF, + }, + { + .name = "power well 3", + .domains = RKL_PW_3_POWER_DOMAINS, + .ops = &hsw_power_well_ops, + .id = ICL_DISP_PW_3, + { + .hsw.regs = &hsw_power_well_regs, + .hsw.idx = ICL_PW_CTL_IDX_PW_3, + .hsw.irq_pipe_mask = BIT(PIPE_B), + .hsw.has_vga = true, + .hsw.has_fuses = true, + }, + }, + { + .name = "power well 4", + .domains = RKL_PW_4_POWER_DOMAINS, + .ops = &hsw_power_well_ops, + .id = DISP_PW_ID_NONE, + { + .hsw.regs = &hsw_power_well_regs, + .hsw.idx = ICL_PW_CTL_IDX_PW_4, + .hsw.has_fuses = true, + .hsw.irq_pipe_mask = BIT(PIPE_C), + } + }, + { + .name = "DDI A IO", + .domains = ICL_DDI_IO_A_POWER_DOMAINS, + .ops = &hsw_power_well_ops, + .id = DISP_PW_ID_NONE, + { + .hsw.regs = &icl_ddi_power_well_regs, + .hsw.idx = ICL_PW_CTL_IDX_DDI_A, + } + }, + { + .name = "DDI B IO", + .domains = ICL_DDI_IO_B_POWER_DOMAINS, + .ops = &hsw_power_well_ops, + .id = DISP_PW_ID_NONE, + { + .hsw.regs = &icl_ddi_power_well_regs, + .hsw.idx = ICL_PW_CTL_IDX_DDI_B, + } + }, + { + .name = "DDI D TC1 IO", + .domains = TGL_DDI_IO_D_TC1_POWER_DOMAINS, + .ops = &hsw_power_well_ops, + .id = DISP_PW_ID_NONE, + { + .hsw.regs = &icl_ddi_power_well_regs, + .hsw.idx = TGL_PW_CTL_IDX_DDI_TC1, + }, + }, + { + .name = "DDI E TC2 IO", + .domains = TGL_DDI_IO_E_TC2_POWER_DOMAINS, + .ops = &hsw_power_well_ops, + .id = DISP_PW_ID_NONE, + { + .hsw.regs = &icl_ddi_power_well_regs, + .hsw.idx = TGL_PW_CTL_IDX_DDI_TC2, + }, + }, + { + .name = "AUX A", + .domains = ICL_AUX_A_IO_POWER_DOMAINS, + .ops = &icl_aux_power_well_ops, + .id = DISP_PW_ID_NONE, + { + .hsw.regs = &icl_aux_power_well_regs, + .hsw.idx = ICL_PW_CTL_IDX_AUX_A, + }, + }, + { + .name = "AUX B", + .domains = ICL_AUX_B_IO_POWER_DOMAINS, + .ops = &icl_aux_power_well_ops, + .id = DISP_PW_ID_NONE, + { + .hsw.regs = &icl_aux_power_well_regs, + .hsw.idx = ICL_PW_CTL_IDX_AUX_B, + }, + }, + { + .name = "AUX D TC1", + .domains = TGL_AUX_D_TC1_IO_POWER_DOMAINS, + .ops = &icl_aux_power_well_ops, + .id = DISP_PW_ID_NONE, + { + .hsw.regs = &icl_aux_power_well_regs, + .hsw.idx = TGL_PW_CTL_IDX_AUX_TC1, + }, + }, + { + .name = "AUX E TC2", + .domains = TGL_AUX_E_TC2_IO_POWER_DOMAINS, + .ops = &icl_aux_power_well_ops, + .id = DISP_PW_ID_NONE, + { + .hsw.regs = &icl_aux_power_well_regs, + .hsw.idx = TGL_PW_CTL_IDX_AUX_TC2, + }, + }, +}; + static int sanitize_disable_power_well_option(const struct drm_i915_private *dev_priv, int disable_power_well) @@ -4322,7 +4513,7 @@ static u32 get_allowed_dc_mask(const struct drm_i915_private *dev_priv, mask = 0; } - if (!i915_modparams.disable_power_well) + if (!dev_priv->params.disable_power_well) max_dc = 0; if (enable_dc >= 0 && enable_dc <= max_dc) { @@ -4365,6 +4556,9 @@ __set_power_wells(struct i915_power_domains *power_domains, const struct i915_power_well_desc *power_well_descs, int power_well_count) { + struct drm_i915_private *i915 = container_of(power_domains, + struct drm_i915_private, + power_domains); u64 power_well_ids = 0; int i; @@ -4384,8 +4578,8 @@ __set_power_wells(struct i915_power_domains *power_domains, if (id == DISP_PW_ID_NONE) continue; - WARN_ON(id >= sizeof(power_well_ids) * 8); - WARN_ON(power_well_ids & BIT_ULL(id)); + drm_WARN_ON(&i915->drm, id >= sizeof(power_well_ids) * 8); + drm_WARN_ON(&i915->drm, power_well_ids & BIT_ULL(id)); power_well_ids |= BIT_ULL(id); } @@ -4408,11 +4602,11 @@ int intel_power_domains_init(struct drm_i915_private *dev_priv) struct i915_power_domains *power_domains = &dev_priv->power_domains; int err; - i915_modparams.disable_power_well = + dev_priv->params.disable_power_well = sanitize_disable_power_well_option(dev_priv, - i915_modparams.disable_power_well); + dev_priv->params.disable_power_well); dev_priv->csr.allowed_dc_mask = - get_allowed_dc_mask(dev_priv, i915_modparams.enable_dc); + get_allowed_dc_mask(dev_priv, dev_priv->params.enable_dc); dev_priv->csr.target_dc_state = sanitize_target_dc_state(dev_priv, DC_STATE_EN_UPTO_DC6); @@ -4428,7 +4622,9 @@ int intel_power_domains_init(struct drm_i915_private *dev_priv) * The enabling order will be from lower to higher indexed wells, * the disabling order is reversed. */ - if (IS_GEN(dev_priv, 12)) { + if (IS_ROCKETLAKE(dev_priv)) { + err = set_power_wells(power_domains, rkl_power_wells); + } else if (IS_GEN(dev_priv, 12)) { err = set_power_wells(power_domains, tgl_power_wells); } else if (IS_GEN(dev_priv, 11)) { err = set_power_wells(power_domains, icl_power_wells); @@ -4491,45 +4687,38 @@ static void intel_power_domains_sync_hw(struct drm_i915_private *dev_priv) mutex_unlock(&power_domains->lock); } -static bool intel_dbuf_slice_set(struct drm_i915_private *dev_priv, - i915_reg_t reg, bool enable) +static void gen9_dbuf_slice_set(struct drm_i915_private *dev_priv, + enum dbuf_slice slice, bool enable) { - u32 val, status; + i915_reg_t reg = DBUF_CTL_S(slice); + bool state; + u32 val; val = intel_de_read(dev_priv, reg); - val = enable ? (val | DBUF_POWER_REQUEST) : (val & ~DBUF_POWER_REQUEST); + if (enable) + val |= DBUF_POWER_REQUEST; + else + val &= ~DBUF_POWER_REQUEST; intel_de_write(dev_priv, reg, val); intel_de_posting_read(dev_priv, reg); udelay(10); - status = intel_de_read(dev_priv, reg) & DBUF_POWER_STATE; - if ((enable && !status) || (!enable && status)) { - drm_err(&dev_priv->drm, "DBus power %s timeout!\n", - enable ? "enable" : "disable"); - return false; - } - return true; + state = intel_de_read(dev_priv, reg) & DBUF_POWER_STATE; + drm_WARN(&dev_priv->drm, enable != state, + "DBuf slice %d power %s timeout!\n", + slice, enable ? "enable" : "disable"); } -static void gen9_dbuf_enable(struct drm_i915_private *dev_priv) +void gen9_dbuf_slices_update(struct drm_i915_private *dev_priv, + u8 req_slices) { - icl_dbuf_slices_update(dev_priv, BIT(DBUF_S1)); -} - -static void gen9_dbuf_disable(struct drm_i915_private *dev_priv) -{ - icl_dbuf_slices_update(dev_priv, 0); -} - -void icl_dbuf_slices_update(struct drm_i915_private *dev_priv, - u8 req_slices) -{ - int i; - int max_slices = INTEL_INFO(dev_priv)->num_supported_dbuf_slices; + int num_slices = INTEL_INFO(dev_priv)->num_supported_dbuf_slices; struct i915_power_domains *power_domains = &dev_priv->power_domains; + enum dbuf_slice slice; - drm_WARN(&dev_priv->drm, hweight8(req_slices) > max_slices, - "Invalid number of dbuf slices requested\n"); + drm_WARN(&dev_priv->drm, req_slices & ~(BIT(num_slices) - 1), + "Invalid set of dbuf slices (0x%x) requested (num dbuf slices %d)\n", + req_slices, num_slices); drm_dbg_kms(&dev_priv->drm, "Updating dbuf slices to 0x%x\n", req_slices); @@ -4543,36 +4732,36 @@ void icl_dbuf_slices_update(struct drm_i915_private *dev_priv, */ mutex_lock(&power_domains->lock); - for (i = 0; i < max_slices; i++) { - intel_dbuf_slice_set(dev_priv, - DBUF_CTL_S(i), - (req_slices & BIT(i)) != 0); - } + for (slice = DBUF_S1; slice < num_slices; slice++) + gen9_dbuf_slice_set(dev_priv, slice, req_slices & BIT(slice)); - dev_priv->enabled_dbuf_slices_mask = req_slices; + dev_priv->dbuf.enabled_slices = req_slices; mutex_unlock(&power_domains->lock); } -static void icl_dbuf_enable(struct drm_i915_private *dev_priv) +static void gen9_dbuf_enable(struct drm_i915_private *dev_priv) { - skl_ddb_get_hw_state(dev_priv); + dev_priv->dbuf.enabled_slices = + intel_enabled_dbuf_slices_mask(dev_priv); + /* * Just power up at least 1 slice, we will * figure out later which slices we have and what we need. */ - icl_dbuf_slices_update(dev_priv, dev_priv->enabled_dbuf_slices_mask | - BIT(DBUF_S1)); + gen9_dbuf_slices_update(dev_priv, BIT(DBUF_S1) | + dev_priv->dbuf.enabled_slices); } -static void icl_dbuf_disable(struct drm_i915_private *dev_priv) +static void gen9_dbuf_disable(struct drm_i915_private *dev_priv) { - icl_dbuf_slices_update(dev_priv, 0); + gen9_dbuf_slices_update(dev_priv, 0); } static void icl_mbus_init(struct drm_i915_private *dev_priv) { - u32 mask, val; + unsigned long abox_regs = INTEL_INFO(dev_priv)->abox_mask; + u32 mask, val, i; mask = MBUS_ABOX_BT_CREDIT_POOL1_MASK | MBUS_ABOX_BT_CREDIT_POOL2_MASK | @@ -4583,11 +4772,16 @@ static void icl_mbus_init(struct drm_i915_private *dev_priv) MBUS_ABOX_B_CREDIT(1) | MBUS_ABOX_BW_CREDIT(1); - intel_de_rmw(dev_priv, MBUS_ABOX_CTL, mask, val); - if (INTEL_GEN(dev_priv) >= 12) { - intel_de_rmw(dev_priv, MBUS_ABOX1_CTL, mask, val); - intel_de_rmw(dev_priv, MBUS_ABOX2_CTL, mask, val); - } + /* + * gen12 platforms that use abox1 and abox2 for pixel data reads still + * expect us to program the abox_ctl0 register as well, even though + * we don't have to program other instance-0 registers like BW_BUDDY. + */ + if (IS_GEN(dev_priv, 12)) + abox_regs |= BIT(0); + + for_each_set_bit(i, &abox_regs, sizeof(abox_regs)) + intel_de_rmw(dev_priv, MBUS_ABOX_CTL(i), mask, val); } static void hsw_assert_cdclk(struct drm_i915_private *dev_priv) @@ -5066,7 +5260,8 @@ static void tgl_bw_buddy_init(struct drm_i915_private *dev_priv) enum intel_dram_type type = dev_priv->dram_info.type; u8 num_channels = dev_priv->dram_info.num_channels; const struct buddy_page_mask *table; - int i; + unsigned long abox_mask = INTEL_INFO(dev_priv)->abox_mask; + int config, i; if (IS_TGL_REVID(dev_priv, TGL_REVID_A0, TGL_REVID_B0)) /* Wa_1409767108: tgl */ @@ -5074,29 +5269,27 @@ static void tgl_bw_buddy_init(struct drm_i915_private *dev_priv) else table = tgl_buddy_page_masks; - for (i = 0; table[i].page_mask != 0; i++) - if (table[i].num_channels == num_channels && - table[i].type == type) + for (config = 0; table[config].page_mask != 0; config++) + if (table[config].num_channels == num_channels && + table[config].type == type) break; - if (table[i].page_mask == 0) { + if (table[config].page_mask == 0) { drm_dbg(&dev_priv->drm, "Unknown memory configuration; disabling address buddy logic.\n"); - intel_de_write(dev_priv, BW_BUDDY1_CTL, BW_BUDDY_DISABLE); - intel_de_write(dev_priv, BW_BUDDY2_CTL, BW_BUDDY_DISABLE); + for_each_set_bit(i, &abox_mask, sizeof(abox_mask)) + intel_de_write(dev_priv, BW_BUDDY_CTL(i), + BW_BUDDY_DISABLE); } else { - intel_de_write(dev_priv, BW_BUDDY1_PAGE_MASK, - table[i].page_mask); - intel_de_write(dev_priv, BW_BUDDY2_PAGE_MASK, - table[i].page_mask); - - /* Wa_22010178259:tgl */ - intel_de_rmw(dev_priv, BW_BUDDY1_CTL, - BW_BUDDY_TLB_REQ_TIMER_MASK, - REG_FIELD_PREP(BW_BUDDY_TLB_REQ_TIMER_MASK, 0x8)); - intel_de_rmw(dev_priv, BW_BUDDY2_CTL, - BW_BUDDY_TLB_REQ_TIMER_MASK, - REG_FIELD_PREP(BW_BUDDY_TLB_REQ_TIMER_MASK, 0x8)); + for_each_set_bit(i, &abox_mask, sizeof(abox_mask)) { + intel_de_write(dev_priv, BW_BUDDY_PAGE_MASK(i), + table[config].page_mask); + + /* Wa_22010178259:tgl,rkl */ + intel_de_rmw(dev_priv, BW_BUDDY_CTL(i), + BW_BUDDY_TLB_REQ_TIMER_MASK, + BW_BUDDY_TLB_REQ_TIMER(0x8)); + } } } @@ -5127,7 +5320,7 @@ static void icl_display_core_init(struct drm_i915_private *dev_priv, intel_cdclk_init_hw(dev_priv); /* 5. Enable DBUF. */ - icl_dbuf_enable(dev_priv); + gen9_dbuf_enable(dev_priv); /* 6. Setup MBUS. */ icl_mbus_init(dev_priv); @@ -5150,7 +5343,7 @@ static void icl_display_core_uninit(struct drm_i915_private *dev_priv) /* 1. Disable all display engine functions -> aready done */ /* 2. Disable DBUF */ - icl_dbuf_disable(dev_priv); + gen9_dbuf_disable(dev_priv); /* 3. Disable CD clock */ intel_cdclk_uninit_hw(dev_priv); @@ -5375,7 +5568,7 @@ void intel_power_domains_init_hw(struct drm_i915_private *i915, bool resume) intel_display_power_get(i915, POWER_DOMAIN_INIT); /* Disable power support if the user asked so. */ - if (!i915_modparams.disable_power_well) + if (!i915->params.disable_power_well) intel_display_power_get(i915, POWER_DOMAIN_INIT); intel_power_domains_sync_hw(i915); @@ -5399,7 +5592,7 @@ void intel_power_domains_driver_remove(struct drm_i915_private *i915) fetch_and_zero(&i915->power_domains.wakeref); /* Remove the refcount we took to keep power well support disabled. */ - if (!i915_modparams.disable_power_well) + if (!i915->params.disable_power_well) intel_display_power_put_unchecked(i915, POWER_DOMAIN_INIT); intel_display_power_flush_work_sync(i915); @@ -5488,7 +5681,7 @@ void intel_power_domains_suspend(struct drm_i915_private *i915, * Even if power well support was disabled we still want to disable * power wells if power domains must be deinitialized for suspend. */ - if (!i915_modparams.disable_power_well) + if (!i915->params.disable_power_well) intel_display_power_put_unchecked(i915, POWER_DOMAIN_INIT); intel_display_power_flush_work(i915); diff --git a/drivers/gpu/drm/i915/display/intel_display_power.h b/drivers/gpu/drm/i915/display/intel_display_power.h index 6c917699293b..54c20c76057e 100644 --- a/drivers/gpu/drm/i915/display/intel_display_power.h +++ b/drivers/gpu/drm/i915/display/intel_display_power.h @@ -314,15 +314,16 @@ intel_display_power_put_async(struct drm_i915_private *i915, enum dbuf_slice { DBUF_S1, DBUF_S2, + I915_MAX_DBUF_SLICES }; +void gen9_dbuf_slices_update(struct drm_i915_private *dev_priv, + u8 req_slices); + #define with_intel_display_power(i915, domain, wf) \ for ((wf) = intel_display_power_get((i915), (domain)); (wf); \ intel_display_power_put_async((i915), (domain), (wf)), (wf) = 0) -void icl_dbuf_slices_update(struct drm_i915_private *dev_priv, - u8 req_slices); - void chv_phy_powergate_lanes(struct intel_encoder *encoder, bool override, unsigned int mask); bool chv_phy_powergate_ch(struct drm_i915_private *dev_priv, enum dpio_phy phy, diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h index 2bf3d4cb4ea9..4b0aaa3081c9 100644 --- a/drivers/gpu/drm/i915/display/intel_display_types.h +++ b/drivers/gpu/drm/i915/display/intel_display_types.h @@ -479,16 +479,6 @@ struct intel_atomic_state { bool dpll_set, modeset; - /* - * Does this transaction change the pipes that are active? This mask - * tracks which CRTC's have changed their active state at the end of - * the transaction (not counting the temporary disable during modesets). - * This mask should only be non-zero when intel_state->modeset is true, - * but the converse is not necessarily true; simply changing a mode may - * not flip the final active status of any CRTC's - */ - u8 active_pipe_changes; - u8 active_pipes; struct intel_shared_dpll_state shared_dpll[I915_NUM_PLLS]; @@ -506,9 +496,6 @@ struct intel_atomic_state { */ bool global_state_changed; - /* Number of enabled DBuf slices */ - u8 enabled_dbuf_slices_mask; - struct i915_sw_fence commit_ready; struct llist_node freed; @@ -643,8 +630,7 @@ struct intel_crtc_scaler_state { int scaler_id; }; -/* drm_mode->private_flags */ -#define I915_MODE_FLAG_INHERITED (1<<0) +/* {crtc,crtc_state}->mode_flags */ /* Flag to get scanline using frame time stamps */ #define I915_MODE_FLAG_GET_SCANLINE_FROM_TIMESTAMP (1<<1) /* Flag to use the scanline counter instead of the pixel counter */ @@ -841,6 +827,7 @@ struct intel_crtc_state { bool update_wm_pre, update_wm_post; /* watermarks are updated */ bool fifo_changed; /* FIFO split is changed */ bool preload_luts; + bool inherited; /* state inherited from BIOS? */ /* Pipe source size (ie. panel fitter input size) * All planes will be positioned inside this space, @@ -956,6 +943,9 @@ struct intel_crtc_state { /* Used by SDVO (and if we ever fix it, HDMI). */ unsigned pixel_multiplier; + /* I915_MODE_FLAG_* */ + u8 mode_flags; + u8 lane_count; /* @@ -1080,6 +1070,9 @@ struct intel_crtc_state { /* Only valid on TGL+ */ enum transcoder mst_master_transcoder; + + /* For DSB related info */ + struct intel_dsb *dsb; }; enum intel_pipe_crc_source { @@ -1118,6 +1111,10 @@ struct intel_crtc { */ bool active; u8 plane_ids_mask; + + /* I915_MODE_FLAG_* */ + u8 mode_flags; + unsigned long long enabled_power_domains; struct intel_overlay *overlay; @@ -1149,9 +1146,6 @@ struct intel_crtc { /* scalers available on this crtc */ int num_scalers; - /* per pipe DSB related info */ - struct intel_dsb dsb; - #ifdef CONFIG_DEBUG_FS struct intel_pipe_crc pipe_crc; #endif @@ -1373,6 +1367,9 @@ struct intel_dp { void (*set_idle_link_train)(struct intel_dp *intel_dp); void (*set_signal_levels)(struct intel_dp *intel_dp); + u8 (*preemph_max)(struct intel_dp *intel_dp); + u8 (*voltage_max)(struct intel_dp *intel_dp); + /* Displayport compliance testing */ struct intel_dp_compliance compliance; diff --git a/drivers/gpu/drm/i915/display/intel_dp.c b/drivers/gpu/drm/i915/display/intel_dp.c index cc525fda441a..3df5d901dd9d 100644 --- a/drivers/gpu/drm/i915/display/intel_dp.c +++ b/drivers/gpu/drm/i915/display/intel_dp.c @@ -409,7 +409,10 @@ static int intel_dp_rate_index(const int *rates, int len, int rate) static void intel_dp_set_common_rates(struct intel_dp *intel_dp) { - WARN_ON(!intel_dp->num_source_rates || !intel_dp->num_sink_rates); + struct drm_i915_private *i915 = dp_to_i915(intel_dp); + + drm_WARN_ON(&i915->drm, + !intel_dp->num_source_rates || !intel_dp->num_sink_rates); intel_dp->num_common_rates = intersect_rates(intel_dp->source_rates, intel_dp->num_source_rates, @@ -418,7 +421,7 @@ static void intel_dp_set_common_rates(struct intel_dp *intel_dp) intel_dp->common_rates); /* Paranoia, there should always be something in common. */ - if (WARN_ON(intel_dp->num_common_rates == 0)) { + if (drm_WARN_ON(&i915->drm, intel_dp->num_common_rates == 0)) { intel_dp->common_rates[0] = 162000; intel_dp->num_common_rates = 1; } @@ -465,6 +468,15 @@ int intel_dp_get_link_train_fallback_values(struct intel_dp *intel_dp, struct drm_i915_private *i915 = dp_to_i915(intel_dp); int index; + /* + * TODO: Enable fallback on MST links once MST link compute can handle + * the fallback params. + */ + if (intel_dp->is_mst) { + drm_err(&i915->drm, "Link Training Unsuccessful\n"); + return -1; + } + index = intel_dp_rate_index(intel_dp->common_rates, intel_dp->num_common_rates, link_rate); @@ -1555,6 +1567,7 @@ static ssize_t intel_dp_aux_transfer(struct drm_dp_aux *aux, struct drm_dp_aux_msg *msg) { struct intel_dp *intel_dp = container_of(aux, struct intel_dp, aux); + struct drm_i915_private *i915 = dp_to_i915(intel_dp); u8 txbuf[20], rxbuf[20]; size_t txsize, rxsize; int ret; @@ -1568,10 +1581,10 @@ intel_dp_aux_transfer(struct drm_dp_aux *aux, struct drm_dp_aux_msg *msg) txsize = msg->size ? HEADER_SIZE + msg->size : BARE_ADDRESS_SIZE; rxsize = 2; /* 0 or 1 data bytes */ - if (WARN_ON(txsize > 20)) + if (drm_WARN_ON(&i915->drm, txsize > 20)) return -E2BIG; - WARN_ON(!msg->buffer != !msg->size); + drm_WARN_ON(&i915->drm, !msg->buffer != !msg->size); if (msg->buffer) memcpy(txbuf + HEADER_SIZE, msg->buffer, msg->size); @@ -1596,7 +1609,7 @@ intel_dp_aux_transfer(struct drm_dp_aux *aux, struct drm_dp_aux_msg *msg) txsize = msg->size ? HEADER_SIZE : BARE_ADDRESS_SIZE; rxsize = msg->size + 1; - if (WARN_ON(rxsize > 20)) + if (drm_WARN_ON(&i915->drm, rxsize > 20)) return -E2BIG; ret = intel_dp_aux_xfer(intel_dp, txbuf, txsize, @@ -1871,10 +1884,11 @@ static void intel_dp_print_rates(struct intel_dp *intel_dp) int intel_dp_max_link_rate(struct intel_dp *intel_dp) { + struct drm_i915_private *i915 = dp_to_i915(intel_dp); int len; len = intel_dp_common_len_rate_limit(intel_dp, intel_dp->max_link_rate); - if (WARN_ON(len <= 0)) + if (drm_WARN_ON(&i915->drm, len <= 0)) return 162000; return intel_dp->common_rates[len - 1]; @@ -1882,10 +1896,11 @@ intel_dp_max_link_rate(struct intel_dp *intel_dp) int intel_dp_rate_select(struct intel_dp *intel_dp, int rate) { + struct drm_i915_private *i915 = dp_to_i915(intel_dp); int i = intel_dp_rate_index(intel_dp->sink_rates, intel_dp->num_sink_rates, rate); - if (WARN_ON(i < 0)) + if (drm_WARN_ON(&i915->drm, i < 0)) i = 0; return i; @@ -3984,70 +3999,24 @@ intel_dp_get_link_status(struct intel_dp *intel_dp, u8 link_status[DP_LINK_STATU DP_LINK_STATUS_SIZE) == DP_LINK_STATUS_SIZE; } -/* These are source-specific values. */ -u8 -intel_dp_voltage_max(struct intel_dp *intel_dp) +static u8 intel_dp_voltage_max_2(struct intel_dp *intel_dp) { - struct drm_i915_private *dev_priv = dp_to_i915(intel_dp); - struct intel_encoder *encoder = &dp_to_dig_port(intel_dp)->base; - enum port port = encoder->port; + return DP_TRAIN_VOLTAGE_SWING_LEVEL_2; +} - if (HAS_DDI(dev_priv)) - return intel_ddi_dp_voltage_max(encoder); - else if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) - return DP_TRAIN_VOLTAGE_SWING_LEVEL_3; - else if (IS_IVYBRIDGE(dev_priv) && port == PORT_A) - return DP_TRAIN_VOLTAGE_SWING_LEVEL_2; - else if (HAS_PCH_CPT(dev_priv) && port != PORT_A) - return DP_TRAIN_VOLTAGE_SWING_LEVEL_3; - else - return DP_TRAIN_VOLTAGE_SWING_LEVEL_2; +static u8 intel_dp_voltage_max_3(struct intel_dp *intel_dp) +{ + return DP_TRAIN_VOLTAGE_SWING_LEVEL_3; } -u8 -intel_dp_pre_emphasis_max(struct intel_dp *intel_dp, u8 voltage_swing) +static u8 intel_dp_pre_empemph_max_2(struct intel_dp *intel_dp) { - struct drm_i915_private *dev_priv = dp_to_i915(intel_dp); - struct intel_encoder *encoder = &dp_to_dig_port(intel_dp)->base; - enum port port = encoder->port; + return DP_TRAIN_PRE_EMPH_LEVEL_2; +} - if (HAS_DDI(dev_priv)) { - return intel_ddi_dp_pre_emphasis_max(encoder, voltage_swing); - } else if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) { - switch (voltage_swing & DP_TRAIN_VOLTAGE_SWING_MASK) { - case DP_TRAIN_VOLTAGE_SWING_LEVEL_0: - return DP_TRAIN_PRE_EMPH_LEVEL_3; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_1: - return DP_TRAIN_PRE_EMPH_LEVEL_2; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_2: - return DP_TRAIN_PRE_EMPH_LEVEL_1; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_3: - default: - return DP_TRAIN_PRE_EMPH_LEVEL_0; - } - } else if (IS_IVYBRIDGE(dev_priv) && port == PORT_A) { - switch (voltage_swing & DP_TRAIN_VOLTAGE_SWING_MASK) { - case DP_TRAIN_VOLTAGE_SWING_LEVEL_0: - return DP_TRAIN_PRE_EMPH_LEVEL_2; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_1: - case DP_TRAIN_VOLTAGE_SWING_LEVEL_2: - return DP_TRAIN_PRE_EMPH_LEVEL_1; - default: - return DP_TRAIN_PRE_EMPH_LEVEL_0; - } - } else { - switch (voltage_swing & DP_TRAIN_VOLTAGE_SWING_MASK) { - case DP_TRAIN_VOLTAGE_SWING_LEVEL_0: - return DP_TRAIN_PRE_EMPH_LEVEL_2; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_1: - return DP_TRAIN_PRE_EMPH_LEVEL_2; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_2: - return DP_TRAIN_PRE_EMPH_LEVEL_1; - case DP_TRAIN_VOLTAGE_SWING_LEVEL_3: - default: - return DP_TRAIN_PRE_EMPH_LEVEL_0; - } - } +static u8 intel_dp_pre_empemph_max_3(struct intel_dp *intel_dp) +{ + return DP_TRAIN_PRE_EMPH_LEVEL_3; } static void vlv_set_signal_levels(struct intel_dp *intel_dp) @@ -4330,6 +4299,7 @@ static u32 ivb_cpu_edp_signal_levels(u8 train_set) case DP_TRAIN_VOLTAGE_SWING_LEVEL_0 | DP_TRAIN_PRE_EMPH_LEVEL_1: return EDP_LINK_TRAIN_400MV_3_5DB_IVB; case DP_TRAIN_VOLTAGE_SWING_LEVEL_0 | DP_TRAIN_PRE_EMPH_LEVEL_2: + case DP_TRAIN_VOLTAGE_SWING_LEVEL_1 | DP_TRAIN_PRE_EMPH_LEVEL_2: return EDP_LINK_TRAIN_400MV_6DB_IVB; case DP_TRAIN_VOLTAGE_SWING_LEVEL_1 | DP_TRAIN_PRE_EMPH_LEVEL_0: @@ -4746,7 +4716,9 @@ intel_dp_sink_can_mst(struct intel_dp *intel_dp) static bool intel_dp_can_mst(struct intel_dp *intel_dp) { - return i915_modparams.enable_dp_mst && + struct drm_i915_private *i915 = dp_to_i915(intel_dp); + + return i915->params.enable_dp_mst && intel_dp->can_mst && intel_dp_sink_can_mst(intel_dp); } @@ -4763,13 +4735,13 @@ intel_dp_configure_mst(struct intel_dp *intel_dp) "[ENCODER:%d:%s] MST support: port: %s, sink: %s, modparam: %s\n", encoder->base.base.id, encoder->base.name, yesno(intel_dp->can_mst), yesno(sink_can_mst), - yesno(i915_modparams.enable_dp_mst)); + yesno(i915->params.enable_dp_mst)); if (!intel_dp->can_mst) return; intel_dp->is_mst = sink_can_mst && - i915_modparams.enable_dp_mst; + i915->params.enable_dp_mst; drm_dp_mst_topology_mgr_set_mst(&intel_dp->mst_mgr, intel_dp->is_mst); @@ -5595,35 +5567,46 @@ update_status: "Could not write test response to sink\n"); } -static int +/** + * intel_dp_check_mst_status - service any pending MST interrupts, check link status + * @intel_dp: Intel DP struct + * + * Read any pending MST interrupts, call MST core to handle these and ack the + * interrupts. Check if the main and AUX link state is ok. + * + * Returns: + * - %true if pending interrupts were serviced (or no interrupts were + * pending) w/o detecting an error condition. + * - %false if an error condition - like AUX failure or a loss of link - is + * detected, which needs servicing from the hotplug work. + */ +static bool intel_dp_check_mst_status(struct intel_dp *intel_dp) { struct drm_i915_private *i915 = dp_to_i915(intel_dp); - bool need_retrain = false; - - if (!intel_dp->is_mst) - return -EINVAL; + bool link_ok = true; - WARN_ON_ONCE(intel_dp->active_mst_links < 0); + drm_WARN_ON_ONCE(&i915->drm, intel_dp->active_mst_links < 0); for (;;) { u8 esi[DP_DPRX_ESI_LEN] = {}; - bool bret, handled; + bool handled; int retry; - bret = intel_dp_get_sink_irq_esi(intel_dp, esi); - if (!bret) { + if (!intel_dp_get_sink_irq_esi(intel_dp, esi)) { drm_dbg_kms(&i915->drm, "failed to get ESI - device may have failed\n"); - return -EINVAL; + link_ok = false; + + break; } /* check link status - esi[10] = 0x200c */ - if (intel_dp->active_mst_links > 0 && !need_retrain && + if (intel_dp->active_mst_links > 0 && link_ok && !drm_dp_channel_eq_ok(&esi[10], intel_dp->lane_count)) { drm_dbg_kms(&i915->drm, "channel EQ not ok, retraining\n"); - need_retrain = true; + link_ok = false; } drm_dbg_kms(&i915->drm, "got esi %3ph\n", esi); @@ -5643,7 +5626,7 @@ intel_dp_check_mst_status(struct intel_dp *intel_dp) } } - return need_retrain; + return link_ok; } static bool @@ -5966,7 +5949,7 @@ intel_dp_detect_dpcd(struct intel_dp *intel_dp) u8 *dpcd = intel_dp->dpcd; u8 type; - if (WARN_ON(intel_dp_is_edp(intel_dp))) + if (drm_WARN_ON(&i915->drm, intel_dp_is_edp(intel_dp))) return connector_status_connected; if (lspcon->active) @@ -6191,7 +6174,17 @@ intel_dp_detect(struct drm_connector *connector, goto out; } - if (intel_dp->reset_link_params) { + /* Read DP Sink DSC Cap DPCD regs for DP v1.4 */ + if (INTEL_GEN(dev_priv) >= 11) + intel_dp_get_dsc_sink_cap(intel_dp); + + intel_dp_configure_mst(intel_dp); + + /* + * TODO: Reset link params when switching to MST mode, until MST + * supports link training fallback params. + */ + if (intel_dp->reset_link_params || intel_dp->is_mst) { /* Initial max link lane count */ intel_dp->max_link_lane_count = intel_dp_max_common_lane_count(intel_dp); @@ -6203,12 +6196,6 @@ intel_dp_detect(struct drm_connector *connector, intel_dp_print_rates(intel_dp); - /* Read DP Sink DSC Cap DPCD regs for DP v1.4 */ - if (INTEL_GEN(dev_priv) >= 11) - intel_dp_get_dsc_sink_cap(intel_dp); - - intel_dp_configure_mst(intel_dp); - if (intel_dp->is_mst) { /* * If we are in MST mode then this connector @@ -7294,35 +7281,10 @@ intel_dp_hpd_pulse(struct intel_digital_port *intel_dig_port, bool long_hpd) } if (intel_dp->is_mst) { - switch (intel_dp_check_mst_status(intel_dp)) { - case -EINVAL: - /* - * If we were in MST mode, and device is not - * there, get out of MST mode - */ - drm_dbg_kms(&i915->drm, - "MST device may have disappeared %d vs %d\n", - intel_dp->is_mst, - intel_dp->mst_mgr.mst_state); - intel_dp->is_mst = false; - drm_dp_mst_topology_mgr_set_mst(&intel_dp->mst_mgr, - intel_dp->is_mst); - - return IRQ_NONE; - case 1: - return IRQ_NONE; - default: - break; - } - } - - if (!intel_dp->is_mst) { - bool handled; - - handled = intel_dp_short_pulse(intel_dp); - - if (!handled) + if (!intel_dp_check_mst_status(intel_dp)) return IRQ_NONE; + } else if (!intel_dp_short_pulse(intel_dp)) { + return IRQ_NONE; } return IRQ_HANDLED; @@ -8366,6 +8328,15 @@ bool intel_dp_init(struct drm_i915_private *dev_priv, else intel_dig_port->dp.set_signal_levels = g4x_set_signal_levels; + if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv) || + (HAS_PCH_SPLIT(dev_priv) && port != PORT_A)) { + intel_dig_port->dp.preemph_max = intel_dp_pre_empemph_max_3; + intel_dig_port->dp.voltage_max = intel_dp_voltage_max_3; + } else { + intel_dig_port->dp.preemph_max = intel_dp_pre_empemph_max_2; + intel_dig_port->dp.voltage_max = intel_dp_voltage_max_2; + } + intel_dig_port->dp.output_reg = output_reg; intel_dig_port->max_lanes = 4; intel_dig_port->dp.regs.dp_tp_ctl = DP_TP_CTL(port); diff --git a/drivers/gpu/drm/i915/display/intel_dp.h b/drivers/gpu/drm/i915/display/intel_dp.h index 1702959ca079..0a8950f744f6 100644 --- a/drivers/gpu/drm/i915/display/intel_dp.h +++ b/drivers/gpu/drm/i915/display/intel_dp.h @@ -92,10 +92,6 @@ intel_dp_program_link_training_pattern(struct intel_dp *intel_dp, void intel_dp_set_signal_levels(struct intel_dp *intel_dp); void intel_dp_set_idle_link_train(struct intel_dp *intel_dp); -u8 -intel_dp_voltage_max(struct intel_dp *intel_dp); -u8 -intel_dp_pre_emphasis_max(struct intel_dp *intel_dp, u8 voltage_swing); void intel_dp_compute_rate(struct intel_dp *intel_dp, int port_clock, u8 *link_bw, u8 *rate_select); bool intel_dp_source_supports_hbr2(struct intel_dp *intel_dp); diff --git a/drivers/gpu/drm/i915/display/intel_dp_aux_backlight.c b/drivers/gpu/drm/i915/display/intel_dp_aux_backlight.c index 0722540d64ad..acbd7eb66cbe 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_aux_backlight.c +++ b/drivers/gpu/drm/i915/display/intel_dp_aux_backlight.c @@ -348,7 +348,7 @@ int intel_dp_aux_init_backlight_funcs(struct intel_connector *intel_connector) struct intel_dp *intel_dp = enc_to_intel_dp(intel_connector->encoder); struct drm_i915_private *i915 = dp_to_i915(intel_dp); - if (i915_modparams.enable_dpcd_backlight == 0 || + if (i915->params.enable_dpcd_backlight == 0 || !intel_dp_aux_display_control_capable(intel_connector)) return -ENODEV; @@ -358,7 +358,7 @@ int intel_dp_aux_init_backlight_funcs(struct intel_connector *intel_connector) */ if (i915->vbt.backlight.type != INTEL_BACKLIGHT_VESA_EDP_AUX_INTERFACE && - i915_modparams.enable_dpcd_backlight != 1 && + i915->params.enable_dpcd_backlight != 1 && !drm_dp_has_quirk(&intel_dp->desc, intel_dp->edid_quirks, DP_QUIRK_FORCE_DPCD_BACKLIGHT)) { drm_info(&i915->drm, diff --git a/drivers/gpu/drm/i915/display/intel_dp_link_training.c b/drivers/gpu/drm/i915/display/intel_dp_link_training.c index e4f1843170b7..b9e4ee2dbddc 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_link_training.c +++ b/drivers/gpu/drm/i915/display/intel_dp_link_training.c @@ -34,6 +34,21 @@ intel_dp_dump_link_status(const u8 link_status[DP_LINK_STATUS_SIZE]) link_status[3], link_status[4], link_status[5]); } +static u8 dp_voltage_max(u8 preemph) +{ + switch (preemph & DP_TRAIN_PRE_EMPHASIS_MASK) { + case DP_TRAIN_PRE_EMPH_LEVEL_0: + return DP_TRAIN_VOLTAGE_SWING_LEVEL_3; + case DP_TRAIN_PRE_EMPH_LEVEL_1: + return DP_TRAIN_VOLTAGE_SWING_LEVEL_2; + case DP_TRAIN_PRE_EMPH_LEVEL_2: + return DP_TRAIN_VOLTAGE_SWING_LEVEL_1; + case DP_TRAIN_PRE_EMPH_LEVEL_3: + default: + return DP_TRAIN_VOLTAGE_SWING_LEVEL_0; + } +} + void intel_dp_get_adjust_train(struct intel_dp *intel_dp, const u8 link_status[DP_LINK_STATUS_SIZE]) { @@ -44,23 +59,19 @@ void intel_dp_get_adjust_train(struct intel_dp *intel_dp, u8 preemph_max; for (lane = 0; lane < intel_dp->lane_count; lane++) { - u8 this_v = drm_dp_get_adjust_request_voltage(link_status, lane); - u8 this_p = drm_dp_get_adjust_request_pre_emphasis(link_status, lane); - - if (this_v > v) - v = this_v; - if (this_p > p) - p = this_p; + v = max(v, drm_dp_get_adjust_request_voltage(link_status, lane)); + p = max(p, drm_dp_get_adjust_request_pre_emphasis(link_status, lane)); } - voltage_max = intel_dp_voltage_max(intel_dp); - if (v >= voltage_max) - v = voltage_max | DP_TRAIN_MAX_SWING_REACHED; - - preemph_max = intel_dp_pre_emphasis_max(intel_dp, v); + preemph_max = intel_dp->preemph_max(intel_dp); if (p >= preemph_max) p = preemph_max | DP_TRAIN_MAX_PRE_EMPHASIS_REACHED; + voltage_max = min(intel_dp->voltage_max(intel_dp), + dp_voltage_max(p)); + if (v >= voltage_max) + v = voltage_max | DP_TRAIN_MAX_SWING_REACHED; + for (lane = 0; lane < 4; lane++) intel_dp->train_set[lane] = v | p; } diff --git a/drivers/gpu/drm/i915/display/intel_dp_mst.c b/drivers/gpu/drm/i915/display/intel_dp_mst.c index 2e6c6375a23b..8273f2e07427 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_mst.c +++ b/drivers/gpu/drm/i915/display/intel_dp_mst.c @@ -317,6 +317,25 @@ intel_dp_mst_atomic_check(struct drm_connector *connector, return ret; } +static void clear_act_sent(struct intel_dp *intel_dp) +{ + struct drm_i915_private *i915 = dp_to_i915(intel_dp); + + intel_de_write(i915, intel_dp->regs.dp_tp_status, + DP_TP_STATUS_ACT_SENT); +} + +static void wait_for_act_sent(struct intel_dp *intel_dp) +{ + struct drm_i915_private *i915 = dp_to_i915(intel_dp); + + if (intel_de_wait_for_set(i915, intel_dp->regs.dp_tp_status, + DP_TP_STATUS_ACT_SENT, 1)) + drm_err(&i915->drm, "Timed out waiting for ACT sent\n"); + + drm_dp_check_act_status(&intel_dp->mst_mgr); +} + static void intel_mst_disable_dp(struct intel_atomic_state *state, struct intel_encoder *encoder, const struct intel_crtc_state *old_crtc_state, @@ -370,6 +389,8 @@ static void intel_mst_post_disable_dp(struct intel_atomic_state *state, drm_dp_update_payload_part2(&intel_dp->mst_mgr); + clear_act_sent(intel_dp); + val = intel_de_read(dev_priv, TRANS_DDI_FUNC_CTL(old_crtc_state->cpu_transcoder)); val &= ~TRANS_DDI_DP_VC_PAYLOAD_ALLOC; @@ -377,11 +398,7 @@ static void intel_mst_post_disable_dp(struct intel_atomic_state *state, TRANS_DDI_FUNC_CTL(old_crtc_state->cpu_transcoder), val); - if (intel_de_wait_for_set(dev_priv, intel_dp->regs.dp_tp_status, - DP_TP_STATUS_ACT_SENT, 1)) - drm_err(&dev_priv->drm, - "Timed out waiting for ACT sent when disabling\n"); - drm_dp_check_act_status(&intel_dp->mst_mgr); + wait_for_act_sent(intel_dp); drm_dp_mst_deallocate_vcpi(&intel_dp->mst_mgr, connector->port); @@ -452,7 +469,6 @@ static void intel_mst_pre_enable_dp(struct intel_atomic_state *state, struct intel_connector *connector = to_intel_connector(conn_state->connector); int ret; - u32 temp; bool first_mst_stream; /* MST encoders are bound to a crtc, not to a connector, @@ -485,8 +501,6 @@ static void intel_mst_pre_enable_dp(struct intel_atomic_state *state, drm_err(&dev_priv->drm, "failed to allocate vcpi\n"); intel_dp->active_mst_links++; - temp = intel_de_read(dev_priv, intel_dp->regs.dp_tp_status); - intel_de_write(dev_priv, intel_dp->regs.dp_tp_status, temp); ret = drm_dp_update_payload_part1(&intel_dp->mst_mgr); @@ -514,19 +528,25 @@ static void intel_mst_enable_dp(struct intel_atomic_state *state, struct intel_digital_port *intel_dig_port = intel_mst->primary; struct intel_dp *intel_dp = &intel_dig_port->dp; struct drm_i915_private *dev_priv = to_i915(encoder->base.dev); + u32 val; drm_WARN_ON(&dev_priv->drm, pipe_config->has_pch_encoder); + clear_act_sent(intel_dp); + intel_ddi_enable_transcoder_func(encoder, pipe_config); + val = intel_de_read(dev_priv, + TRANS_DDI_FUNC_CTL(pipe_config->cpu_transcoder)); + val |= TRANS_DDI_DP_VC_PAYLOAD_ALLOC; + intel_de_write(dev_priv, + TRANS_DDI_FUNC_CTL(pipe_config->cpu_transcoder), + val); + drm_dbg_kms(&dev_priv->drm, "active links %d\n", intel_dp->active_mst_links); - if (intel_de_wait_for_set(dev_priv, intel_dp->regs.dp_tp_status, - DP_TP_STATUS_ACT_SENT, 1)) - drm_err(&dev_priv->drm, "Timed out waiting for ACT sent\n"); - - drm_dp_check_act_status(&intel_dp->mst_mgr); + wait_for_act_sent(intel_dp); drm_dp_update_payload_part2(&intel_dp->mst_mgr); diff --git a/drivers/gpu/drm/i915/display/intel_dsb.c b/drivers/gpu/drm/i915/display/intel_dsb.c index 29fec6a92d17..24e6d63e2d47 100644 --- a/drivers/gpu/drm/i915/display/intel_dsb.c +++ b/drivers/gpu/drm/i915/display/intel_dsb.c @@ -34,152 +34,52 @@ #define DSB_BYTE_EN_SHIFT 20 #define DSB_REG_VALUE_MASK 0xfffff -static bool is_dsb_busy(struct intel_dsb *dsb) +static bool is_dsb_busy(struct drm_i915_private *i915, enum pipe pipe, + enum dsb_id id) { - struct intel_crtc *crtc = container_of(dsb, typeof(*crtc), dsb); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - enum pipe pipe = crtc->pipe; - - return DSB_STATUS & intel_de_read(dev_priv, DSB_CTRL(pipe, dsb->id)); + return DSB_STATUS & intel_de_read(i915, DSB_CTRL(pipe, id)); } -static bool intel_dsb_enable_engine(struct intel_dsb *dsb) +static bool intel_dsb_enable_engine(struct drm_i915_private *i915, + enum pipe pipe, enum dsb_id id) { - struct intel_crtc *crtc = container_of(dsb, typeof(*crtc), dsb); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - enum pipe pipe = crtc->pipe; u32 dsb_ctrl; - dsb_ctrl = intel_de_read(dev_priv, DSB_CTRL(pipe, dsb->id)); + dsb_ctrl = intel_de_read(i915, DSB_CTRL(pipe, id)); if (DSB_STATUS & dsb_ctrl) { - drm_dbg_kms(&dev_priv->drm, "DSB engine is busy.\n"); + drm_dbg_kms(&i915->drm, "DSB engine is busy.\n"); return false; } dsb_ctrl |= DSB_ENABLE; - intel_de_write(dev_priv, DSB_CTRL(pipe, dsb->id), dsb_ctrl); + intel_de_write(i915, DSB_CTRL(pipe, id), dsb_ctrl); - intel_de_posting_read(dev_priv, DSB_CTRL(pipe, dsb->id)); + intel_de_posting_read(i915, DSB_CTRL(pipe, id)); return true; } -static bool intel_dsb_disable_engine(struct intel_dsb *dsb) +static bool intel_dsb_disable_engine(struct drm_i915_private *i915, + enum pipe pipe, enum dsb_id id) { - struct intel_crtc *crtc = container_of(dsb, typeof(*crtc), dsb); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - enum pipe pipe = crtc->pipe; u32 dsb_ctrl; - dsb_ctrl = intel_de_read(dev_priv, DSB_CTRL(pipe, dsb->id)); + dsb_ctrl = intel_de_read(i915, DSB_CTRL(pipe, id)); if (DSB_STATUS & dsb_ctrl) { - drm_dbg_kms(&dev_priv->drm, "DSB engine is busy.\n"); + drm_dbg_kms(&i915->drm, "DSB engine is busy.\n"); return false; } dsb_ctrl &= ~DSB_ENABLE; - intel_de_write(dev_priv, DSB_CTRL(pipe, dsb->id), dsb_ctrl); + intel_de_write(i915, DSB_CTRL(pipe, id), dsb_ctrl); - intel_de_posting_read(dev_priv, DSB_CTRL(pipe, dsb->id)); + intel_de_posting_read(i915, DSB_CTRL(pipe, id)); return true; } /** - * intel_dsb_get() - Allocate DSB context and return a DSB instance. - * @crtc: intel_crtc structure to get pipe info. - * - * This function provides handle of a DSB instance, for the further DSB - * operations. - * - * Returns: address of Intel_dsb instance requested for. - * Failure: Returns the same DSB instance, but without a command buffer. - */ - -struct intel_dsb * -intel_dsb_get(struct intel_crtc *crtc) -{ - struct drm_device *dev = crtc->base.dev; - struct drm_i915_private *i915 = to_i915(dev); - struct intel_dsb *dsb = &crtc->dsb; - struct drm_i915_gem_object *obj; - struct i915_vma *vma; - u32 *buf; - intel_wakeref_t wakeref; - - if (!HAS_DSB(i915)) - return dsb; - - if (dsb->refcount++ != 0) - return dsb; - - wakeref = intel_runtime_pm_get(&i915->runtime_pm); - - obj = i915_gem_object_create_internal(i915, DSB_BUF_SIZE); - if (IS_ERR(obj)) { - drm_err(&i915->drm, "Gem object creation failed\n"); - goto out; - } - - vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, 0); - if (IS_ERR(vma)) { - drm_err(&i915->drm, "Vma creation failed\n"); - i915_gem_object_put(obj); - goto out; - } - - buf = i915_gem_object_pin_map(vma->obj, I915_MAP_WC); - if (IS_ERR(buf)) { - drm_err(&i915->drm, "Command buffer creation failed\n"); - goto out; - } - - dsb->id = DSB1; - dsb->vma = vma; - dsb->cmd_buf = buf; - -out: - /* - * On error dsb->cmd_buf will continue to be NULL, making the writes - * pass-through. Leave the dangling ref to be removed later by the - * corresponding intel_dsb_put(): the important error message will - * already be logged above. - */ - - intel_runtime_pm_put(&i915->runtime_pm, wakeref); - - return dsb; -} - -/** - * intel_dsb_put() - To destroy DSB context. - * @dsb: intel_dsb structure. - * - * This function destroys the DSB context allocated by a dsb_get(), by - * unpinning and releasing the VMA object associated with it. - */ - -void intel_dsb_put(struct intel_dsb *dsb) -{ - struct intel_crtc *crtc = container_of(dsb, typeof(*crtc), dsb); - struct drm_i915_private *i915 = to_i915(crtc->base.dev); - - if (!HAS_DSB(i915)) - return; - - if (drm_WARN_ON(&i915->drm, dsb->refcount == 0)) - return; - - if (--dsb->refcount == 0) { - i915_vma_unpin_and_release(&dsb->vma, I915_VMA_RELEASE_MAP); - dsb->cmd_buf = NULL; - dsb->free_pos = 0; - dsb->ins_start_offset = 0; - } -} - -/** * intel_dsb_indexed_reg_write() -Write to the DSB context for auto * increment register. - * @dsb: intel_dsb structure. + * @crtc_state: intel_crtc_state structure * @reg: register address. * @val: value. * @@ -189,19 +89,20 @@ void intel_dsb_put(struct intel_dsb *dsb) * is done through mmio write. */ -void intel_dsb_indexed_reg_write(struct intel_dsb *dsb, i915_reg_t reg, - u32 val) +void intel_dsb_indexed_reg_write(const struct intel_crtc_state *crtc_state, + i915_reg_t reg, u32 val) { - struct intel_crtc *crtc = container_of(dsb, typeof(*crtc), dsb); + struct intel_dsb *dsb = crtc_state->dsb; + struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - u32 *buf = dsb->cmd_buf; + u32 *buf; u32 reg_val; - if (!buf) { + if (!dsb) { intel_de_write(dev_priv, reg, val); return; } - + buf = dsb->cmd_buf; if (drm_WARN_ON(&dev_priv->drm, dsb->free_pos >= DSB_BUF_SIZE)) { drm_dbg_kms(&dev_priv->drm, "DSB buffer overflow\n"); return; @@ -256,7 +157,7 @@ void intel_dsb_indexed_reg_write(struct intel_dsb *dsb, i915_reg_t reg, /** * intel_dsb_reg_write() -Write to the DSB context for normal * register. - * @dsb: intel_dsb structure. + * @crtc_state: intel_crtc_state structure * @reg: register address. * @val: value. * @@ -265,17 +166,21 @@ void intel_dsb_indexed_reg_write(struct intel_dsb *dsb, i915_reg_t reg, * and rest all erroneous condition register programming is done * through mmio write. */ -void intel_dsb_reg_write(struct intel_dsb *dsb, i915_reg_t reg, u32 val) +void intel_dsb_reg_write(const struct intel_crtc_state *crtc_state, + i915_reg_t reg, u32 val) { - struct intel_crtc *crtc = container_of(dsb, typeof(*crtc), dsb); + struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - u32 *buf = dsb->cmd_buf; + struct intel_dsb *dsb; + u32 *buf; - if (!buf) { + dsb = crtc_state->dsb; + if (!dsb) { intel_de_write(dev_priv, reg, val); return; } + buf = dsb->cmd_buf; if (drm_WARN_ON(&dev_priv->drm, dsb->free_pos >= DSB_BUF_SIZE)) { drm_dbg_kms(&dev_priv->drm, "DSB buffer overflow\n"); return; @@ -290,26 +195,27 @@ void intel_dsb_reg_write(struct intel_dsb *dsb, i915_reg_t reg, u32 val) /** * intel_dsb_commit() - Trigger workload execution of DSB. - * @dsb: intel_dsb structure. + * @crtc_state: intel_crtc_state structure * * This function is used to do actual write to hardware using DSB. * On errors, fall back to MMIO. Also this function help to reset the context. */ -void intel_dsb_commit(struct intel_dsb *dsb) +void intel_dsb_commit(const struct intel_crtc_state *crtc_state) { - struct intel_crtc *crtc = container_of(dsb, typeof(*crtc), dsb); + struct intel_dsb *dsb = crtc_state->dsb; + struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); struct drm_device *dev = crtc->base.dev; struct drm_i915_private *dev_priv = to_i915(dev); enum pipe pipe = crtc->pipe; u32 tail; - if (!dsb->free_pos) + if (!(dsb && dsb->free_pos)) return; - if (!intel_dsb_enable_engine(dsb)) + if (!intel_dsb_enable_engine(dev_priv, pipe, dsb->id)) goto reset; - if (is_dsb_busy(dsb)) { + if (is_dsb_busy(dev_priv, pipe, dsb->id)) { drm_err(&dev_priv->drm, "HEAD_PTR write failed - dsb engine is busy.\n"); goto reset; @@ -322,7 +228,7 @@ void intel_dsb_commit(struct intel_dsb *dsb) memset(&dsb->cmd_buf[dsb->free_pos], 0, (tail - dsb->free_pos * 4)); - if (is_dsb_busy(dsb)) { + if (is_dsb_busy(dev_priv, pipe, dsb->id)) { drm_err(&dev_priv->drm, "TAIL_PTR write failed - dsb engine is busy.\n"); goto reset; @@ -332,7 +238,7 @@ void intel_dsb_commit(struct intel_dsb *dsb) i915_ggtt_offset(dsb->vma), tail); intel_de_write(dev_priv, DSB_TAIL(pipe, dsb->id), i915_ggtt_offset(dsb->vma) + tail); - if (wait_for(!is_dsb_busy(dsb), 1)) { + if (wait_for(!is_dsb_busy(dev_priv, pipe, dsb->id), 1)) { drm_err(&dev_priv->drm, "Timed out waiting for DSB workload completion.\n"); goto reset; @@ -341,5 +247,79 @@ void intel_dsb_commit(struct intel_dsb *dsb) reset: dsb->free_pos = 0; dsb->ins_start_offset = 0; - intel_dsb_disable_engine(dsb); + intel_dsb_disable_engine(dev_priv, pipe, dsb->id); +} + +/** + * intel_dsb_prepare() - Allocate, pin and map the DSB command buffer. + * @crtc_state: intel_crtc_state structure to prepare associated dsb instance. + * + * This function prepare the command buffer which is used to store dsb + * instructions with data. + */ +void intel_dsb_prepare(struct intel_crtc_state *crtc_state) +{ + struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); + struct intel_dsb *dsb; + struct drm_i915_gem_object *obj; + struct i915_vma *vma; + u32 *buf; + intel_wakeref_t wakeref; + + if (!HAS_DSB(i915)) + return; + + dsb = kmalloc(sizeof(*dsb), GFP_KERNEL); + + wakeref = intel_runtime_pm_get(&i915->runtime_pm); + + obj = i915_gem_object_create_internal(i915, DSB_BUF_SIZE); + if (IS_ERR(obj)) { + drm_err(&i915->drm, "Gem object creation failed\n"); + kfree(dsb); + goto out; + } + + vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, 0); + if (IS_ERR(vma)) { + drm_err(&i915->drm, "Vma creation failed\n"); + i915_gem_object_put(obj); + kfree(dsb); + goto out; + } + + buf = i915_gem_object_pin_map(vma->obj, I915_MAP_WC); + if (IS_ERR(buf)) { + drm_err(&i915->drm, "Command buffer creation failed\n"); + i915_vma_unpin_and_release(&vma, I915_VMA_RELEASE_MAP); + kfree(dsb); + goto out; + } + + dsb->id = DSB1; + dsb->vma = vma; + dsb->cmd_buf = buf; + dsb->free_pos = 0; + dsb->ins_start_offset = 0; + crtc_state->dsb = dsb; +out: + intel_runtime_pm_put(&i915->runtime_pm, wakeref); +} + +/** + * intel_dsb_cleanup() - To cleanup DSB context. + * @crtc_state: intel_crtc_state structure to cleanup associated dsb instance. + * + * This function cleanup the DSB context by unpinning and releasing + * the VMA object associated with it. + */ +void intel_dsb_cleanup(struct intel_crtc_state *crtc_state) +{ + if (!crtc_state->dsb) + return; + + i915_vma_unpin_and_release(&crtc_state->dsb->vma, I915_VMA_RELEASE_MAP); + kfree(crtc_state->dsb); + crtc_state->dsb = NULL; } diff --git a/drivers/gpu/drm/i915/display/intel_dsb.h b/drivers/gpu/drm/i915/display/intel_dsb.h index 395ef9ce558e..654a11f24b80 100644 --- a/drivers/gpu/drm/i915/display/intel_dsb.h +++ b/drivers/gpu/drm/i915/display/intel_dsb.h @@ -10,7 +10,7 @@ #include "i915_reg.h" -struct intel_crtc; +struct intel_crtc_state; struct i915_vma; enum dsb_id { @@ -22,7 +22,6 @@ enum dsb_id { }; struct intel_dsb { - long refcount; enum dsb_id id; u32 *cmd_buf; struct i915_vma *vma; @@ -41,12 +40,12 @@ struct intel_dsb { u32 ins_start_offset; }; -struct intel_dsb * -intel_dsb_get(struct intel_crtc *crtc); -void intel_dsb_put(struct intel_dsb *dsb); -void intel_dsb_reg_write(struct intel_dsb *dsb, i915_reg_t reg, u32 val); -void intel_dsb_indexed_reg_write(struct intel_dsb *dsb, i915_reg_t reg, - u32 val); -void intel_dsb_commit(struct intel_dsb *dsb); +void intel_dsb_prepare(struct intel_crtc_state *crtc_state); +void intel_dsb_cleanup(struct intel_crtc_state *crtc_state); +void intel_dsb_reg_write(const struct intel_crtc_state *crtc_state, + i915_reg_t reg, u32 val); +void intel_dsb_indexed_reg_write(const struct intel_crtc_state *crtc_state, + i915_reg_t reg, u32 val); +void intel_dsb_commit(const struct intel_crtc_state *crtc_state); #endif diff --git a/drivers/gpu/drm/i915/display/intel_fbc.c b/drivers/gpu/drm/i915/display/intel_fbc.c index 1c26673acb2d..30649e17cfb7 100644 --- a/drivers/gpu/drm/i915/display/intel_fbc.c +++ b/drivers/gpu/drm/i915/display/intel_fbc.c @@ -740,7 +740,7 @@ static bool intel_fbc_can_enable(struct drm_i915_private *dev_priv) return false; } - if (!i915_modparams.enable_fbc) { + if (!dev_priv->params.enable_fbc) { fbc->no_fbc_reason = "disabled per module param or by default"; return false; } @@ -1017,7 +1017,7 @@ static void __intel_fbc_post_update(struct intel_crtc *crtc) fbc->flip_pending = false; - if (!i915_modparams.enable_fbc) { + if (!dev_priv->params.enable_fbc) { intel_fbc_deactivate(dev_priv, "disabled at runtime per module param"); __intel_fbc_disable(dev_priv); @@ -1370,8 +1370,8 @@ void intel_fbc_handle_fifo_underrun_irq(struct drm_i915_private *dev_priv) */ static int intel_sanitize_fbc_option(struct drm_i915_private *dev_priv) { - if (i915_modparams.enable_fbc >= 0) - return !!i915_modparams.enable_fbc; + if (dev_priv->params.enable_fbc >= 0) + return !!dev_priv->params.enable_fbc; if (!HAS_FBC(dev_priv)) return 0; @@ -1415,9 +1415,9 @@ void intel_fbc_init(struct drm_i915_private *dev_priv) if (need_fbc_vtd_wa(dev_priv)) mkwrite_device_info(dev_priv)->display.has_fbc = false; - i915_modparams.enable_fbc = intel_sanitize_fbc_option(dev_priv); + dev_priv->params.enable_fbc = intel_sanitize_fbc_option(dev_priv); drm_dbg_kms(&dev_priv->drm, "Sanitized enable_fbc value: %d\n", - i915_modparams.enable_fbc); + dev_priv->params.enable_fbc); if (!HAS_FBC(dev_priv)) { fbc->no_fbc_reason = "unsupported by this chipset"; diff --git a/drivers/gpu/drm/i915/display/intel_hdcp.c b/drivers/gpu/drm/i915/display/intel_hdcp.c index 2cbc4619b4ce..815b054bb167 100644 --- a/drivers/gpu/drm/i915/display/intel_hdcp.c +++ b/drivers/gpu/drm/i915/display/intel_hdcp.c @@ -1923,8 +1923,11 @@ static bool is_hdcp2_supported(struct drm_i915_private *dev_priv) if (!IS_ENABLED(CONFIG_INTEL_MEI_HDCP)) return false; - return (INTEL_GEN(dev_priv) >= 10 || IS_GEMINILAKE(dev_priv) || - IS_KABYLAKE(dev_priv) || IS_COFFEELAKE(dev_priv)); + return (INTEL_GEN(dev_priv) >= 10 || + IS_GEMINILAKE(dev_priv) || + IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)); } void intel_hdcp_component_init(struct drm_i915_private *dev_priv) diff --git a/drivers/gpu/drm/i915/display/intel_hdmi.c b/drivers/gpu/drm/i915/display/intel_hdmi.c index 010f37240710..a31a98d26882 100644 --- a/drivers/gpu/drm/i915/display/intel_hdmi.c +++ b/drivers/gpu/drm/i915/display/intel_hdmi.c @@ -3082,6 +3082,24 @@ static u8 mcc_port_to_ddc_pin(struct drm_i915_private *dev_priv, enum port port) return ddc_pin; } +static u8 rkl_port_to_ddc_pin(struct drm_i915_private *dev_priv, enum port port) +{ + enum phy phy = intel_port_to_phy(dev_priv, port); + + WARN_ON(port == PORT_C); + + /* + * Pin mapping for RKL depends on which PCH is present. With TGP, the + * final two outputs use type-c pins, even though they're actually + * combo outputs. With CMP, the traditional DDI A-D pins are used for + * all outputs. + */ + if (INTEL_PCH_TYPE(dev_priv) >= PCH_TGP && phy >= PHY_C) + return GMBUS_PIN_9_TC1_ICP + phy - PHY_C; + + return GMBUS_PIN_1_BXT + phy; +} + static u8 g4x_port_to_ddc_pin(struct drm_i915_private *dev_priv, enum port port) { @@ -3119,7 +3137,9 @@ static u8 intel_hdmi_ddc_pin(struct intel_encoder *encoder) return ddc_pin; } - if (HAS_PCH_MCC(dev_priv)) + if (IS_ROCKETLAKE(dev_priv)) + ddc_pin = rkl_port_to_ddc_pin(dev_priv, port); + else if (HAS_PCH_MCC(dev_priv)) ddc_pin = mcc_port_to_ddc_pin(dev_priv, port); else if (INTEL_PCH_TYPE(dev_priv) >= PCH_ICP) ddc_pin = icl_port_to_ddc_pin(dev_priv, port); diff --git a/drivers/gpu/drm/i915/display/intel_hotplug.c b/drivers/gpu/drm/i915/display/intel_hotplug.c index 664f88354101..2e94c1413c02 100644 --- a/drivers/gpu/drm/i915/display/intel_hotplug.c +++ b/drivers/gpu/drm/i915/display/intel_hotplug.c @@ -89,6 +89,15 @@ enum hpd_pin intel_hpd_pin_default(struct drm_i915_private *dev_priv, { enum phy phy = intel_port_to_phy(dev_priv, port); + /* + * RKL + TGP PCH is a special case; we effectively choose the hpd_pin + * based on the DDI rather than the PHY (i.e., the last two outputs + * shold be HPD_PORT_{D,E} rather than {C,D}. Note that this differs + * from the behavior of both TGL+TGP and RKL+CMP. + */ + if (IS_ROCKETLAKE(dev_priv) && HAS_PCH_TGP(dev_priv)) + return HPD_PORT_A + port - PORT_A; + switch (phy) { case PHY_F: return IS_CNL_WITH_PORT_F(dev_priv) ? HPD_PORT_E : HPD_PORT_F; diff --git a/drivers/gpu/drm/i915/display/intel_lvds.c b/drivers/gpu/drm/i915/display/intel_lvds.c index 872f2a489339..1888611244db 100644 --- a/drivers/gpu/drm/i915/display/intel_lvds.c +++ b/drivers/gpu/drm/i915/display/intel_lvds.c @@ -784,8 +784,8 @@ static bool compute_is_dual_link_lvds(struct intel_lvds_encoder *lvds_encoder) struct drm_i915_private *dev_priv = to_i915(dev); /* use the module option value if specified */ - if (i915_modparams.lvds_channel_mode > 0) - return i915_modparams.lvds_channel_mode == 2; + if (dev_priv->params.lvds_channel_mode > 0) + return dev_priv->params.lvds_channel_mode == 2; /* single channel LVDS is limited to 112 MHz */ if (lvds_encoder->attached_connector->panel.fixed_mode->clock > 112999) diff --git a/drivers/gpu/drm/i915/display/intel_opregion.c b/drivers/gpu/drm/i915/display/intel_opregion.c index cc6b00959586..de995362f428 100644 --- a/drivers/gpu/drm/i915/display/intel_opregion.c +++ b/drivers/gpu/drm/i915/display/intel_opregion.c @@ -801,7 +801,7 @@ static int intel_load_vbt_firmware(struct drm_i915_private *dev_priv) { struct intel_opregion *opregion = &dev_priv->opregion; const struct firmware *fw = NULL; - const char *name = i915_modparams.vbt_firmware; + const char *name = dev_priv->params.vbt_firmware; int ret; if (!name || !*name) diff --git a/drivers/gpu/drm/i915/display/intel_overlay.c b/drivers/gpu/drm/i915/display/intel_overlay.c index 66711e62fa71..52b4f6193b4c 100644 --- a/drivers/gpu/drm/i915/display/intel_overlay.c +++ b/drivers/gpu/drm/i915/display/intel_overlay.c @@ -100,12 +100,15 @@ #define CLK_RGB24_MASK 0x0 #define CLK_RGB16_MASK 0x070307 #define CLK_RGB15_MASK 0x070707 -#define CLK_RGB8I_MASK 0xffffff +#define RGB30_TO_COLORKEY(c) \ + ((((c) & 0x3fc00000) >> 6) | (((c) & 0x000ff000) >> 4) | (((c) & 0x000003fc) >> 2)) #define RGB16_TO_COLORKEY(c) \ - (((c & 0xF800) << 8) | ((c & 0x07E0) << 5) | ((c & 0x001F) << 3)) + ((((c) & 0xf800) << 8) | (((c) & 0x07e0) << 5) | (((c) & 0x001f) << 3)) #define RGB15_TO_COLORKEY(c) \ - (((c & 0x7c00) << 9) | ((c & 0x03E0) << 6) | ((c & 0x001F) << 3)) + ((((c) & 0x7c00) << 9) | (((c) & 0x03e0) << 6) | (((c) & 0x001f) << 3)) +#define RGB8I_TO_COLORKEY(c) \ + ((((c) & 0xff) << 16) | (((c) & 0xff) << 8) | (((c) & 0xff) << 0)) /* overlay flip addr flag */ #define OFC_UPDATE 0x1 @@ -682,8 +685,8 @@ static void update_colorkey(struct intel_overlay *overlay, switch (format) { case DRM_FORMAT_C8: - key = 0; - flags |= CLK_RGB8I_MASK; + key = RGB8I_TO_COLORKEY(key); + flags |= CLK_RGB24_MASK; break; case DRM_FORMAT_XRGB1555: key = RGB15_TO_COLORKEY(key); @@ -693,6 +696,11 @@ static void update_colorkey(struct intel_overlay *overlay, key = RGB16_TO_COLORKEY(key); flags |= CLK_RGB16_MASK; break; + case DRM_FORMAT_XRGB2101010: + case DRM_FORMAT_XBGR2101010: + key = RGB30_TO_COLORKEY(key); + flags |= CLK_RGB24_MASK; + break; default: flags |= CLK_RGB24_MASK; break; @@ -777,9 +785,15 @@ static int intel_overlay_do_put_image(struct intel_overlay *overlay, i915_gem_object_flush_frontbuffer(new_bo, ORIGIN_DIRTYFB); if (!overlay->active) { - u32 oconfig; - - oconfig = OCONF_CC_OUT_8BIT; + const struct intel_crtc_state *crtc_state = + overlay->crtc->config; + u32 oconfig = 0; + + if (crtc_state->gamma_enable && + crtc_state->gamma_mode == GAMMA_MODE_MODE_8BIT) + oconfig |= OCONF_CC_OUT_8BIT; + if (crtc_state->gamma_enable) + oconfig |= OCONF_GAMMA2_ENABLE; if (IS_GEN(dev_priv, 4)) oconfig |= OCONF_CSC_MODE_BT709; oconfig |= pipe == 0 ? diff --git a/drivers/gpu/drm/i915/display/intel_panel.c b/drivers/gpu/drm/i915/display/intel_panel.c index 3c5056dbf607..aaed9eb3b56c 100644 --- a/drivers/gpu/drm/i915/display/intel_panel.c +++ b/drivers/gpu/drm/i915/display/intel_panel.c @@ -521,10 +521,10 @@ static u32 intel_panel_compute_brightness(struct intel_connector *connector, drm_WARN_ON(&dev_priv->drm, panel->backlight.max == 0); - if (i915_modparams.invert_brightness < 0) + if (dev_priv->params.invert_brightness < 0) return val; - if (i915_modparams.invert_brightness > 0 || + if (dev_priv->params.invert_brightness > 0 || dev_priv->quirks & QUIRK_INVERT_BRIGHTNESS) { return panel->backlight.max - val + panel->backlight.min; } diff --git a/drivers/gpu/drm/i915/display/intel_psr.c b/drivers/gpu/drm/i915/display/intel_psr.c index b7a2c102648a..86bf7a76f93d 100644 --- a/drivers/gpu/drm/i915/display/intel_psr.c +++ b/drivers/gpu/drm/i915/display/intel_psr.c @@ -83,7 +83,7 @@ static bool psr_global_enabled(struct drm_i915_private *i915) { switch (i915->psr.debug & I915_PSR_DEBUG_MODE_MASK) { case I915_PSR_DEBUG_DEFAULT: - return i915_modparams.enable_psr; + return i915->params.enable_psr; case I915_PSR_DEBUG_DISABLE: return false; default: @@ -426,6 +426,12 @@ static u32 intel_psr1_get_tp_time(struct intel_dp *intel_dp) if (INTEL_GEN(dev_priv) >= 11) val |= EDP_PSR_TP4_TIME_0US; + if (dev_priv->params.psr_safest_params) { + val |= EDP_PSR_TP1_TIME_2500us; + val |= EDP_PSR_TP2_TP3_TIME_2500us; + goto check_tp3_sel; + } + if (dev_priv->vbt.psr.tp1_wakeup_time_us == 0) val |= EDP_PSR_TP1_TIME_0us; else if (dev_priv->vbt.psr.tp1_wakeup_time_us <= 100) @@ -444,6 +450,7 @@ static u32 intel_psr1_get_tp_time(struct intel_dp *intel_dp) else val |= EDP_PSR_TP2_TP3_TIME_2500us; +check_tp3_sel: if (intel_dp_source_supports_hbr2(intel_dp) && drm_dp_tps3_supported(intel_dp->dpcd)) val |= EDP_PSR_TP1_TP3_SEL; @@ -495,18 +502,13 @@ static void hsw_activate_psr1(struct intel_dp *intel_dp) intel_de_write(dev_priv, EDP_PSR_CTL(dev_priv->psr.transcoder), val); } -static void hsw_activate_psr2(struct intel_dp *intel_dp) +static u32 intel_psr2_get_tp_time(struct intel_dp *intel_dp) { struct drm_i915_private *dev_priv = dp_to_i915(intel_dp); - u32 val; - - val = psr_compute_idle_frames(intel_dp) << EDP_PSR2_IDLE_FRAME_SHIFT; - - val |= EDP_PSR2_ENABLE | EDP_SU_TRACK_ENABLE; - if (INTEL_GEN(dev_priv) >= 10 || IS_GEMINILAKE(dev_priv)) - val |= EDP_Y_COORDINATE_ENABLE; + u32 val = 0; - val |= EDP_PSR2_FRAME_BEFORE_SU(dev_priv->psr.sink_sync_latency + 1); + if (dev_priv->params.psr_safest_params) + return EDP_PSR2_TP2_TIME_2500us; if (dev_priv->vbt.psr.psr2_tp2_tp3_wakeup_time_us >= 0 && dev_priv->vbt.psr.psr2_tp2_tp3_wakeup_time_us <= 50) @@ -518,6 +520,39 @@ static void hsw_activate_psr2(struct intel_dp *intel_dp) else val |= EDP_PSR2_TP2_TIME_2500us; + return val; +} + +static void hsw_activate_psr2(struct intel_dp *intel_dp) +{ + struct drm_i915_private *dev_priv = dp_to_i915(intel_dp); + u32 val; + + val = psr_compute_idle_frames(intel_dp) << EDP_PSR2_IDLE_FRAME_SHIFT; + + val |= EDP_PSR2_ENABLE | EDP_SU_TRACK_ENABLE; + if (INTEL_GEN(dev_priv) >= 10 || IS_GEMINILAKE(dev_priv)) + val |= EDP_Y_COORDINATE_ENABLE; + + val |= EDP_PSR2_FRAME_BEFORE_SU(dev_priv->psr.sink_sync_latency + 1); + val |= intel_psr2_get_tp_time(intel_dp); + + if (INTEL_GEN(dev_priv) >= 12) { + /* + * TODO: 7 lines of IO_BUFFER_WAKE and FAST_WAKE are default + * values from BSpec. In order to setting an optimal power + * consumption, lower than 4k resoluition mode needs to decrese + * IO_BUFFER_WAKE and FAST_WAKE. And higher than 4K resolution + * mode needs to increase IO_BUFFER_WAKE and FAST_WAKE. + */ + val |= TGL_EDP_PSR2_BLOCK_COUNT_NUM_2; + val |= TGL_EDP_PSR2_IO_BUFFER_WAKE(7); + val |= TGL_EDP_PSR2_FAST_WAKE(7); + } else if (INTEL_GEN(dev_priv) >= 9) { + val |= EDP_PSR2_IO_BUFFER_WAKE(7); + val |= EDP_PSR2_FAST_WAKE(7); + } + /* * PSR2 HW is incorrectly using EDP_PSR_TP1_TP3_SEL and BSpec is * recommending keep this bit unset while PSR2 is enabled. @@ -647,6 +682,21 @@ static bool intel_psr2_config_valid(struct intel_dp *intel_dp, } /* + * Some platforms lack PSR2 HW tracking and instead require manual + * tracking by software. In this case, the driver is required to track + * the areas that need updates and program hardware to send selective + * updates. + * + * So until the software tracking is implemented, PSR2 needs to be + * disabled for platforms without PSR2 HW tracking. + */ + if (!HAS_PSR_HW_TRACKING(dev_priv)) { + drm_dbg_kms(&dev_priv->drm, + "No PSR2 HW tracking in the platform\n"); + return false; + } + + /* * DSC and PSR2 cannot be enabled simultaneously. If a requested * resolution requires DSC to be enabled, priority is given to DSC * over PSR2. @@ -1450,9 +1500,9 @@ void intel_psr_init(struct drm_i915_private *dev_priv) */ dev_priv->hsw_psr_mmio_adjust = _SRD_CTL_EDP - _HSW_EDP_PSR_BASE; - if (i915_modparams.enable_psr == -1) + if (dev_priv->params.enable_psr == -1) if (INTEL_GEN(dev_priv) < 9 || !dev_priv->vbt.psr.enable) - i915_modparams.enable_psr = 0; + dev_priv->params.enable_psr = 0; /* Set link_standby x link_off defaults */ if (IS_HASWELL(dev_priv) || IS_BROADWELL(dev_priv)) diff --git a/drivers/gpu/drm/i915/display/intel_sdvo.c b/drivers/gpu/drm/i915/display/intel_sdvo.c index bc6c26818e15..773523dcd107 100644 --- a/drivers/gpu/drm/i915/display/intel_sdvo.c +++ b/drivers/gpu/drm/i915/display/intel_sdvo.c @@ -411,6 +411,7 @@ static const char *sdvo_cmd_name(u8 cmd) static void intel_sdvo_debug_write(struct intel_sdvo *intel_sdvo, u8 cmd, const void *args, int args_len) { + struct drm_i915_private *dev_priv = to_i915(intel_sdvo->base.base.dev); const char *cmd_name; int i, pos = 0; char buffer[64]; @@ -431,7 +432,7 @@ static void intel_sdvo_debug_write(struct intel_sdvo *intel_sdvo, u8 cmd, else BUF_PRINT("(%02X)", cmd); - WARN_ON(pos >= sizeof(buffer) - 1); + drm_WARN_ON(&dev_priv->drm, pos >= sizeof(buffer) - 1); #undef BUF_PRINT DRM_DEBUG_KMS("%s: W: %02X %s\n", SDVO_NAME(intel_sdvo), cmd, buffer); @@ -533,6 +534,7 @@ static bool intel_sdvo_write_cmd(struct intel_sdvo *intel_sdvo, u8 cmd, static bool intel_sdvo_read_response(struct intel_sdvo *intel_sdvo, void *response, int response_len) { + struct drm_i915_private *dev_priv = to_i915(intel_sdvo->base.base.dev); const char *cmd_status; u8 retry = 15; /* 5 quick checks, followed by 10 long checks */ u8 status; @@ -597,7 +599,7 @@ static bool intel_sdvo_read_response(struct intel_sdvo *intel_sdvo, BUF_PRINT(" %02X", ((u8 *)response)[i]); } - WARN_ON(pos >= sizeof(buffer) - 1); + drm_WARN_ON(&dev_priv->drm, pos >= sizeof(buffer) - 1); #undef BUF_PRINT DRM_DEBUG_KMS("%s: R: %s\n", SDVO_NAME(intel_sdvo), buffer); @@ -1081,6 +1083,7 @@ static bool intel_sdvo_compute_avi_infoframe(struct intel_sdvo *intel_sdvo, struct intel_crtc_state *crtc_state, struct drm_connector_state *conn_state) { + struct drm_i915_private *dev_priv = to_i915(intel_sdvo->base.base.dev); struct hdmi_avi_infoframe *frame = &crtc_state->infoframes.avi.avi; const struct drm_display_mode *adjusted_mode = &crtc_state->hw.adjusted_mode; @@ -1106,7 +1109,7 @@ static bool intel_sdvo_compute_avi_infoframe(struct intel_sdvo *intel_sdvo, HDMI_QUANTIZATION_RANGE_FULL); ret = hdmi_avi_infoframe_check(frame); - if (WARN_ON(ret)) + if (drm_WARN_ON(&dev_priv->drm, ret)) return false; return true; @@ -1115,6 +1118,7 @@ static bool intel_sdvo_compute_avi_infoframe(struct intel_sdvo *intel_sdvo, static bool intel_sdvo_set_avi_infoframe(struct intel_sdvo *intel_sdvo, const struct intel_crtc_state *crtc_state) { + struct drm_i915_private *dev_priv = to_i915(intel_sdvo->base.base.dev); u8 sdvo_data[HDMI_INFOFRAME_SIZE(AVI)]; const union hdmi_infoframe *frame = &crtc_state->infoframes.avi; ssize_t len; @@ -1123,11 +1127,12 @@ static bool intel_sdvo_set_avi_infoframe(struct intel_sdvo *intel_sdvo, intel_hdmi_infoframe_enable(HDMI_INFOFRAME_TYPE_AVI)) == 0) return true; - if (WARN_ON(frame->any.type != HDMI_INFOFRAME_TYPE_AVI)) + if (drm_WARN_ON(&dev_priv->drm, + frame->any.type != HDMI_INFOFRAME_TYPE_AVI)) return false; len = hdmi_infoframe_pack_only(frame, sdvo_data, sizeof(sdvo_data)); - if (WARN_ON(len < 0)) + if (drm_WARN_ON(&dev_priv->drm, len < 0)) return false; return intel_sdvo_write_infoframe(intel_sdvo, SDVO_HBUF_INDEX_AVI_IF, @@ -1237,6 +1242,7 @@ intel_sdvo_get_preferred_input_mode(struct intel_sdvo *intel_sdvo, static void i9xx_adjust_sdvo_tv_clock(struct intel_crtc_state *pipe_config) { + struct drm_i915_private *dev_priv = to_i915(pipe_config->uapi.crtc->dev); unsigned dotclock = pipe_config->port_clock; struct dpll *clock = &pipe_config->dpll; @@ -1257,7 +1263,8 @@ static void i9xx_adjust_sdvo_tv_clock(struct intel_crtc_state *pipe_config) clock->m1 = 12; clock->m2 = 8; } else { - WARN(1, "SDVO TV clock out of range: %i\n", dotclock); + drm_WARN(&dev_priv->drm, 1, + "SDVO TV clock out of range: %i\n", dotclock); } pipe_config->clock_set = true; @@ -2293,7 +2300,7 @@ intel_sdvo_connector_atomic_get_property(struct drm_connector *connector, return 0; } - WARN_ON(1); + drm_WARN_ON(connector->dev, 1); *val = 0; } else if (property == intel_sdvo_connector->top || property == intel_sdvo_connector->bottom) diff --git a/drivers/gpu/drm/i915/display/intel_sprite.c b/drivers/gpu/drm/i915/display/intel_sprite.c index 0000ec7055f7..3cd461bf9131 100644 --- a/drivers/gpu/drm/i915/display/intel_sprite.c +++ b/drivers/gpu/drm/i915/display/intel_sprite.c @@ -333,6 +333,21 @@ int intel_plane_check_src_coordinates(struct intel_plane_state *plane_state) return 0; } +static u8 icl_nv12_y_plane_mask(struct drm_i915_private *i915) +{ + if (IS_ROCKETLAKE(i915)) + return BIT(PLANE_SPRITE2) | BIT(PLANE_SPRITE3); + else + return BIT(PLANE_SPRITE4) | BIT(PLANE_SPRITE5); +} + +bool icl_is_nv12_y_plane(struct drm_i915_private *dev_priv, + enum plane_id plane_id) +{ + return INTEL_GEN(dev_priv) >= 11 && + icl_nv12_y_plane_mask(dev_priv) & BIT(plane_id); +} + bool icl_is_hdr_plane(struct drm_i915_private *dev_priv, enum plane_id plane_id) { return INTEL_GEN(dev_priv) >= 11 && @@ -3003,7 +3018,7 @@ static const u32 *icl_get_plane_formats(struct drm_i915_private *dev_priv, if (icl_is_hdr_plane(dev_priv, plane_id)) { *num_formats = ARRAY_SIZE(icl_hdr_plane_formats); return icl_hdr_plane_formats; - } else if (icl_is_nv12_y_plane(plane_id)) { + } else if (icl_is_nv12_y_plane(dev_priv, plane_id)) { *num_formats = ARRAY_SIZE(icl_sdr_y_plane_formats); return icl_sdr_y_plane_formats; } else { @@ -3046,6 +3061,7 @@ skl_universal_plane_create(struct drm_i915_private *dev_priv, struct intel_plane *plane; enum drm_plane_type plane_type; unsigned int supported_rotations; + unsigned int supported_csc; const u64 *modifiers; const u32 *formats; int num_formats; @@ -3120,9 +3136,13 @@ skl_universal_plane_create(struct drm_i915_private *dev_priv, DRM_MODE_ROTATE_0, supported_rotations); + supported_csc = BIT(DRM_COLOR_YCBCR_BT601) | BIT(DRM_COLOR_YCBCR_BT709); + + if (INTEL_GEN(dev_priv) >= 10 || IS_GEMINILAKE(dev_priv)) + supported_csc |= BIT(DRM_COLOR_YCBCR_BT2020); + drm_plane_create_color_properties(&plane->base, - BIT(DRM_COLOR_YCBCR_BT601) | - BIT(DRM_COLOR_YCBCR_BT709), + supported_csc, BIT(DRM_COLOR_YCBCR_LIMITED_RANGE) | BIT(DRM_COLOR_YCBCR_FULL_RANGE), DRM_COLOR_YCBCR_BT709, diff --git a/drivers/gpu/drm/i915/display/intel_sprite.h b/drivers/gpu/drm/i915/display/intel_sprite.h index 5eeaa92420d1..cd2104ba1ca1 100644 --- a/drivers/gpu/drm/i915/display/intel_sprite.h +++ b/drivers/gpu/drm/i915/display/intel_sprite.h @@ -32,21 +32,14 @@ struct intel_plane * skl_universal_plane_create(struct drm_i915_private *dev_priv, enum pipe pipe, enum plane_id plane_id); -static inline bool icl_is_nv12_y_plane(enum plane_id id) -{ - /* Don't need to do a gen check, these planes are only available on gen11 */ - if (id == PLANE_SPRITE4 || id == PLANE_SPRITE5) - return true; - - return false; -} - static inline u8 icl_hdr_plane_mask(void) { return BIT(PLANE_PRIMARY) | BIT(PLANE_SPRITE0) | BIT(PLANE_SPRITE1); } +bool icl_is_nv12_y_plane(struct drm_i915_private *dev_priv, + enum plane_id plane_id); bool icl_is_hdr_plane(struct drm_i915_private *dev_priv, enum plane_id plane_id); int ivb_plane_min_cdclk(const struct intel_crtc_state *crtc_state, diff --git a/drivers/gpu/drm/i915/display/intel_tc.c b/drivers/gpu/drm/i915/display/intel_tc.c index b161c15baf86..5b5dc86a5737 100644 --- a/drivers/gpu/drm/i915/display/intel_tc.c +++ b/drivers/gpu/drm/i915/display/intel_tc.c @@ -360,12 +360,12 @@ static void icl_tc_phy_connect(struct intel_digital_port *dig_port, } if (!icl_tc_phy_set_safe_mode(dig_port, false) && - !WARN_ON(dig_port->tc_legacy_port)) + !drm_WARN_ON(&i915->drm, dig_port->tc_legacy_port)) goto out_set_tbt_alt_mode; max_lanes = intel_tc_port_fia_max_lane_count(dig_port); if (dig_port->tc_legacy_port) { - WARN_ON(max_lanes != 4); + drm_WARN_ON(&i915->drm, max_lanes != 4); dig_port->tc_mode = TC_PORT_LEGACY; return; @@ -445,18 +445,20 @@ static bool icl_tc_phy_is_connected(struct intel_digital_port *dig_port) static enum tc_port_mode intel_tc_port_get_current_mode(struct intel_digital_port *dig_port) { + struct drm_i915_private *i915 = to_i915(dig_port->base.base.dev); u32 live_status_mask = tc_port_live_status_mask(dig_port); bool in_safe_mode = icl_tc_phy_is_in_safe_mode(dig_port); enum tc_port_mode mode; - if (in_safe_mode || WARN_ON(!icl_tc_phy_status_complete(dig_port))) + if (in_safe_mode || + drm_WARN_ON(&i915->drm, !icl_tc_phy_status_complete(dig_port))) return TC_PORT_TBT_ALT; mode = dig_port->tc_legacy_port ? TC_PORT_LEGACY : TC_PORT_DP_ALT; if (live_status_mask) { enum tc_port_mode live_mode = fls(live_status_mask) - 1; - if (!WARN_ON(live_mode == TC_PORT_TBT_ALT)) + if (!drm_WARN_ON(&i915->drm, live_mode == TC_PORT_TBT_ALT)) mode = live_mode; } @@ -505,7 +507,9 @@ static void intel_tc_port_link_init_refcount(struct intel_digital_port *dig_port, int refcount) { - WARN_ON(dig_port->tc_link_refcount); + struct drm_i915_private *i915 = to_i915(dig_port->base.base.dev); + + drm_WARN_ON(&i915->drm, dig_port->tc_link_refcount); dig_port->tc_link_refcount = refcount; } diff --git a/drivers/gpu/drm/i915/display/intel_tv.c b/drivers/gpu/drm/i915/display/intel_tv.c index abc67207f2f3..777032d9697b 100644 --- a/drivers/gpu/drm/i915/display/intel_tv.c +++ b/drivers/gpu/drm/i915/display/intel_tv.c @@ -1158,7 +1158,7 @@ intel_tv_get_config(struct intel_encoder *encoder, /* pixel counter doesn't work on i965gm TV output */ if (IS_I965GM(dev_priv)) - adjusted_mode->private_flags |= + pipe_config->mode_flags |= I915_MODE_FLAG_USE_SCANLINE_COUNTER; } @@ -1328,7 +1328,7 @@ intel_tv_compute_config(struct intel_encoder *encoder, /* pixel counter doesn't work on i965gm TV output */ if (IS_I965GM(dev_priv)) - adjusted_mode->private_flags |= + pipe_config->mode_flags |= I915_MODE_FLAG_USE_SCANLINE_COUNTER; return 0; diff --git a/drivers/gpu/drm/i915/display/intel_vdsc.c b/drivers/gpu/drm/i915/display/intel_vdsc.c index 95ad87d4ccb3..d145fe2bed81 100644 --- a/drivers/gpu/drm/i915/display/intel_vdsc.c +++ b/drivers/gpu/drm/i915/display/intel_vdsc.c @@ -476,13 +476,13 @@ intel_dsc_power_domain(const struct intel_crtc_state *crtc_state) * POWER_DOMAIN_TRANSCODER_VDSC_PW2 power domain in two cases: * * - ICL eDP/DSI transcoder - * - TGL pipe A + * - Gen12+ (except RKL) pipe A * * For any other pipe, VDSC/joining uses the power well associated with * the pipe in use. Hence another reference on the pipe power domain * will suffice. (Except no VDSC/joining on ICL pipe A.) */ - if (INTEL_GEN(i915) >= 12 && pipe == PIPE_A) + if (INTEL_GEN(i915) >= 12 && !IS_ROCKETLAKE(i915) && pipe == PIPE_A) return POWER_DOMAIN_TRANSCODER_VDSC_PW2; else if (is_pipe_dsc(crtc_state)) return POWER_DOMAIN_PIPE(pipe); diff --git a/drivers/gpu/drm/i915/display/vlv_dsi.c b/drivers/gpu/drm/i915/display/vlv_dsi.c index f582ab52f0b0..052e0b31a2da 100644 --- a/drivers/gpu/drm/i915/display/vlv_dsi.c +++ b/drivers/gpu/drm/i915/display/vlv_dsi.c @@ -298,7 +298,7 @@ static int intel_dsi_compute_config(struct intel_encoder *encoder, if (IS_GEN9_LP(dev_priv)) { /* Enable Frame time stamp based scanline reporting */ - adjusted_mode->private_flags |= + pipe_config->mode_flags |= I915_MODE_FLAG_GET_SCANLINE_FROM_TIMESTAMP; /* Dual link goes to DSI transcoder A. */ @@ -1097,8 +1097,8 @@ static void bxt_dsi_get_pipe_config(struct intel_encoder *encoder, pipe_config->pipe_bpp = bdw_get_pipemisc_bpp(crtc); /* Enable Frame time stamo based scanline reporting */ - adjusted_mode->private_flags |= - I915_MODE_FLAG_GET_SCANLINE_FROM_TIMESTAMP; + pipe_config->mode_flags |= + I915_MODE_FLAG_GET_SCANLINE_FROM_TIMESTAMP; /* In terms of pixels */ adjusted_mode->crtc_hdisplay = diff --git a/drivers/gpu/drm/i915/gem/i915_gem_context.c b/drivers/gpu/drm/i915/gem/i915_gem_context.c index 30c229fcb404..5c13809dc3c8 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_context.c @@ -650,7 +650,7 @@ static void context_close(struct i915_gem_context *ctx) * context close. */ if (!i915_gem_context_is_persistent(ctx) || - !i915_modparams.enable_hangcheck) + !ctx->i915->params.enable_hangcheck) kill_context(ctx); i915_gem_context_put(ctx); @@ -667,7 +667,7 @@ static int __context_set_persistence(struct i915_gem_context *ctx, bool state) * reset] are allowed to survive past termination. We require * hangcheck to ensure that the persistent requests are healthy. */ - if (!i915_modparams.enable_hangcheck) + if (!ctx->i915->params.enable_hangcheck) return -EINVAL; i915_gem_context_set_persistence(ctx); diff --git a/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c b/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c index 7db5a793739d..2679380159fc 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c @@ -217,6 +217,7 @@ static void i915_gem_object_put_pages_dmabuf(struct drm_i915_gem_object *obj, } static const struct drm_i915_gem_object_ops i915_gem_object_dmabuf_ops = { + .name = "i915_gem_object_dmabuf", .get_pages = i915_gem_object_get_pages_dmabuf, .put_pages = i915_gem_object_put_pages_dmabuf, }; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index db8eb1c6afe9..c38ab51e82f0 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -45,13 +45,6 @@ struct eb_vma_array { struct eb_vma vma[]; }; -enum { - FORCE_CPU_RELOC = 1, - FORCE_GTT_RELOC, - FORCE_GPU_RELOC, -#define DBG_FORCE_RELOC 0 /* choose one of the above! */ -}; - #define __EXEC_OBJECT_HAS_PIN BIT(31) #define __EXEC_OBJECT_HAS_FENCE BIT(30) #define __EXEC_OBJECT_NEEDS_MAP BIT(29) @@ -260,8 +253,6 @@ struct i915_execbuffer { */ struct reloc_cache { struct drm_mm_node node; /** temporary GTT binding */ - unsigned long vaddr; /** Current kmap address */ - unsigned long page; /** Currently mapped page index */ unsigned int gen; /** Cached value of INTEL_GEN */ bool use_64bit_reloc : 1; bool has_llc : 1; @@ -605,23 +596,6 @@ eb_add_vma(struct i915_execbuffer *eb, } } -static inline int use_cpu_reloc(const struct reloc_cache *cache, - const struct drm_i915_gem_object *obj) -{ - if (!i915_gem_object_has_struct_page(obj)) - return false; - - if (DBG_FORCE_RELOC == FORCE_CPU_RELOC) - return true; - - if (DBG_FORCE_RELOC == FORCE_GTT_RELOC) - return false; - - return (cache->has_llc || - obj->cache_dirty || - obj->cache_level != I915_CACHE_NONE); -} - static int eb_reserve_vma(const struct i915_execbuffer *eb, struct eb_vma *ev, u64 pin_flags) @@ -945,8 +919,6 @@ relocation_target(const struct drm_i915_gem_relocation_entry *reloc, static void reloc_cache_init(struct reloc_cache *cache, struct drm_i915_private *i915) { - cache->page = -1; - cache->vaddr = 0; /* Must be a variable in the struct to allow GCC to unroll. */ cache->gen = INTEL_GEN(i915); cache->has_llc = HAS_LLC(i915); @@ -958,25 +930,6 @@ static void reloc_cache_init(struct reloc_cache *cache, cache->target = NULL; } -static inline void *unmask_page(unsigned long p) -{ - return (void *)(uintptr_t)(p & PAGE_MASK); -} - -static inline unsigned int unmask_flags(unsigned long p) -{ - return p & ~PAGE_MASK; -} - -#define KMAP 0x4 /* after CLFLUSH_FLAGS */ - -static inline struct i915_ggtt *cache_to_ggtt(struct reloc_cache *cache) -{ - struct drm_i915_private *i915 = - container_of(cache, struct i915_execbuffer, reloc_cache)->i915; - return &i915->ggtt; -} - #define RELOC_TAIL 4 static int reloc_gpu_chain(struct reloc_cache *cache) @@ -1089,181 +1042,6 @@ static int reloc_gpu_flush(struct reloc_cache *cache) return err; } -static void reloc_cache_reset(struct reloc_cache *cache) -{ - void *vaddr; - - if (!cache->vaddr) - return; - - vaddr = unmask_page(cache->vaddr); - if (cache->vaddr & KMAP) { - if (cache->vaddr & CLFLUSH_AFTER) - mb(); - - kunmap_atomic(vaddr); - i915_gem_object_finish_access((struct drm_i915_gem_object *)cache->node.mm); - } else { - struct i915_ggtt *ggtt = cache_to_ggtt(cache); - - intel_gt_flush_ggtt_writes(ggtt->vm.gt); - io_mapping_unmap_atomic((void __iomem *)vaddr); - - if (drm_mm_node_allocated(&cache->node)) { - ggtt->vm.clear_range(&ggtt->vm, - cache->node.start, - cache->node.size); - mutex_lock(&ggtt->vm.mutex); - drm_mm_remove_node(&cache->node); - mutex_unlock(&ggtt->vm.mutex); - } else { - i915_vma_unpin((struct i915_vma *)cache->node.mm); - } - } - - cache->vaddr = 0; - cache->page = -1; -} - -static void *reloc_kmap(struct drm_i915_gem_object *obj, - struct reloc_cache *cache, - unsigned long page) -{ - void *vaddr; - - if (cache->vaddr) { - kunmap_atomic(unmask_page(cache->vaddr)); - } else { - unsigned int flushes; - int err; - - err = i915_gem_object_prepare_write(obj, &flushes); - if (err) - return ERR_PTR(err); - - BUILD_BUG_ON(KMAP & CLFLUSH_FLAGS); - BUILD_BUG_ON((KMAP | CLFLUSH_FLAGS) & PAGE_MASK); - - cache->vaddr = flushes | KMAP; - cache->node.mm = (void *)obj; - if (flushes) - mb(); - } - - vaddr = kmap_atomic(i915_gem_object_get_dirty_page(obj, page)); - cache->vaddr = unmask_flags(cache->vaddr) | (unsigned long)vaddr; - cache->page = page; - - return vaddr; -} - -static void *reloc_iomap(struct drm_i915_gem_object *obj, - struct reloc_cache *cache, - unsigned long page) -{ - struct i915_ggtt *ggtt = cache_to_ggtt(cache); - unsigned long offset; - void *vaddr; - - if (cache->vaddr) { - intel_gt_flush_ggtt_writes(ggtt->vm.gt); - io_mapping_unmap_atomic((void __force __iomem *) unmask_page(cache->vaddr)); - } else { - struct i915_vma *vma; - int err; - - if (i915_gem_object_is_tiled(obj)) - return ERR_PTR(-EINVAL); - - if (use_cpu_reloc(cache, obj)) - return NULL; - - i915_gem_object_lock(obj); - err = i915_gem_object_set_to_gtt_domain(obj, true); - i915_gem_object_unlock(obj); - if (err) - return ERR_PTR(err); - - vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, - PIN_MAPPABLE | - PIN_NONBLOCK /* NOWARN */ | - PIN_NOEVICT); - if (IS_ERR(vma)) { - memset(&cache->node, 0, sizeof(cache->node)); - mutex_lock(&ggtt->vm.mutex); - err = drm_mm_insert_node_in_range - (&ggtt->vm.mm, &cache->node, - PAGE_SIZE, 0, I915_COLOR_UNEVICTABLE, - 0, ggtt->mappable_end, - DRM_MM_INSERT_LOW); - mutex_unlock(&ggtt->vm.mutex); - if (err) /* no inactive aperture space, use cpu reloc */ - return NULL; - } else { - cache->node.start = vma->node.start; - cache->node.mm = (void *)vma; - } - } - - offset = cache->node.start; - if (drm_mm_node_allocated(&cache->node)) { - ggtt->vm.insert_page(&ggtt->vm, - i915_gem_object_get_dma_address(obj, page), - offset, I915_CACHE_NONE, 0); - } else { - offset += page << PAGE_SHIFT; - } - - vaddr = (void __force *)io_mapping_map_atomic_wc(&ggtt->iomap, - offset); - cache->page = page; - cache->vaddr = (unsigned long)vaddr; - - return vaddr; -} - -static void *reloc_vaddr(struct drm_i915_gem_object *obj, - struct reloc_cache *cache, - unsigned long page) -{ - void *vaddr; - - if (cache->page == page) { - vaddr = unmask_page(cache->vaddr); - } else { - vaddr = NULL; - if ((cache->vaddr & KMAP) == 0) - vaddr = reloc_iomap(obj, cache, page); - if (!vaddr) - vaddr = reloc_kmap(obj, cache, page); - } - - return vaddr; -} - -static void clflush_write32(u32 *addr, u32 value, unsigned int flushes) -{ - if (unlikely(flushes & (CLFLUSH_BEFORE | CLFLUSH_AFTER))) { - if (flushes & CLFLUSH_BEFORE) { - clflushopt(addr); - mb(); - } - - *addr = value; - - /* - * Writes to the same cacheline are serialised by the CPU - * (including clflush). On the write path, we only require - * that it hits memory in an orderly fashion and place - * mb barriers at the start and end of the relocation phase - * to ensure ordering of clflush wrt to the system. - */ - if (flushes & CLFLUSH_AFTER) - clflushopt(addr); - } else - *addr = value; -} - static int reloc_move_to_gpu(struct i915_request *rq, struct i915_vma *vma) { struct drm_i915_gem_object *obj = vma->obj; @@ -1429,17 +1207,6 @@ static u32 *reloc_gpu(struct i915_execbuffer *eb, return cmd; } -static inline bool use_reloc_gpu(struct i915_vma *vma) -{ - if (DBG_FORCE_RELOC == FORCE_GPU_RELOC) - return true; - - if (DBG_FORCE_RELOC) - return false; - - return !dma_resv_test_signaled_rcu(vma->resv, true); -} - static unsigned long vma_phys_addr(struct i915_vma *vma, u32 offset) { struct page *page; @@ -1454,10 +1221,10 @@ static unsigned long vma_phys_addr(struct i915_vma *vma, u32 offset) return addr + offset_in_page(offset); } -static bool __reloc_entry_gpu(struct i915_execbuffer *eb, - struct i915_vma *vma, - u64 offset, - u64 target_addr) +static int __reloc_entry_gpu(struct i915_execbuffer *eb, + struct i915_vma *vma, + u64 offset, + u64 target_addr) { const unsigned int gen = eb->reloc_cache.gen; unsigned int len; @@ -1473,7 +1240,7 @@ static bool __reloc_entry_gpu(struct i915_execbuffer *eb, batch = reloc_gpu(eb, vma, len); if (IS_ERR(batch)) - return false; + return PTR_ERR(batch); addr = gen8_canonical_addr(vma->node.start + offset); if (gen >= 8) { @@ -1522,55 +1289,21 @@ static bool __reloc_entry_gpu(struct i915_execbuffer *eb, *batch++ = target_addr; } - return true; -} - -static bool reloc_entry_gpu(struct i915_execbuffer *eb, - struct i915_vma *vma, - u64 offset, - u64 target_addr) -{ - if (eb->reloc_cache.vaddr) - return false; - - if (!use_reloc_gpu(vma)) - return false; - - return __reloc_entry_gpu(eb, vma, offset, target_addr); + return 0; } static u64 -relocate_entry(struct i915_vma *vma, +relocate_entry(struct i915_execbuffer *eb, + struct i915_vma *vma, const struct drm_i915_gem_relocation_entry *reloc, - struct i915_execbuffer *eb, const struct i915_vma *target) { u64 target_addr = relocation_target(reloc, target); - u64 offset = reloc->offset; - - if (!reloc_entry_gpu(eb, vma, offset, target_addr)) { - bool wide = eb->reloc_cache.use_64bit_reloc; - void *vaddr; - -repeat: - vaddr = reloc_vaddr(vma->obj, - &eb->reloc_cache, - offset >> PAGE_SHIFT); - if (IS_ERR(vaddr)) - return PTR_ERR(vaddr); - - GEM_BUG_ON(!IS_ALIGNED(offset, sizeof(u32))); - clflush_write32(vaddr + offset_in_page(offset), - lower_32_bits(target_addr), - eb->reloc_cache.vaddr); - - if (wide) { - offset += sizeof(u32); - target_addr >>= 32; - wide = false; - goto repeat; - } - } + int err; + + err = __reloc_entry_gpu(eb, vma, reloc->offset, target_addr); + if (err) + return err; return target->node.start | UPDATE; } @@ -1626,8 +1359,7 @@ eb_relocate_entry(struct i915_execbuffer *eb, err = i915_vma_bind(target->vma, target->vma->obj->cache_level, PIN_GLOBAL, NULL); - if (WARN_ONCE(err, - "Unexpected failure to bind target VMA!")) + if (err) return err; } } @@ -1636,8 +1368,7 @@ eb_relocate_entry(struct i915_execbuffer *eb, * If the relocation already has the right value in it, no * more work needs to be done. */ - if (!DBG_FORCE_RELOC && - gen8_canonical_addr(target->vma->node.start) == reloc->presumed_offset) + if (gen8_canonical_addr(target->vma->node.start) == reloc->presumed_offset) return 0; /* Check that the relocation address is valid... */ @@ -1669,7 +1400,7 @@ eb_relocate_entry(struct i915_execbuffer *eb, ev->flags &= ~EXEC_OBJECT_ASYNC; /* and update the user's relocation entry */ - return relocate_entry(ev->vma, reloc, eb, target->vma); + return relocate_entry(eb, ev->vma, reloc, target->vma); } static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) @@ -1707,10 +1438,8 @@ static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) * this is bad and so lockdep complains vehemently. */ copied = __copy_from_user(r, urelocs, count * sizeof(r[0])); - if (unlikely(copied)) { - remain = -EFAULT; - goto out; - } + if (unlikely(copied)) + return -EFAULT; remain -= count; do { @@ -1718,8 +1447,7 @@ static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) if (likely(offset == 0)) { } else if ((s64)offset < 0) { - remain = (int)offset; - goto out; + return (int)offset; } else { /* * Note that reporting an error now @@ -1749,9 +1477,8 @@ static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) } while (r++, --count); urelocs += ARRAY_SIZE(stack); } while (remain); -out: - reloc_cache_reset(&eb->reloc_cache); - return remain; + + return 0; } static int eb_relocate(struct i915_execbuffer *eb) @@ -1911,8 +1638,8 @@ static int i915_reset_gen7_sol_offsets(struct i915_request *rq) u32 *cs; int i; - if (!IS_GEN(rq->i915, 7) || rq->engine->id != RCS0) { - drm_dbg(&rq->i915->drm, "sol reset is gen7/rcs only\n"); + if (!IS_GEN(rq->engine->i915, 7) || rq->engine->id != RCS0) { + drm_dbg(&rq->engine->i915->drm, "sol reset is gen7/rcs only\n"); return -EINVAL; } @@ -2659,7 +2386,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, eb.i915 = i915; eb.file = file; eb.args = args; - if (DBG_FORCE_RELOC || !(args->flags & I915_EXEC_NO_RELOC)) + if (!(args->flags & I915_EXEC_NO_RELOC)) args->flags |= __EXEC_HAS_RELOC; eb.exec = exec; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_internal.c b/drivers/gpu/drm/i915/gem/i915_gem_internal.c index cbbff81aa0af..ad22f42541bd 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_internal.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_internal.c @@ -137,6 +137,7 @@ static void i915_gem_object_put_pages_internal(struct drm_i915_gem_object *obj, } static const struct drm_i915_gem_object_ops i915_gem_object_internal_ops = { + .name = "i915_gem_object_internal", .flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE | I915_GEM_OBJECT_IS_SHRINKABLE, .get_pages = i915_gem_object_get_pages_internal, diff --git a/drivers/gpu/drm/i915/gem/i915_gem_lmem.c b/drivers/gpu/drm/i915/gem/i915_gem_lmem.c index 70543c83df06..932ee21e6609 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_lmem.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_lmem.c @@ -9,6 +9,7 @@ #include "i915_drv.h" const struct drm_i915_gem_object_ops i915_gem_lmem_obj_ops = { + .name = "i915_gem_object_lmem", .flags = I915_GEM_OBJECT_HAS_IOMEM, .get_pages = i915_gem_object_get_pages_buddy, diff --git a/drivers/gpu/drm/i915/gem/i915_gem_mman.c b/drivers/gpu/drm/i915/gem/i915_gem_mman.c index fe45bd4d63a5..fe27c5b344e3 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_mman.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_mman.c @@ -216,12 +216,12 @@ static vm_fault_t i915_error_to_vmf_fault(int err) case -ENXIO: /* unable to access backing store (on device) */ return VM_FAULT_SIGBUS; - case -ENOSPC: /* shmemfs allocation failure */ case -ENOMEM: /* our allocation failure */ return VM_FAULT_OOM; case 0: case -EAGAIN: + case -ENOSPC: /* transient failure to evict? */ case -ERESTARTSYS: case -EINTR: case -EBUSY: diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object.c b/drivers/gpu/drm/i915/gem/i915_gem_object.c index 99356c00c19e..b6ec5b50d93b 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_object.c @@ -53,7 +53,7 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj, const struct drm_i915_gem_object_ops *ops, struct lock_class_key *key) { - __mutex_init(&obj->mm.lock, "obj->mm.lock", key); + __mutex_init(&obj->mm.lock, ops->name ?: "obj->mm.lock", key); spin_lock_init(&obj->vma.lock); INIT_LIST_HEAD(&obj->vma.list); @@ -72,6 +72,10 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj, obj->mm.madv = I915_MADV_WILLNEED; INIT_RADIX_TREE(&obj->mm.get_page.radix, GFP_KERNEL | __GFP_NOWARN); mutex_init(&obj->mm.get_page.lock); + + if (IS_ENABLED(CONFIG_LOCKDEP) && i915_gem_object_is_shrinkable(obj)) + i915_gem_shrinker_taints_mutex(to_i915(obj->base.dev), + &obj->mm.lock); } /** diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c index f457d7130491..bfdb32d46877 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c @@ -126,6 +126,17 @@ void intel_emit_vma_release(struct intel_context *ce, struct i915_vma *vma) intel_engine_pm_put(ce->engine); } +static int +move_obj_to_gpu(struct drm_i915_gem_object *obj, + struct i915_request *rq, + bool write) +{ + if (obj->cache_dirty & ~obj->cache_coherent) + i915_gem_clflush_object(obj, 0); + + return i915_request_await_object(rq, obj, write); +} + int i915_gem_object_fill_blt(struct drm_i915_gem_object *obj, struct intel_context *ce, u32 value) @@ -143,12 +154,6 @@ int i915_gem_object_fill_blt(struct drm_i915_gem_object *obj, if (unlikely(err)) return err; - if (obj->cache_dirty & ~obj->cache_coherent) { - i915_gem_object_lock(obj); - i915_gem_clflush_object(obj, 0); - i915_gem_object_unlock(obj); - } - batch = intel_emit_vma_fill_blt(ce, vma, value); if (IS_ERR(batch)) { err = PTR_ERR(batch); @@ -165,27 +170,22 @@ int i915_gem_object_fill_blt(struct drm_i915_gem_object *obj, if (unlikely(err)) goto out_request; - err = i915_request_await_object(rq, obj, true); - if (unlikely(err)) - goto out_request; - - if (ce->engine->emit_init_breadcrumb) { - err = ce->engine->emit_init_breadcrumb(rq); - if (unlikely(err)) - goto out_request; - } - i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, true); + err = move_obj_to_gpu(vma->obj, rq, true); if (err == 0) err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); i915_vma_unlock(vma); if (unlikely(err)) goto out_request; - err = ce->engine->emit_bb_start(rq, - batch->node.start, batch->node.size, - 0); + if (ce->engine->emit_init_breadcrumb) + err = ce->engine->emit_init_breadcrumb(rq); + + if (likely(!err)) + err = ce->engine->emit_bb_start(rq, + batch->node.start, + batch->node.size, + 0); out_request: if (unlikely(err)) i915_request_set_error_once(rq, err); @@ -317,16 +317,6 @@ out_pm: return ERR_PTR(err); } -static int move_to_gpu(struct i915_vma *vma, struct i915_request *rq, bool write) -{ - struct drm_i915_gem_object *obj = vma->obj; - - if (obj->cache_dirty & ~obj->cache_coherent) - i915_gem_clflush_object(obj, 0); - - return i915_request_await_object(rq, obj, write); -} - int i915_gem_object_copy_blt(struct drm_i915_gem_object *src, struct drm_i915_gem_object *dst, struct intel_context *ce) @@ -375,7 +365,7 @@ int i915_gem_object_copy_blt(struct drm_i915_gem_object *src, goto out_request; for (i = 0; i < ARRAY_SIZE(vma); i++) { - err = move_to_gpu(vma[i], rq, i); + err = move_obj_to_gpu(vma[i]->obj, rq, i); if (unlikely(err)) goto out_unlock; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object_types.h b/drivers/gpu/drm/i915/gem/i915_gem_object_types.h index 54ee658bb168..b1f82a11aef2 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object_types.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_object_types.h @@ -61,6 +61,8 @@ struct drm_i915_gem_object_ops { int (*dmabuf_export)(struct drm_i915_gem_object *obj); void (*release)(struct drm_i915_gem_object *obj); + + const char *name; /* friendly name for debug, e.g. lockdep classes */ }; enum i915_mmap_type { diff --git a/drivers/gpu/drm/i915/gem/i915_gem_phys.c b/drivers/gpu/drm/i915/gem/i915_gem_phys.c index 7fe9831aa9ba..28147aab47b9 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_phys.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_phys.c @@ -27,7 +27,7 @@ static int i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj) void *dst; int i; - if (WARN_ON(i915_gem_object_needs_bit17_swizzle(obj))) + if (GEM_WARN_ON(i915_gem_object_needs_bit17_swizzle(obj))) return -EINVAL; /* @@ -140,6 +140,7 @@ static void phys_release(struct drm_i915_gem_object *obj) } static const struct drm_i915_gem_object_ops i915_gem_phys_ops = { + .name = "i915_gem_object_phys", .get_pages = i915_gem_object_get_pages_phys, .put_pages = i915_gem_object_put_pages_phys, diff --git a/drivers/gpu/drm/i915/gem/i915_gem_shmem.c b/drivers/gpu/drm/i915/gem/i915_gem_shmem.c index 7aff3514d97a..38113d3c0138 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_shmem.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_shmem.c @@ -147,8 +147,7 @@ rebuild_st: last_pfn = page_to_pfn(page); /* Check that the i965g/gm workaround works. */ - drm_WARN_ON(&i915->drm, - (gfp & __GFP_DMA32) && (last_pfn >= 0x00100000UL)); + GEM_BUG_ON(gfp & __GFP_DMA32 && last_pfn >= 0x00100000UL); } if (sg) { /* loop terminated early; short sg table */ sg_page_sizes |= sg->length; @@ -430,6 +429,7 @@ static void shmem_release(struct drm_i915_gem_object *obj) } const struct drm_i915_gem_object_ops i915_gem_shmem_ops = { + .name = "i915_gem_object_shmem", .flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE | I915_GEM_OBJECT_IS_SHRINKABLE, diff --git a/drivers/gpu/drm/i915/gem/i915_gem_stolen.c b/drivers/gpu/drm/i915/gem/i915_gem_stolen.c index dc250278bd2c..e0f21f12d3ce 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_stolen.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_stolen.c @@ -566,6 +566,7 @@ i915_gem_object_release_stolen(struct drm_i915_gem_object *obj) } static const struct drm_i915_gem_object_ops i915_gem_object_stolen_ops = { + .name = "i915_gem_object_stolen", .get_pages = i915_gem_object_get_pages_stolen, .put_pages = i915_gem_object_put_pages_stolen, .release = i915_gem_object_release_stolen, diff --git a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c index c31a6744daee..9c53eb883400 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c @@ -235,7 +235,7 @@ i915_gem_userptr_init__mmu_notifier(struct drm_i915_gem_object *obj, if (flags & I915_USERPTR_UNSYNCHRONIZED) return capable(CAP_SYS_ADMIN) ? 0 : -EPERM; - if (WARN_ON(obj->userptr.mm == NULL)) + if (GEM_WARN_ON(!obj->userptr.mm)) return -EINVAL; mn = i915_mmu_notifier_find(obj->userptr.mm); @@ -712,6 +712,7 @@ i915_gem_userptr_dmabuf_export(struct drm_i915_gem_object *obj) } static const struct drm_i915_gem_object_ops i915_gem_userptr_ops = { + .name = "i915_gem_object_userptr", .flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE | I915_GEM_OBJECT_IS_SHRINKABLE | I915_GEM_OBJECT_NO_MMAP | diff --git a/drivers/gpu/drm/i915/gem/selftests/huge_gem_object.c b/drivers/gpu/drm/i915/gem/selftests/huge_gem_object.c index 2b46c6530da9..a768ec61e966 100644 --- a/drivers/gpu/drm/i915/gem/selftests/huge_gem_object.c +++ b/drivers/gpu/drm/i915/gem/selftests/huge_gem_object.c @@ -88,6 +88,7 @@ static void huge_put_pages(struct drm_i915_gem_object *obj, } static const struct drm_i915_gem_object_ops huge_ops = { + .name = "huge-gem", .flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE, .get_pages = huge_get_pages, .put_pages = huge_put_pages, diff --git a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c index c9988b6d5c88..8291ede6902c 100644 --- a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c +++ b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c @@ -139,6 +139,7 @@ static void put_huge_pages(struct drm_i915_gem_object *obj, } static const struct drm_i915_gem_object_ops huge_page_ops = { + .name = "huge-gem", .flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE | I915_GEM_OBJECT_IS_SHRINKABLE, .get_pages = get_huge_pages, @@ -283,12 +284,14 @@ static void fake_put_huge_pages(struct drm_i915_gem_object *obj, } static const struct drm_i915_gem_object_ops fake_ops = { + .name = "fake-gem", .flags = I915_GEM_OBJECT_IS_SHRINKABLE, .get_pages = fake_get_huge_pages, .put_pages = fake_put_huge_pages, }; static const struct drm_i915_gem_object_ops fake_ops_single = { + .name = "fake-gem", .flags = I915_GEM_OBJECT_IS_SHRINKABLE, .get_pages = fake_get_huge_pages_single, .put_pages = fake_put_huge_pages, @@ -1409,147 +1412,6 @@ out: return err; } -static int igt_ppgtt_pin_update(void *arg) -{ - struct i915_gem_context *ctx = arg; - struct drm_i915_private *dev_priv = ctx->i915; - unsigned long supported = INTEL_INFO(dev_priv)->page_sizes; - struct drm_i915_gem_object *obj; - struct i915_gem_engines_iter it; - struct i915_address_space *vm; - struct intel_context *ce; - struct i915_vma *vma; - unsigned int flags = PIN_USER | PIN_OFFSET_FIXED; - unsigned int n; - int first, last; - int err = 0; - - /* - * Make sure there's no funny business when doing a PIN_UPDATE -- in the - * past we had a subtle issue with being able to incorrectly do multiple - * alloc va ranges on the same object when doing a PIN_UPDATE, which - * resulted in some pretty nasty bugs, though only when using - * huge-gtt-pages. - */ - - vm = i915_gem_context_get_vm_rcu(ctx); - if (!i915_vm_is_4lvl(vm)) { - pr_info("48b PPGTT not supported, skipping\n"); - goto out_vm; - } - - first = ilog2(I915_GTT_PAGE_SIZE_64K); - last = ilog2(I915_GTT_PAGE_SIZE_2M); - - for_each_set_bit_from(first, &supported, last + 1) { - unsigned int page_size = BIT(first); - - obj = i915_gem_object_create_internal(dev_priv, page_size); - if (IS_ERR(obj)) { - err = PTR_ERR(obj); - goto out_vm; - } - - vma = i915_vma_instance(obj, vm, NULL); - if (IS_ERR(vma)) { - err = PTR_ERR(vma); - goto out_put; - } - - err = i915_vma_pin(vma, SZ_2M, 0, flags); - if (err) - goto out_put; - - if (vma->page_sizes.sg < page_size) { - pr_info("Unable to allocate page-size %x, finishing test early\n", - page_size); - goto out_unpin; - } - - err = igt_check_page_sizes(vma); - if (err) - goto out_unpin; - - if (vma->page_sizes.gtt != page_size) { - dma_addr_t addr = i915_gem_object_get_dma_address(obj, 0); - - /* - * The only valid reason for this to ever fail would be - * if the dma-mapper screwed us over when we did the - * dma_map_sg(), since it has the final say over the dma - * address. - */ - if (IS_ALIGNED(addr, page_size)) { - pr_err("page_sizes.gtt=%u, expected=%u\n", - vma->page_sizes.gtt, page_size); - err = -EINVAL; - } else { - pr_info("dma address misaligned, finishing test early\n"); - } - - goto out_unpin; - } - - err = i915_vma_bind(vma, I915_CACHE_NONE, PIN_UPDATE, NULL); - if (err) - goto out_unpin; - - i915_vma_unpin(vma); - i915_gem_object_put(obj); - } - - obj = i915_gem_object_create_internal(dev_priv, PAGE_SIZE); - if (IS_ERR(obj)) { - err = PTR_ERR(obj); - goto out_vm; - } - - vma = i915_vma_instance(obj, vm, NULL); - if (IS_ERR(vma)) { - err = PTR_ERR(vma); - goto out_put; - } - - err = i915_vma_pin(vma, 0, 0, flags); - if (err) - goto out_put; - - /* - * Make sure we don't end up with something like where the pde is still - * pointing to the 2M page, and the pt we just filled-in is dangling -- - * we can check this by writing to the first page where it would then - * land in the now stale 2M page. - */ - - n = 0; - for_each_gem_engine(ce, i915_gem_context_lock_engines(ctx), it) { - if (!intel_engine_can_store_dword(ce->engine)) - continue; - - err = gpu_write(ce, vma, n++, 0xdeadbeaf); - if (err) - break; - } - i915_gem_context_unlock_engines(ctx); - if (err) - goto out_unpin; - - while (n--) { - err = cpu_check(obj, n, 0xdeadbeaf); - if (err) - goto out_unpin; - } - -out_unpin: - i915_vma_unpin(vma); -out_put: - i915_gem_object_put(obj); -out_vm: - i915_vm_put(vm); - - return err; -} - static int igt_tmpfs_fallback(void *arg) { struct i915_gem_context *ctx = arg; @@ -1760,7 +1622,6 @@ int i915_gem_huge_page_live_selftests(struct drm_i915_private *i915) { static const struct i915_subtest tests[] = { SUBTEST(igt_shrink_thp), - SUBTEST(igt_ppgtt_pin_update), SUBTEST(igt_tmpfs_fallback), SUBTEST(igt_ppgtt_smoke_huge), SUBTEST(igt_ppgtt_sanity_check), diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c index 8fe3ad2ee34e..299c29e9ad86 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c @@ -702,8 +702,5 @@ int i915_gem_client_blt_live_selftests(struct drm_i915_private *i915) if (intel_gt_is_wedged(&i915->gt)) return 0; - if (!HAS_ENGINE(i915, BCS0)) - return 0; - return i915_live_subtests(tests, i915); } diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c index a49016f8ee0d..57c14d3340cd 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c @@ -37,20 +37,14 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, return err; /* 8-Byte aligned */ - if (!__reloc_entry_gpu(eb, vma, - offsets[0] * sizeof(u32), - 0)) { - err = -EIO; + err = __reloc_entry_gpu(eb, vma, offsets[0] * sizeof(u32), 0); + if (err) goto unpin_vma; - } /* !8-Byte aligned */ - if (!__reloc_entry_gpu(eb, vma, - offsets[1] * sizeof(u32), - 1)) { - err = -EIO; + err = __reloc_entry_gpu(eb, vma, offsets[1] * sizeof(u32), 1); + if (err) goto unpin_vma; - } /* Skip to the end of the cmd page */ i = PAGE_SIZE / sizeof(u32) - RELOC_TAIL - 1; @@ -60,12 +54,9 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, eb->reloc_cache.rq_size += i; /* Force batch chaining */ - if (!__reloc_entry_gpu(eb, vma, - offsets[2] * sizeof(u32), - 2)) { - err = -EIO; + err = __reloc_entry_gpu(eb, vma, offsets[2] * sizeof(u32), 2); + if (err) goto unpin_vma; - } GEM_BUG_ON(!eb->reloc_cache.rq); rq = i915_request_get(eb->reloc_cache.rq); diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_object_blt.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_object_blt.c index 31549ad83fa6..23b6e11bbc3e 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_object_blt.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_object_blt.c @@ -193,7 +193,7 @@ err_src: } struct igt_thread_arg { - struct drm_i915_private *i915; + struct intel_engine_cs *engine; struct i915_gem_context *ctx; struct file *file; struct rnd_state prng; @@ -203,7 +203,7 @@ struct igt_thread_arg { static int igt_fill_blt_thread(void *arg) { struct igt_thread_arg *thread = arg; - struct drm_i915_private *i915 = thread->i915; + struct intel_engine_cs *engine = thread->engine; struct rnd_state *prng = &thread->prng; struct drm_i915_gem_object *obj; struct i915_gem_context *ctx; @@ -215,7 +215,7 @@ static int igt_fill_blt_thread(void *arg) ctx = thread->ctx; if (!ctx) { - ctx = live_context(i915, thread->file); + ctx = live_context_for_engine(engine, thread->file); if (IS_ERR(ctx)) return PTR_ERR(ctx); @@ -223,7 +223,7 @@ static int igt_fill_blt_thread(void *arg) ctx->sched.priority = I915_USER_PRIORITY(prio); } - ce = i915_gem_context_get_engine(ctx, BCS0); + ce = i915_gem_context_get_engine(ctx, 0); GEM_BUG_ON(IS_ERR(ce)); /* @@ -256,7 +256,7 @@ static int igt_fill_blt_thread(void *arg) pr_debug("%s with phys_sz= %x, sz=%x, val=%x\n", __func__, phys_sz, sz, val); - obj = huge_gem_object(i915, phys_sz, sz); + obj = huge_gem_object(engine->i915, phys_sz, sz); if (IS_ERR(obj)) { err = PTR_ERR(obj); goto err_flush; @@ -321,7 +321,7 @@ err_flush: static int igt_copy_blt_thread(void *arg) { struct igt_thread_arg *thread = arg; - struct drm_i915_private *i915 = thread->i915; + struct intel_engine_cs *engine = thread->engine; struct rnd_state *prng = &thread->prng; struct drm_i915_gem_object *src, *dst; struct i915_gem_context *ctx; @@ -333,7 +333,7 @@ static int igt_copy_blt_thread(void *arg) ctx = thread->ctx; if (!ctx) { - ctx = live_context(i915, thread->file); + ctx = live_context_for_engine(engine, thread->file); if (IS_ERR(ctx)) return PTR_ERR(ctx); @@ -341,7 +341,7 @@ static int igt_copy_blt_thread(void *arg) ctx->sched.priority = I915_USER_PRIORITY(prio); } - ce = i915_gem_context_get_engine(ctx, BCS0); + ce = i915_gem_context_get_engine(ctx, 0); GEM_BUG_ON(IS_ERR(ce)); /* @@ -374,7 +374,7 @@ static int igt_copy_blt_thread(void *arg) pr_debug("%s with phys_sz= %x, sz=%x, val=%x\n", __func__, phys_sz, sz, val); - src = huge_gem_object(i915, phys_sz, sz); + src = huge_gem_object(engine->i915, phys_sz, sz); if (IS_ERR(src)) { err = PTR_ERR(src); goto err_flush; @@ -394,7 +394,7 @@ static int igt_copy_blt_thread(void *arg) if (!(src->cache_coherent & I915_BO_CACHE_COHERENT_FOR_READ)) src->cache_dirty = true; - dst = huge_gem_object(i915, phys_sz, sz); + dst = huge_gem_object(engine->i915, phys_sz, sz); if (IS_ERR(dst)) { err = PTR_ERR(dst); goto err_put_src; @@ -456,7 +456,7 @@ err_flush: return err; } -static int igt_threaded_blt(struct drm_i915_private *i915, +static int igt_threaded_blt(struct intel_engine_cs *engine, int (*blt_fn)(void *arg), unsigned int flags) #define SINGLE_CTX BIT(0) @@ -477,14 +477,14 @@ static int igt_threaded_blt(struct drm_i915_private *i915, if (!thread) goto out_tsk; - thread[0].file = mock_file(i915); + thread[0].file = mock_file(engine->i915); if (IS_ERR(thread[0].file)) { err = PTR_ERR(thread[0].file); goto out_thread; } if (flags & SINGLE_CTX) { - thread[0].ctx = live_context(i915, thread[0].file); + thread[0].ctx = live_context_for_engine(engine, thread[0].file); if (IS_ERR(thread[0].ctx)) { err = PTR_ERR(thread[0].ctx); goto out_file; @@ -492,7 +492,7 @@ static int igt_threaded_blt(struct drm_i915_private *i915, } for (i = 0; i < n_cpus; ++i) { - thread[i].i915 = i915; + thread[i].engine = engine; thread[i].file = thread[0].file; thread[i].ctx = thread[0].ctx; thread[i].n_cpus = n_cpus; @@ -532,24 +532,40 @@ out_tsk: return err; } +static int test_copy_engines(struct drm_i915_private *i915, + int (*fn)(void *arg), + unsigned int flags) +{ + struct intel_engine_cs *engine; + int ret; + + for_each_uabi_class_engine(engine, I915_ENGINE_CLASS_COPY, i915) { + ret = igt_threaded_blt(engine, fn, flags); + if (ret) + return ret; + } + + return 0; +} + static int igt_fill_blt(void *arg) { - return igt_threaded_blt(arg, igt_fill_blt_thread, 0); + return test_copy_engines(arg, igt_fill_blt_thread, 0); } static int igt_fill_blt_ctx0(void *arg) { - return igt_threaded_blt(arg, igt_fill_blt_thread, SINGLE_CTX); + return test_copy_engines(arg, igt_fill_blt_thread, SINGLE_CTX); } static int igt_copy_blt(void *arg) { - return igt_threaded_blt(arg, igt_copy_blt_thread, 0); + return test_copy_engines(arg, igt_copy_blt_thread, 0); } static int igt_copy_blt_ctx0(void *arg) { - return igt_threaded_blt(arg, igt_copy_blt_thread, SINGLE_CTX); + return test_copy_engines(arg, igt_copy_blt_thread, SINGLE_CTX); } int i915_gem_object_blt_live_selftests(struct drm_i915_private *i915) @@ -564,9 +580,6 @@ int i915_gem_object_blt_live_selftests(struct drm_i915_private *i915) if (intel_gt_is_wedged(&i915->gt)) return 0; - if (!HAS_ENGINE(i915, BCS0)) - return 0; - return i915_live_subtests(tests, i915); } diff --git a/drivers/gpu/drm/i915/gem/selftests/mock_context.c b/drivers/gpu/drm/i915/gem/selftests/mock_context.c index e7e3c620f542..aa0d06cf1903 100644 --- a/drivers/gpu/drm/i915/gem/selftests/mock_context.c +++ b/drivers/gpu/drm/i915/gem/selftests/mock_context.c @@ -100,6 +100,43 @@ err_ctx: } struct i915_gem_context * +live_context_for_engine(struct intel_engine_cs *engine, struct file *file) +{ + struct i915_gem_engines *engines; + struct i915_gem_context *ctx; + struct intel_context *ce; + + engines = alloc_engines(1); + if (!engines) + return ERR_PTR(-ENOMEM); + + ctx = live_context(engine->i915, file); + if (IS_ERR(ctx)) { + __free_engines(engines, 0); + return ctx; + } + + ce = intel_context_create(engine); + if (IS_ERR(ce)) { + __free_engines(engines, 0); + return ERR_CAST(ce); + } + + intel_context_set_gem(ce, ctx); + engines->engines[0] = ce; + engines->num_engines = 1; + + mutex_lock(&ctx->engines_mutex); + i915_gem_context_set_user_engines(ctx); + engines = rcu_replace_pointer(ctx->engines, engines, 1); + mutex_unlock(&ctx->engines_mutex); + + engines_idle_release(ctx, engines); + + return ctx; +} + +struct i915_gem_context * kernel_context(struct drm_i915_private *i915) { struct i915_gem_context *ctx; diff --git a/drivers/gpu/drm/i915/gem/selftests/mock_context.h b/drivers/gpu/drm/i915/gem/selftests/mock_context.h index fb83d2f09212..2a6121d33352 100644 --- a/drivers/gpu/drm/i915/gem/selftests/mock_context.h +++ b/drivers/gpu/drm/i915/gem/selftests/mock_context.h @@ -9,6 +9,7 @@ struct file; struct drm_i915_private; +struct intel_engine_cs; void mock_init_contexts(struct drm_i915_private *i915); @@ -21,6 +22,9 @@ void mock_context_close(struct i915_gem_context *ctx); struct i915_gem_context * live_context(struct drm_i915_private *i915, struct file *file); +struct i915_gem_context * +live_context_for_engine(struct intel_engine_cs *engine, struct file *file); + struct i915_gem_context *kernel_context(struct drm_i915_private *i915); void kernel_context_close(struct i915_gem_context *ctx); diff --git a/drivers/gpu/drm/i915/gt/gen2_engine_cs.c b/drivers/gpu/drm/i915/gt/gen2_engine_cs.c new file mode 100644 index 000000000000..b491a64919c8 --- /dev/null +++ b/drivers/gpu/drm/i915/gt/gen2_engine_cs.c @@ -0,0 +1,329 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright © 2020 Intel Corporation + */ + +#include "gen2_engine_cs.h" +#include "i915_drv.h" +#include "intel_engine.h" +#include "intel_gpu_commands.h" +#include "intel_gt.h" +#include "intel_gt_irq.h" +#include "intel_ring.h" + +int gen2_emit_flush(struct i915_request *rq, u32 mode) +{ + unsigned int num_store_dw = 12; + u32 cmd, *cs; + + cmd = MI_FLUSH; + if (mode & EMIT_INVALIDATE) + cmd |= MI_READ_FLUSH; + + cs = intel_ring_begin(rq, 2 + 4 * num_store_dw); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = cmd; + while (num_store_dw--) { + *cs++ = MI_STORE_DWORD_INDEX; + *cs++ = I915_GEM_HWS_SCRATCH * sizeof(u32); + *cs++ = 0; + *cs++ = MI_FLUSH | MI_NO_WRITE_FLUSH; + } + *cs++ = cmd; + + intel_ring_advance(rq, cs); + + return 0; +} + +int gen4_emit_flush_rcs(struct i915_request *rq, u32 mode) +{ + u32 cmd, *cs; + int i; + + /* + * read/write caches: + * + * I915_GEM_DOMAIN_RENDER is always invalidated, but is + * only flushed if MI_NO_WRITE_FLUSH is unset. On 965, it is + * also flushed at 2d versus 3d pipeline switches. + * + * read-only caches: + * + * I915_GEM_DOMAIN_SAMPLER is flushed on pre-965 if + * MI_READ_FLUSH is set, and is always flushed on 965. + * + * I915_GEM_DOMAIN_COMMAND may not exist? + * + * I915_GEM_DOMAIN_INSTRUCTION, which exists on 965, is + * invalidated when MI_EXE_FLUSH is set. + * + * I915_GEM_DOMAIN_VERTEX, which exists on 965, is + * invalidated with every MI_FLUSH. + * + * TLBs: + * + * On 965, TLBs associated with I915_GEM_DOMAIN_COMMAND + * and I915_GEM_DOMAIN_CPU in are invalidated at PTE write and + * I915_GEM_DOMAIN_RENDER and I915_GEM_DOMAIN_SAMPLER + * are flushed at any MI_FLUSH. + */ + + cmd = MI_FLUSH; + if (mode & EMIT_INVALIDATE) { + cmd |= MI_EXE_FLUSH; + if (IS_G4X(rq->engine->i915) || IS_GEN(rq->engine->i915, 5)) + cmd |= MI_INVALIDATE_ISP; + } + + i = 2; + if (mode & EMIT_INVALIDATE) + i += 20; + + cs = intel_ring_begin(rq, i); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = cmd; + + /* + * A random delay to let the CS invalidate take effect? Without this + * delay, the GPU relocation path fails as the CS does not see + * the updated contents. Just as important, if we apply the flushes + * to the EMIT_FLUSH branch (i.e. immediately after the relocation + * write and before the invalidate on the next batch), the relocations + * still fail. This implies that is a delay following invalidation + * that is required to reset the caches as opposed to a delay to + * ensure the memory is written. + */ + if (mode & EMIT_INVALIDATE) { + *cs++ = GFX_OP_PIPE_CONTROL(4) | PIPE_CONTROL_QW_WRITE; + *cs++ = intel_gt_scratch_offset(rq->engine->gt, + INTEL_GT_SCRATCH_FIELD_DEFAULT) | + PIPE_CONTROL_GLOBAL_GTT; + *cs++ = 0; + *cs++ = 0; + + for (i = 0; i < 12; i++) + *cs++ = MI_FLUSH; + + *cs++ = GFX_OP_PIPE_CONTROL(4) | PIPE_CONTROL_QW_WRITE; + *cs++ = intel_gt_scratch_offset(rq->engine->gt, + INTEL_GT_SCRATCH_FIELD_DEFAULT) | + PIPE_CONTROL_GLOBAL_GTT; + *cs++ = 0; + *cs++ = 0; + } + + *cs++ = cmd; + + intel_ring_advance(rq, cs); + + return 0; +} + +int gen4_emit_flush_vcs(struct i915_request *rq, u32 mode) +{ + u32 *cs; + + cs = intel_ring_begin(rq, 2); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = MI_FLUSH; + *cs++ = MI_NOOP; + intel_ring_advance(rq, cs); + + return 0; +} + +static u32 *__gen2_emit_breadcrumb(struct i915_request *rq, u32 *cs, + int flush, int post) +{ + GEM_BUG_ON(i915_request_active_timeline(rq)->hwsp_ggtt != rq->engine->status_page.vma); + GEM_BUG_ON(offset_in_page(i915_request_active_timeline(rq)->hwsp_offset) != I915_GEM_HWS_SEQNO_ADDR); + + *cs++ = MI_FLUSH; + + while (flush--) { + *cs++ = MI_STORE_DWORD_INDEX; + *cs++ = I915_GEM_HWS_SCRATCH * sizeof(u32); + *cs++ = rq->fence.seqno; + } + + while (post--) { + *cs++ = MI_STORE_DWORD_INDEX; + *cs++ = I915_GEM_HWS_SEQNO_ADDR; + *cs++ = rq->fence.seqno; + } + + *cs++ = MI_USER_INTERRUPT; + + rq->tail = intel_ring_offset(rq, cs); + assert_ring_tail_valid(rq->ring, rq->tail); + + return cs; +} + +u32 *gen3_emit_breadcrumb(struct i915_request *rq, u32 *cs) +{ + return __gen2_emit_breadcrumb(rq, cs, 16, 8); +} + +u32 *gen5_emit_breadcrumb(struct i915_request *rq, u32 *cs) +{ + return __gen2_emit_breadcrumb(rq, cs, 8, 8); +} + +/* Just userspace ABI convention to limit the wa batch bo to a resonable size */ +#define I830_BATCH_LIMIT SZ_256K +#define I830_TLB_ENTRIES (2) +#define I830_WA_SIZE max(I830_TLB_ENTRIES * SZ_4K, I830_BATCH_LIMIT) +int i830_emit_bb_start(struct i915_request *rq, + u64 offset, u32 len, + unsigned int dispatch_flags) +{ + u32 *cs, cs_offset = + intel_gt_scratch_offset(rq->engine->gt, + INTEL_GT_SCRATCH_FIELD_DEFAULT); + + GEM_BUG_ON(rq->engine->gt->scratch->size < I830_WA_SIZE); + + cs = intel_ring_begin(rq, 6); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + /* Evict the invalid PTE TLBs */ + *cs++ = COLOR_BLT_CMD | BLT_WRITE_RGBA; + *cs++ = BLT_DEPTH_32 | BLT_ROP_COLOR_COPY | 4096; + *cs++ = I830_TLB_ENTRIES << 16 | 4; /* load each page */ + *cs++ = cs_offset; + *cs++ = 0xdeadbeef; + *cs++ = MI_NOOP; + intel_ring_advance(rq, cs); + + if ((dispatch_flags & I915_DISPATCH_PINNED) == 0) { + if (len > I830_BATCH_LIMIT) + return -ENOSPC; + + cs = intel_ring_begin(rq, 6 + 2); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + /* + * Blit the batch (which has now all relocs applied) to the + * stable batch scratch bo area (so that the CS never + * stumbles over its tlb invalidation bug) ... + */ + *cs++ = SRC_COPY_BLT_CMD | BLT_WRITE_RGBA | (6 - 2); + *cs++ = BLT_DEPTH_32 | BLT_ROP_SRC_COPY | 4096; + *cs++ = DIV_ROUND_UP(len, 4096) << 16 | 4096; + *cs++ = cs_offset; + *cs++ = 4096; + *cs++ = offset; + + *cs++ = MI_FLUSH; + *cs++ = MI_NOOP; + intel_ring_advance(rq, cs); + + /* ... and execute it. */ + offset = cs_offset; + } + + if (!(dispatch_flags & I915_DISPATCH_SECURE)) + offset |= MI_BATCH_NON_SECURE; + + cs = intel_ring_begin(rq, 2); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT; + *cs++ = offset; + intel_ring_advance(rq, cs); + + return 0; +} + +int gen3_emit_bb_start(struct i915_request *rq, + u64 offset, u32 len, + unsigned int dispatch_flags) +{ + u32 *cs; + + if (!(dispatch_flags & I915_DISPATCH_SECURE)) + offset |= MI_BATCH_NON_SECURE; + + cs = intel_ring_begin(rq, 2); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT; + *cs++ = offset; + intel_ring_advance(rq, cs); + + return 0; +} + +int gen4_emit_bb_start(struct i915_request *rq, + u64 offset, u32 length, + unsigned int dispatch_flags) +{ + u32 security; + u32 *cs; + + security = MI_BATCH_NON_SECURE_I965; + if (dispatch_flags & I915_DISPATCH_SECURE) + security = 0; + + cs = intel_ring_begin(rq, 2); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT | security; + *cs++ = offset; + intel_ring_advance(rq, cs); + + return 0; +} + +void gen2_irq_enable(struct intel_engine_cs *engine) +{ + struct drm_i915_private *i915 = engine->i915; + + i915->irq_mask &= ~engine->irq_enable_mask; + intel_uncore_write16(&i915->uncore, GEN2_IMR, i915->irq_mask); + ENGINE_POSTING_READ16(engine, RING_IMR); +} + +void gen2_irq_disable(struct intel_engine_cs *engine) +{ + struct drm_i915_private *i915 = engine->i915; + + i915->irq_mask |= engine->irq_enable_mask; + intel_uncore_write16(&i915->uncore, GEN2_IMR, i915->irq_mask); +} + +void gen3_irq_enable(struct intel_engine_cs *engine) +{ + engine->i915->irq_mask &= ~engine->irq_enable_mask; + intel_uncore_write(engine->uncore, GEN2_IMR, engine->i915->irq_mask); + intel_uncore_posting_read_fw(engine->uncore, GEN2_IMR); +} + +void gen3_irq_disable(struct intel_engine_cs *engine) +{ + engine->i915->irq_mask |= engine->irq_enable_mask; + intel_uncore_write(engine->uncore, GEN2_IMR, engine->i915->irq_mask); +} + +void gen5_irq_enable(struct intel_engine_cs *engine) +{ + gen5_gt_enable_irq(engine->gt, engine->irq_enable_mask); +} + +void gen5_irq_disable(struct intel_engine_cs *engine) +{ + gen5_gt_disable_irq(engine->gt, engine->irq_enable_mask); +} diff --git a/drivers/gpu/drm/i915/gt/gen2_engine_cs.h b/drivers/gpu/drm/i915/gt/gen2_engine_cs.h new file mode 100644 index 000000000000..a5cd64a65c9e --- /dev/null +++ b/drivers/gpu/drm/i915/gt/gen2_engine_cs.h @@ -0,0 +1,38 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright © 2020 Intel Corporation + */ + +#ifndef __GEN2_ENGINE_CS_H__ +#define __GEN2_ENGINE_CS_H__ + +#include <linux/types.h> + +struct i915_request; +struct intel_engine_cs; + +int gen2_emit_flush(struct i915_request *rq, u32 mode); +int gen4_emit_flush_rcs(struct i915_request *rq, u32 mode); +int gen4_emit_flush_vcs(struct i915_request *rq, u32 mode); + +u32 *gen3_emit_breadcrumb(struct i915_request *rq, u32 *cs); +u32 *gen5_emit_breadcrumb(struct i915_request *rq, u32 *cs); + +int i830_emit_bb_start(struct i915_request *rq, + u64 offset, u32 len, + unsigned int dispatch_flags); +int gen3_emit_bb_start(struct i915_request *rq, + u64 offset, u32 len, + unsigned int dispatch_flags); +int gen4_emit_bb_start(struct i915_request *rq, + u64 offset, u32 length, + unsigned int dispatch_flags); + +void gen2_irq_enable(struct intel_engine_cs *engine); +void gen2_irq_disable(struct intel_engine_cs *engine); +void gen3_irq_enable(struct intel_engine_cs *engine); +void gen3_irq_disable(struct intel_engine_cs *engine); +void gen5_irq_enable(struct intel_engine_cs *engine); +void gen5_irq_disable(struct intel_engine_cs *engine); + +#endif /* __GEN2_ENGINE_CS_H__ */ diff --git a/drivers/gpu/drm/i915/gt/gen6_engine_cs.c b/drivers/gpu/drm/i915/gt/gen6_engine_cs.c new file mode 100644 index 000000000000..ce38d1bcaba3 --- /dev/null +++ b/drivers/gpu/drm/i915/gt/gen6_engine_cs.c @@ -0,0 +1,455 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright © 2020 Intel Corporation + */ + +#include "gen6_engine_cs.h" +#include "intel_engine.h" +#include "intel_gpu_commands.h" +#include "intel_gt.h" +#include "intel_gt_irq.h" +#include "intel_gt_pm_irq.h" +#include "intel_ring.h" + +#define HWS_SCRATCH_ADDR (I915_GEM_HWS_SCRATCH * sizeof(u32)) + +/* + * Emits a PIPE_CONTROL with a non-zero post-sync operation, for + * implementing two workarounds on gen6. From section 1.4.7.1 + * "PIPE_CONTROL" of the Sandy Bridge PRM volume 2 part 1: + * + * [DevSNB-C+{W/A}] Before any depth stall flush (including those + * produced by non-pipelined state commands), software needs to first + * send a PIPE_CONTROL with no bits set except Post-Sync Operation != + * 0. + * + * [Dev-SNB{W/A}]: Before a PIPE_CONTROL with Write Cache Flush Enable + * =1, a PIPE_CONTROL with any non-zero post-sync-op is required. + * + * And the workaround for these two requires this workaround first: + * + * [Dev-SNB{W/A}]: Pipe-control with CS-stall bit set must be sent + * BEFORE the pipe-control with a post-sync op and no write-cache + * flushes. + * + * And this last workaround is tricky because of the requirements on + * that bit. From section 1.4.7.2.3 "Stall" of the Sandy Bridge PRM + * volume 2 part 1: + * + * "1 of the following must also be set: + * - Render Target Cache Flush Enable ([12] of DW1) + * - Depth Cache Flush Enable ([0] of DW1) + * - Stall at Pixel Scoreboard ([1] of DW1) + * - Depth Stall ([13] of DW1) + * - Post-Sync Operation ([13] of DW1) + * - Notify Enable ([8] of DW1)" + * + * The cache flushes require the workaround flush that triggered this + * one, so we can't use it. Depth stall would trigger the same. + * Post-sync nonzero is what triggered this second workaround, so we + * can't use that one either. Notify enable is IRQs, which aren't + * really our business. That leaves only stall at scoreboard. + */ +static int +gen6_emit_post_sync_nonzero_flush(struct i915_request *rq) +{ + u32 scratch_addr = + intel_gt_scratch_offset(rq->engine->gt, + INTEL_GT_SCRATCH_FIELD_RENDER_FLUSH); + u32 *cs; + + cs = intel_ring_begin(rq, 6); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = GFX_OP_PIPE_CONTROL(5); + *cs++ = PIPE_CONTROL_CS_STALL | PIPE_CONTROL_STALL_AT_SCOREBOARD; + *cs++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT; + *cs++ = 0; /* low dword */ + *cs++ = 0; /* high dword */ + *cs++ = MI_NOOP; + intel_ring_advance(rq, cs); + + cs = intel_ring_begin(rq, 6); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = GFX_OP_PIPE_CONTROL(5); + *cs++ = PIPE_CONTROL_QW_WRITE; + *cs++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT; + *cs++ = 0; + *cs++ = 0; + *cs++ = MI_NOOP; + intel_ring_advance(rq, cs); + + return 0; +} + +int gen6_emit_flush_rcs(struct i915_request *rq, u32 mode) +{ + u32 scratch_addr = + intel_gt_scratch_offset(rq->engine->gt, + INTEL_GT_SCRATCH_FIELD_RENDER_FLUSH); + u32 *cs, flags = 0; + int ret; + + /* Force SNB workarounds for PIPE_CONTROL flushes */ + ret = gen6_emit_post_sync_nonzero_flush(rq); + if (ret) + return ret; + + /* + * Just flush everything. Experiments have shown that reducing the + * number of bits based on the write domains has little performance + * impact. And when rearranging requests, the order of flushes is + * unknown. + */ + if (mode & EMIT_FLUSH) { + flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH; + flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH; + /* + * Ensure that any following seqno writes only happen + * when the render cache is indeed flushed. + */ + flags |= PIPE_CONTROL_CS_STALL; + } + if (mode & EMIT_INVALIDATE) { + flags |= PIPE_CONTROL_TLB_INVALIDATE; + flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE; + /* + * TLB invalidate requires a post-sync write. + */ + flags |= PIPE_CONTROL_QW_WRITE | PIPE_CONTROL_CS_STALL; + } + + cs = intel_ring_begin(rq, 4); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = GFX_OP_PIPE_CONTROL(4); + *cs++ = flags; + *cs++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT; + *cs++ = 0; + intel_ring_advance(rq, cs); + + return 0; +} + +u32 *gen6_emit_breadcrumb_rcs(struct i915_request *rq, u32 *cs) +{ + /* First we do the gen6_emit_post_sync_nonzero_flush w/a */ + *cs++ = GFX_OP_PIPE_CONTROL(4); + *cs++ = PIPE_CONTROL_CS_STALL | PIPE_CONTROL_STALL_AT_SCOREBOARD; + *cs++ = 0; + *cs++ = 0; + + *cs++ = GFX_OP_PIPE_CONTROL(4); + *cs++ = PIPE_CONTROL_QW_WRITE; + *cs++ = intel_gt_scratch_offset(rq->engine->gt, + INTEL_GT_SCRATCH_FIELD_DEFAULT) | + PIPE_CONTROL_GLOBAL_GTT; + *cs++ = 0; + + /* Finally we can flush and with it emit the breadcrumb */ + *cs++ = GFX_OP_PIPE_CONTROL(4); + *cs++ = (PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH | + PIPE_CONTROL_DEPTH_CACHE_FLUSH | + PIPE_CONTROL_DC_FLUSH_ENABLE | + PIPE_CONTROL_QW_WRITE | + PIPE_CONTROL_CS_STALL); + *cs++ = i915_request_active_timeline(rq)->hwsp_offset | + PIPE_CONTROL_GLOBAL_GTT; + *cs++ = rq->fence.seqno; + + *cs++ = MI_USER_INTERRUPT; + *cs++ = MI_NOOP; + + rq->tail = intel_ring_offset(rq, cs); + assert_ring_tail_valid(rq->ring, rq->tail); + + return cs; +} + +static int mi_flush_dw(struct i915_request *rq, u32 flags) +{ + u32 cmd, *cs; + + cs = intel_ring_begin(rq, 4); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + cmd = MI_FLUSH_DW; + + /* + * We always require a command barrier so that subsequent + * commands, such as breadcrumb interrupts, are strictly ordered + * wrt the contents of the write cache being flushed to memory + * (and thus being coherent from the CPU). + */ + cmd |= MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW; + + /* + * Bspec vol 1c.3 - blitter engine command streamer: + * "If ENABLED, all TLBs will be invalidated once the flush + * operation is complete. This bit is only valid when the + * Post-Sync Operation field is a value of 1h or 3h." + */ + cmd |= flags; + + *cs++ = cmd; + *cs++ = HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT; + *cs++ = 0; + *cs++ = MI_NOOP; + + intel_ring_advance(rq, cs); + + return 0; +} + +static int gen6_flush_dw(struct i915_request *rq, u32 mode, u32 invflags) +{ + return mi_flush_dw(rq, mode & EMIT_INVALIDATE ? invflags : 0); +} + +int gen6_emit_flush_xcs(struct i915_request *rq, u32 mode) +{ + return gen6_flush_dw(rq, mode, MI_INVALIDATE_TLB); +} + +int gen6_emit_flush_vcs(struct i915_request *rq, u32 mode) +{ + return gen6_flush_dw(rq, mode, MI_INVALIDATE_TLB | MI_INVALIDATE_BSD); +} + +int gen6_emit_bb_start(struct i915_request *rq, + u64 offset, u32 len, + unsigned int dispatch_flags) +{ + u32 security; + u32 *cs; + + security = MI_BATCH_NON_SECURE_I965; + if (dispatch_flags & I915_DISPATCH_SECURE) + security = 0; + + cs = intel_ring_begin(rq, 2); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + cs = __gen6_emit_bb_start(cs, offset, security); + intel_ring_advance(rq, cs); + + return 0; +} + +int +hsw_emit_bb_start(struct i915_request *rq, + u64 offset, u32 len, + unsigned int dispatch_flags) +{ + u32 security; + u32 *cs; + + security = MI_BATCH_PPGTT_HSW | MI_BATCH_NON_SECURE_HSW; + if (dispatch_flags & I915_DISPATCH_SECURE) + security = 0; + + cs = intel_ring_begin(rq, 2); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + cs = __gen6_emit_bb_start(cs, offset, security); + intel_ring_advance(rq, cs); + + return 0; +} + +static int gen7_stall_cs(struct i915_request *rq) +{ + u32 *cs; + + cs = intel_ring_begin(rq, 4); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = GFX_OP_PIPE_CONTROL(4); + *cs++ = PIPE_CONTROL_CS_STALL | PIPE_CONTROL_STALL_AT_SCOREBOARD; + *cs++ = 0; + *cs++ = 0; + intel_ring_advance(rq, cs); + + return 0; +} + +int gen7_emit_flush_rcs(struct i915_request *rq, u32 mode) +{ + u32 scratch_addr = + intel_gt_scratch_offset(rq->engine->gt, + INTEL_GT_SCRATCH_FIELD_RENDER_FLUSH); + u32 *cs, flags = 0; + + /* + * Ensure that any following seqno writes only happen when the render + * cache is indeed flushed. + * + * Workaround: 4th PIPE_CONTROL command (except the ones with only + * read-cache invalidate bits set) must have the CS_STALL bit set. We + * don't try to be clever and just set it unconditionally. + */ + flags |= PIPE_CONTROL_CS_STALL; + + /* + * CS_STALL suggests at least a post-sync write. + */ + flags |= PIPE_CONTROL_QW_WRITE; + flags |= PIPE_CONTROL_GLOBAL_GTT_IVB; + + /* + * Just flush everything. Experiments have shown that reducing the + * number of bits based on the write domains has little performance + * impact. + */ + if (mode & EMIT_FLUSH) { + flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH; + flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH; + flags |= PIPE_CONTROL_DC_FLUSH_ENABLE; + flags |= PIPE_CONTROL_FLUSH_ENABLE; + } + if (mode & EMIT_INVALIDATE) { + flags |= PIPE_CONTROL_TLB_INVALIDATE; + flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_MEDIA_STATE_CLEAR; + + /* + * Workaround: we must issue a pipe_control with CS-stall bit + * set before a pipe_control command that has the state cache + * invalidate bit set. + */ + gen7_stall_cs(rq); + } + + cs = intel_ring_begin(rq, 4); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = GFX_OP_PIPE_CONTROL(4); + *cs++ = flags; + *cs++ = scratch_addr; + *cs++ = 0; + intel_ring_advance(rq, cs); + + return 0; +} + +u32 *gen7_emit_breadcrumb_rcs(struct i915_request *rq, u32 *cs) +{ + *cs++ = GFX_OP_PIPE_CONTROL(4); + *cs++ = (PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH | + PIPE_CONTROL_DEPTH_CACHE_FLUSH | + PIPE_CONTROL_DC_FLUSH_ENABLE | + PIPE_CONTROL_FLUSH_ENABLE | + PIPE_CONTROL_QW_WRITE | + PIPE_CONTROL_GLOBAL_GTT_IVB | + PIPE_CONTROL_CS_STALL); + *cs++ = i915_request_active_timeline(rq)->hwsp_offset; + *cs++ = rq->fence.seqno; + + *cs++ = MI_USER_INTERRUPT; + *cs++ = MI_NOOP; + + rq->tail = intel_ring_offset(rq, cs); + assert_ring_tail_valid(rq->ring, rq->tail); + + return cs; +} + +u32 *gen6_emit_breadcrumb_xcs(struct i915_request *rq, u32 *cs) +{ + GEM_BUG_ON(i915_request_active_timeline(rq)->hwsp_ggtt != rq->engine->status_page.vma); + GEM_BUG_ON(offset_in_page(i915_request_active_timeline(rq)->hwsp_offset) != I915_GEM_HWS_SEQNO_ADDR); + + *cs++ = MI_FLUSH_DW | MI_FLUSH_DW_OP_STOREDW | MI_FLUSH_DW_STORE_INDEX; + *cs++ = I915_GEM_HWS_SEQNO_ADDR | MI_FLUSH_DW_USE_GTT; + *cs++ = rq->fence.seqno; + + *cs++ = MI_USER_INTERRUPT; + + rq->tail = intel_ring_offset(rq, cs); + assert_ring_tail_valid(rq->ring, rq->tail); + + return cs; +} + +#define GEN7_XCS_WA 32 +u32 *gen7_emit_breadcrumb_xcs(struct i915_request *rq, u32 *cs) +{ + int i; + + GEM_BUG_ON(i915_request_active_timeline(rq)->hwsp_ggtt != rq->engine->status_page.vma); + GEM_BUG_ON(offset_in_page(i915_request_active_timeline(rq)->hwsp_offset) != I915_GEM_HWS_SEQNO_ADDR); + + *cs++ = MI_FLUSH_DW | MI_INVALIDATE_TLB | + MI_FLUSH_DW_OP_STOREDW | MI_FLUSH_DW_STORE_INDEX; + *cs++ = I915_GEM_HWS_SEQNO_ADDR | MI_FLUSH_DW_USE_GTT; + *cs++ = rq->fence.seqno; + + for (i = 0; i < GEN7_XCS_WA; i++) { + *cs++ = MI_STORE_DWORD_INDEX; + *cs++ = I915_GEM_HWS_SEQNO_ADDR; + *cs++ = rq->fence.seqno; + } + + *cs++ = MI_FLUSH_DW; + *cs++ = 0; + *cs++ = 0; + + *cs++ = MI_USER_INTERRUPT; + *cs++ = MI_NOOP; + + rq->tail = intel_ring_offset(rq, cs); + assert_ring_tail_valid(rq->ring, rq->tail); + + return cs; +} +#undef GEN7_XCS_WA + +void gen6_irq_enable(struct intel_engine_cs *engine) +{ + ENGINE_WRITE(engine, RING_IMR, + ~(engine->irq_enable_mask | engine->irq_keep_mask)); + + /* Flush/delay to ensure the RING_IMR is active before the GT IMR */ + ENGINE_POSTING_READ(engine, RING_IMR); + + gen5_gt_enable_irq(engine->gt, engine->irq_enable_mask); +} + +void gen6_irq_disable(struct intel_engine_cs *engine) +{ + ENGINE_WRITE(engine, RING_IMR, ~engine->irq_keep_mask); + gen5_gt_disable_irq(engine->gt, engine->irq_enable_mask); +} + +void hsw_irq_enable_vecs(struct intel_engine_cs *engine) +{ + ENGINE_WRITE(engine, RING_IMR, ~engine->irq_enable_mask); + + /* Flush/delay to ensure the RING_IMR is active before the GT IMR */ + ENGINE_POSTING_READ(engine, RING_IMR); + + gen6_gt_pm_unmask_irq(engine->gt, engine->irq_enable_mask); +} + +void hsw_irq_disable_vecs(struct intel_engine_cs *engine) +{ + ENGINE_WRITE(engine, RING_IMR, ~0); + gen6_gt_pm_mask_irq(engine->gt, engine->irq_enable_mask); +} diff --git a/drivers/gpu/drm/i915/gt/gen6_engine_cs.h b/drivers/gpu/drm/i915/gt/gen6_engine_cs.h new file mode 100644 index 000000000000..76c6bc9f3bde --- /dev/null +++ b/drivers/gpu/drm/i915/gt/gen6_engine_cs.h @@ -0,0 +1,39 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright © 2020 Intel Corporation + */ + +#ifndef __GEN6_ENGINE_CS_H__ +#define __GEN6_ENGINE_CS_H__ + +#include <linux/types.h> + +#include "intel_gpu_commands.h" + +struct i915_request; +struct intel_engine_cs; + +int gen6_emit_flush_rcs(struct i915_request *rq, u32 mode); +int gen6_emit_flush_vcs(struct i915_request *rq, u32 mode); +int gen6_emit_flush_xcs(struct i915_request *rq, u32 mode); +u32 *gen6_emit_breadcrumb_rcs(struct i915_request *rq, u32 *cs); +u32 *gen6_emit_breadcrumb_xcs(struct i915_request *rq, u32 *cs); + +int gen7_emit_flush_rcs(struct i915_request *rq, u32 mode); +u32 *gen7_emit_breadcrumb_rcs(struct i915_request *rq, u32 *cs); +u32 *gen7_emit_breadcrumb_xcs(struct i915_request *rq, u32 *cs); + +int gen6_emit_bb_start(struct i915_request *rq, + u64 offset, u32 len, + unsigned int dispatch_flags); +int hsw_emit_bb_start(struct i915_request *rq, + u64 offset, u32 len, + unsigned int dispatch_flags); + +void gen6_irq_enable(struct intel_engine_cs *engine); +void gen6_irq_disable(struct intel_engine_cs *engine); + +void hsw_irq_enable_vecs(struct intel_engine_cs *engine); +void hsw_irq_disable_vecs(struct intel_engine_cs *engine); + +#endif /* __GEN6_ENGINE_CS_H__ */ diff --git a/drivers/gpu/drm/i915/gt/intel_context_sseu.c b/drivers/gpu/drm/i915/gt/intel_context_sseu.c index 487299cb91f2..27ae48049239 100644 --- a/drivers/gpu/drm/i915/gt/intel_context_sseu.c +++ b/drivers/gpu/drm/i915/gt/intel_context_sseu.c @@ -30,7 +30,7 @@ static int gen8_emit_rpcs_config(struct i915_request *rq, *cs++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT; *cs++ = lower_32_bits(offset); *cs++ = upper_32_bits(offset); - *cs++ = intel_sseu_make_rpcs(rq->i915, &sseu); + *cs++ = intel_sseu_make_rpcs(rq->engine->i915, &sseu); intel_ring_advance(rq, cs); diff --git a/drivers/gpu/drm/i915/gt/intel_engine.h b/drivers/gpu/drm/i915/gt/intel_engine.h index 9bf6d4989968..a9249a23903a 100644 --- a/drivers/gpu/drm/i915/gt/intel_engine.h +++ b/drivers/gpu/drm/i915/gt/intel_engine.h @@ -187,7 +187,6 @@ intel_write_status_page(struct intel_engine_cs *engine, int reg, u32 value) #define I915_GEM_HWS_SEQNO 0x40 #define I915_GEM_HWS_SEQNO_ADDR (I915_GEM_HWS_SEQNO * sizeof(u32)) #define I915_GEM_HWS_SCRATCH 0x80 -#define I915_GEM_HWS_SCRATCH_ADDR (I915_GEM_HWS_SCRATCH * sizeof(u32)) #define I915_HWS_CSB_BUF0_INDEX 0x10 #define I915_HWS_CSB_WRITE_INDEX 0x1f @@ -335,7 +334,8 @@ void intel_engine_dump(struct intel_engine_cs *engine, struct drm_printer *m, const char *header, ...); -ktime_t intel_engine_get_busy_time(struct intel_engine_cs *engine); +ktime_t intel_engine_get_busy_time(struct intel_engine_cs *engine, + ktime_t *now); struct i915_request * intel_engine_find_active_request(struct intel_engine_cs *engine); diff --git a/drivers/gpu/drm/i915/gt/intel_engine_cs.c b/drivers/gpu/drm/i915/gt/intel_engine_cs.c index 8691eb61e185..7bf2f76212f0 100644 --- a/drivers/gpu/drm/i915/gt/intel_engine_cs.c +++ b/drivers/gpu/drm/i915/gt/intel_engine_cs.c @@ -414,12 +414,12 @@ void intel_engines_release(struct intel_gt *gt) /* Decouple the backend; but keep the layout for late GPU resets */ for_each_engine(engine, gt, id) { - intel_wakeref_wait_for_idle(&engine->wakeref); - GEM_BUG_ON(intel_engine_pm_is_awake(engine)); - if (!engine->release) continue; + intel_wakeref_wait_for_idle(&engine->wakeref); + GEM_BUG_ON(intel_engine_pm_is_awake(engine)); + engine->release(engine); engine->release = NULL; @@ -661,7 +661,6 @@ static int measure_breadcrumb_dw(struct intel_context *ce) if (!frame) return -ENOMEM; - frame->rq.i915 = engine->i915; frame->rq.engine = engine; frame->rq.context = ce; rcu_assign_pointer(frame->rq.timeline, ce->timeline); @@ -1095,19 +1094,21 @@ void intel_engine_flush_submission(struct intel_engine_cs *engine) { struct tasklet_struct *t = &engine->execlists.tasklet; - if (__tasklet_is_scheduled(t)) { - local_bh_disable(); - if (tasklet_trylock(t)) { - /* Must wait for any GPU reset in progress. */ - if (__tasklet_is_enabled(t)) - t->func(t->data); - tasklet_unlock(t); - } - local_bh_enable(); - } + if (!t->func) + return; - /* Otherwise flush the tasklet if it was running on another cpu */ - tasklet_unlock_wait(t); + /* Synchronise and wait for the tasklet on another CPU */ + tasklet_kill(t); + + /* Having cancelled the tasklet, ensure that is run */ + local_bh_disable(); + if (tasklet_trylock(t)) { + /* Must wait for any GPU reset in progress. */ + if (__tasklet_is_enabled(t)) + t->func(t->data); + tasklet_unlock(t); + } + local_bh_enable(); } /** @@ -1194,8 +1195,7 @@ bool intel_engine_can_store_dword(struct intel_engine_cs *engine) } } -static int print_sched_attr(struct drm_i915_private *i915, - const struct i915_sched_attr *attr, +static int print_sched_attr(const struct i915_sched_attr *attr, char *buf, int x, int len) { if (attr->priority == I915_PRIORITY_INVALID) @@ -1215,7 +1215,7 @@ static void print_request(struct drm_printer *m, char buf[80] = ""; int x = 0; - x = print_sched_attr(rq->i915, &rq->sched.attr, buf, x, sizeof(buf)); + x = print_sched_attr(&rq->sched.attr, buf, x, sizeof(buf)); drm_printf(m, "%s %llx:%llx%s%s %s @ %dms: %s\n", prefix, @@ -1423,9 +1423,11 @@ static void intel_engine_print_registers(struct intel_engine_cs *engine, int len; len = scnprintf(hdr, sizeof(hdr), - "\t\tActive[%d]: ccid:%08x, ", + "\t\tActive[%d]: ccid:%08x%s%s, ", (int)(port - execlists->active), - rq->context->lrc.ccid); + rq->context->lrc.ccid, + intel_context_is_closed(rq->context) ? "!" : "", + intel_context_is_banned(rq->context) ? "*" : ""); len += print_ring(hdr + len, sizeof(hdr) - len, rq); scnprintf(hdr + len, sizeof(hdr) - len, "rq: "); print_request(m, rq, hdr); @@ -1435,9 +1437,11 @@ static void intel_engine_print_registers(struct intel_engine_cs *engine, int len; len = scnprintf(hdr, sizeof(hdr), - "\t\tPending[%d]: ccid:%08x, ", + "\t\tPending[%d]: ccid:%08x%s%s, ", (int)(port - execlists->pending), - rq->context->lrc.ccid); + rq->context->lrc.ccid, + intel_context_is_closed(rq->context) ? "!" : "", + intel_context_is_banned(rq->context) ? "*" : ""); len += print_ring(hdr + len, sizeof(hdr) - len, rq); scnprintf(hdr + len, sizeof(hdr) - len, "rq: "); print_request(m, rq, hdr); @@ -1506,6 +1510,7 @@ void intel_engine_dump(struct intel_engine_cs *engine, struct i915_request *rq; intel_wakeref_t wakeref; unsigned long flags; + ktime_t dummy; if (header) { va_list ap; @@ -1523,6 +1528,12 @@ void intel_engine_dump(struct intel_engine_cs *engine, yesno(!llist_empty(&engine->barrier_tasks))); drm_printf(m, "\tLatency: %luus\n", ewma__engine_latency_read(&engine->latency)); + if (intel_engine_supports_stats(engine)) + drm_printf(m, "\tRuntime: %llums\n", + ktime_to_ms(intel_engine_get_busy_time(engine, + &dummy))); + drm_printf(m, "\tForcewake: %x domains, %d active\n", + engine->fw_domain, atomic_read(&engine->fw_active)); rcu_read_lock(); rq = READ_ONCE(engine->heartbeat.systole); @@ -1589,7 +1600,8 @@ void intel_engine_dump(struct intel_engine_cs *engine, intel_engine_print_breadcrumbs(engine, m); } -static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine) +static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine, + ktime_t *now) { ktime_t total = engine->stats.total; @@ -1597,9 +1609,9 @@ static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine) * If the engine is executing something at the moment * add it to the total. */ + *now = ktime_get(); if (atomic_read(&engine->stats.active)) - total = ktime_add(total, - ktime_sub(ktime_get(), engine->stats.start)); + total = ktime_add(total, ktime_sub(*now, engine->stats.start)); return total; } @@ -1607,17 +1619,18 @@ static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine) /** * intel_engine_get_busy_time() - Return current accumulated engine busyness * @engine: engine to report on + * @now: monotonic timestamp of sampling * * Returns accumulated time @engine was busy since engine stats were enabled. */ -ktime_t intel_engine_get_busy_time(struct intel_engine_cs *engine) +ktime_t intel_engine_get_busy_time(struct intel_engine_cs *engine, ktime_t *now) { unsigned int seq; ktime_t total; do { seq = read_seqbegin(&engine->stats.lock); - total = __intel_engine_get_busy_time(engine); + total = __intel_engine_get_busy_time(engine, now); } while (read_seqretry(&engine->stats.lock, seq)); return total; diff --git a/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.c b/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.c index 5136c8bf112d..8db7e93abde5 100644 --- a/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.c +++ b/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.c @@ -4,6 +4,7 @@ * Copyright © 2019 Intel Corporation */ +#include "i915_drv.h" #include "i915_request.h" #include "intel_context.h" @@ -48,8 +49,10 @@ static void show_heartbeat(const struct i915_request *rq, struct drm_printer p = drm_debug_printer("heartbeat"); intel_engine_dump(engine, &p, - "%s heartbeat {prio:%d} not ticking\n", + "%s heartbeat {seqno:%llx:%lld, prio:%d} not ticking\n", engine->name, + rq->fence.context, + rq->fence.seqno, rq->sched.attr.priority); } @@ -63,6 +66,9 @@ static void heartbeat(struct work_struct *wrk) struct intel_context *ce = engine->kernel_context; struct i915_request *rq; + /* Just in case everything has gone horribly wrong, give it a kick */ + intel_engine_flush_submission(engine); + rq = engine->heartbeat.systole; if (rq && i915_request_completed(rq)) { i915_request_put(rq); @@ -76,8 +82,19 @@ static void heartbeat(struct work_struct *wrk) goto out; if (engine->heartbeat.systole) { - if (engine->schedule && - rq->sched.attr.priority < I915_PRIORITY_BARRIER) { + if (!i915_sw_fence_signaled(&rq->submit)) { + /* + * Not yet submitted, system is stalled. + * + * This more often happens for ring submission, + * where all contexts are funnelled into a common + * ringbuffer. If one context is blocked on an + * external fence, not only is it not submitted, + * but all other contexts, including the kernel + * context are stuck waiting for the signal. + */ + } else if (engine->schedule && + rq->sched.attr.priority < I915_PRIORITY_BARRIER) { /* * Gradually raise the priority of the heartbeat to * give high priority work [which presumably desires @@ -117,7 +134,7 @@ static void heartbeat(struct work_struct *wrk) goto unlock; idle_pulse(engine, rq); - if (i915_modparams.enable_hangcheck) + if (engine->i915->params.enable_hangcheck) engine->heartbeat.systole = i915_request_get(rq); __i915_request_commit(rq); diff --git a/drivers/gpu/drm/i915/gt/intel_engine_types.h b/drivers/gpu/drm/i915/gt/intel_engine_types.h index 2b6cdf47d428..073c3769e8cc 100644 --- a/drivers/gpu/drm/i915/gt/intel_engine_types.h +++ b/drivers/gpu/drm/i915/gt/intel_engine_types.h @@ -24,6 +24,7 @@ #include "i915_selftest.h" #include "intel_sseu.h" #include "intel_timeline_types.h" +#include "intel_uncore.h" #include "intel_wakeref.h" #include "intel_workarounds_types.h" @@ -313,6 +314,16 @@ struct intel_engine_cs { u32 context_size; u32 mmio_base; + /* + * Some w/a require forcewake to be held (which prevents RC6) while + * a particular engine is active. If so, we set fw_domain to which + * domains need to be held for the duration of request activity, + * and 0 if none. We try to limit the duration of the hold as much + * as possible. + */ + enum forcewake_domains fw_domain; + atomic_t fw_active; + unsigned long context_tag; struct rb_node uabi_node; diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt.c b/drivers/gpu/drm/i915/gt/intel_ggtt.c index 66165b10256e..323c328d444a 100644 --- a/drivers/gpu/drm/i915/gt/intel_ggtt.c +++ b/drivers/gpu/drm/i915/gt/intel_ggtt.c @@ -108,13 +108,32 @@ static bool needs_idle_maps(struct drm_i915_private *i915) void i915_ggtt_suspend(struct i915_ggtt *ggtt) { - struct i915_vma *vma; + struct i915_vma *vma, *vn; + int open; + + mutex_lock(&ggtt->vm.mutex); + + /* Skip rewriting PTE on VMA unbind. */ + open = atomic_xchg(&ggtt->vm.open, 0); - list_for_each_entry(vma, &ggtt->vm.bound_list, vm_link) + list_for_each_entry_safe(vma, vn, &ggtt->vm.bound_list, vm_link) { + GEM_BUG_ON(!drm_mm_node_allocated(&vma->node)); i915_vma_wait_for_bind(vma); + if (i915_vma_is_pinned(vma)) + continue; + + if (!i915_vma_is_bound(vma, I915_VMA_GLOBAL_BIND)) { + __i915_vma_evict(vma); + drm_mm_remove_node(&vma->node); + } + } + ggtt->vm.clear_range(&ggtt->vm, 0, ggtt->vm.total); ggtt->invalidate(ggtt); + atomic_set(&ggtt->vm.open, open); + + mutex_unlock(&ggtt->vm.mutex); intel_gt_check_and_clear_faults(ggtt->vm.gt); } @@ -424,22 +443,17 @@ static int ggtt_bind_vma(struct i915_vma *vma, struct drm_i915_gem_object *obj = vma->obj; u32 pte_flags; + if (i915_vma_is_bound(vma, ~flags & I915_VMA_BIND_MASK)) + return 0; + /* Applicable to VLV (gen8+ do not support RO in the GGTT) */ pte_flags = 0; if (i915_gem_object_is_readonly(obj)) pte_flags |= PTE_READ_ONLY; vma->vm->insert_entries(vma->vm, vma, cache_level, pte_flags); - vma->page_sizes.gtt = I915_GTT_PAGE_SIZE; - /* - * Without aliasing PPGTT there's no difference between - * GLOBAL/LOCAL_BIND, it's all the same ptes. Hence unconditionally - * upgrade to both bound if we bind either to avoid double-binding. - */ - atomic_or(I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND, &vma->flags); - return 0; } @@ -1166,6 +1180,11 @@ void i915_ggtt_disable_guc(struct i915_ggtt *ggtt) ggtt->invalidate(ggtt); } +static unsigned int clear_bind(struct i915_vma *vma) +{ + return atomic_fetch_and(~I915_VMA_BIND_MASK, &vma->flags); +} + void i915_ggtt_resume(struct i915_ggtt *ggtt) { struct i915_vma *vma; @@ -1183,14 +1202,11 @@ void i915_ggtt_resume(struct i915_ggtt *ggtt) /* clflush objects bound into the GGTT and rebind them. */ list_for_each_entry(vma, &ggtt->vm.bound_list, vm_link) { struct drm_i915_gem_object *obj = vma->obj; + unsigned int was_bound = clear_bind(vma); - if (!i915_vma_is_bound(vma, I915_VMA_GLOBAL_BIND)) - continue; - - clear_bit(I915_VMA_GLOBAL_BIND_BIT, __i915_vma_flags(vma)); WARN_ON(i915_vma_bind(vma, obj ? obj->cache_level : 0, - PIN_GLOBAL, NULL)); + was_bound, NULL)); if (obj) { /* only used during resume => exclusive access */ flush |= fetch_and_zero(&obj->write_domain); obj->read_domains |= I915_GEM_DOMAIN_GTT; diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c index f069551e412f..ebc29b6ee86c 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt.c +++ b/drivers/gpu/drm/i915/gt/intel_gt.c @@ -616,6 +616,11 @@ void intel_gt_driver_unregister(struct intel_gt *gt) void intel_gt_driver_release(struct intel_gt *gt) { struct i915_address_space *vm; + intel_wakeref_t wakeref; + + /* Scrub all HW state upon release */ + with_intel_runtime_pm(gt->uncore->rpm, wakeref) + __intel_gt_reset(gt, ALL_ENGINES); vm = fetch_and_zero(>->vm); if (vm) /* FIXME being called twice on error paths :( */ diff --git a/drivers/gpu/drm/i915/gt/intel_gt_buffer_pool.c b/drivers/gpu/drm/i915/gt/intel_gt_buffer_pool.c index 1495054a4305..418ae184cecf 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt_buffer_pool.c +++ b/drivers/gpu/drm/i915/gt/intel_gt_buffer_pool.c @@ -212,8 +212,9 @@ void intel_gt_flush_buffer_pool(struct intel_gt *gt) { struct intel_gt_buffer_pool *pool = >->buffer_pool; - if (cancel_delayed_work_sync(&pool->work)) + do { pool_free_imm(pool); + } while (cancel_delayed_work_sync(&pool->work)); } void intel_gt_fini_buffer_pool(struct intel_gt *gt) diff --git a/drivers/gpu/drm/i915/gt/intel_gt_pm.c b/drivers/gpu/drm/i915/gt/intel_gt_pm.c index 6bdb434a442d..f1d5333f9456 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt_pm.c +++ b/drivers/gpu/drm/i915/gt/intel_gt_pm.c @@ -214,8 +214,8 @@ int intel_gt_resume(struct intel_gt *gt) /* Only when the HW is re-initialised, can we replay the requests */ err = intel_gt_init_hw(gt); if (err) { - drm_err(>->i915->drm, - "Failed to initialize GPU, declaring it wedged!\n"); + i915_probe_error(gt->i915, + "Failed to initialize GPU, declaring it wedged!\n"); goto err_wedged; } diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c index 7c3d8ef4a47c..e866b8d721ed 100644 --- a/drivers/gpu/drm/i915/gt/intel_lrc.c +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c @@ -446,6 +446,9 @@ static int queue_prio(const struct intel_engine_execlists *execlists) * we have to flip the index value to become priority. */ p = to_priolist(rb); + if (!I915_USER_PRIORITY_SHIFT) + return p->priority; + return ((p->priority + 1) << I915_USER_PRIORITY_SHIFT) - ffs(p->used); } @@ -1377,6 +1380,8 @@ __execlists_schedule_in(struct i915_request *rq) ce->lrc.ccid |= engine->execlists.ccid; __intel_gt_pm_get(engine->gt); + if (engine->fw_domain && !atomic_fetch_inc(&engine->fw_active)) + intel_uncore_forcewake_get(engine->uncore, engine->fw_domain); execlists_context_status_change(rq, INTEL_CONTEXT_SCHEDULE_IN); intel_engine_context_in(engine); @@ -1409,8 +1414,8 @@ static void kick_siblings(struct i915_request *rq, struct intel_context *ce) struct virtual_engine *ve = container_of(ce, typeof(*ve), context); struct i915_request *next = READ_ONCE(ve->request); - if (next && next->execution_mask & ~rq->execution_mask) - tasklet_schedule(&ve->base.execlists.tasklet); + if (next == rq || (next && next->execution_mask & ~rq->execution_mask)) + tasklet_hi_schedule(&ve->base.execlists.tasklet); } static inline void @@ -1445,6 +1450,8 @@ __execlists_schedule_out(struct i915_request *rq, intel_context_update_runtime(ce); intel_engine_context_out(engine); execlists_context_status_change(rq, INTEL_CONTEXT_SCHEDULE_OUT); + if (engine->fw_domain && !atomic_dec_return(&engine->fw_active)) + intel_uncore_forcewake_put(engine->uncore, engine->fw_domain); intel_gt_pm_put_async(engine->gt); /* @@ -1636,9 +1643,9 @@ assert_pending_valid(const struct intel_engine_execlists *execlists, ccid = ce->lrc.ccid; /* - * Sentinels are supposed to be lonely so they flush the - * current exection off the HW. Check that they are the - * only request in the pending submission. + * Sentinels are supposed to be the last request so they flush + * the current execution off the HW. Check that they are the only + * request in the pending submission. */ if (sentinel) { GEM_TRACE_ERR("%s: context:%llx after sentinel in pending[%zd]\n", @@ -1647,15 +1654,7 @@ assert_pending_valid(const struct intel_engine_execlists *execlists, port - execlists->pending); return false; } - sentinel = i915_request_has_sentinel(rq); - if (sentinel && port != execlists->pending) { - GEM_TRACE_ERR("%s: sentinel context:%llx not in prime position[%zd]\n", - engine->name, - ce->timeline->fence_context, - port - execlists->pending); - return false; - } /* Hold tightly onto the lock to prevent concurrent retires! */ if (!spin_trylock_irqsave(&rq->lock, flags)) @@ -1967,7 +1966,7 @@ static int switch_prio(struct intel_engine_cs *engine, const struct i915_request *rq) { if (list_is_last(&rq->sched.link, &engine->active.requests)) - return INT_MIN; + return engine->execlists.queue_priority_hint; return rq_prio(list_next_entry(rq, sched.link)); } @@ -2445,6 +2444,7 @@ done: set_preempt_timeout(engine, *active); execlists_submit_ports(engine); } else { + start_timeslice(engine, execlists->queue_priority_hint); skip_submit: ring_set_paused(engine, 0); } @@ -3170,13 +3170,6 @@ static void __submit_queue_imm(struct intel_engine_cs *engine) if (reset_in_progress(execlists)) return; /* defer until we restart the engine following reset */ - /* Hopefully we clear execlists->pending[] to let us through */ - if (READ_ONCE(execlists->pending[0]) && - tasklet_trylock(&execlists->tasklet)) { - process_csb(engine); - tasklet_unlock(&execlists->tasklet); - } - __execlists_submission_tasklet(engine); } @@ -3199,11 +3192,25 @@ static bool ancestor_on_hold(const struct intel_engine_cs *engine, return !list_empty(&engine->active.hold) && hold_request(rq); } +static void flush_csb(struct intel_engine_cs *engine) +{ + struct intel_engine_execlists *el = &engine->execlists; + + if (READ_ONCE(el->pending[0]) && tasklet_trylock(&el->tasklet)) { + if (!reset_in_progress(el)) + process_csb(engine); + tasklet_unlock(&el->tasklet); + } +} + static void execlists_submit_request(struct i915_request *request) { struct intel_engine_cs *engine = request->engine; unsigned long flags; + /* Hopefully we clear execlists->pending[] to let us through */ + flush_csb(engine); + /* Will be called from irq-context when using foreign fences. */ spin_lock_irqsave(&engine->active.lock, flags); @@ -3536,7 +3543,7 @@ static int emit_pdps(struct i915_request *rq) int err, i; u32 *cs; - GEM_BUG_ON(intel_vgpu_active(rq->i915)); + GEM_BUG_ON(intel_vgpu_active(rq->engine->i915)); /* * Beware ye of the dragons, this sequence is magic! @@ -4515,11 +4522,11 @@ static int gen8_emit_flush_render(struct i915_request *request, * On GEN9: before VF_CACHE_INVALIDATE we need to emit a NULL * pipe control. */ - if (IS_GEN(request->i915, 9)) + if (IS_GEN(request->engine->i915, 9)) vf_flush_wa = true; /* WaForGAMHang:kbl */ - if (IS_KBL_REVID(request->i915, 0, KBL_REVID_B0)) + if (IS_KBL_REVID(request->engine->i915, 0, KBL_REVID_B0)) dc_flush_wa = true; } @@ -5598,7 +5605,7 @@ static void virtual_submit_request(struct i915_request *rq) GEM_BUG_ON(!list_empty(virtual_queue(ve))); list_move_tail(&rq->sched.link, virtual_queue(ve)); - tasklet_schedule(&ve->base.execlists.tasklet); + tasklet_hi_schedule(&ve->base.execlists.tasklet); } spin_unlock_irqrestore(&ve->base.active.lock, flags); diff --git a/drivers/gpu/drm/i915/gt/intel_renderstate.c b/drivers/gpu/drm/i915/gt/intel_renderstate.c index f59e7875cc5e..6db23389e427 100644 --- a/drivers/gpu/drm/i915/gt/intel_renderstate.c +++ b/drivers/gpu/drm/i915/gt/intel_renderstate.c @@ -61,7 +61,7 @@ render_state_get_rodata(const struct intel_engine_cs *engine) #define OUT_BATCH(batch, i, val) \ do { \ if ((i) >= PAGE_SIZE / sizeof(u32)) \ - goto err; \ + goto out; \ (batch)[(i)++] = (val); \ } while(0) @@ -70,15 +70,12 @@ static int render_state_setup(struct intel_renderstate *so, { const struct intel_renderstate_rodata *rodata = so->rodata; unsigned int i = 0, reloc_index = 0; - unsigned int needs_clflush; + int ret = -EINVAL; u32 *d; - int ret; - ret = i915_gem_object_prepare_write(so->vma->obj, &needs_clflush); - if (ret) - return ret; - - d = kmap_atomic(i915_gem_object_get_dirty_page(so->vma->obj, 0)); + d = i915_gem_object_pin_map(so->vma->obj, I915_MAP_WB); + if (IS_ERR(d)) + return PTR_ERR(d); while (i < rodata->batch_items) { u32 s = rodata->batch[i]; @@ -89,7 +86,7 @@ static int render_state_setup(struct intel_renderstate *so, if (HAS_64BIT_RELOC(i915)) { if (i + 1 >= rodata->batch_items || rodata->batch[i + 1] != 0) - goto err; + goto out; d[i++] = s; s = upper_32_bits(r); @@ -103,7 +100,7 @@ static int render_state_setup(struct intel_renderstate *so, if (rodata->reloc[reloc_index] != -1) { drm_err(&i915->drm, "only %d relocs resolved\n", reloc_index); - goto err; + goto out; } so->batch_offset = i915_ggtt_offset(so->vma); @@ -150,19 +147,11 @@ static int render_state_setup(struct intel_renderstate *so, */ so->aux_size = ALIGN(so->aux_size, 8); - if (needs_clflush) - drm_clflush_virt_range(d, i * sizeof(u32)); - kunmap_atomic(d); - ret = 0; out: - i915_gem_object_finish_access(so->vma->obj); + __i915_gem_object_flush_map(so->vma->obj, 0, i * sizeof(u32)); + i915_gem_object_unpin_map(so->vma->obj); return ret; - -err: - kunmap_atomic(d); - ret = -EINVAL; - goto out; } #undef OUT_BATCH diff --git a/drivers/gpu/drm/i915/gt/intel_reset.c b/drivers/gpu/drm/i915/gt/intel_reset.c index 39070b514e65..0156f1f5c736 100644 --- a/drivers/gpu/drm/i915/gt/intel_reset.c +++ b/drivers/gpu/drm/i915/gt/intel_reset.c @@ -638,7 +638,7 @@ int __intel_gt_reset(struct intel_gt *gt, intel_engine_mask_t engine_mask) bool intel_has_gpu_reset(const struct intel_gt *gt) { - if (!i915_modparams.reset) + if (!gt->i915->params.reset) return NULL; return intel_get_gpu_reset(gt); @@ -646,7 +646,7 @@ bool intel_has_gpu_reset(const struct intel_gt *gt) bool intel_has_reset_engine(const struct intel_gt *gt) { - if (i915_modparams.reset < 2) + if (gt->i915->params.reset < 2) return false; return INTEL_INFO(gt->i915)->has_reset_engine; @@ -1038,7 +1038,7 @@ void intel_gt_reset(struct intel_gt *gt, awake = reset_prepare(gt); if (!intel_has_gpu_reset(gt)) { - if (i915_modparams.reset) + if (gt->i915->params.reset) drm_err(>->i915->drm, "GPU reset not supported\n"); else drm_dbg(>->i915->drm, "GPU reset disabled\n"); diff --git a/drivers/gpu/drm/i915/gt/intel_ring_submission.c b/drivers/gpu/drm/i915/gt/intel_ring_submission.c index ca7286e58409..68a08486fc87 100644 --- a/drivers/gpu/drm/i915/gt/intel_ring_submission.c +++ b/drivers/gpu/drm/i915/gt/intel_ring_submission.c @@ -27,21 +27,15 @@ * */ -#include <linux/log2.h> - -#include "gem/i915_gem_context.h" - +#include "gen2_engine_cs.h" +#include "gen6_engine_cs.h" #include "gen6_ppgtt.h" #include "gen7_renderclear.h" #include "i915_drv.h" -#include "i915_trace.h" #include "intel_context.h" #include "intel_gt.h" -#include "intel_gt_irq.h" -#include "intel_gt_pm_irq.h" #include "intel_reset.h" #include "intel_ring.h" -#include "intel_workarounds.h" #include "shmem_utils.h" /* Rough estimate of the typical request size, performing a flush, @@ -49,436 +43,6 @@ */ #define LEGACY_REQUEST_SIZE 200 -static int -gen2_render_ring_flush(struct i915_request *rq, u32 mode) -{ - unsigned int num_store_dw; - u32 cmd, *cs; - - cmd = MI_FLUSH; - num_store_dw = 0; - if (mode & EMIT_INVALIDATE) - cmd |= MI_READ_FLUSH; - if (mode & EMIT_FLUSH) - num_store_dw = 4; - - cs = intel_ring_begin(rq, 2 + 3 * num_store_dw); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = cmd; - while (num_store_dw--) { - *cs++ = MI_STORE_DWORD_IMM | MI_MEM_VIRTUAL; - *cs++ = intel_gt_scratch_offset(rq->engine->gt, - INTEL_GT_SCRATCH_FIELD_DEFAULT); - *cs++ = 0; - } - *cs++ = MI_FLUSH | MI_NO_WRITE_FLUSH; - - intel_ring_advance(rq, cs); - - return 0; -} - -static int -gen4_render_ring_flush(struct i915_request *rq, u32 mode) -{ - u32 cmd, *cs; - int i; - - /* - * read/write caches: - * - * I915_GEM_DOMAIN_RENDER is always invalidated, but is - * only flushed if MI_NO_WRITE_FLUSH is unset. On 965, it is - * also flushed at 2d versus 3d pipeline switches. - * - * read-only caches: - * - * I915_GEM_DOMAIN_SAMPLER is flushed on pre-965 if - * MI_READ_FLUSH is set, and is always flushed on 965. - * - * I915_GEM_DOMAIN_COMMAND may not exist? - * - * I915_GEM_DOMAIN_INSTRUCTION, which exists on 965, is - * invalidated when MI_EXE_FLUSH is set. - * - * I915_GEM_DOMAIN_VERTEX, which exists on 965, is - * invalidated with every MI_FLUSH. - * - * TLBs: - * - * On 965, TLBs associated with I915_GEM_DOMAIN_COMMAND - * and I915_GEM_DOMAIN_CPU in are invalidated at PTE write and - * I915_GEM_DOMAIN_RENDER and I915_GEM_DOMAIN_SAMPLER - * are flushed at any MI_FLUSH. - */ - - cmd = MI_FLUSH; - if (mode & EMIT_INVALIDATE) { - cmd |= MI_EXE_FLUSH; - if (IS_G4X(rq->i915) || IS_GEN(rq->i915, 5)) - cmd |= MI_INVALIDATE_ISP; - } - - i = 2; - if (mode & EMIT_INVALIDATE) - i += 20; - - cs = intel_ring_begin(rq, i); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = cmd; - - /* - * A random delay to let the CS invalidate take effect? Without this - * delay, the GPU relocation path fails as the CS does not see - * the updated contents. Just as important, if we apply the flushes - * to the EMIT_FLUSH branch (i.e. immediately after the relocation - * write and before the invalidate on the next batch), the relocations - * still fail. This implies that is a delay following invalidation - * that is required to reset the caches as opposed to a delay to - * ensure the memory is written. - */ - if (mode & EMIT_INVALIDATE) { - *cs++ = GFX_OP_PIPE_CONTROL(4) | PIPE_CONTROL_QW_WRITE; - *cs++ = intel_gt_scratch_offset(rq->engine->gt, - INTEL_GT_SCRATCH_FIELD_DEFAULT) | - PIPE_CONTROL_GLOBAL_GTT; - *cs++ = 0; - *cs++ = 0; - - for (i = 0; i < 12; i++) - *cs++ = MI_FLUSH; - - *cs++ = GFX_OP_PIPE_CONTROL(4) | PIPE_CONTROL_QW_WRITE; - *cs++ = intel_gt_scratch_offset(rq->engine->gt, - INTEL_GT_SCRATCH_FIELD_DEFAULT) | - PIPE_CONTROL_GLOBAL_GTT; - *cs++ = 0; - *cs++ = 0; - } - - *cs++ = cmd; - - intel_ring_advance(rq, cs); - - return 0; -} - -/* - * Emits a PIPE_CONTROL with a non-zero post-sync operation, for - * implementing two workarounds on gen6. From section 1.4.7.1 - * "PIPE_CONTROL" of the Sandy Bridge PRM volume 2 part 1: - * - * [DevSNB-C+{W/A}] Before any depth stall flush (including those - * produced by non-pipelined state commands), software needs to first - * send a PIPE_CONTROL with no bits set except Post-Sync Operation != - * 0. - * - * [Dev-SNB{W/A}]: Before a PIPE_CONTROL with Write Cache Flush Enable - * =1, a PIPE_CONTROL with any non-zero post-sync-op is required. - * - * And the workaround for these two requires this workaround first: - * - * [Dev-SNB{W/A}]: Pipe-control with CS-stall bit set must be sent - * BEFORE the pipe-control with a post-sync op and no write-cache - * flushes. - * - * And this last workaround is tricky because of the requirements on - * that bit. From section 1.4.7.2.3 "Stall" of the Sandy Bridge PRM - * volume 2 part 1: - * - * "1 of the following must also be set: - * - Render Target Cache Flush Enable ([12] of DW1) - * - Depth Cache Flush Enable ([0] of DW1) - * - Stall at Pixel Scoreboard ([1] of DW1) - * - Depth Stall ([13] of DW1) - * - Post-Sync Operation ([13] of DW1) - * - Notify Enable ([8] of DW1)" - * - * The cache flushes require the workaround flush that triggered this - * one, so we can't use it. Depth stall would trigger the same. - * Post-sync nonzero is what triggered this second workaround, so we - * can't use that one either. Notify enable is IRQs, which aren't - * really our business. That leaves only stall at scoreboard. - */ -static int -gen6_emit_post_sync_nonzero_flush(struct i915_request *rq) -{ - u32 scratch_addr = - intel_gt_scratch_offset(rq->engine->gt, - INTEL_GT_SCRATCH_FIELD_RENDER_FLUSH); - u32 *cs; - - cs = intel_ring_begin(rq, 6); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = GFX_OP_PIPE_CONTROL(5); - *cs++ = PIPE_CONTROL_CS_STALL | PIPE_CONTROL_STALL_AT_SCOREBOARD; - *cs++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT; - *cs++ = 0; /* low dword */ - *cs++ = 0; /* high dword */ - *cs++ = MI_NOOP; - intel_ring_advance(rq, cs); - - cs = intel_ring_begin(rq, 6); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = GFX_OP_PIPE_CONTROL(5); - *cs++ = PIPE_CONTROL_QW_WRITE; - *cs++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT; - *cs++ = 0; - *cs++ = 0; - *cs++ = MI_NOOP; - intel_ring_advance(rq, cs); - - return 0; -} - -static int -gen6_render_ring_flush(struct i915_request *rq, u32 mode) -{ - u32 scratch_addr = - intel_gt_scratch_offset(rq->engine->gt, - INTEL_GT_SCRATCH_FIELD_RENDER_FLUSH); - u32 *cs, flags = 0; - int ret; - - /* Force SNB workarounds for PIPE_CONTROL flushes */ - ret = gen6_emit_post_sync_nonzero_flush(rq); - if (ret) - return ret; - - /* Just flush everything. Experiments have shown that reducing the - * number of bits based on the write domains has little performance - * impact. - */ - if (mode & EMIT_FLUSH) { - flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH; - flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH; - /* - * Ensure that any following seqno writes only happen - * when the render cache is indeed flushed. - */ - flags |= PIPE_CONTROL_CS_STALL; - } - if (mode & EMIT_INVALIDATE) { - flags |= PIPE_CONTROL_TLB_INVALIDATE; - flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE; - /* - * TLB invalidate requires a post-sync write. - */ - flags |= PIPE_CONTROL_QW_WRITE | PIPE_CONTROL_CS_STALL; - } - - cs = intel_ring_begin(rq, 4); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = GFX_OP_PIPE_CONTROL(4); - *cs++ = flags; - *cs++ = scratch_addr | PIPE_CONTROL_GLOBAL_GTT; - *cs++ = 0; - intel_ring_advance(rq, cs); - - return 0; -} - -static u32 *gen6_rcs_emit_breadcrumb(struct i915_request *rq, u32 *cs) -{ - /* First we do the gen6_emit_post_sync_nonzero_flush w/a */ - *cs++ = GFX_OP_PIPE_CONTROL(4); - *cs++ = PIPE_CONTROL_CS_STALL | PIPE_CONTROL_STALL_AT_SCOREBOARD; - *cs++ = 0; - *cs++ = 0; - - *cs++ = GFX_OP_PIPE_CONTROL(4); - *cs++ = PIPE_CONTROL_QW_WRITE; - *cs++ = intel_gt_scratch_offset(rq->engine->gt, - INTEL_GT_SCRATCH_FIELD_DEFAULT) | - PIPE_CONTROL_GLOBAL_GTT; - *cs++ = 0; - - /* Finally we can flush and with it emit the breadcrumb */ - *cs++ = GFX_OP_PIPE_CONTROL(4); - *cs++ = (PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH | - PIPE_CONTROL_DEPTH_CACHE_FLUSH | - PIPE_CONTROL_DC_FLUSH_ENABLE | - PIPE_CONTROL_QW_WRITE | - PIPE_CONTROL_CS_STALL); - *cs++ = i915_request_active_timeline(rq)->hwsp_offset | - PIPE_CONTROL_GLOBAL_GTT; - *cs++ = rq->fence.seqno; - - *cs++ = MI_USER_INTERRUPT; - *cs++ = MI_NOOP; - - rq->tail = intel_ring_offset(rq, cs); - assert_ring_tail_valid(rq->ring, rq->tail); - - return cs; -} - -static int -gen7_render_ring_cs_stall_wa(struct i915_request *rq) -{ - u32 *cs; - - cs = intel_ring_begin(rq, 4); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = GFX_OP_PIPE_CONTROL(4); - *cs++ = PIPE_CONTROL_CS_STALL | PIPE_CONTROL_STALL_AT_SCOREBOARD; - *cs++ = 0; - *cs++ = 0; - intel_ring_advance(rq, cs); - - return 0; -} - -static int -gen7_render_ring_flush(struct i915_request *rq, u32 mode) -{ - u32 scratch_addr = - intel_gt_scratch_offset(rq->engine->gt, - INTEL_GT_SCRATCH_FIELD_RENDER_FLUSH); - u32 *cs, flags = 0; - - /* - * Ensure that any following seqno writes only happen when the render - * cache is indeed flushed. - * - * Workaround: 4th PIPE_CONTROL command (except the ones with only - * read-cache invalidate bits set) must have the CS_STALL bit set. We - * don't try to be clever and just set it unconditionally. - */ - flags |= PIPE_CONTROL_CS_STALL; - - /* - * CS_STALL suggests at least a post-sync write. - */ - flags |= PIPE_CONTROL_QW_WRITE; - flags |= PIPE_CONTROL_GLOBAL_GTT_IVB; - - /* Just flush everything. Experiments have shown that reducing the - * number of bits based on the write domains has little performance - * impact. - */ - if (mode & EMIT_FLUSH) { - flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH; - flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH; - flags |= PIPE_CONTROL_DC_FLUSH_ENABLE; - flags |= PIPE_CONTROL_FLUSH_ENABLE; - } - if (mode & EMIT_INVALIDATE) { - flags |= PIPE_CONTROL_TLB_INVALIDATE; - flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_MEDIA_STATE_CLEAR; - - /* Workaround: we must issue a pipe_control with CS-stall bit - * set before a pipe_control command that has the state cache - * invalidate bit set. */ - gen7_render_ring_cs_stall_wa(rq); - } - - cs = intel_ring_begin(rq, 4); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = GFX_OP_PIPE_CONTROL(4); - *cs++ = flags; - *cs++ = scratch_addr; - *cs++ = 0; - intel_ring_advance(rq, cs); - - return 0; -} - -static u32 *gen7_rcs_emit_breadcrumb(struct i915_request *rq, u32 *cs) -{ - *cs++ = GFX_OP_PIPE_CONTROL(4); - *cs++ = (PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH | - PIPE_CONTROL_DEPTH_CACHE_FLUSH | - PIPE_CONTROL_DC_FLUSH_ENABLE | - PIPE_CONTROL_FLUSH_ENABLE | - PIPE_CONTROL_QW_WRITE | - PIPE_CONTROL_GLOBAL_GTT_IVB | - PIPE_CONTROL_CS_STALL); - *cs++ = i915_request_active_timeline(rq)->hwsp_offset; - *cs++ = rq->fence.seqno; - - *cs++ = MI_USER_INTERRUPT; - *cs++ = MI_NOOP; - - rq->tail = intel_ring_offset(rq, cs); - assert_ring_tail_valid(rq->ring, rq->tail); - - return cs; -} - -static u32 *gen6_xcs_emit_breadcrumb(struct i915_request *rq, u32 *cs) -{ - GEM_BUG_ON(i915_request_active_timeline(rq)->hwsp_ggtt != rq->engine->status_page.vma); - GEM_BUG_ON(offset_in_page(i915_request_active_timeline(rq)->hwsp_offset) != I915_GEM_HWS_SEQNO_ADDR); - - *cs++ = MI_FLUSH_DW | MI_FLUSH_DW_OP_STOREDW | MI_FLUSH_DW_STORE_INDEX; - *cs++ = I915_GEM_HWS_SEQNO_ADDR | MI_FLUSH_DW_USE_GTT; - *cs++ = rq->fence.seqno; - - *cs++ = MI_USER_INTERRUPT; - - rq->tail = intel_ring_offset(rq, cs); - assert_ring_tail_valid(rq->ring, rq->tail); - - return cs; -} - -#define GEN7_XCS_WA 32 -static u32 *gen7_xcs_emit_breadcrumb(struct i915_request *rq, u32 *cs) -{ - int i; - - GEM_BUG_ON(i915_request_active_timeline(rq)->hwsp_ggtt != rq->engine->status_page.vma); - GEM_BUG_ON(offset_in_page(i915_request_active_timeline(rq)->hwsp_offset) != I915_GEM_HWS_SEQNO_ADDR); - - *cs++ = MI_FLUSH_DW | MI_INVALIDATE_TLB | - MI_FLUSH_DW_OP_STOREDW | MI_FLUSH_DW_STORE_INDEX; - *cs++ = I915_GEM_HWS_SEQNO_ADDR | MI_FLUSH_DW_USE_GTT; - *cs++ = rq->fence.seqno; - - for (i = 0; i < GEN7_XCS_WA; i++) { - *cs++ = MI_STORE_DWORD_INDEX; - *cs++ = I915_GEM_HWS_SEQNO_ADDR; - *cs++ = rq->fence.seqno; - } - - *cs++ = MI_FLUSH_DW; - *cs++ = 0; - *cs++ = 0; - - *cs++ = MI_USER_INTERRUPT; - *cs++ = MI_NOOP; - - rq->tail = intel_ring_offset(rq, cs); - assert_ring_tail_valid(rq->ring, rq->tail); - - return cs; -} -#undef GEN7_XCS_WA - static void set_hwstam(struct intel_engine_cs *engine, u32 mask) { /* @@ -865,32 +429,6 @@ static void reset_finish(struct intel_engine_cs *engine) { } -static int rcs_resume(struct intel_engine_cs *engine) -{ - struct drm_i915_private *i915 = engine->i915; - struct intel_uncore *uncore = engine->uncore; - - /* - * Disable CONSTANT_BUFFER before it is loaded from the context - * image. For as it is loaded, it is executed and the stored - * address may no longer be valid, leading to a GPU hang. - * - * This imposes the requirement that userspace reload their - * CONSTANT_BUFFER on every batch, fortunately a requirement - * they are already accustomed to from before contexts were - * enabled. - */ - if (IS_GEN(i915, 4)) - intel_uncore_write(uncore, ECOSKPD, - _MASKED_BIT_ENABLE(ECO_CONSTANT_BUFFER_SR_DISABLE)); - - if (IS_GEN_RANGE(i915, 6, 7)) - intel_uncore_write(uncore, INSTPM, - _MASKED_BIT_ENABLE(INSTPM_FORCE_ORDERING)); - - return xcs_resume(engine); -} - static void reset_cancel(struct intel_engine_cs *engine) { struct i915_request *request; @@ -918,255 +456,6 @@ static void i9xx_submit_request(struct i915_request *request) intel_ring_set_tail(request->ring, request->tail)); } -static u32 *i9xx_emit_breadcrumb(struct i915_request *rq, u32 *cs) -{ - GEM_BUG_ON(i915_request_active_timeline(rq)->hwsp_ggtt != rq->engine->status_page.vma); - GEM_BUG_ON(offset_in_page(i915_request_active_timeline(rq)->hwsp_offset) != I915_GEM_HWS_SEQNO_ADDR); - - *cs++ = MI_FLUSH; - - *cs++ = MI_STORE_DWORD_INDEX; - *cs++ = I915_GEM_HWS_SEQNO_ADDR; - *cs++ = rq->fence.seqno; - - *cs++ = MI_USER_INTERRUPT; - *cs++ = MI_NOOP; - - rq->tail = intel_ring_offset(rq, cs); - assert_ring_tail_valid(rq->ring, rq->tail); - - return cs; -} - -#define GEN5_WA_STORES 8 /* must be at least 1! */ -static u32 *gen5_emit_breadcrumb(struct i915_request *rq, u32 *cs) -{ - int i; - - GEM_BUG_ON(i915_request_active_timeline(rq)->hwsp_ggtt != rq->engine->status_page.vma); - GEM_BUG_ON(offset_in_page(i915_request_active_timeline(rq)->hwsp_offset) != I915_GEM_HWS_SEQNO_ADDR); - - *cs++ = MI_FLUSH; - - BUILD_BUG_ON(GEN5_WA_STORES < 1); - for (i = 0; i < GEN5_WA_STORES; i++) { - *cs++ = MI_STORE_DWORD_INDEX; - *cs++ = I915_GEM_HWS_SEQNO_ADDR; - *cs++ = rq->fence.seqno; - } - - *cs++ = MI_USER_INTERRUPT; - - rq->tail = intel_ring_offset(rq, cs); - assert_ring_tail_valid(rq->ring, rq->tail); - - return cs; -} -#undef GEN5_WA_STORES - -static void -gen5_irq_enable(struct intel_engine_cs *engine) -{ - gen5_gt_enable_irq(engine->gt, engine->irq_enable_mask); -} - -static void -gen5_irq_disable(struct intel_engine_cs *engine) -{ - gen5_gt_disable_irq(engine->gt, engine->irq_enable_mask); -} - -static void -i9xx_irq_enable(struct intel_engine_cs *engine) -{ - engine->i915->irq_mask &= ~engine->irq_enable_mask; - intel_uncore_write(engine->uncore, GEN2_IMR, engine->i915->irq_mask); - intel_uncore_posting_read_fw(engine->uncore, GEN2_IMR); -} - -static void -i9xx_irq_disable(struct intel_engine_cs *engine) -{ - engine->i915->irq_mask |= engine->irq_enable_mask; - intel_uncore_write(engine->uncore, GEN2_IMR, engine->i915->irq_mask); -} - -static void -i8xx_irq_enable(struct intel_engine_cs *engine) -{ - struct drm_i915_private *i915 = engine->i915; - - i915->irq_mask &= ~engine->irq_enable_mask; - intel_uncore_write16(&i915->uncore, GEN2_IMR, i915->irq_mask); - ENGINE_POSTING_READ16(engine, RING_IMR); -} - -static void -i8xx_irq_disable(struct intel_engine_cs *engine) -{ - struct drm_i915_private *i915 = engine->i915; - - i915->irq_mask |= engine->irq_enable_mask; - intel_uncore_write16(&i915->uncore, GEN2_IMR, i915->irq_mask); -} - -static int -bsd_ring_flush(struct i915_request *rq, u32 mode) -{ - u32 *cs; - - cs = intel_ring_begin(rq, 2); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = MI_FLUSH; - *cs++ = MI_NOOP; - intel_ring_advance(rq, cs); - return 0; -} - -static void -gen6_irq_enable(struct intel_engine_cs *engine) -{ - ENGINE_WRITE(engine, RING_IMR, - ~(engine->irq_enable_mask | engine->irq_keep_mask)); - - /* Flush/delay to ensure the RING_IMR is active before the GT IMR */ - ENGINE_POSTING_READ(engine, RING_IMR); - - gen5_gt_enable_irq(engine->gt, engine->irq_enable_mask); -} - -static void -gen6_irq_disable(struct intel_engine_cs *engine) -{ - ENGINE_WRITE(engine, RING_IMR, ~engine->irq_keep_mask); - gen5_gt_disable_irq(engine->gt, engine->irq_enable_mask); -} - -static void -hsw_vebox_irq_enable(struct intel_engine_cs *engine) -{ - ENGINE_WRITE(engine, RING_IMR, ~engine->irq_enable_mask); - - /* Flush/delay to ensure the RING_IMR is active before the GT IMR */ - ENGINE_POSTING_READ(engine, RING_IMR); - - gen6_gt_pm_unmask_irq(engine->gt, engine->irq_enable_mask); -} - -static void -hsw_vebox_irq_disable(struct intel_engine_cs *engine) -{ - ENGINE_WRITE(engine, RING_IMR, ~0); - gen6_gt_pm_mask_irq(engine->gt, engine->irq_enable_mask); -} - -static int -i965_emit_bb_start(struct i915_request *rq, - u64 offset, u32 length, - unsigned int dispatch_flags) -{ - u32 *cs; - - cs = intel_ring_begin(rq, 2); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT | (dispatch_flags & - I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE_I965); - *cs++ = offset; - intel_ring_advance(rq, cs); - - return 0; -} - -/* Just userspace ABI convention to limit the wa batch bo to a resonable size */ -#define I830_BATCH_LIMIT SZ_256K -#define I830_TLB_ENTRIES (2) -#define I830_WA_SIZE max(I830_TLB_ENTRIES*4096, I830_BATCH_LIMIT) -static int -i830_emit_bb_start(struct i915_request *rq, - u64 offset, u32 len, - unsigned int dispatch_flags) -{ - u32 *cs, cs_offset = - intel_gt_scratch_offset(rq->engine->gt, - INTEL_GT_SCRATCH_FIELD_DEFAULT); - - GEM_BUG_ON(rq->engine->gt->scratch->size < I830_WA_SIZE); - - cs = intel_ring_begin(rq, 6); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - /* Evict the invalid PTE TLBs */ - *cs++ = COLOR_BLT_CMD | BLT_WRITE_RGBA; - *cs++ = BLT_DEPTH_32 | BLT_ROP_COLOR_COPY | 4096; - *cs++ = I830_TLB_ENTRIES << 16 | 4; /* load each page */ - *cs++ = cs_offset; - *cs++ = 0xdeadbeef; - *cs++ = MI_NOOP; - intel_ring_advance(rq, cs); - - if ((dispatch_flags & I915_DISPATCH_PINNED) == 0) { - if (len > I830_BATCH_LIMIT) - return -ENOSPC; - - cs = intel_ring_begin(rq, 6 + 2); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - /* Blit the batch (which has now all relocs applied) to the - * stable batch scratch bo area (so that the CS never - * stumbles over its tlb invalidation bug) ... - */ - *cs++ = SRC_COPY_BLT_CMD | BLT_WRITE_RGBA | (6 - 2); - *cs++ = BLT_DEPTH_32 | BLT_ROP_SRC_COPY | 4096; - *cs++ = DIV_ROUND_UP(len, 4096) << 16 | 4096; - *cs++ = cs_offset; - *cs++ = 4096; - *cs++ = offset; - - *cs++ = MI_FLUSH; - *cs++ = MI_NOOP; - intel_ring_advance(rq, cs); - - /* ... and execute it. */ - offset = cs_offset; - } - - cs = intel_ring_begin(rq, 2); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT; - *cs++ = offset | (dispatch_flags & I915_DISPATCH_SECURE ? 0 : - MI_BATCH_NON_SECURE); - intel_ring_advance(rq, cs); - - return 0; -} - -static int -i915_emit_bb_start(struct i915_request *rq, - u64 offset, u32 len, - unsigned int dispatch_flags) -{ - u32 *cs; - - cs = intel_ring_begin(rq, 2); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT; - *cs++ = offset | (dispatch_flags & I915_DISPATCH_SECURE ? 0 : - MI_BATCH_NON_SECURE); - intel_ring_advance(rq, cs); - - return 0; -} - static void __ring_context_fini(struct intel_context *ce) { i915_vma_put(ce->state); @@ -1356,8 +645,8 @@ static inline int mi_set_context(struct i915_request *rq, struct intel_context *ce, u32 flags) { - struct drm_i915_private *i915 = rq->i915; struct intel_engine_cs *engine = rq->engine; + struct drm_i915_private *i915 = engine->i915; enum intel_engine_id id; const int num_engines = IS_HASWELL(i915) ? RUNTIME_INFO(i915)->num_engines - 1 : 0; @@ -1471,7 +760,7 @@ static inline int mi_set_context(struct i915_request *rq, static int remap_l3_slice(struct i915_request *rq, int slice) { - u32 *cs, *remap_info = rq->i915->l3_parity.remap_info[slice]; + u32 *cs, *remap_info = rq->engine->i915->l3_parity.remap_info[slice]; int i; if (!remap_info) @@ -1582,7 +871,7 @@ static int switch_context(struct i915_request *rq) void **residuals = NULL; int ret; - GEM_BUG_ON(HAS_EXECLISTS(rq->i915)); + GEM_BUG_ON(HAS_EXECLISTS(engine->i915)); if (engine->wa_ctx.vma && ce != engine->kernel_context) { if (engine->wa_ctx.vma->private != ce) { @@ -1704,99 +993,6 @@ static void gen6_bsd_submit_request(struct i915_request *request) intel_uncore_forcewake_put(uncore, FORCEWAKE_ALL); } -static int mi_flush_dw(struct i915_request *rq, u32 flags) -{ - u32 cmd, *cs; - - cs = intel_ring_begin(rq, 4); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - cmd = MI_FLUSH_DW; - - /* - * We always require a command barrier so that subsequent - * commands, such as breadcrumb interrupts, are strictly ordered - * wrt the contents of the write cache being flushed to memory - * (and thus being coherent from the CPU). - */ - cmd |= MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW; - - /* - * Bspec vol 1c.3 - blitter engine command streamer: - * "If ENABLED, all TLBs will be invalidated once the flush - * operation is complete. This bit is only valid when the - * Post-Sync Operation field is a value of 1h or 3h." - */ - cmd |= flags; - - *cs++ = cmd; - *cs++ = I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT; - *cs++ = 0; - *cs++ = MI_NOOP; - - intel_ring_advance(rq, cs); - - return 0; -} - -static int gen6_flush_dw(struct i915_request *rq, u32 mode, u32 invflags) -{ - return mi_flush_dw(rq, mode & EMIT_INVALIDATE ? invflags : 0); -} - -static int gen6_bsd_ring_flush(struct i915_request *rq, u32 mode) -{ - return gen6_flush_dw(rq, mode, MI_INVALIDATE_TLB | MI_INVALIDATE_BSD); -} - -static int -hsw_emit_bb_start(struct i915_request *rq, - u64 offset, u32 len, - unsigned int dispatch_flags) -{ - u32 *cs; - - cs = intel_ring_begin(rq, 2); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = MI_BATCH_BUFFER_START | (dispatch_flags & I915_DISPATCH_SECURE ? - 0 : MI_BATCH_PPGTT_HSW | MI_BATCH_NON_SECURE_HSW); - /* bit0-7 is the length on GEN6+ */ - *cs++ = offset; - intel_ring_advance(rq, cs); - - return 0; -} - -static int -gen6_emit_bb_start(struct i915_request *rq, - u64 offset, u32 len, - unsigned int dispatch_flags) -{ - u32 *cs; - - cs = intel_ring_begin(rq, 2); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = MI_BATCH_BUFFER_START | (dispatch_flags & I915_DISPATCH_SECURE ? - 0 : MI_BATCH_NON_SECURE_I965); - /* bit0-7 is the length on GEN6+ */ - *cs++ = offset; - intel_ring_advance(rq, cs); - - return 0; -} - -/* Blitter support (SandyBridge+) */ - -static int gen6_ring_flush(struct i915_request *rq, u32 mode) -{ - return gen6_flush_dw(rq, mode, MI_INVALIDATE_TLB); -} - static void i9xx_set_default_submission(struct intel_engine_cs *engine) { engine->submit_request = i9xx_submit_request; @@ -1843,11 +1039,11 @@ static void setup_irq(struct intel_engine_cs *engine) engine->irq_enable = gen5_irq_enable; engine->irq_disable = gen5_irq_disable; } else if (INTEL_GEN(i915) >= 3) { - engine->irq_enable = i9xx_irq_enable; - engine->irq_disable = i9xx_irq_disable; + engine->irq_enable = gen3_irq_enable; + engine->irq_disable = gen3_irq_disable; } else { - engine->irq_enable = i8xx_irq_enable; - engine->irq_disable = i8xx_irq_disable; + engine->irq_enable = gen2_irq_enable; + engine->irq_disable = gen2_irq_disable; } } @@ -1874,7 +1070,7 @@ static void setup_common(struct intel_engine_cs *engine) * equivalent to our next initial bread so we can elide * engine->emit_init_breadcrumb(). */ - engine->emit_fini_breadcrumb = i9xx_emit_breadcrumb; + engine->emit_fini_breadcrumb = gen3_emit_breadcrumb; if (IS_GEN(i915, 5)) engine->emit_fini_breadcrumb = gen5_emit_breadcrumb; @@ -1883,11 +1079,11 @@ static void setup_common(struct intel_engine_cs *engine) if (INTEL_GEN(i915) >= 6) engine->emit_bb_start = gen6_emit_bb_start; else if (INTEL_GEN(i915) >= 4) - engine->emit_bb_start = i965_emit_bb_start; + engine->emit_bb_start = gen4_emit_bb_start; else if (IS_I830(i915) || IS_I845G(i915)) engine->emit_bb_start = i830_emit_bb_start; else - engine->emit_bb_start = i915_emit_bb_start; + engine->emit_bb_start = gen3_emit_bb_start; } static void setup_rcs(struct intel_engine_cs *engine) @@ -1900,25 +1096,23 @@ static void setup_rcs(struct intel_engine_cs *engine) engine->irq_enable_mask = GT_RENDER_USER_INTERRUPT; if (INTEL_GEN(i915) >= 7) { - engine->emit_flush = gen7_render_ring_flush; - engine->emit_fini_breadcrumb = gen7_rcs_emit_breadcrumb; + engine->emit_flush = gen7_emit_flush_rcs; + engine->emit_fini_breadcrumb = gen7_emit_breadcrumb_rcs; } else if (IS_GEN(i915, 6)) { - engine->emit_flush = gen6_render_ring_flush; - engine->emit_fini_breadcrumb = gen6_rcs_emit_breadcrumb; + engine->emit_flush = gen6_emit_flush_rcs; + engine->emit_fini_breadcrumb = gen6_emit_breadcrumb_rcs; } else if (IS_GEN(i915, 5)) { - engine->emit_flush = gen4_render_ring_flush; + engine->emit_flush = gen4_emit_flush_rcs; } else { if (INTEL_GEN(i915) < 4) - engine->emit_flush = gen2_render_ring_flush; + engine->emit_flush = gen2_emit_flush; else - engine->emit_flush = gen4_render_ring_flush; + engine->emit_flush = gen4_emit_flush_rcs; engine->irq_enable_mask = I915_USER_INTERRUPT; } if (IS_HASWELL(i915)) engine->emit_bb_start = hsw_emit_bb_start; - - engine->resume = rcs_resume; } static void setup_vcs(struct intel_engine_cs *engine) @@ -1929,15 +1123,15 @@ static void setup_vcs(struct intel_engine_cs *engine) /* gen6 bsd needs a special wa for tail updates */ if (IS_GEN(i915, 6)) engine->set_default_submission = gen6_bsd_set_default_submission; - engine->emit_flush = gen6_bsd_ring_flush; + engine->emit_flush = gen6_emit_flush_vcs; engine->irq_enable_mask = GT_BSD_USER_INTERRUPT; if (IS_GEN(i915, 6)) - engine->emit_fini_breadcrumb = gen6_xcs_emit_breadcrumb; + engine->emit_fini_breadcrumb = gen6_emit_breadcrumb_xcs; else - engine->emit_fini_breadcrumb = gen7_xcs_emit_breadcrumb; + engine->emit_fini_breadcrumb = gen7_emit_breadcrumb_xcs; } else { - engine->emit_flush = bsd_ring_flush; + engine->emit_flush = gen4_emit_flush_vcs; if (IS_GEN(i915, 5)) engine->irq_enable_mask = ILK_BSD_USER_INTERRUPT; else @@ -1949,13 +1143,13 @@ static void setup_bcs(struct intel_engine_cs *engine) { struct drm_i915_private *i915 = engine->i915; - engine->emit_flush = gen6_ring_flush; + engine->emit_flush = gen6_emit_flush_xcs; engine->irq_enable_mask = GT_BLT_USER_INTERRUPT; if (IS_GEN(i915, 6)) - engine->emit_fini_breadcrumb = gen6_xcs_emit_breadcrumb; + engine->emit_fini_breadcrumb = gen6_emit_breadcrumb_xcs; else - engine->emit_fini_breadcrumb = gen7_xcs_emit_breadcrumb; + engine->emit_fini_breadcrumb = gen7_emit_breadcrumb_xcs; } static void setup_vecs(struct intel_engine_cs *engine) @@ -1964,12 +1158,12 @@ static void setup_vecs(struct intel_engine_cs *engine) GEM_BUG_ON(INTEL_GEN(i915) < 7); - engine->emit_flush = gen6_ring_flush; + engine->emit_flush = gen6_emit_flush_xcs; engine->irq_enable_mask = PM_VEBOX_USER_INTERRUPT; - engine->irq_enable = hsw_vebox_irq_enable; - engine->irq_disable = hsw_vebox_irq_disable; + engine->irq_enable = hsw_irq_enable_vecs; + engine->irq_disable = hsw_irq_disable_vecs; - engine->emit_fini_breadcrumb = gen7_xcs_emit_breadcrumb; + engine->emit_fini_breadcrumb = gen7_emit_breadcrumb_xcs; } static int gen7_ctx_switch_bb_setup(struct intel_engine_cs * const engine, diff --git a/drivers/gpu/drm/i915/gt/intel_rps.c b/drivers/gpu/drm/i915/gt/intel_rps.c index 2f59fc6df3c2..296391deeb94 100644 --- a/drivers/gpu/drm/i915/gt/intel_rps.c +++ b/drivers/gpu/drm/i915/gt/intel_rps.c @@ -51,15 +51,16 @@ static void rps_timer(struct timer_list *t) { struct intel_rps *rps = from_timer(rps, t, timer); struct intel_engine_cs *engine; + ktime_t dt, last, timestamp; enum intel_engine_id id; s64 max_busy[3] = {}; - ktime_t dt, last; + timestamp = 0; for_each_engine(engine, rps_to_gt(rps), id) { s64 busy; int i; - dt = intel_engine_get_busy_time(engine); + dt = intel_engine_get_busy_time(engine, ×tamp); last = engine->stats.rps; engine->stats.rps = dt; @@ -69,16 +70,14 @@ static void rps_timer(struct timer_list *t) swap(busy, max_busy[i]); } } - - dt = ktime_get(); last = rps->pm_timestamp; - rps->pm_timestamp = dt; + rps->pm_timestamp = timestamp; if (intel_rps_is_active(rps)) { s64 busy; int i; - dt = ktime_sub(dt, last); + dt = ktime_sub(timestamp, last); /* * Our goal is to evaluate each engine independently, so we run diff --git a/drivers/gpu/drm/i915/gt/intel_workarounds.c b/drivers/gpu/drm/i915/gt/intel_workarounds.c index 85d2bef51524..2da366821dda 100644 --- a/drivers/gpu/drm/i915/gt/intel_workarounds.c +++ b/drivers/gpu/drm/i915/gt/intel_workarounds.c @@ -205,6 +205,18 @@ wa_masked_dis(struct i915_wa_list *wal, i915_reg_t reg, u32 val) #define WA_SET_FIELD_MASKED(addr, mask, value) \ wa_write_masked_or(wal, (addr), 0, _MASKED_FIELD((mask), (value))) +static void gen6_ctx_workarounds_init(struct intel_engine_cs *engine, + struct i915_wa_list *wal) +{ + WA_SET_BIT_MASKED(INSTPM, INSTPM_FORCE_ORDERING); +} + +static void gen7_ctx_workarounds_init(struct intel_engine_cs *engine, + struct i915_wa_list *wal) +{ + WA_SET_BIT_MASKED(INSTPM, INSTPM_FORCE_ORDERING); +} + static void gen8_ctx_workarounds_init(struct intel_engine_cs *engine, struct i915_wa_list *wal) { @@ -355,7 +367,10 @@ static void gen9_ctx_workarounds_init(struct intel_engine_cs *engine, HDC_FORCE_NON_COHERENT); /* WaDisableSamplerPowerBypassForSOPingPong:skl,bxt,kbl,cfl */ - if (IS_SKYLAKE(i915) || IS_KABYLAKE(i915) || IS_COFFEELAKE(i915)) + if (IS_SKYLAKE(i915) || + IS_KABYLAKE(i915) || + IS_COFFEELAKE(i915) || + IS_COMETLAKE(i915)) WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, GEN8_SAMPLER_POWER_BYPASS_DIS); @@ -600,11 +615,14 @@ static void tgl_ctx_workarounds_init(struct intel_engine_cs *engine, * Wa_1604555607:gen12 and Wa_1608008084:gen12 * FF_MODE2 register will return the wrong value when read. The default * value for this register is zero for all fields and there are no bit - * masks. So instead of doing a RMW we should just write the TDS timer - * value for Wa_1604555607. + * masks. So instead of doing a RMW we should just write the GS Timer + * and TDS timer values for Wa_1604555607 and Wa_16011163337. */ - wa_add(wal, FF_MODE2, FF_MODE2_TDS_TIMER_MASK, - FF_MODE2_TDS_TIMER_128, 0); + wa_add(wal, + FF_MODE2, + FF_MODE2_GS_TIMER_MASK | FF_MODE2_TDS_TIMER_MASK, + FF_MODE2_GS_TIMER_224 | FF_MODE2_TDS_TIMER_128, + 0); /* WaDisableGPGPUMidThreadPreemption:tgl */ WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1, @@ -630,7 +648,7 @@ __intel_engine_init_ctx_wa(struct intel_engine_cs *engine, icl_ctx_workarounds_init(engine, wal); else if (IS_CANNONLAKE(i915)) cnl_ctx_workarounds_init(engine, wal); - else if (IS_COFFEELAKE(i915)) + else if (IS_COFFEELAKE(i915) || IS_COMETLAKE(i915)) cfl_ctx_workarounds_init(engine, wal); else if (IS_GEMINILAKE(i915)) glk_ctx_workarounds_init(engine, wal); @@ -644,6 +662,10 @@ __intel_engine_init_ctx_wa(struct intel_engine_cs *engine, chv_ctx_workarounds_init(engine, wal); else if (IS_BROADWELL(i915)) bdw_ctx_workarounds_init(engine, wal); + else if (IS_GEN(i915, 7)) + gen7_ctx_workarounds_init(engine, wal); + else if (IS_GEN(i915, 6)) + gen6_ctx_workarounds_init(engine, wal); else if (INTEL_GEN(i915) < 8) return; else @@ -917,7 +939,7 @@ static void gen9_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) { /* WaDisableKillLogic:bxt,skl,kbl */ - if (!IS_COFFEELAKE(i915)) + if (!IS_COFFEELAKE(i915) && !IS_COMETLAKE(i915)) wa_write_or(wal, GAM_ECOCHK, ECOCHK_DIS_TLB); @@ -1180,7 +1202,7 @@ gt_init_workarounds(struct drm_i915_private *i915, struct i915_wa_list *wal) icl_gt_workarounds_init(i915, wal); else if (IS_CANNONLAKE(i915)) cnl_gt_workarounds_init(i915, wal); - else if (IS_COFFEELAKE(i915)) + else if (IS_COFFEELAKE(i915) || IS_COMETLAKE(i915)) cfl_gt_workarounds_init(i915, wal); else if (IS_GEMINILAKE(i915)) glk_gt_workarounds_init(i915, wal); @@ -1428,6 +1450,18 @@ static void cfl_whitelist_build(struct intel_engine_cs *engine) RING_FORCE_TO_NONPRIV_RANGE_4); } +static void cml_whitelist_build(struct intel_engine_cs *engine) +{ + struct i915_wa_list *w = &engine->whitelist; + + if (engine->class != RENDER_CLASS) + whitelist_reg_ext(w, + RING_CTX_TIMESTAMP(engine->mmio_base), + RING_FORCE_TO_NONPRIV_ACCESS_RD); + + cfl_whitelist_build(engine); +} + static void cnl_whitelist_build(struct intel_engine_cs *engine) { struct i915_wa_list *w = &engine->whitelist; @@ -1478,9 +1512,15 @@ static void icl_whitelist_build(struct intel_engine_cs *engine) /* hucStatus2RegOffset */ whitelist_reg_ext(w, _MMIO(0x23B0 + engine->mmio_base), RING_FORCE_TO_NONPRIV_ACCESS_RD); + whitelist_reg_ext(w, + RING_CTX_TIMESTAMP(engine->mmio_base), + RING_FORCE_TO_NONPRIV_ACCESS_RD); break; default: + whitelist_reg_ext(w, + RING_CTX_TIMESTAMP(engine->mmio_base), + RING_FORCE_TO_NONPRIV_ACCESS_RD); break; } } @@ -1512,6 +1552,9 @@ static void tgl_whitelist_build(struct intel_engine_cs *engine) whitelist_reg(w, HIZ_CHICKEN); break; default: + whitelist_reg_ext(w, + RING_CTX_TIMESTAMP(engine->mmio_base), + RING_FORCE_TO_NONPRIV_ACCESS_RD); break; } } @@ -1529,6 +1572,8 @@ void intel_engine_init_whitelist(struct intel_engine_cs *engine) icl_whitelist_build(engine); else if (IS_CANNONLAKE(i915)) cnl_whitelist_build(engine); + else if (IS_COMETLAKE(i915)) + cml_whitelist_build(engine); else if (IS_COFFEELAKE(i915)) cfl_whitelist_build(engine); else if (IS_GEMINILAKE(i915)) @@ -1725,6 +1770,12 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal) wa_write_or(wal, GEN7_FF_THREAD_MODE, GEN12_FF_TESSELATION_DOP_GATE_DISABLE); + + /* Wa_22010271021:ehl */ + if (IS_ELKHARTLAKE(i915)) + wa_masked_en(wal, + GEN9_CS_DEBUG_MODE1, + FF_DOP_CLOCK_GATE_DISABLE); } if (IS_GEN_RANGE(i915, 9, 12)) { @@ -1734,7 +1785,10 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal) GEN9_FFSC_PERCTX_PREEMPT_CTRL); } - if (IS_SKYLAKE(i915) || IS_KABYLAKE(i915) || IS_COFFEELAKE(i915)) { + if (IS_SKYLAKE(i915) || + IS_KABYLAKE(i915) || + IS_COFFEELAKE(i915) || + IS_COMETLAKE(i915)) { /* WaEnableGapsTsvCreditFix:skl,kbl,cfl */ wa_write_or(wal, GEN8_GARBCNTL, @@ -1818,6 +1872,21 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal) 0, _MASKED_BIT_ENABLE(VS_TIMER_DISPATCH), /* XXX bit doesn't stick on Broadwater */ IS_I965G(i915) ? 0 : VS_TIMER_DISPATCH); + + if (IS_GEN(i915, 4)) + /* + * Disable CONSTANT_BUFFER before it is loaded from the context + * image. For as it is loaded, it is executed and the stored + * address may no longer be valid, leading to a GPU hang. + * + * This imposes the requirement that userspace reload their + * CONSTANT_BUFFER on every batch, fortunately a requirement + * they are already accustomed to from before contexts were + * enabled. + */ + wa_add(wal, ECOSKPD, + 0, _MASKED_BIT_ENABLE(ECO_CONSTANT_BUFFER_SR_DISABLE), + 0 /* XXX bit doesn't stick on Broadwater */); } static void @@ -1932,7 +2001,7 @@ wa_list_srm(struct i915_request *rq, const struct i915_wa_list *wal, struct i915_vma *vma) { - struct drm_i915_private *i915 = rq->i915; + struct drm_i915_private *i915 = rq->engine->i915; unsigned int i, count = 0; const struct i915_wa *wa; u32 srm, *cs; @@ -2021,7 +2090,7 @@ static int engine_wa_list_verify(struct intel_context *ce, err = 0; for (i = 0, wa = wal->list; i < wal->count; i++, wa++) { - if (mcr_range(rq->i915, i915_mmio_reg_offset(wa->reg))) + if (mcr_range(rq->engine->i915, i915_mmio_reg_offset(wa->reg))) continue; if (!wa_verify(wa, results[i], wal->name, from)) diff --git a/drivers/gpu/drm/i915/gt/selftest_engine_cs.c b/drivers/gpu/drm/i915/gt/selftest_engine_cs.c index f88e445a1cae..729c3c7b11e2 100644 --- a/drivers/gpu/drm/i915/gt/selftest_engine_cs.c +++ b/drivers/gpu/drm/i915/gt/selftest_engine_cs.c @@ -49,7 +49,7 @@ static int write_timestamp(struct i915_request *rq, int slot) return PTR_ERR(cs); cmd = MI_STORE_REGISTER_MEM | MI_USE_GGTT; - if (INTEL_GEN(rq->i915) >= 8) + if (INTEL_GEN(rq->engine->i915) >= 8) cmd++; *cs++ = cmd; *cs++ = i915_mmio_reg_offset(RING_TIMESTAMP(rq->engine->mmio_base)); diff --git a/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.c b/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.c index 697114dd1f47..73243ba59c7d 100644 --- a/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.c +++ b/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.c @@ -10,6 +10,7 @@ #include "intel_gt_requests.h" #include "i915_selftest.h" +#include "selftest_engine_heartbeat.h" static int timeline_sync(struct intel_timeline *tl) { @@ -142,24 +143,6 @@ out: return err; } -static void engine_heartbeat_disable(struct intel_engine_cs *engine, - unsigned long *saved) -{ - *saved = engine->props.heartbeat_interval_ms; - engine->props.heartbeat_interval_ms = 0; - - intel_engine_pm_get(engine); - intel_engine_park_heartbeat(engine); -} - -static void engine_heartbeat_enable(struct intel_engine_cs *engine, - unsigned long saved) -{ - intel_engine_pm_put(engine); - - engine->props.heartbeat_interval_ms = saved; -} - static int live_idle_flush(void *arg) { struct intel_gt *gt = arg; @@ -170,11 +153,9 @@ static int live_idle_flush(void *arg) /* Check that we can flush the idle barriers */ for_each_engine(engine, gt, id) { - unsigned long heartbeat; - - engine_heartbeat_disable(engine, &heartbeat); + st_engine_heartbeat_disable(engine); err = __live_idle_pulse(engine, intel_engine_flush_barriers); - engine_heartbeat_enable(engine, heartbeat); + st_engine_heartbeat_enable(engine); if (err) break; } @@ -192,11 +173,9 @@ static int live_idle_pulse(void *arg) /* Check that heartbeat pulses flush the idle barriers */ for_each_engine(engine, gt, id) { - unsigned long heartbeat; - - engine_heartbeat_disable(engine, &heartbeat); + st_engine_heartbeat_disable(engine); err = __live_idle_pulse(engine, intel_engine_pulse); - engine_heartbeat_enable(engine, heartbeat); + st_engine_heartbeat_enable(engine); if (err && err != -ENODEV) break; @@ -386,11 +365,27 @@ int intel_heartbeat_live_selftests(struct drm_i915_private *i915) if (intel_gt_is_wedged(&i915->gt)) return 0; - saved_hangcheck = i915_modparams.enable_hangcheck; - i915_modparams.enable_hangcheck = INT_MAX; + saved_hangcheck = i915->params.enable_hangcheck; + i915->params.enable_hangcheck = INT_MAX; err = intel_gt_live_subtests(tests, &i915->gt); - i915_modparams.enable_hangcheck = saved_hangcheck; + i915->params.enable_hangcheck = saved_hangcheck; return err; } + +void st_engine_heartbeat_disable(struct intel_engine_cs *engine) +{ + engine->props.heartbeat_interval_ms = 0; + + intel_engine_pm_get(engine); + intel_engine_park_heartbeat(engine); +} + +void st_engine_heartbeat_enable(struct intel_engine_cs *engine) +{ + intel_engine_pm_put(engine); + + engine->props.heartbeat_interval_ms = + engine->defaults.heartbeat_interval_ms; +} diff --git a/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.h b/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.h new file mode 100644 index 000000000000..cd27113d5400 --- /dev/null +++ b/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.h @@ -0,0 +1,14 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright © 2020 Intel Corporation + */ + +#ifndef SELFTEST_ENGINE_HEARTBEAT_H +#define SELFTEST_ENGINE_HEARTBEAT_H + +struct intel_engine_cs; + +void st_engine_heartbeat_disable(struct intel_engine_cs *engine); +void st_engine_heartbeat_enable(struct intel_engine_cs *engine); + +#endif /* SELFTEST_ENGINE_HEARTBEAT_H */ diff --git a/drivers/gpu/drm/i915/gt/selftest_engine_pm.c b/drivers/gpu/drm/i915/gt/selftest_engine_pm.c index cbf6b0735272..b08fc5390e8a 100644 --- a/drivers/gpu/drm/i915/gt/selftest_engine_pm.c +++ b/drivers/gpu/drm/i915/gt/selftest_engine_pm.c @@ -6,7 +6,107 @@ #include "i915_selftest.h" #include "selftest_engine.h" +#include "selftest_engine_heartbeat.h" #include "selftests/igt_atomic.h" +#include "selftests/igt_flush_test.h" +#include "selftests/igt_spinner.h" + +static int live_engine_busy_stats(void *arg) +{ + struct intel_gt *gt = arg; + struct intel_engine_cs *engine; + enum intel_engine_id id; + struct igt_spinner spin; + int err = 0; + + /* + * Check that if an engine supports busy-stats, they tell the truth. + */ + + if (igt_spinner_init(&spin, gt)) + return -ENOMEM; + + GEM_BUG_ON(intel_gt_pm_is_awake(gt)); + for_each_engine(engine, gt, id) { + struct i915_request *rq; + ktime_t de, dt; + ktime_t t[2]; + + if (!intel_engine_supports_stats(engine)) + continue; + + if (!intel_engine_can_store_dword(engine)) + continue; + + if (intel_gt_pm_wait_for_idle(gt)) { + err = -EBUSY; + break; + } + + st_engine_heartbeat_disable(engine); + + ENGINE_TRACE(engine, "measuring idle time\n"); + preempt_disable(); + de = intel_engine_get_busy_time(engine, &t[0]); + udelay(100); + de = ktime_sub(intel_engine_get_busy_time(engine, &t[1]), de); + preempt_enable(); + dt = ktime_sub(t[1], t[0]); + if (de < 0 || de > 10) { + pr_err("%s: reported %lldns [%d%%] busyness while sleeping [for %lldns]\n", + engine->name, + de, (int)div64_u64(100 * de, dt), dt); + GEM_TRACE_DUMP(); + err = -EINVAL; + goto end; + } + + /* 100% busy */ + rq = igt_spinner_create_request(&spin, + engine->kernel_context, + MI_NOOP); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto end; + } + i915_request_add(rq); + + if (!igt_wait_for_spinner(&spin, rq)) { + intel_gt_set_wedged(engine->gt); + err = -ETIME; + goto end; + } + + ENGINE_TRACE(engine, "measuring busy time\n"); + preempt_disable(); + de = intel_engine_get_busy_time(engine, &t[0]); + udelay(100); + de = ktime_sub(intel_engine_get_busy_time(engine, &t[1]), de); + preempt_enable(); + dt = ktime_sub(t[1], t[0]); + if (100 * de < 95 * dt || 95 * de > 100 * dt) { + pr_err("%s: reported %lldns [%d%%] busyness while spinning [for %lldns]\n", + engine->name, + de, (int)div64_u64(100 * de, dt), dt); + GEM_TRACE_DUMP(); + err = -EINVAL; + goto end; + } + +end: + st_engine_heartbeat_enable(engine); + igt_spinner_end(&spin); + if (igt_flush_test(gt->i915)) + err = -EIO; + if (err) + break; + } + + igt_spinner_fini(&spin); + if (igt_flush_test(gt->i915)) + err = -EIO; + return err; +} static int live_engine_pm(void *arg) { @@ -77,6 +177,7 @@ static int live_engine_pm(void *arg) int live_engine_pm_selftests(struct intel_gt *gt) { static const struct i915_subtest tests[] = { + SUBTEST(live_engine_busy_stats), SUBTEST(live_engine_pm), }; diff --git a/drivers/gpu/drm/i915/gt/selftest_gt_pm.c b/drivers/gpu/drm/i915/gt/selftest_gt_pm.c index 242181a5214c..6180a47c1b51 100644 --- a/drivers/gpu/drm/i915/gt/selftest_gt_pm.c +++ b/drivers/gpu/drm/i915/gt/selftest_gt_pm.c @@ -5,10 +5,141 @@ * Copyright © 2019 Intel Corporation */ +#include <linux/sort.h> + +#include "intel_gt_clock_utils.h" + #include "selftest_llc.h" #include "selftest_rc6.h" #include "selftest_rps.h" +static int cmp_u64(const void *A, const void *B) +{ + const u64 *a = A, *b = B; + + if (a < b) + return -1; + else if (a > b) + return 1; + else + return 0; +} + +static int cmp_u32(const void *A, const void *B) +{ + const u32 *a = A, *b = B; + + if (a < b) + return -1; + else if (a > b) + return 1; + else + return 0; +} + +static void measure_clocks(struct intel_engine_cs *engine, + u32 *out_cycles, ktime_t *out_dt) +{ + ktime_t dt[5]; + u32 cycles[5]; + int i; + + for (i = 0; i < 5; i++) { + preempt_disable(); + cycles[i] = -ENGINE_READ_FW(engine, RING_TIMESTAMP); + dt[i] = ktime_get(); + + udelay(1000); + + dt[i] = ktime_sub(ktime_get(), dt[i]); + cycles[i] += ENGINE_READ_FW(engine, RING_TIMESTAMP); + preempt_enable(); + } + + /* Use the median of both cycle/dt; close enough */ + sort(cycles, 5, sizeof(*cycles), cmp_u32, NULL); + *out_cycles = (cycles[1] + 2 * cycles[2] + cycles[3]) / 4; + + sort(dt, 5, sizeof(*dt), cmp_u64, NULL); + *out_dt = div_u64(dt[1] + 2 * dt[2] + dt[3], 4); +} + +static int live_gt_clocks(void *arg) +{ + struct intel_gt *gt = arg; + struct intel_engine_cs *engine; + enum intel_engine_id id; + int err = 0; + + if (!RUNTIME_INFO(gt->i915)->cs_timestamp_frequency_hz) { /* unknown */ + pr_info("CS_TIMESTAMP frequency unknown\n"); + return 0; + } + + if (INTEL_GEN(gt->i915) < 4) /* Any CS_TIMESTAMP? */ + return 0; + + if (IS_GEN(gt->i915, 5)) + /* + * XXX CS_TIMESTAMP low dword is dysfunctional? + * + * Ville's experiments indicate the high dword still works, + * but at a correspondingly reduced frequency. + */ + return 0; + + if (IS_GEN(gt->i915, 4)) + /* + * XXX CS_TIMESTAMP appears gibberish + * + * Ville's experiments indicate that it mostly appears 'stuck' + * in that we see the register report the same cycle count + * for a couple of reads. + */ + return 0; + + intel_gt_pm_get(gt); + intel_uncore_forcewake_get(gt->uncore, FORCEWAKE_ALL); + + for_each_engine(engine, gt, id) { + u32 cycles; + u32 expected; + u64 time; + u64 dt; + + if (INTEL_GEN(engine->i915) < 7 && engine->id != RCS0) + continue; + + measure_clocks(engine, &cycles, &dt); + + time = i915_cs_timestamp_ticks_to_ns(engine->i915, cycles); + expected = i915_cs_timestamp_ns_to_ticks(engine->i915, dt); + + pr_info("%s: TIMESTAMP %d cycles [%lldns] in %lldns [%d cycles], using CS clock frequency of %uKHz\n", + engine->name, cycles, time, dt, expected, + RUNTIME_INFO(engine->i915)->cs_timestamp_frequency_hz / 1000); + + if (9 * time < 8 * dt || 8 * time > 9 * dt) { + pr_err("%s: CS ticks did not match walltime!\n", + engine->name); + err = -EINVAL; + break; + } + + if (9 * expected < 8 * cycles || 8 * expected > 9 * cycles) { + pr_err("%s: walltime did not match CS ticks!\n", + engine->name); + err = -EINVAL; + break; + } + } + + intel_uncore_forcewake_put(gt->uncore, FORCEWAKE_ALL); + intel_gt_pm_put(gt); + + return err; +} + static int live_gt_resume(void *arg) { struct intel_gt *gt = arg; @@ -52,6 +183,7 @@ static int live_gt_resume(void *arg) int intel_gt_pm_live_selftests(struct drm_i915_private *i915) { static const struct i915_subtest tests[] = { + SUBTEST(live_gt_clocks), SUBTEST(live_rc6_manual), SUBTEST(live_rps_clock_interval), SUBTEST(live_rps_control), diff --git a/drivers/gpu/drm/i915/gt/selftest_hangcheck.c b/drivers/gpu/drm/i915/gt/selftest_hangcheck.c index 4aa4cc917d8b..fb5ebf930ab2 100644 --- a/drivers/gpu/drm/i915/gt/selftest_hangcheck.c +++ b/drivers/gpu/drm/i915/gt/selftest_hangcheck.c @@ -29,6 +29,7 @@ #include "intel_gt.h" #include "intel_engine_heartbeat.h" #include "intel_engine_pm.h" +#include "selftest_engine_heartbeat.h" #include "i915_selftest.h" #include "selftests/i915_random.h" @@ -203,12 +204,12 @@ hang_create_request(struct hang *h, struct intel_engine_cs *engine) *batch++ = lower_32_bits(hws_address(hws, rq)); *batch++ = upper_32_bits(hws_address(hws, rq)); *batch++ = rq->fence.seqno; - *batch++ = MI_ARB_CHECK; + *batch++ = MI_NOOP; memset(batch, 0, 1024); batch += 1024 / sizeof(*batch); - *batch++ = MI_ARB_CHECK; + *batch++ = MI_NOOP; *batch++ = MI_BATCH_BUFFER_START | 1 << 8 | 1; *batch++ = lower_32_bits(vma->node.start); *batch++ = upper_32_bits(vma->node.start); @@ -217,12 +218,12 @@ hang_create_request(struct hang *h, struct intel_engine_cs *engine) *batch++ = 0; *batch++ = lower_32_bits(hws_address(hws, rq)); *batch++ = rq->fence.seqno; - *batch++ = MI_ARB_CHECK; + *batch++ = MI_NOOP; memset(batch, 0, 1024); batch += 1024 / sizeof(*batch); - *batch++ = MI_ARB_CHECK; + *batch++ = MI_NOOP; *batch++ = MI_BATCH_BUFFER_START | 1 << 8; *batch++ = lower_32_bits(vma->node.start); } else if (INTEL_GEN(gt->i915) >= 4) { @@ -230,24 +231,24 @@ hang_create_request(struct hang *h, struct intel_engine_cs *engine) *batch++ = 0; *batch++ = lower_32_bits(hws_address(hws, rq)); *batch++ = rq->fence.seqno; - *batch++ = MI_ARB_CHECK; + *batch++ = MI_NOOP; memset(batch, 0, 1024); batch += 1024 / sizeof(*batch); - *batch++ = MI_ARB_CHECK; + *batch++ = MI_NOOP; *batch++ = MI_BATCH_BUFFER_START | 2 << 6; *batch++ = lower_32_bits(vma->node.start); } else { *batch++ = MI_STORE_DWORD_IMM | MI_MEM_VIRTUAL; *batch++ = lower_32_bits(hws_address(hws, rq)); *batch++ = rq->fence.seqno; - *batch++ = MI_ARB_CHECK; + *batch++ = MI_NOOP; memset(batch, 0, 1024); batch += 1024 / sizeof(*batch); - *batch++ = MI_ARB_CHECK; + *batch++ = MI_NOOP; *batch++ = MI_BATCH_BUFFER_START | 2 << 6; *batch++ = lower_32_bits(vma->node.start); } @@ -310,22 +311,6 @@ static bool wait_until_running(struct hang *h, struct i915_request *rq) 1000)); } -static void engine_heartbeat_disable(struct intel_engine_cs *engine) -{ - engine->props.heartbeat_interval_ms = 0; - - intel_engine_pm_get(engine); - intel_engine_park_heartbeat(engine); -} - -static void engine_heartbeat_enable(struct intel_engine_cs *engine) -{ - intel_engine_pm_put(engine); - - engine->props.heartbeat_interval_ms = - engine->defaults.heartbeat_interval_ms; -} - static int igt_hang_sanitycheck(void *arg) { struct intel_gt *gt = arg; @@ -482,7 +467,7 @@ static int igt_reset_nop_engine(void *arg) reset_engine_count = i915_reset_engine_count(global, engine); count = 0; - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); set_bit(I915_RESET_ENGINE + id, >->reset.flags); do { int i; @@ -499,6 +484,20 @@ static int igt_reset_nop_engine(void *arg) rq = intel_context_create_request(ce); if (IS_ERR(rq)) { + struct drm_printer p = + drm_info_printer(gt->i915->drm.dev); + intel_engine_dump(engine, &p, + "%s(%s): failed to submit request\n", + __func__, + engine->name); + + GEM_TRACE("%s(%s): failed to submit request\n", + __func__, + engine->name); + GEM_TRACE_DUMP(); + + intel_gt_set_wedged(gt); + err = PTR_ERR(rq); break; } @@ -526,7 +525,7 @@ static int igt_reset_nop_engine(void *arg) } } while (time_before(jiffies, end_time)); clear_bit(I915_RESET_ENGINE + id, >->reset.flags); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); pr_info("%s(%s): %d resets\n", __func__, engine->name, count); @@ -576,7 +575,7 @@ static int __igt_reset_engine(struct intel_gt *gt, bool active) reset_count = i915_reset_count(global); reset_engine_count = i915_reset_engine_count(global, engine); - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); set_bit(I915_RESET_ENGINE + id, >->reset.flags); do { if (active) { @@ -628,7 +627,7 @@ static int __igt_reset_engine(struct intel_gt *gt, bool active) } } while (time_before(jiffies, end_time)); clear_bit(I915_RESET_ENGINE + id, >->reset.flags); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (err) break; @@ -805,10 +804,10 @@ static int __igt_reset_engines(struct intel_gt *gt, threads[tmp].resets = i915_reset_engine_count(global, other); - if (!(flags & TEST_OTHERS)) + if (other == engine && !(flags & TEST_SELF)) continue; - if (other == engine && !(flags & TEST_SELF)) + if (other != engine && !(flags & TEST_OTHERS)) continue; threads[tmp].engine = other; @@ -827,7 +826,7 @@ static int __igt_reset_engines(struct intel_gt *gt, yield(); /* start all threads before we begin */ - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); set_bit(I915_RESET_ENGINE + id, >->reset.flags); do { struct i915_request *rq = NULL; @@ -866,13 +865,29 @@ static int __igt_reset_engines(struct intel_gt *gt, count++; if (rq) { + if (rq->fence.error != -EIO) { + pr_err("i915_reset_engine(%s:%s):" + " failed to reset request %llx:%lld\n", + engine->name, test_name, + rq->fence.context, + rq->fence.seqno); + i915_request_put(rq); + + GEM_TRACE_DUMP(); + intel_gt_set_wedged(gt); + err = -EIO; + break; + } + if (i915_request_wait(rq, 0, HZ / 5) < 0) { struct drm_printer p = drm_info_printer(gt->i915->drm.dev); pr_err("i915_reset_engine(%s:%s):" - " failed to complete request after reset\n", - engine->name, test_name); + " failed to complete request %llx:%lld after reset\n", + engine->name, test_name, + rq->fence.context, + rq->fence.seqno); intel_engine_dump(engine, &p, "%s\n", engine->name); i915_request_put(rq); @@ -901,7 +916,7 @@ static int __igt_reset_engines(struct intel_gt *gt, } } while (time_before(jiffies, end_time)); clear_bit(I915_RESET_ENGINE + id, >->reset.flags); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); pr_info("i915_reset_engine(%s:%s): %lu resets\n", engine->name, test_name, count); @@ -983,7 +998,7 @@ static int igt_reset_engines(void *arg) }, { "self-priority", - TEST_OTHERS | TEST_ACTIVE | TEST_PRIORITY | TEST_SELF, + TEST_ACTIVE | TEST_PRIORITY | TEST_SELF, }, { } }; diff --git a/drivers/gpu/drm/i915/gt/selftest_lrc.c b/drivers/gpu/drm/i915/gt/selftest_lrc.c index 924bc01ef526..daa4aabab9a7 100644 --- a/drivers/gpu/drm/i915/gt/selftest_lrc.c +++ b/drivers/gpu/drm/i915/gt/selftest_lrc.c @@ -9,6 +9,7 @@ #include "gem/i915_gem_pm.h" #include "gt/intel_engine_heartbeat.h" #include "gt/intel_reset.h" +#include "gt/selftest_engine_heartbeat.h" #include "i915_selftest.h" #include "selftests/i915_random.h" @@ -51,22 +52,6 @@ static struct i915_vma *create_scratch(struct intel_gt *gt) return vma; } -static void engine_heartbeat_disable(struct intel_engine_cs *engine) -{ - engine->props.heartbeat_interval_ms = 0; - - intel_engine_pm_get(engine); - intel_engine_park_heartbeat(engine); -} - -static void engine_heartbeat_enable(struct intel_engine_cs *engine) -{ - intel_engine_pm_put(engine); - - engine->props.heartbeat_interval_ms = - engine->defaults.heartbeat_interval_ms; -} - static bool is_active(struct i915_request *rq) { if (i915_request_is_active(rq)) @@ -75,7 +60,7 @@ static bool is_active(struct i915_request *rq) if (i915_request_on_hold(rq)) return true; - if (i915_request_started(rq)) + if (i915_request_has_initial_breadcrumb(rq) && i915_request_started(rq)) return true; return false; @@ -234,7 +219,7 @@ static int live_unlite_restore(struct intel_gt *gt, int prio) err = -EIO; break; } - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); for (n = 0; n < ARRAY_SIZE(ce); n++) { struct intel_context *tmp; @@ -332,7 +317,7 @@ static int live_unlite_restore(struct intel_gt *gt, int prio) i915_request_put(rq[0]); err_ce: - tasklet_kill(&engine->execlists.tasklet); /* flush submission */ + intel_engine_flush_submission(engine); igt_spinner_end(&spin); for (n = 0; n < ARRAY_SIZE(ce); n++) { if (IS_ERR_OR_NULL(ce[n])) @@ -342,7 +327,7 @@ err_ce: intel_context_put(ce[n]); } - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (igt_live_test_end(&t)) err = -EIO; if (err) @@ -363,6 +348,156 @@ static int live_unlite_preempt(void *arg) return live_unlite_restore(arg, I915_USER_PRIORITY(I915_PRIORITY_MAX)); } +static int live_unlite_ring(void *arg) +{ + struct intel_gt *gt = arg; + struct intel_engine_cs *engine; + struct igt_spinner spin; + enum intel_engine_id id; + int err = 0; + + /* + * Setup a preemption event that will cause almost the entire ring + * to be unwound, potentially fooling our intel_ring_direction() + * into emitting a forward lite-restore instead of the rollback. + */ + + if (igt_spinner_init(&spin, gt)) + return -ENOMEM; + + for_each_engine(engine, gt, id) { + struct intel_context *ce[2] = {}; + struct i915_request *rq; + struct igt_live_test t; + int n; + + if (!intel_engine_has_preemption(engine)) + continue; + + if (!intel_engine_can_store_dword(engine)) + continue; + + if (igt_live_test_begin(&t, gt->i915, __func__, engine->name)) { + err = -EIO; + break; + } + st_engine_heartbeat_disable(engine); + + for (n = 0; n < ARRAY_SIZE(ce); n++) { + struct intel_context *tmp; + + tmp = intel_context_create(engine); + if (IS_ERR(tmp)) { + err = PTR_ERR(tmp); + goto err_ce; + } + + err = intel_context_pin(tmp); + if (err) { + intel_context_put(tmp); + goto err_ce; + } + + memset32(tmp->ring->vaddr, + 0xdeadbeef, /* trigger a hang if executed */ + tmp->ring->vma->size / sizeof(u32)); + + ce[n] = tmp; + } + + /* Create max prio spinner, followed by N low prio nops */ + rq = igt_spinner_create_request(&spin, ce[0], MI_ARB_CHECK); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err_ce; + } + + i915_request_get(rq); + rq->sched.attr.priority = I915_PRIORITY_BARRIER; + i915_request_add(rq); + + if (!igt_wait_for_spinner(&spin, rq)) { + intel_gt_set_wedged(gt); + i915_request_put(rq); + err = -ETIME; + goto err_ce; + } + + /* Fill the ring, until we will cause a wrap */ + n = 0; + while (intel_ring_direction(ce[0]->ring, + rq->wa_tail, + ce[0]->ring->tail) <= 0) { + struct i915_request *tmp; + + tmp = intel_context_create_request(ce[0]); + if (IS_ERR(tmp)) { + err = PTR_ERR(tmp); + i915_request_put(rq); + goto err_ce; + } + + i915_request_add(tmp); + intel_engine_flush_submission(engine); + n++; + } + intel_engine_flush_submission(engine); + pr_debug("%s: Filled ring with %d nop tails {size:%x, tail:%x, emit:%x, rq.tail:%x}\n", + engine->name, n, + ce[0]->ring->size, + ce[0]->ring->tail, + ce[0]->ring->emit, + rq->tail); + GEM_BUG_ON(intel_ring_direction(ce[0]->ring, + rq->tail, + ce[0]->ring->tail) <= 0); + i915_request_put(rq); + + /* Create a second ring to preempt the first ring after rq[0] */ + rq = intel_context_create_request(ce[1]); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err_ce; + } + + rq->sched.attr.priority = I915_PRIORITY_BARRIER; + i915_request_get(rq); + i915_request_add(rq); + + err = wait_for_submit(engine, rq, HZ / 2); + i915_request_put(rq); + if (err) { + pr_err("%s: preemption request was not submitted\n", + engine->name); + err = -ETIME; + } + + pr_debug("%s: ring[0]:{ tail:%x, emit:%x }, ring[1]:{ tail:%x, emit:%x }\n", + engine->name, + ce[0]->ring->tail, ce[0]->ring->emit, + ce[1]->ring->tail, ce[1]->ring->emit); + +err_ce: + intel_engine_flush_submission(engine); + igt_spinner_end(&spin); + for (n = 0; n < ARRAY_SIZE(ce); n++) { + if (IS_ERR_OR_NULL(ce[n])) + break; + + intel_context_unpin(ce[n]); + intel_context_put(ce[n]); + } + st_engine_heartbeat_enable(engine); + if (igt_live_test_end(&t)) + err = -EIO; + if (err) + break; + } + + igt_spinner_fini(&spin); + return err; +} + static int live_pin_rewind(void *arg) { struct intel_gt *gt = arg; @@ -471,7 +606,7 @@ static int live_hold_reset(void *arg) break; } - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); rq = igt_spinner_create_request(&spin, ce, MI_ARB_CHECK); if (IS_ERR(rq)) { @@ -531,7 +666,7 @@ static int live_hold_reset(void *arg) i915_request_put(rq); out: - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); intel_context_put(ce); if (err) break; @@ -578,7 +713,7 @@ static int live_error_interrupt(void *arg) const struct error_phase *p; int err = 0; - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); for (p = phases; p->error[0] != GOOD; p++) { struct i915_request *client[ARRAY_SIZE(phases->error)]; @@ -677,7 +812,7 @@ out: } } - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (err) { intel_gt_set_wedged(gt); return err; @@ -845,10 +980,11 @@ static int live_timeslice_preempt(void *arg) { struct intel_gt *gt = arg; struct drm_i915_gem_object *obj; + struct intel_engine_cs *engine; + enum intel_engine_id id; struct i915_vma *vma; void *vaddr; int err = 0; - int count; /* * If a request takes too long, we would like to give other users @@ -885,26 +1021,21 @@ static int live_timeslice_preempt(void *arg) if (err) goto err_pin; - for_each_prime_number_from(count, 1, 16) { - struct intel_engine_cs *engine; - enum intel_engine_id id; - - for_each_engine(engine, gt, id) { - if (!intel_engine_has_preemption(engine)) - continue; + for_each_engine(engine, gt, id) { + if (!intel_engine_has_preemption(engine)) + continue; - memset(vaddr, 0, PAGE_SIZE); + memset(vaddr, 0, PAGE_SIZE); - engine_heartbeat_disable(engine); - err = slice_semaphore_queue(engine, vma, count); - engine_heartbeat_enable(engine); - if (err) - goto err_pin; + st_engine_heartbeat_disable(engine); + err = slice_semaphore_queue(engine, vma, 5); + st_engine_heartbeat_enable(engine); + if (err) + goto err_pin; - if (igt_flush_test(gt->i915)) { - err = -EIO; - goto err_pin; - } + if (igt_flush_test(gt->i915)) { + err = -EIO; + goto err_pin; } } @@ -1020,7 +1151,7 @@ static int live_timeslice_rewind(void *arg) * Expect execution/evaluation order XZY */ - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); timeslice = xchg(&engine->props.timeslice_duration_ms, 1); slot = memset32(engine->status_page.addr + 1000, 0, 4); @@ -1031,18 +1162,18 @@ static int live_timeslice_rewind(void *arg) goto err; } - rq[0] = create_rewinder(ce, NULL, slot, X); - if (IS_ERR(rq[0])) { + rq[A1] = create_rewinder(ce, NULL, slot, X); + if (IS_ERR(rq[A1])) { intel_context_put(ce); goto err; } - rq[1] = create_rewinder(ce, NULL, slot, Y); + rq[A2] = create_rewinder(ce, NULL, slot, Y); intel_context_put(ce); - if (IS_ERR(rq[1])) + if (IS_ERR(rq[A2])) goto err; - err = wait_for_submit(engine, rq[1], HZ / 2); + err = wait_for_submit(engine, rq[A2], HZ / 2); if (err) { pr_err("%s: failed to submit first context\n", engine->name); @@ -1055,12 +1186,12 @@ static int live_timeslice_rewind(void *arg) goto err; } - rq[2] = create_rewinder(ce, rq[0], slot, Z); + rq[B1] = create_rewinder(ce, rq[A1], slot, Z); intel_context_put(ce); if (IS_ERR(rq[2])) goto err; - err = wait_for_submit(engine, rq[2], HZ / 2); + err = wait_for_submit(engine, rq[B1], HZ / 2); if (err) { pr_err("%s: failed to submit second context\n", engine->name); @@ -1068,6 +1199,7 @@ static int live_timeslice_rewind(void *arg) } /* ELSP[] = { { A:rq1, A:rq2 }, { B:rq1 } } */ + ENGINE_TRACE(engine, "forcing tasklet for rewind\n"); if (i915_request_is_active(rq[A2])) { /* semaphore yielded! */ /* Wait for the timeslice to kick in */ del_timer(&engine->execlists.timer); @@ -1114,7 +1246,7 @@ err: wmb(); engine->props.timeslice_duration_ms = timeslice; - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); for (i = 0; i < 3; i++) i915_request_put(rq[i]); if (igt_flush_test(gt->i915)) @@ -1140,9 +1272,17 @@ static struct i915_request *nop_request(struct intel_engine_cs *engine) return rq; } -static long timeslice_threshold(const struct intel_engine_cs *engine) +static long slice_timeout(struct intel_engine_cs *engine) { - return 2 * msecs_to_jiffies_timeout(timeslice(engine)) + 1; + long timeout; + + /* Enough time for a timeslice to kick in, and kick out */ + timeout = 2 * msecs_to_jiffies_timeout(timeslice(engine)); + + /* Enough time for the nop request to complete */ + timeout += HZ / 5; + + return timeout + 1; } static int live_timeslice_queue(void *arg) @@ -1198,7 +1338,7 @@ static int live_timeslice_queue(void *arg) if (!intel_engine_has_preemption(engine)) continue; - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); memset(vaddr, 0, PAGE_SIZE); /* ELSP[0]: semaphore wait */ @@ -1243,24 +1383,8 @@ static int live_timeslice_queue(void *arg) intel_engine_flush_submission(engine); } while (READ_ONCE(engine->execlists.pending[0])); - if (!READ_ONCE(engine->execlists.timer.expires) && - execlists_active(&engine->execlists) == rq && - !i915_request_completed(rq)) { - struct drm_printer p = - drm_info_printer(gt->i915->drm.dev); - - GEM_TRACE_ERR("%s: Failed to enable timeslicing!\n", - engine->name); - intel_engine_dump(engine, &p, - "%s\n", engine->name); - GEM_TRACE_DUMP(); - - memset(vaddr, 0xff, PAGE_SIZE); - err = -EINVAL; - } - /* Timeslice every jiffy, so within 2 we should signal */ - if (i915_request_wait(rq, 0, timeslice_threshold(engine)) < 0) { + if (i915_request_wait(rq, 0, slice_timeout(engine)) < 0) { struct drm_printer p = drm_info_printer(gt->i915->drm.dev); @@ -1275,7 +1399,7 @@ static int live_timeslice_queue(void *arg) err_rq: i915_request_put(rq); err_heartbeat: - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (err) break; } @@ -1321,7 +1445,7 @@ static int live_timeslice_nopreempt(void *arg) break; } - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); timeslice = xchg(&engine->props.timeslice_duration_ms, 1); /* Create an unpreemptible spinner */ @@ -1349,7 +1473,7 @@ static int live_timeslice_nopreempt(void *arg) ce = intel_context_create(engine); if (IS_ERR(ce)) { - err = PTR_ERR(rq); + err = PTR_ERR(ce); goto out_spin; } @@ -1379,7 +1503,7 @@ static int live_timeslice_nopreempt(void *arg) * allow the maximum priority barrier through. Wait long * enough to see if it is timesliced in by mistake. */ - if (i915_request_wait(rq, 0, timeslice_threshold(engine)) >= 0) { + if (i915_request_wait(rq, 0, slice_timeout(engine)) >= 0) { pr_err("%s: I915_PRIORITY_BARRIER request completed, bypassing no-preempt request\n", engine->name); err = -EINVAL; @@ -1390,7 +1514,7 @@ out_spin: igt_spinner_end(&spin); out_heartbeat: xchg(&engine->props.timeslice_duration_ms, timeslice); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (err) break; @@ -2294,7 +2418,7 @@ static int live_suppress_self_preempt(void *arg) if (igt_flush_test(gt->i915)) goto err_wedged; - intel_engine_pm_get(engine); + st_engine_heartbeat_disable(engine); engine->execlists.preempt_hang.count = 0; rq_a = spinner_create_request(&a.spin, @@ -2302,14 +2426,14 @@ static int live_suppress_self_preempt(void *arg) MI_NOOP); if (IS_ERR(rq_a)) { err = PTR_ERR(rq_a); - intel_engine_pm_put(engine); + st_engine_heartbeat_enable(engine); goto err_client_b; } i915_request_add(rq_a); if (!igt_wait_for_spinner(&a.spin, rq_a)) { pr_err("First client failed to start\n"); - intel_engine_pm_put(engine); + st_engine_heartbeat_enable(engine); goto err_wedged; } @@ -2321,7 +2445,7 @@ static int live_suppress_self_preempt(void *arg) MI_NOOP); if (IS_ERR(rq_b)) { err = PTR_ERR(rq_b); - intel_engine_pm_put(engine); + st_engine_heartbeat_enable(engine); goto err_client_b; } i915_request_add(rq_b); @@ -2332,7 +2456,7 @@ static int live_suppress_self_preempt(void *arg) if (!igt_wait_for_spinner(&b.spin, rq_b)) { pr_err("Second client failed to start\n"); - intel_engine_pm_put(engine); + st_engine_heartbeat_enable(engine); goto err_wedged; } @@ -2346,12 +2470,12 @@ static int live_suppress_self_preempt(void *arg) engine->name, engine->execlists.preempt_hang.count, depth); - intel_engine_pm_put(engine); + st_engine_heartbeat_enable(engine); err = -EINVAL; goto err_client_b; } - intel_engine_pm_put(engine); + st_engine_heartbeat_enable(engine); if (igt_flush_test(gt->i915)) goto err_wedged; } @@ -2371,183 +2495,6 @@ err_wedged: goto err_client_b; } -static int __i915_sw_fence_call -dummy_notify(struct i915_sw_fence *fence, enum i915_sw_fence_notify state) -{ - return NOTIFY_DONE; -} - -static struct i915_request *dummy_request(struct intel_engine_cs *engine) -{ - struct i915_request *rq; - - rq = kzalloc(sizeof(*rq), GFP_KERNEL); - if (!rq) - return NULL; - - rq->engine = engine; - - spin_lock_init(&rq->lock); - INIT_LIST_HEAD(&rq->fence.cb_list); - rq->fence.lock = &rq->lock; - rq->fence.ops = &i915_fence_ops; - - i915_sched_node_init(&rq->sched); - - /* mark this request as permanently incomplete */ - rq->fence.seqno = 1; - BUILD_BUG_ON(sizeof(rq->fence.seqno) != 8); /* upper 32b == 0 */ - rq->hwsp_seqno = (u32 *)&rq->fence.seqno + 1; - GEM_BUG_ON(i915_request_completed(rq)); - - i915_sw_fence_init(&rq->submit, dummy_notify); - set_bit(I915_FENCE_FLAG_ACTIVE, &rq->fence.flags); - - spin_lock_init(&rq->lock); - rq->fence.lock = &rq->lock; - INIT_LIST_HEAD(&rq->fence.cb_list); - - return rq; -} - -static void dummy_request_free(struct i915_request *dummy) -{ - /* We have to fake the CS interrupt to kick the next request */ - i915_sw_fence_commit(&dummy->submit); - - i915_request_mark_complete(dummy); - dma_fence_signal(&dummy->fence); - - i915_sched_node_fini(&dummy->sched); - i915_sw_fence_fini(&dummy->submit); - - dma_fence_free(&dummy->fence); -} - -static int live_suppress_wait_preempt(void *arg) -{ - struct intel_gt *gt = arg; - struct preempt_client client[4]; - struct i915_request *rq[ARRAY_SIZE(client)] = {}; - struct intel_engine_cs *engine; - enum intel_engine_id id; - int err = -ENOMEM; - int i; - - /* - * Waiters are given a little priority nudge, but not enough - * to actually cause any preemption. Double check that we do - * not needlessly generate preempt-to-idle cycles. - */ - - if (!HAS_LOGICAL_RING_PREEMPTION(gt->i915)) - return 0; - - if (preempt_client_init(gt, &client[0])) /* ELSP[0] */ - return -ENOMEM; - if (preempt_client_init(gt, &client[1])) /* ELSP[1] */ - goto err_client_0; - if (preempt_client_init(gt, &client[2])) /* head of queue */ - goto err_client_1; - if (preempt_client_init(gt, &client[3])) /* bystander */ - goto err_client_2; - - for_each_engine(engine, gt, id) { - int depth; - - if (!intel_engine_has_preemption(engine)) - continue; - - if (!engine->emit_init_breadcrumb) - continue; - - for (depth = 0; depth < ARRAY_SIZE(client); depth++) { - struct i915_request *dummy; - - engine->execlists.preempt_hang.count = 0; - - dummy = dummy_request(engine); - if (!dummy) - goto err_client_3; - - for (i = 0; i < ARRAY_SIZE(client); i++) { - struct i915_request *this; - - this = spinner_create_request(&client[i].spin, - client[i].ctx, engine, - MI_NOOP); - if (IS_ERR(this)) { - err = PTR_ERR(this); - goto err_wedged; - } - - /* Disable NEWCLIENT promotion */ - __i915_active_fence_set(&i915_request_timeline(this)->last_request, - &dummy->fence); - - rq[i] = i915_request_get(this); - i915_request_add(this); - } - - dummy_request_free(dummy); - - GEM_BUG_ON(i915_request_completed(rq[0])); - if (!igt_wait_for_spinner(&client[0].spin, rq[0])) { - pr_err("%s: First client failed to start\n", - engine->name); - goto err_wedged; - } - GEM_BUG_ON(!i915_request_started(rq[0])); - - if (i915_request_wait(rq[depth], - I915_WAIT_PRIORITY, - 1) != -ETIME) { - pr_err("%s: Waiter depth:%d completed!\n", - engine->name, depth); - goto err_wedged; - } - - for (i = 0; i < ARRAY_SIZE(client); i++) { - igt_spinner_end(&client[i].spin); - i915_request_put(rq[i]); - rq[i] = NULL; - } - - if (igt_flush_test(gt->i915)) - goto err_wedged; - - if (engine->execlists.preempt_hang.count) { - pr_err("%s: Preemption recorded x%d, depth %d; should have been suppressed!\n", - engine->name, - engine->execlists.preempt_hang.count, - depth); - err = -EINVAL; - goto err_client_3; - } - } - } - - err = 0; -err_client_3: - preempt_client_fini(&client[3]); -err_client_2: - preempt_client_fini(&client[2]); -err_client_1: - preempt_client_fini(&client[1]); -err_client_0: - preempt_client_fini(&client[0]); - return err; - -err_wedged: - for (i = 0; i < ARRAY_SIZE(client); i++) { - igt_spinner_end(&client[i].spin); - i915_request_put(rq[i]); - } - intel_gt_set_wedged(gt); - err = -EIO; - goto err_client_3; -} - static int live_chain_preempt(void *arg) { struct intel_gt *gt = arg; @@ -2796,6 +2743,168 @@ err_ce: return err; } +static int __live_preempt_ring(struct intel_engine_cs *engine, + struct igt_spinner *spin, + int queue_sz, int ring_sz) +{ + struct intel_context *ce[2] = {}; + struct i915_request *rq; + struct igt_live_test t; + int err = 0; + int n; + + if (igt_live_test_begin(&t, engine->i915, __func__, engine->name)) + return -EIO; + + for (n = 0; n < ARRAY_SIZE(ce); n++) { + struct intel_context *tmp; + + tmp = intel_context_create(engine); + if (IS_ERR(tmp)) { + err = PTR_ERR(tmp); + goto err_ce; + } + + tmp->ring = __intel_context_ring_size(ring_sz); + + err = intel_context_pin(tmp); + if (err) { + intel_context_put(tmp); + goto err_ce; + } + + memset32(tmp->ring->vaddr, + 0xdeadbeef, /* trigger a hang if executed */ + tmp->ring->vma->size / sizeof(u32)); + + ce[n] = tmp; + } + + rq = igt_spinner_create_request(spin, ce[0], MI_ARB_CHECK); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err_ce; + } + + i915_request_get(rq); + rq->sched.attr.priority = I915_PRIORITY_BARRIER; + i915_request_add(rq); + + if (!igt_wait_for_spinner(spin, rq)) { + intel_gt_set_wedged(engine->gt); + i915_request_put(rq); + err = -ETIME; + goto err_ce; + } + + /* Fill the ring, until we will cause a wrap */ + n = 0; + while (ce[0]->ring->tail - rq->wa_tail <= queue_sz) { + struct i915_request *tmp; + + tmp = intel_context_create_request(ce[0]); + if (IS_ERR(tmp)) { + err = PTR_ERR(tmp); + i915_request_put(rq); + goto err_ce; + } + + i915_request_add(tmp); + intel_engine_flush_submission(engine); + n++; + } + intel_engine_flush_submission(engine); + pr_debug("%s: Filled %d with %d nop tails {size:%x, tail:%x, emit:%x, rq.tail:%x}\n", + engine->name, queue_sz, n, + ce[0]->ring->size, + ce[0]->ring->tail, + ce[0]->ring->emit, + rq->tail); + i915_request_put(rq); + + /* Create a second request to preempt the first ring */ + rq = intel_context_create_request(ce[1]); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err_ce; + } + + rq->sched.attr.priority = I915_PRIORITY_BARRIER; + i915_request_get(rq); + i915_request_add(rq); + + err = wait_for_submit(engine, rq, HZ / 2); + i915_request_put(rq); + if (err) { + pr_err("%s: preemption request was not submited\n", + engine->name); + err = -ETIME; + } + + pr_debug("%s: ring[0]:{ tail:%x, emit:%x }, ring[1]:{ tail:%x, emit:%x }\n", + engine->name, + ce[0]->ring->tail, ce[0]->ring->emit, + ce[1]->ring->tail, ce[1]->ring->emit); + +err_ce: + intel_engine_flush_submission(engine); + igt_spinner_end(spin); + for (n = 0; n < ARRAY_SIZE(ce); n++) { + if (IS_ERR_OR_NULL(ce[n])) + break; + + intel_context_unpin(ce[n]); + intel_context_put(ce[n]); + } + if (igt_live_test_end(&t)) + err = -EIO; + return err; +} + +static int live_preempt_ring(void *arg) +{ + struct intel_gt *gt = arg; + struct intel_engine_cs *engine; + struct igt_spinner spin; + enum intel_engine_id id; + int err = 0; + + /* + * Check that we rollback large chunks of a ring in order to do a + * preemption event. Similar to live_unlite_ring, but looking at + * ring size rather than the impact of intel_ring_direction(). + */ + + if (igt_spinner_init(&spin, gt)) + return -ENOMEM; + + for_each_engine(engine, gt, id) { + int n; + + if (!intel_engine_has_preemption(engine)) + continue; + + if (!intel_engine_can_store_dword(engine)) + continue; + + st_engine_heartbeat_disable(engine); + + for (n = 0; n <= 3; n++) { + err = __live_preempt_ring(engine, &spin, + n * SZ_4K / 4, SZ_4K); + if (err) + break; + } + + st_engine_heartbeat_enable(engine); + if (err) + break; + } + + igt_spinner_fini(&spin); + return err; +} + static int live_preempt_gang(void *arg) { struct intel_gt *gt = arg; @@ -2840,16 +2949,8 @@ static int live_preempt_gang(void *arg) /* Submit each spinner at increasing priority */ engine->schedule(rq, &attr); - - if (prio <= I915_PRIORITY_MAX) - continue; - - if (prio > (INT_MAX >> I915_USER_PRIORITY_SHIFT)) - break; - - if (__igt_timeout(end_time, NULL)) - break; - } while (1); + } while (prio <= I915_PRIORITY_MAX && + !__igt_timeout(end_time, NULL)); pr_debug("%s: Preempt chain of %d requests\n", engine->name, prio); @@ -3417,7 +3518,7 @@ static int smoke_crescendo_thread(void *arg) return err; count++; - } while (!__igt_timeout(end_time, NULL)); + } while (count < smoke->ncontext && !__igt_timeout(end_time, NULL)); smoke->count = count; return 0; @@ -3493,7 +3594,7 @@ static int smoke_random(struct preempt_smoke *smoke, unsigned int flags) count++; } - } while (!__igt_timeout(end_time, NULL)); + } while (count < smoke->ncontext && !__igt_timeout(end_time, NULL)); pr_info("Submitted %lu random:%x requests across %d engines and %d contexts\n", count, flags, @@ -3506,7 +3607,7 @@ static int live_preempt_smoke(void *arg) struct preempt_smoke smoke = { .gt = arg, .prng = I915_RND_STATE_INITIALIZER(i915_selftest.random_seed), - .ncontext = 1024, + .ncontext = 256, }; const unsigned int phase[] = { 0, BATCH }; struct igt_live_test t; @@ -3706,13 +3807,43 @@ out: return err; } +static unsigned int +__select_siblings(struct intel_gt *gt, + unsigned int class, + struct intel_engine_cs **siblings, + bool (*filter)(const struct intel_engine_cs *)) +{ + unsigned int n = 0; + unsigned int inst; + + for (inst = 0; inst <= MAX_ENGINE_INSTANCE; inst++) { + if (!gt->engine_class[class][inst]) + continue; + + if (filter && !filter(gt->engine_class[class][inst])) + continue; + + siblings[n++] = gt->engine_class[class][inst]; + } + + return n; +} + +static unsigned int +select_siblings(struct intel_gt *gt, + unsigned int class, + struct intel_engine_cs **siblings) +{ + return __select_siblings(gt, class, siblings, NULL); +} + static int live_virtual_engine(void *arg) { struct intel_gt *gt = arg; struct intel_engine_cs *siblings[MAX_ENGINE_INSTANCE + 1]; struct intel_engine_cs *engine; enum intel_engine_id id; - unsigned int class, inst; + unsigned int class; int err; if (intel_uc_uses_guc_submission(>->uc)) @@ -3730,13 +3861,7 @@ static int live_virtual_engine(void *arg) for (class = 0; class <= MAX_ENGINE_CLASS; class++) { int nsibling, n; - nsibling = 0; - for (inst = 0; inst <= MAX_ENGINE_INSTANCE; inst++) { - if (!gt->engine_class[class][inst]) - continue; - - siblings[nsibling++] = gt->engine_class[class][inst]; - } + nsibling = select_siblings(gt, class, siblings); if (nsibling < 2) continue; @@ -3845,7 +3970,7 @@ static int live_virtual_mask(void *arg) { struct intel_gt *gt = arg; struct intel_engine_cs *siblings[MAX_ENGINE_INSTANCE + 1]; - unsigned int class, inst; + unsigned int class; int err; if (intel_uc_uses_guc_submission(>->uc)) @@ -3854,17 +3979,178 @@ static int live_virtual_mask(void *arg) for (class = 0; class <= MAX_ENGINE_CLASS; class++) { unsigned int nsibling; - nsibling = 0; - for (inst = 0; inst <= MAX_ENGINE_INSTANCE; inst++) { - if (!gt->engine_class[class][inst]) - break; + nsibling = select_siblings(gt, class, siblings); + if (nsibling < 2) + continue; + + err = mask_virtual_engine(gt, siblings, nsibling); + if (err) + return err; + } + + return 0; +} + +static int slicein_virtual_engine(struct intel_gt *gt, + struct intel_engine_cs **siblings, + unsigned int nsibling) +{ + const long timeout = slice_timeout(siblings[0]); + struct intel_context *ce; + struct i915_request *rq; + struct igt_spinner spin; + unsigned int n; + int err = 0; + + /* + * Virtual requests must take part in timeslicing on the target engines. + */ + + if (igt_spinner_init(&spin, gt)) + return -ENOMEM; + + for (n = 0; n < nsibling; n++) { + ce = intel_context_create(siblings[n]); + if (IS_ERR(ce)) { + err = PTR_ERR(ce); + goto out; + } + + rq = igt_spinner_create_request(&spin, ce, MI_ARB_CHECK); + intel_context_put(ce); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto out; + } + + i915_request_add(rq); + } + + ce = intel_execlists_create_virtual(siblings, nsibling); + if (IS_ERR(ce)) { + err = PTR_ERR(ce); + goto out; + } + + rq = intel_context_create_request(ce); + intel_context_put(ce); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto out; + } + + i915_request_get(rq); + i915_request_add(rq); + if (i915_request_wait(rq, 0, timeout) < 0) { + GEM_TRACE_ERR("%s(%s) failed to slice in virtual request\n", + __func__, rq->engine->name); + GEM_TRACE_DUMP(); + intel_gt_set_wedged(gt); + err = -EIO; + } + i915_request_put(rq); + +out: + igt_spinner_end(&spin); + if (igt_flush_test(gt->i915)) + err = -EIO; + igt_spinner_fini(&spin); + return err; +} + +static int sliceout_virtual_engine(struct intel_gt *gt, + struct intel_engine_cs **siblings, + unsigned int nsibling) +{ + const long timeout = slice_timeout(siblings[0]); + struct intel_context *ce; + struct i915_request *rq; + struct igt_spinner spin; + unsigned int n; + int err = 0; + + /* + * Virtual requests must allow others a fair timeslice. + */ + + if (igt_spinner_init(&spin, gt)) + return -ENOMEM; + + /* XXX We do not handle oversubscription and fairness with normal rq */ + for (n = 0; n < nsibling; n++) { + ce = intel_execlists_create_virtual(siblings, nsibling); + if (IS_ERR(ce)) { + err = PTR_ERR(ce); + goto out; + } + + rq = igt_spinner_create_request(&spin, ce, MI_ARB_CHECK); + intel_context_put(ce); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto out; + } + + i915_request_add(rq); + } + + for (n = 0; !err && n < nsibling; n++) { + ce = intel_context_create(siblings[n]); + if (IS_ERR(ce)) { + err = PTR_ERR(ce); + goto out; + } + + rq = intel_context_create_request(ce); + intel_context_put(ce); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto out; + } - siblings[nsibling++] = gt->engine_class[class][inst]; + i915_request_get(rq); + i915_request_add(rq); + if (i915_request_wait(rq, 0, timeout) < 0) { + GEM_TRACE_ERR("%s(%s) failed to slice out virtual request\n", + __func__, siblings[n]->name); + GEM_TRACE_DUMP(); + intel_gt_set_wedged(gt); + err = -EIO; } + i915_request_put(rq); + } + +out: + igt_spinner_end(&spin); + if (igt_flush_test(gt->i915)) + err = -EIO; + igt_spinner_fini(&spin); + return err; +} + +static int live_virtual_slice(void *arg) +{ + struct intel_gt *gt = arg; + struct intel_engine_cs *siblings[MAX_ENGINE_INSTANCE + 1]; + unsigned int class; + int err; + + if (intel_uc_uses_guc_submission(>->uc)) + return 0; + + for (class = 0; class <= MAX_ENGINE_CLASS; class++) { + unsigned int nsibling; + + nsibling = __select_siblings(gt, class, siblings, + intel_engine_has_timeslices); if (nsibling < 2) continue; - err = mask_virtual_engine(gt, siblings, nsibling); + err = slicein_virtual_engine(gt, siblings, nsibling); + if (err) + return err; + + err = sliceout_virtual_engine(gt, siblings, nsibling); if (err) return err; } @@ -3982,7 +4268,7 @@ static int live_virtual_preserved(void *arg) { struct intel_gt *gt = arg; struct intel_engine_cs *siblings[MAX_ENGINE_INSTANCE + 1]; - unsigned int class, inst; + unsigned int class; /* * Check that the context image retains non-privileged (user) registers @@ -4000,13 +4286,7 @@ static int live_virtual_preserved(void *arg) for (class = 0; class <= MAX_ENGINE_CLASS; class++) { int nsibling, err; - nsibling = 0; - for (inst = 0; inst <= MAX_ENGINE_INSTANCE; inst++) { - if (!gt->engine_class[class][inst]) - continue; - - siblings[nsibling++] = gt->engine_class[class][inst]; - } + nsibling = select_siblings(gt, class, siblings); if (nsibling < 2) continue; @@ -4217,7 +4497,7 @@ static int live_virtual_bond(void *arg) }; struct intel_gt *gt = arg; struct intel_engine_cs *siblings[MAX_ENGINE_INSTANCE + 1]; - unsigned int class, inst; + unsigned int class; int err; if (intel_uc_uses_guc_submission(>->uc)) @@ -4227,14 +4507,7 @@ static int live_virtual_bond(void *arg) const struct phase *p; int nsibling; - nsibling = 0; - for (inst = 0; inst <= MAX_ENGINE_INSTANCE; inst++) { - if (!gt->engine_class[class][inst]) - break; - - GEM_BUG_ON(nsibling == ARRAY_SIZE(siblings)); - siblings[nsibling++] = gt->engine_class[class][inst]; - } + nsibling = select_siblings(gt, class, siblings); if (nsibling < 2) continue; @@ -4280,7 +4553,7 @@ static int reset_virtual_engine(struct intel_gt *gt, } for (n = 0; n < nsibling; n++) - engine_heartbeat_disable(siblings[n]); + st_engine_heartbeat_disable(siblings[n]); rq = igt_spinner_create_request(&spin, ve, MI_ARB_CHECK); if (IS_ERR(rq)) { @@ -4351,7 +4624,7 @@ out_rq: i915_request_put(rq); out_heartbeat: for (n = 0; n < nsibling; n++) - engine_heartbeat_enable(siblings[n]); + st_engine_heartbeat_enable(siblings[n]); intel_context_put(ve); out_spin: @@ -4363,7 +4636,7 @@ static int live_virtual_reset(void *arg) { struct intel_gt *gt = arg; struct intel_engine_cs *siblings[MAX_ENGINE_INSTANCE + 1]; - unsigned int class, inst; + unsigned int class; /* * Check that we handle a reset event within a virtual engine. @@ -4381,13 +4654,7 @@ static int live_virtual_reset(void *arg) for (class = 0; class <= MAX_ENGINE_CLASS; class++) { int nsibling, err; - nsibling = 0; - for (inst = 0; inst <= MAX_ENGINE_INSTANCE; inst++) { - if (!gt->engine_class[class][inst]) - continue; - - siblings[nsibling++] = gt->engine_class[class][inst]; - } + nsibling = select_siblings(gt, class, siblings); if (nsibling < 2) continue; @@ -4405,6 +4672,7 @@ int intel_execlists_live_selftests(struct drm_i915_private *i915) SUBTEST(live_sanitycheck), SUBTEST(live_unlite_switch), SUBTEST(live_unlite_preempt), + SUBTEST(live_unlite_ring), SUBTEST(live_pin_rewind), SUBTEST(live_hold_reset), SUBTEST(live_error_interrupt), @@ -4418,8 +4686,8 @@ int intel_execlists_live_selftests(struct drm_i915_private *i915) SUBTEST(live_nopreempt), SUBTEST(live_preempt_cancel), SUBTEST(live_suppress_self_preempt), - SUBTEST(live_suppress_wait_preempt), SUBTEST(live_chain_preempt), + SUBTEST(live_preempt_ring), SUBTEST(live_preempt_gang), SUBTEST(live_preempt_timeout), SUBTEST(live_preempt_user), @@ -4427,6 +4695,7 @@ int intel_execlists_live_selftests(struct drm_i915_private *i915) SUBTEST(live_virtual_engine), SUBTEST(live_virtual_mask), SUBTEST(live_virtual_preserved), + SUBTEST(live_virtual_slice), SUBTEST(live_virtual_bond), SUBTEST(live_virtual_reset), }; @@ -5030,7 +5299,7 @@ static int live_lrc_gpr(void *arg) return PTR_ERR(scratch); for_each_engine(engine, gt, id) { - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); err = __live_lrc_gpr(engine, scratch, false); if (err) @@ -5041,7 +5310,7 @@ static int live_lrc_gpr(void *arg) goto err; err: - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (igt_flush_test(gt->i915)) err = -EIO; if (err) @@ -5190,7 +5459,7 @@ static int live_lrc_timestamp(void *arg) for_each_engine(data.engine, gt, id) { int i, err = 0; - engine_heartbeat_disable(data.engine); + st_engine_heartbeat_disable(data.engine); for (i = 0; i < ARRAY_SIZE(data.ce); i++) { struct intel_context *tmp; @@ -5223,7 +5492,7 @@ static int live_lrc_timestamp(void *arg) } err: - engine_heartbeat_enable(data.engine); + st_engine_heartbeat_enable(data.engine); for (i = 0; i < ARRAY_SIZE(data.ce); i++) { if (!data.ce[i]) break; diff --git a/drivers/gpu/drm/i915/gt/selftest_mocs.c b/drivers/gpu/drm/i915/gt/selftest_mocs.c index 63f87d8608c3..b25eba50c88e 100644 --- a/drivers/gpu/drm/i915/gt/selftest_mocs.c +++ b/drivers/gpu/drm/i915/gt/selftest_mocs.c @@ -157,7 +157,7 @@ static int read_mocs_table(struct i915_request *rq, { u32 addr; - if (HAS_GLOBAL_MOCS_REGISTERS(rq->i915)) + if (HAS_GLOBAL_MOCS_REGISTERS(rq->engine->i915)) addr = global_mocs_offset(); else addr = mocs_offset(rq->engine); diff --git a/drivers/gpu/drm/i915/gt/selftest_rc6.c b/drivers/gpu/drm/i915/gt/selftest_rc6.c index 2dc460624bbc..3c8434846fa1 100644 --- a/drivers/gpu/drm/i915/gt/selftest_rc6.c +++ b/drivers/gpu/drm/i915/gt/selftest_rc6.c @@ -132,7 +132,7 @@ static const u32 *__live_rc6_ctx(struct intel_context *ce) } cmd = MI_STORE_REGISTER_MEM | MI_USE_GGTT; - if (INTEL_GEN(rq->i915) >= 8) + if (INTEL_GEN(rq->engine->i915) >= 8) cmd++; *cs++ = cmd; @@ -197,10 +197,10 @@ int live_rc6_ctx_wa(void *arg) int pass; for (pass = 0; pass < 2; pass++) { + struct i915_gpu_error *error = >->i915->gpu_error; struct intel_context *ce; unsigned int resets = - i915_reset_engine_count(>->i915->gpu_error, - engine); + i915_reset_engine_count(error, engine); const u32 *res; /* Use a sacrifical context */ @@ -230,8 +230,7 @@ int live_rc6_ctx_wa(void *arg) engine->name, READ_ONCE(*res)); if (resets != - i915_reset_engine_count(>->i915->gpu_error, - engine)) { + i915_reset_engine_count(error, engine)) { pr_err("%s: GPU reset required\n", engine->name); add_taint_for_CI(TAINT_WARN); diff --git a/drivers/gpu/drm/i915/gt/selftest_rps.c b/drivers/gpu/drm/i915/gt/selftest_rps.c index 5049c3dd08a6..bb753f0c12eb 100644 --- a/drivers/gpu/drm/i915/gt/selftest_rps.c +++ b/drivers/gpu/drm/i915/gt/selftest_rps.c @@ -12,6 +12,7 @@ #include "intel_gt_clock_utils.h" #include "intel_gt_pm.h" #include "intel_rc6.h" +#include "selftest_engine_heartbeat.h" #include "selftest_rps.h" #include "selftests/igt_flush_test.h" #include "selftests/igt_spinner.h" @@ -20,22 +21,6 @@ /* Try to isolate the impact of cstates from determing frequency response */ #define CPU_LATENCY 0 /* -1 to disable pm_qos, 0 to disable cstates */ -static void engine_heartbeat_disable(struct intel_engine_cs *engine) -{ - engine->props.heartbeat_interval_ms = 0; - - intel_engine_pm_get(engine); - intel_engine_park_heartbeat(engine); -} - -static void engine_heartbeat_enable(struct intel_engine_cs *engine) -{ - intel_engine_pm_put(engine); - - engine->props.heartbeat_interval_ms = - engine->defaults.heartbeat_interval_ms; -} - static void dummy_rps_work(struct work_struct *wrk) { } @@ -249,13 +234,13 @@ int live_rps_clock_interval(void *arg) if (!intel_engine_can_store_dword(engine)) continue; - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); rq = igt_spinner_create_request(&spin, engine->kernel_context, MI_NOOP); if (IS_ERR(rq)) { - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); err = PTR_ERR(rq); break; } @@ -266,7 +251,7 @@ int live_rps_clock_interval(void *arg) pr_err("%s: RPS spinner did not start\n", engine->name); igt_spinner_end(&spin); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); intel_gt_set_wedged(engine->gt); err = -EIO; break; @@ -322,7 +307,7 @@ int live_rps_clock_interval(void *arg) intel_uncore_forcewake_put(gt->uncore, FORCEWAKE_ALL); igt_spinner_end(&spin); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (err == 0) { u64 time = intel_gt_pm_interval_to_ns(gt, cycles); @@ -408,7 +393,7 @@ int live_rps_control(void *arg) if (!intel_engine_can_store_dword(engine)) continue; - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); rq = igt_spinner_create_request(&spin, engine->kernel_context, @@ -424,7 +409,7 @@ int live_rps_control(void *arg) pr_err("%s: RPS spinner did not start\n", engine->name); igt_spinner_end(&spin); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); intel_gt_set_wedged(engine->gt); err = -EIO; break; @@ -434,7 +419,7 @@ int live_rps_control(void *arg) pr_err("%s: could not set minimum frequency [%x], only %x!\n", engine->name, rps->min_freq, read_cagf(rps)); igt_spinner_end(&spin); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); show_pstate_limits(rps); err = -EINVAL; break; @@ -451,7 +436,7 @@ int live_rps_control(void *arg) pr_err("%s: could not restore minimum frequency [%x], only %x!\n", engine->name, rps->min_freq, read_cagf(rps)); igt_spinner_end(&spin); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); show_pstate_limits(rps); err = -EINVAL; break; @@ -466,7 +451,7 @@ int live_rps_control(void *arg) min_dt = ktime_sub(ktime_get(), min_dt); igt_spinner_end(&spin); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); pr_info("%s: range:[%x:%uMHz, %x:%uMHz] limit:[%x:%uMHz], %x:%x response %lluns:%lluns\n", engine->name, @@ -637,14 +622,14 @@ int live_rps_frequency_cs(void *arg) int freq; } min, max; - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); vma = create_spin_counter(engine, engine->kernel_context->vm, false, &cancel, &cntr); if (IS_ERR(vma)) { err = PTR_ERR(vma); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); break; } @@ -725,7 +710,7 @@ err_vma: i915_vma_unpin(vma); i915_vma_put(vma); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (igt_flush_test(gt->i915)) err = -EIO; if (err) @@ -779,14 +764,14 @@ int live_rps_frequency_srm(void *arg) int freq; } min, max; - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); vma = create_spin_counter(engine, engine->kernel_context->vm, true, &cancel, &cntr); if (IS_ERR(vma)) { err = PTR_ERR(vma); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); break; } @@ -866,7 +851,7 @@ err_vma: i915_vma_unpin(vma); i915_vma_put(vma); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (igt_flush_test(gt->i915)) err = -EIO; if (err) @@ -1061,11 +1046,11 @@ int live_rps_interrupt(void *arg) intel_gt_pm_wait_for_idle(engine->gt); GEM_BUG_ON(intel_rps_is_active(rps)); - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); err = __rps_up_interrupt(rps, engine, &spin); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (err) goto out; @@ -1074,13 +1059,13 @@ int live_rps_interrupt(void *arg) /* Keep the engine awake but idle and check for DOWN */ if (pm_events & GEN6_PM_RP_DOWN_THRESHOLD) { - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); intel_rc6_disable(>->rc6); err = __rps_down_interrupt(rps, engine); intel_rc6_enable(>->rc6); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (err) goto out; } @@ -1165,13 +1150,13 @@ int live_rps_power(void *arg) if (!intel_engine_can_store_dword(engine)) continue; - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); rq = igt_spinner_create_request(&spin, engine->kernel_context, MI_NOOP); if (IS_ERR(rq)) { - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); err = PTR_ERR(rq); break; } @@ -1182,7 +1167,7 @@ int live_rps_power(void *arg) pr_err("%s: RPS spinner did not start\n", engine->name); igt_spinner_end(&spin); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); intel_gt_set_wedged(engine->gt); err = -EIO; break; @@ -1195,7 +1180,7 @@ int live_rps_power(void *arg) min.power = measure_power_at(rps, &min.freq); igt_spinner_end(&spin); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); pr_info("%s: min:%llumW @ %uMHz, max:%llumW @ %uMHz\n", engine->name, @@ -1252,6 +1237,11 @@ int live_rps_dynamic(void *arg) if (igt_spinner_init(&spin, gt)) return -ENOMEM; + if (intel_rps_has_interrupts(rps)) + pr_info("RPS has interrupt support\n"); + if (intel_rps_uses_timer(rps)) + pr_info("RPS has timer support\n"); + for_each_engine(engine, gt, id) { struct i915_request *rq; struct { diff --git a/drivers/gpu/drm/i915/gt/selftest_timeline.c b/drivers/gpu/drm/i915/gt/selftest_timeline.c index ef1c35073dc0..fcdee951579b 100644 --- a/drivers/gpu/drm/i915/gt/selftest_timeline.c +++ b/drivers/gpu/drm/i915/gt/selftest_timeline.c @@ -12,6 +12,7 @@ #include "intel_gt.h" #include "intel_gt_requests.h" #include "intel_ring.h" +#include "selftest_engine_heartbeat.h" #include "../selftests/i915_random.h" #include "../i915_selftest.h" @@ -426,12 +427,12 @@ static int emit_ggtt_store_dw(struct i915_request *rq, u32 addr, u32 value) if (IS_ERR(cs)) return PTR_ERR(cs); - if (INTEL_GEN(rq->i915) >= 8) { + if (INTEL_GEN(rq->engine->i915) >= 8) { *cs++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT; *cs++ = addr; *cs++ = 0; *cs++ = value; - } else if (INTEL_GEN(rq->i915) >= 4) { + } else if (INTEL_GEN(rq->engine->i915) >= 4) { *cs++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT; *cs++ = 0; *cs++ = addr; @@ -751,22 +752,6 @@ out_free: return err; } -static void engine_heartbeat_disable(struct intel_engine_cs *engine) -{ - engine->props.heartbeat_interval_ms = 0; - - intel_engine_pm_get(engine); - intel_engine_park_heartbeat(engine); -} - -static void engine_heartbeat_enable(struct intel_engine_cs *engine) -{ - intel_engine_pm_put(engine); - - engine->props.heartbeat_interval_ms = - engine->defaults.heartbeat_interval_ms; -} - static int live_hwsp_rollover_kernel(void *arg) { struct intel_gt *gt = arg; @@ -785,7 +770,7 @@ static int live_hwsp_rollover_kernel(void *arg) struct i915_request *rq[3] = {}; int i; - engine_heartbeat_disable(engine); + st_engine_heartbeat_disable(engine); if (intel_gt_wait_for_idle(gt, HZ / 2)) { err = -EIO; goto out; @@ -836,7 +821,7 @@ static int live_hwsp_rollover_kernel(void *arg) out: for (i = 0; i < ARRAY_SIZE(rq); i++) i915_request_put(rq[i]); - engine_heartbeat_enable(engine); + st_engine_heartbeat_enable(engine); if (err) break; } diff --git a/drivers/gpu/drm/i915/gt/selftest_workarounds.c b/drivers/gpu/drm/i915/gt/selftest_workarounds.c index 32785463ec9e..febc9e6692ba 100644 --- a/drivers/gpu/drm/i915/gt/selftest_workarounds.c +++ b/drivers/gpu/drm/i915/gt/selftest_workarounds.c @@ -417,6 +417,20 @@ static bool wo_register(struct intel_engine_cs *engine, u32 reg) return false; } +static bool timestamp(const struct intel_engine_cs *engine, u32 reg) +{ + reg = (reg - engine->mmio_base) & ~RING_FORCE_TO_NONPRIV_ACCESS_MASK; + switch (reg) { + case 0x358: + case 0x35c: + case 0x3a8: + return true; + + default: + return false; + } +} + static bool ro_register(u32 reg) { if ((reg & RING_FORCE_TO_NONPRIV_ACCESS_MASK) == @@ -497,6 +511,9 @@ static int check_dirty_whitelist(struct intel_context *ce) if (wo_register(engine, reg)) continue; + if (timestamp(engine, reg)) + continue; /* timestamps are expected to autoincrement */ + ro_reg = ro_register(reg); /* Clear non priv flags */ diff --git a/drivers/gpu/drm/i915/gt/shaders/README b/drivers/gpu/drm/i915/gt/shaders/README new file mode 100644 index 000000000000..e7e96d7073c7 --- /dev/null +++ b/drivers/gpu/drm/i915/gt/shaders/README @@ -0,0 +1,46 @@ +ASM sources for auto generated shaders +====================================== + +The i915/gt/hsw_clear_kernel.c and i915/gt/ivb_clear_kernel.c files contain +pre-compiled batch chunks that will clear any residual render cache during +context switch. + +They are generated from their respective platform ASM files present on +i915/gt/shaders/clear_kernel directory. + +The generated .c files should never be modified directly. Instead, any modification +needs to be done on the on their respective ASM files and build instructions below +needes to be followed. + +Building +======== + +Environment +----------- + +IGT GPU tool scripts and the Mesa's i965 instruction assembler tool are used +on building. + +Please make sure your Mesa tool is compiled with "-Dtools=intel" and +"-Ddri-drivers=i965", and run this script from IGT source root directory" + +The instructions bellow assume: + * IGT gpu tools source code is located on your home directory (~) as ~/igt + * Mesa source code is located on your home directory (~) as ~/mesa + and built under the ~/mesa/build directory + * Linux kernel source code is under your home directory (~) as ~/linux + +Instructions +------------ + +~ $ cp ~/linux/drivers/gpu/drm/i915/gt/shaders/clear_kernel/ivb.asm \ + ~/igt/lib/i915/shaders/clear_kernel/ivb.asm +~ $ cd ~/igt +igt $ ./scripts/generate_clear_kernel.sh -g ivb \ + -m ~/mesa/build/src/intel/tools/i965_asm + +~ $ cp ~/linux/drivers/gpu/drm/i915/gt/shaders/clear_kernel/hsw.asm \ + ~/igt/lib/i915/shaders/clear_kernel/hsw.asm +~ $ cd ~/igt +igt $ ./scripts/generate_clear_kernel.sh -g hsw \ + -m ~/mesa/build/src/intel/tools/i965_asm
\ No newline at end of file diff --git a/drivers/gpu/drm/i915/gt/shaders/clear_kernel/hsw.asm b/drivers/gpu/drm/i915/gt/shaders/clear_kernel/hsw.asm new file mode 100644 index 000000000000..5fdf384bb621 --- /dev/null +++ b/drivers/gpu/drm/i915/gt/shaders/clear_kernel/hsw.asm @@ -0,0 +1,119 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright © 2020 Intel Corporation + */ + +/* + * Kernel for PAVP buffer clear. + * + * 1. Clear all 64 GRF registers assigned to the kernel with designated value; + * 2. Write 32x16 block of all "0" to render target buffer which indirectly clears + * 512 bytes of Render Cache. + */ + +/* Store designated "clear GRF" value */ +mov(1) f0.1<1>UW g1.2<0,1,0>UW { align1 1N }; + +/** + * Curbe Format + * + * DW 1.0 - Block Offset to write Render Cache + * DW 1.1 [15:0] - Clear Word + * DW 1.2 - Delay iterations + * DW 1.3 - Enable Instrumentation (only for debug) + * DW 1.4 - Rsvd (intended for context ID) + * DW 1.5 - [31:16]:SliceCount, [15:0]:SubSlicePerSliceCount + * DW 1.6 - Rsvd MBZ (intended for Enable Wait on Total Thread Count) + * DW 1.7 - Rsvd MBZ (inteded for Total Thread Count) + * + * Binding Table + * + * BTI 0: 2D Surface to help clear L3 (Render/Data Cache) + * BTI 1: Wait/Instrumentation Buffer + * Size : (SliceCount * SubSliceCount * 16 EUs/SubSlice) rows * (16 threads/EU) cols (Format R32_UINT) + * Expected to be initialized to 0 by driver/another kernel + * Layout: + * RowN: Histogram for EU-N: (SliceID*SubSlicePerSliceCount + SSID)*16 + EUID [assume max 16 EUs / SS] + * Col-k[DW-k]: Threads Executed on ThreadID-k for EU-N + */ +add(1) g1.2<1>UD g1.2<0,1,0>UD 0x00000001UD { align1 1N }; /* Loop count to delay kernel: Init to (g1.2 + 1) */ +cmp.z.f0.0(1) null<1>UD g1.3<0,1,0>UD 0x00000000UD { align1 1N }; +(+f0.0) jmpi(1) 352D { align1 WE_all 1N }; + +/** + * State Register has info on where this thread is running + * IVB: sr0.0 :: [15:13]: MBZ, 12: HSID (Half-Slice ID), [11:8]EUID, [2:0] ThreadSlotID + * HSW: sr0.0 :: 15: MBZ, [14:13]: SliceID, 12: HSID (Half-Slice ID), [11:8]EUID, [2:0] ThreadSlotID + */ +mov(8) g3<1>UD 0x00000000UD { align1 1Q }; +shr(1) g3<1>D sr0<0,1,0>D 12D { align1 1N }; +and(1) g3<1>D g3<0,1,0>D 1D { align1 1N }; /* g3 has HSID */ +shr(1) g3.1<1>D sr0<0,1,0>D 13D { align1 1N }; +and(1) g3.1<1>D g3.1<0,1,0>D 3D { align1 1N }; /* g3.1 has sliceID */ +mul(1) g3.5<1>D g3.1<0,1,0>D g1.10<0,1,0>UW { align1 1N }; +add(1) g3<1>D g3<0,1,0>D g3.5<0,1,0>D { align1 1N }; /* g3 = sliceID * SubSlicePerSliceCount + HSID */ +shr(1) g3.2<1>D sr0<0,1,0>D 8D { align1 1N }; +and(1) g3.2<1>D g3.2<0,1,0>D 15D { align1 1N }; /* g3.2 = EUID */ +mul(1) g3.4<1>D g3<0,1,0>D 16D { align1 1N }; +add(1) g3.2<1>D g3.2<0,1,0>D g3.4<0,1,0>D { align1 1N }; /* g3.2 now points to EU row number (Y-pixel = V address ) in instrumentation surf */ + +mov(8) g5<1>UD 0x00000000UD { align1 1Q }; +and(1) g3.3<1>D sr0<0,1,0>D 7D { align1 1N }; +mul(1) g3.3<1>D g3.3<0,1,0>D 4D { align1 1N }; + +mov(8) g4<1>UD g0<8,8,1>UD { align1 1Q }; /* Initialize message header with g0 */ +mov(1) g4<1>UD g3.3<0,1,0>UD { align1 1N }; /* Block offset */ +mov(1) g4.1<1>UD g3.2<0,1,0>UD { align1 1N }; /* Block offset */ +mov(1) g4.2<1>UD 0x00000003UD { align1 1N }; /* Block size (1 row x 4 bytes) */ +and(1) g4.3<1>UD g4.3<0,1,0>UW 0xffffffffUD { align1 1N }; + +/* Media block read to fetch current value at specified location in instrumentation buffer */ +sendc(8) g5<1>UD g4<8,8,1>F 0x02190001 + + render MsgDesc: media block read MsgCtrl = 0x0 Surface = 1 mlen 1 rlen 1 { align1 1Q }; +add(1) g5<1>D g5<0,1,0>D 1D { align1 1N }; + +/* Media block write for updated value at specified location in instrumentation buffer */ +sendc(8) g5<1>UD g4<8,8,1>F 0x040a8001 + render MsgDesc: media block write MsgCtrl = 0x0 Surface = 1 mlen 2 rlen 0 { align1 1Q }; + +/* Delay thread for specified parameter */ +add.nz.f0.0(1) g1.2<1>UD g1.2<0,1,0>UD -1D { align1 1N }; +(+f0.0) jmpi(1) -32D { align1 WE_all 1N }; + +/* Store designated "clear GRF" value */ +mov(1) f0.1<1>UW g1.2<0,1,0>UW { align1 1N }; + +/* Initialize looping parameters */ +mov(1) a0<1>D 0D { align1 1N }; /* Initialize a0.0:w=0 */ +mov(1) a0.4<1>W 127W { align1 1N }; /* Loop count. Each loop contains 16 GRF's */ + +/* Write 32x16 all "0" block */ +mov(8) g2<1>UD g0<8,8,1>UD { align1 1Q }; +mov(8) g127<1>UD g0<8,8,1>UD { align1 1Q }; +mov(2) g2<1>UD g1<2,2,1>UW { align1 1N }; +mov(1) g2.2<1>UD 0x000f000fUD { align1 1N }; /* Block size (16x16) */ +and(1) g2.3<1>UD g2.3<0,1,0>UW 0xffffffefUD { align1 1N }; +mov(16) g3<1>UD 0x00000000UD { align1 1H }; +mov(16) g4<1>UD 0x00000000UD { align1 1H }; +mov(16) g5<1>UD 0x00000000UD { align1 1H }; +mov(16) g6<1>UD 0x00000000UD { align1 1H }; +mov(16) g7<1>UD 0x00000000UD { align1 1H }; +mov(16) g8<1>UD 0x00000000UD { align1 1H }; +mov(16) g9<1>UD 0x00000000UD { align1 1H }; +mov(16) g10<1>UD 0x00000000UD { align1 1H }; +sendc(8) null<1>UD g2<8,8,1>F 0x120a8000 + render MsgDesc: media block write MsgCtrl = 0x0 Surface = 0 mlen 9 rlen 0 { align1 1Q }; +add(1) g2<1>UD g1<0,1,0>UW 0x0010UW { align1 1N }; +sendc(8) null<1>UD g2<8,8,1>F 0x120a8000 + render MsgDesc: media block write MsgCtrl = 0x0 Surface = 0 mlen 9 rlen 0 { align1 1Q }; + +/* Now, clear all GRF registers */ +add.nz.f0.0(1) a0.4<1>W a0.4<0,1,0>W -1W { align1 1N }; +mov(16) g[a0]<1>UW f0.1<0,1,0>UW { align1 1H }; +add(1) a0<1>D a0<0,1,0>D 32D { align1 1N }; +(+f0.0) jmpi(1) -64D { align1 WE_all 1N }; + +/* Terminante the thread */ +sendc(8) null<1>UD g127<8,8,1>F 0x82000010 + thread_spawner MsgDesc: mlen 1 rlen 0 { align1 1Q EOT }; diff --git a/drivers/gpu/drm/i915/gt/shaders/clear_kernel/ivb.asm b/drivers/gpu/drm/i915/gt/shaders/clear_kernel/ivb.asm new file mode 100644 index 000000000000..97c7ac9e3854 --- /dev/null +++ b/drivers/gpu/drm/i915/gt/shaders/clear_kernel/ivb.asm @@ -0,0 +1,117 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright © 2020 Intel Corporation + */ + +/* + * Kernel for PAVP buffer clear. + * + * 1. Clear all 64 GRF registers assigned to the kernel with designated value; + * 2. Write 32x16 block of all "0" to render target buffer which indirectly clears + * 512 bytes of Render Cache. + */ + +/* Store designated "clear GRF" value */ +mov(1) f0.1<1>UW g1.2<0,1,0>UW { align1 1N }; + +/** + * Curbe Format + * + * DW 1.0 - Block Offset to write Render Cache + * DW 1.1 [15:0] - Clear Word + * DW 1.2 - Delay iterations + * DW 1.3 - Enable Instrumentation (only for debug) + * DW 1.4 - Rsvd (intended for context ID) + * DW 1.5 - [31:16]:SliceCount, [15:0]:SubSlicePerSliceCount + * DW 1.6 - Rsvd MBZ (intended for Enable Wait on Total Thread Count) + * DW 1.7 - Rsvd MBZ (inteded for Total Thread Count) + * + * Binding Table + * + * BTI 0: 2D Surface to help clear L3 (Render/Data Cache) + * BTI 1: Wait/Instrumentation Buffer + * Size : (SliceCount * SubSliceCount * 16 EUs/SubSlice) rows * (16 threads/EU) cols (Format R32_UINT) + * Expected to be initialized to 0 by driver/another kernel + * Layout : + * RowN: Histogram for EU-N: (SliceID*SubSlicePerSliceCount + SSID)*16 + EUID [assume max 16 EUs / SS] + * Col-k[DW-k]: Threads Executed on ThreadID-k for EU-N + */ +add(1) g1.2<1>UD g1.2<0,1,0>UD 0x00000001UD { align1 1N }; /* Loop count to delay kernel: Init to (g1.2 + 1) */ +cmp.z.f0.0(1) null<1>UD g1.3<0,1,0>UD 0x00000000UD { align1 1N }; +(+f0.0) jmpi(1) 44D { align1 WE_all 1N }; + +/** + * State Register has info on where this thread is running + * IVB: sr0.0 :: [15:13]: MBZ, 12: HSID (Half-Slice ID), [11:8]EUID, [2:0] ThreadSlotID + * HSW: sr0.0 :: 15: MBZ, [14:13]: SliceID, 12: HSID (Half-Slice ID), [11:8]EUID, [2:0] ThreadSlotID + */ +mov(8) g3<1>UD 0x00000000UD { align1 1Q }; +shr(1) g3<1>D sr0<0,1,0>D 12D { align1 1N }; +and(1) g3<1>D g3<0,1,0>D 1D { align1 1N }; /* g3 has HSID */ +shr(1) g3.1<1>D sr0<0,1,0>D 13D { align1 1N }; +and(1) g3.1<1>D g3.1<0,1,0>D 3D { align1 1N }; /* g3.1 has sliceID */ +mul(1) g3.5<1>D g3.1<0,1,0>D g1.10<0,1,0>UW { align1 1N }; +add(1) g3<1>D g3<0,1,0>D g3.5<0,1,0>D { align1 1N }; /* g3 = sliceID * SubSlicePerSliceCount + HSID */ +shr(1) g3.2<1>D sr0<0,1,0>D 8D { align1 1N }; +and(1) g3.2<1>D g3.2<0,1,0>D 15D { align1 1N }; /* g3.2 = EUID */ +mul(1) g3.4<1>D g3<0,1,0>D 16D { align1 1N }; +add(1) g3.2<1>D g3.2<0,1,0>D g3.4<0,1,0>D { align1 1N }; /* g3.2 now points to EU row number (Y-pixel = V address ) in instrumentation surf */ + +mov(8) g5<1>UD 0x00000000UD { align1 1Q }; +and(1) g3.3<1>D sr0<0,1,0>D 7D { align1 1N }; +mul(1) g3.3<1>D g3.3<0,1,0>D 4D { align1 1N }; + +mov(8) g4<1>UD g0<8,8,1>UD { align1 1Q }; /* Initialize message header with g0 */ +mov(1) g4<1>UD g3.3<0,1,0>UD { align1 1N }; /* Block offset */ +mov(1) g4.1<1>UD g3.2<0,1,0>UD { align1 1N }; /* Block offset */ +mov(1) g4.2<1>UD 0x00000003UD { align1 1N }; /* Block size (1 row x 4 bytes) */ +and(1) g4.3<1>UD g4.3<0,1,0>UW 0xffffffffUD { align1 1N }; + +/* Media block read to fetch current value at specified location in instrumentation buffer */ +sendc(8) g5<1>UD g4<8,8,1>F 0x02190001 + render MsgDesc: media block read MsgCtrl = 0x0 Surface = 1 mlen 1 rlen 1 { align1 1Q }; +add(1) g5<1>D g5<0,1,0>D 1D { align1 1N }; + +/* Media block write for updated value at specified location in instrumentation buffer */ +sendc(8) g5<1>UD g4<8,8,1>F 0x040a8001 + render MsgDesc: media block write MsgCtrl = 0x0 Surface = 1 mlen 2 rlen 0 { align1 1Q }; +/* Delay thread for specified parameter */ +add.nz.f0.0(1) g1.2<1>UD g1.2<0,1,0>UD -1D { align1 1N }; +(+f0.0) jmpi(1) -4D { align1 WE_all 1N }; + +/* Store designated "clear GRF" value */ +mov(1) f0.1<1>UW g1.2<0,1,0>UW { align1 1N }; + +/* Initialize looping parameters */ +mov(1) a0<1>D 0D { align1 1N }; /* Initialize a0.0:w=0 */ +mov(1) a0.4<1>W 127W { align1 1N }; /* Loop count. Each loop contains 16 GRF's */ + +/* Write 32x16 all "0" block */ +mov(8) g2<1>UD g0<8,8,1>UD { align1 1Q }; +mov(8) g127<1>UD g0<8,8,1>UD { align1 1Q }; +mov(2) g2<1>UD g1<2,2,1>UW { align1 1N }; +mov(1) g2.2<1>UD 0x000f000fUD { align1 1N }; /* Block size (16x16) */ +and(1) g2.3<1>UD g2.3<0,1,0>UW 0xffffffefUD { align1 1N }; +mov(16) g3<1>UD 0x00000000UD { align1 1H }; +mov(16) g4<1>UD 0x00000000UD { align1 1H }; +mov(16) g5<1>UD 0x00000000UD { align1 1H }; +mov(16) g6<1>UD 0x00000000UD { align1 1H }; +mov(16) g7<1>UD 0x00000000UD { align1 1H }; +mov(16) g8<1>UD 0x00000000UD { align1 1H }; +mov(16) g9<1>UD 0x00000000UD { align1 1H }; +mov(16) g10<1>UD 0x00000000UD { align1 1H }; +sendc(8) null<1>UD g2<8,8,1>F 0x120a8000 + render MsgDesc: media block write MsgCtrl = 0x0 Surface = 0 mlen 9 rlen 0 { align1 1Q }; +add(1) g2<1>UD g1<0,1,0>UW 0x0010UW { align1 1N }; +sendc(8) null<1>UD g2<8,8,1>F 0x120a8000 + render MsgDesc: media block write MsgCtrl = 0x0 Surface = 0 mlen 9 rlen 0 { align1 1Q }; + +/* Now, clear all GRF registers */ +add.nz.f0.0(1) a0.4<1>W a0.4<0,1,0>W -1W { align1 1N }; +mov(16) g[a0]<1>UW f0.1<0,1,0>UW { align1 1H }; +add(1) a0<1>D a0<0,1,0>D 32D { align1 1N }; +(+f0.0) jmpi(1) -8D { align1 WE_all 1N }; + +/* Terminante the thread */ +sendc(8) null<1>UD g127<8,8,1>F 0x82000010 + thread_spawner MsgDesc: mlen 1 rlen 0 { align1 1Q EOT }; diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_log.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_log.c index fb10f3597ea5..9bbe8a795cb8 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_log.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_log.c @@ -424,25 +424,28 @@ static void guc_log_capture_logs(struct intel_guc_log *log) static u32 __get_default_log_level(struct intel_guc_log *log) { + struct intel_guc *guc = log_to_guc(log); + struct drm_i915_private *i915 = guc_to_gt(guc)->i915; + /* A negative value means "use platform/config default" */ - if (i915_modparams.guc_log_level < 0) { + if (i915->params.guc_log_level < 0) { return (IS_ENABLED(CONFIG_DRM_I915_DEBUG) || IS_ENABLED(CONFIG_DRM_I915_DEBUG_GEM)) ? GUC_LOG_LEVEL_MAX : GUC_LOG_LEVEL_NON_VERBOSE; } - if (i915_modparams.guc_log_level > GUC_LOG_LEVEL_MAX) { + if (i915->params.guc_log_level > GUC_LOG_LEVEL_MAX) { DRM_WARN("Incompatible option detected: %s=%d, %s!\n", - "guc_log_level", i915_modparams.guc_log_level, + "guc_log_level", i915->params.guc_log_level, "verbosity too high"); return (IS_ENABLED(CONFIG_DRM_I915_DEBUG) || IS_ENABLED(CONFIG_DRM_I915_DEBUG_GEM)) ? GUC_LOG_LEVEL_MAX : GUC_LOG_LEVEL_DISABLED; } - GEM_BUG_ON(i915_modparams.guc_log_level < GUC_LOG_LEVEL_DISABLED); - GEM_BUG_ON(i915_modparams.guc_log_level > GUC_LOG_LEVEL_MAX); - return i915_modparams.guc_log_level; + GEM_BUG_ON(i915->params.guc_log_level < GUC_LOG_LEVEL_DISABLED); + GEM_BUG_ON(i915->params.guc_log_level > GUC_LOG_LEVEL_MAX); + return i915->params.guc_log_level; } int intel_guc_log_create(struct intel_guc_log *log) diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c index 94eb63f309ce..fdfeb4b9b0f5 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c @@ -660,10 +660,12 @@ void intel_guc_submission_disable(struct intel_guc *guc) static bool __guc_submission_selected(struct intel_guc *guc) { + struct drm_i915_private *i915 = guc_to_gt(guc)->i915; + if (!intel_guc_submission_is_supported(guc)) return false; - return i915_modparams.enable_guc & ENABLE_GUC_SUBMISSION; + return i915->params.enable_guc & ENABLE_GUC_SUBMISSION; } void intel_guc_submission_init_early(struct intel_guc *guc) diff --git a/drivers/gpu/drm/i915/gt/uc/intel_uc.c b/drivers/gpu/drm/i915/gt/uc/intel_uc.c index f518fe05c6f9..1c2d6358826c 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_uc.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_uc.c @@ -47,15 +47,15 @@ static void __confirm_options(struct intel_uc *uc) drm_dbg(&i915->drm, "enable_guc=%d (guc:%s submission:%s huc:%s)\n", - i915_modparams.enable_guc, + i915->params.enable_guc, yesno(intel_uc_wants_guc(uc)), yesno(intel_uc_wants_guc_submission(uc)), yesno(intel_uc_wants_huc(uc))); - if (i915_modparams.enable_guc == -1) + if (i915->params.enable_guc == -1) return; - if (i915_modparams.enable_guc == 0) { + if (i915->params.enable_guc == 0) { GEM_BUG_ON(intel_uc_wants_guc(uc)); GEM_BUG_ON(intel_uc_wants_guc_submission(uc)); GEM_BUG_ON(intel_uc_wants_huc(uc)); @@ -65,25 +65,25 @@ static void __confirm_options(struct intel_uc *uc) if (!intel_uc_supports_guc(uc)) drm_info(&i915->drm, "Incompatible option enable_guc=%d - %s\n", - i915_modparams.enable_guc, "GuC is not supported!"); + i915->params.enable_guc, "GuC is not supported!"); - if (i915_modparams.enable_guc & ENABLE_GUC_LOAD_HUC && + if (i915->params.enable_guc & ENABLE_GUC_LOAD_HUC && !intel_uc_supports_huc(uc)) drm_info(&i915->drm, "Incompatible option enable_guc=%d - %s\n", - i915_modparams.enable_guc, "HuC is not supported!"); + i915->params.enable_guc, "HuC is not supported!"); - if (i915_modparams.enable_guc & ENABLE_GUC_SUBMISSION && + if (i915->params.enable_guc & ENABLE_GUC_SUBMISSION && !intel_uc_supports_guc_submission(uc)) drm_info(&i915->drm, "Incompatible option enable_guc=%d - %s\n", - i915_modparams.enable_guc, "GuC submission is N/A"); + i915->params.enable_guc, "GuC submission is N/A"); - if (i915_modparams.enable_guc & ~(ENABLE_GUC_SUBMISSION | + if (i915->params.enable_guc & ~(ENABLE_GUC_SUBMISSION | ENABLE_GUC_LOAD_HUC)) drm_info(&i915->drm, "Incompatible option enable_guc=%d - %s\n", - i915_modparams.enable_guc, "undocumented flag"); + i915->params.enable_guc, "undocumented flag"); } void intel_uc_init_early(struct intel_uc *uc) diff --git a/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c b/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c index e1caae93996d..59b27aba15c6 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c @@ -47,12 +47,15 @@ void intel_uc_fw_change_status(struct intel_uc_fw *uc_fw, * TGL 35.2 is interface-compatible with 33.0 for previous Gens. The deltas * between 33.0 and 35.2 are only related to new additions to support new Gen12 * features. + * + * Note that RKL uses the same firmware as TGL. */ #define INTEL_UC_FIRMWARE_DEFS(fw_def, guc_def, huc_def) \ + fw_def(ROCKETLAKE, 0, guc_def(tgl, 35, 2, 0), huc_def(tgl, 7, 0, 12)) \ fw_def(TIGERLAKE, 0, guc_def(tgl, 35, 2, 0), huc_def(tgl, 7, 0, 12)) \ fw_def(ELKHARTLAKE, 0, guc_def(ehl, 33, 0, 4), huc_def(ehl, 9, 0, 0)) \ fw_def(ICELAKE, 0, guc_def(icl, 33, 0, 0), huc_def(icl, 9, 0, 0)) \ - fw_def(COFFEELAKE, 5, guc_def(cml, 33, 0, 0), huc_def(cml, 4, 0, 0)) \ + fw_def(COMETLAKE, 5, guc_def(cml, 33, 0, 0), huc_def(cml, 4, 0, 0)) \ fw_def(COFFEELAKE, 0, guc_def(kbl, 33, 0, 0), huc_def(kbl, 4, 0, 0)) \ fw_def(GEMINILAKE, 0, guc_def(glk, 33, 0, 0), huc_def(glk, 4, 0, 0)) \ fw_def(KABYLAKE, 0, guc_def(kbl, 33, 0, 0), huc_def(kbl, 4, 0, 0)) \ @@ -112,11 +115,13 @@ struct __packed uc_fw_platform_requirement { }, static void -__uc_fw_auto_select(struct intel_uc_fw *uc_fw, enum intel_platform p, u8 rev) +__uc_fw_auto_select(struct drm_i915_private *i915, struct intel_uc_fw *uc_fw) { static const struct uc_fw_platform_requirement fw_blobs[] = { INTEL_UC_FIRMWARE_DEFS(MAKE_FW_LIST, GUC_FW_BLOB, HUC_FW_BLOB) }; + enum intel_platform p = INTEL_INFO(i915)->platform; + u8 rev = INTEL_REVID(i915); int i; for (i = 0; i < ARRAY_SIZE(fw_blobs) && p <= fw_blobs[i].p; i++) { @@ -151,35 +156,35 @@ __uc_fw_auto_select(struct intel_uc_fw *uc_fw, enum intel_platform p, u8 rev) } /* We don't want to enable GuC/HuC on pre-Gen11 by default */ - if (i915_modparams.enable_guc == -1 && p < INTEL_ICELAKE) + if (i915->params.enable_guc == -1 && p < INTEL_ICELAKE) uc_fw->path = NULL; } -static const char *__override_guc_firmware_path(void) +static const char *__override_guc_firmware_path(struct drm_i915_private *i915) { - if (i915_modparams.enable_guc & (ENABLE_GUC_SUBMISSION | - ENABLE_GUC_LOAD_HUC)) - return i915_modparams.guc_firmware_path; + if (i915->params.enable_guc & (ENABLE_GUC_SUBMISSION | + ENABLE_GUC_LOAD_HUC)) + return i915->params.guc_firmware_path; return ""; } -static const char *__override_huc_firmware_path(void) +static const char *__override_huc_firmware_path(struct drm_i915_private *i915) { - if (i915_modparams.enable_guc & ENABLE_GUC_LOAD_HUC) - return i915_modparams.huc_firmware_path; + if (i915->params.enable_guc & ENABLE_GUC_LOAD_HUC) + return i915->params.huc_firmware_path; return ""; } -static void __uc_fw_user_override(struct intel_uc_fw *uc_fw) +static void __uc_fw_user_override(struct drm_i915_private *i915, struct intel_uc_fw *uc_fw) { const char *path = NULL; switch (uc_fw->type) { case INTEL_UC_FW_TYPE_GUC: - path = __override_guc_firmware_path(); + path = __override_guc_firmware_path(i915); break; case INTEL_UC_FW_TYPE_HUC: - path = __override_huc_firmware_path(); + path = __override_huc_firmware_path(i915); break; } @@ -213,10 +218,8 @@ void intel_uc_fw_init_early(struct intel_uc_fw *uc_fw, uc_fw->type = type; if (HAS_GT_UC(i915)) { - __uc_fw_auto_select(uc_fw, - INTEL_INFO(i915)->platform, - INTEL_REVID(i915)); - __uc_fw_user_override(uc_fw); + __uc_fw_auto_select(i915, uc_fw); + __uc_fw_user_override(i915, uc_fw); } intel_uc_fw_change_status(uc_fw, uc_fw->path ? *uc_fw->path ? diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c index 8b87f130f7f1..f1940939260a 100644 --- a/drivers/gpu/drm/i915/gvt/cmd_parser.c +++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c @@ -1904,19 +1904,10 @@ static int perform_bb_shadow(struct parser_exec_state *s) goto err_free_bb; } - ret = i915_gem_object_prepare_write(bb->obj, &bb->clflush); - if (ret) - goto err_free_obj; - bb->va = i915_gem_object_pin_map(bb->obj, I915_MAP_WB); if (IS_ERR(bb->va)) { ret = PTR_ERR(bb->va); - goto err_finish_shmem_access; - } - - if (bb->clflush & CLFLUSH_BEFORE) { - drm_clflush_virt_range(bb->va, bb->obj->base.size); - bb->clflush &= ~CLFLUSH_BEFORE; + goto err_free_obj; } ret = copy_gma_to_hva(s->vgpu, mm, @@ -1935,7 +1926,6 @@ static int perform_bb_shadow(struct parser_exec_state *s) INIT_LIST_HEAD(&bb->list); list_add(&bb->list, &s->workload->shadow_bb); - bb->accessing = true; bb->bb_start_cmd_va = s->ip_va; if ((s->buf_type == BATCH_BUFFER_INSTRUCTION) && (!s->is_ctx_wa)) @@ -1956,8 +1946,6 @@ static int perform_bb_shadow(struct parser_exec_state *s) return 0; err_unmap: i915_gem_object_unpin_map(bb->obj); -err_finish_shmem_access: - i915_gem_object_finish_access(bb->obj); err_free_obj: i915_gem_object_put(bb->obj); err_free_bb: diff --git a/drivers/gpu/drm/i915/gvt/display.c b/drivers/gpu/drm/i915/gvt/display.c index a1696e9ce4b6..7ba16ddfe75f 100644 --- a/drivers/gpu/drm/i915/gvt/display.c +++ b/drivers/gpu/drm/i915/gvt/display.c @@ -199,8 +199,10 @@ static void emulate_monitor_status_change(struct intel_vgpu *vgpu) SDE_PORTC_HOTPLUG_CPT | SDE_PORTD_HOTPLUG_CPT); - if (IS_SKYLAKE(dev_priv) || IS_KABYLAKE(dev_priv) || - IS_COFFEELAKE(dev_priv)) { + if (IS_SKYLAKE(dev_priv) || + IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) { vgpu_vreg_t(vgpu, SDEISR) &= ~(SDE_PORTA_HOTPLUG_SPT | SDE_PORTE_HOTPLUG_SPT); vgpu_vreg_t(vgpu, SKL_FUSE_STATUS) |= @@ -314,8 +316,10 @@ static void emulate_monitor_status_change(struct intel_vgpu *vgpu) vgpu_vreg_t(vgpu, SFUSE_STRAP) |= SFUSE_STRAP_DDID_DETECTED; } - if ((IS_SKYLAKE(dev_priv) || IS_KABYLAKE(dev_priv) || - IS_COFFEELAKE(dev_priv)) && + if ((IS_SKYLAKE(dev_priv) || + IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) && intel_vgpu_has_monitor_on_port(vgpu, PORT_E)) { vgpu_vreg_t(vgpu, SDEISR) |= SDE_PORTE_HOTPLUG_SPT; } @@ -498,8 +502,10 @@ void intel_vgpu_emulate_hotplug(struct intel_vgpu *vgpu, bool connected) struct drm_i915_private *i915 = vgpu->gvt->gt->i915; /* TODO: add more platforms support */ - if (IS_SKYLAKE(i915) || IS_KABYLAKE(i915) || - IS_COFFEELAKE(i915)) { + if (IS_SKYLAKE(i915) || + IS_KABYLAKE(i915) || + IS_COFFEELAKE(i915) || + IS_COMETLAKE(i915)) { if (connected) { vgpu_vreg_t(vgpu, SFUSE_STRAP) |= SFUSE_STRAP_DDID_DETECTED; @@ -527,8 +533,10 @@ void intel_vgpu_clean_display(struct intel_vgpu *vgpu) { struct drm_i915_private *dev_priv = vgpu->gvt->gt->i915; - if (IS_SKYLAKE(dev_priv) || IS_KABYLAKE(dev_priv) || - IS_COFFEELAKE(dev_priv)) + if (IS_SKYLAKE(dev_priv) || + IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) clean_virtual_dp_monitor(vgpu, PORT_D); else clean_virtual_dp_monitor(vgpu, PORT_B); @@ -551,8 +559,10 @@ int intel_vgpu_init_display(struct intel_vgpu *vgpu, u64 resolution) intel_vgpu_init_i2c_edid(vgpu); - if (IS_SKYLAKE(dev_priv) || IS_KABYLAKE(dev_priv) || - IS_COFFEELAKE(dev_priv)) + if (IS_SKYLAKE(dev_priv) || + IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) return setup_virtual_dp_monitor(vgpu, PORT_D, GVT_DP_D, resolution); else diff --git a/drivers/gpu/drm/i915/gvt/dmabuf.c b/drivers/gpu/drm/i915/gvt/dmabuf.c index 37fc460414a8..c3eb3838fe88 100644 --- a/drivers/gpu/drm/i915/gvt/dmabuf.c +++ b/drivers/gpu/drm/i915/gvt/dmabuf.c @@ -198,6 +198,7 @@ static void vgpu_gem_release(struct drm_i915_gem_object *gem_obj) } static const struct drm_i915_gem_object_ops intel_vgpu_gem_ops = { + .name = "i915_gem_object_vgpu", .flags = I915_GEM_OBJECT_IS_PROXY, .get_pages = vgpu_gem_get_pages, .put_pages = vgpu_gem_put_pages, diff --git a/drivers/gpu/drm/i915/gvt/edid.c b/drivers/gpu/drm/i915/gvt/edid.c index 190651df5db1..22247805c345 100644 --- a/drivers/gpu/drm/i915/gvt/edid.c +++ b/drivers/gpu/drm/i915/gvt/edid.c @@ -149,7 +149,7 @@ static int gmbus0_mmio_write(struct intel_vgpu *vgpu, if (IS_BROXTON(i915)) port = bxt_get_port_from_gmbus0(pin_select); - else if (IS_COFFEELAKE(i915)) + else if (IS_COFFEELAKE(i915) || IS_COMETLAKE(i915)) port = cnp_get_port_from_gmbus0(pin_select); else port = get_port_from_gmbus0(pin_select); diff --git a/drivers/gpu/drm/i915/gvt/handlers.c b/drivers/gpu/drm/i915/gvt/handlers.c index 3e88e3b5c43a..26cae4846c82 100644 --- a/drivers/gpu/drm/i915/gvt/handlers.c +++ b/drivers/gpu/drm/i915/gvt/handlers.c @@ -59,7 +59,7 @@ unsigned long intel_gvt_get_device_type(struct intel_gvt *gvt) return D_KBL; else if (IS_BROXTON(i915)) return D_BXT; - else if (IS_COFFEELAKE(i915)) + else if (IS_COFFEELAKE(i915) || IS_COMETLAKE(i915)) return D_CFL; return 0; @@ -1435,7 +1435,8 @@ static int mailbox_write(struct intel_vgpu *vgpu, unsigned int offset, case GEN9_PCODE_READ_MEM_LATENCY: if (IS_SKYLAKE(vgpu->gvt->gt->i915) || IS_KABYLAKE(vgpu->gvt->gt->i915) || - IS_COFFEELAKE(vgpu->gvt->gt->i915)) { + IS_COFFEELAKE(vgpu->gvt->gt->i915) || + IS_COMETLAKE(vgpu->gvt->gt->i915)) { /** * "Read memory latency" command on gen9. * Below memory latency values are read @@ -1460,7 +1461,8 @@ static int mailbox_write(struct intel_vgpu *vgpu, unsigned int offset, case SKL_PCODE_CDCLK_CONTROL: if (IS_SKYLAKE(vgpu->gvt->gt->i915) || IS_KABYLAKE(vgpu->gvt->gt->i915) || - IS_COFFEELAKE(vgpu->gvt->gt->i915)) + IS_COFFEELAKE(vgpu->gvt->gt->i915) || + IS_COMETLAKE(vgpu->gvt->gt->i915)) *data0 = SKL_CDCLK_READY_FOR_CHANGE; break; case GEN6_PCODE_READ_RC6VIDS: @@ -1722,7 +1724,8 @@ static int ring_mode_mmio_write(struct intel_vgpu *vgpu, unsigned int offset, int ret; (*(u32 *)p_data) &= ~_MASKED_BIT_ENABLE(1); - if (IS_COFFEELAKE(vgpu->gvt->gt->i915)) + if (IS_COFFEELAKE(vgpu->gvt->gt->i915) || + IS_COMETLAKE(vgpu->gvt->gt->i915)) (*(u32 *)p_data) &= ~_MASKED_BIT_ENABLE(2); write_vreg(vgpu, offset, p_data, bytes); @@ -1731,7 +1734,8 @@ static int ring_mode_mmio_write(struct intel_vgpu *vgpu, unsigned int offset, return 0; } - if (IS_COFFEELAKE(vgpu->gvt->gt->i915) && + if ((IS_COFFEELAKE(vgpu->gvt->gt->i915) || + IS_COMETLAKE(vgpu->gvt->gt->i915)) && data & _MASKED_BIT_ENABLE(2)) { enter_failsafe_mode(vgpu, GVT_FAILSAFE_UNSUPPORTED_GUEST); return 0; @@ -3393,7 +3397,8 @@ int intel_gvt_setup_mmio_info(struct intel_gvt *gvt) goto err; } else if (IS_SKYLAKE(i915) || IS_KABYLAKE(i915) || - IS_COFFEELAKE(i915)) { + IS_COFFEELAKE(i915) || + IS_COMETLAKE(i915)) { ret = init_bdw_mmio_info(gvt); if (ret) goto err; diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c index 0fb1df71c637..3c3b9842bbbd 100644 --- a/drivers/gpu/drm/i915/gvt/scheduler.c +++ b/drivers/gpu/drm/i915/gvt/scheduler.c @@ -348,7 +348,7 @@ static int copy_workload_to_ring_buffer(struct intel_vgpu_workload *workload) u32 *cs; int err; - if (IS_GEN(req->i915, 9) && is_inhibit_context(req->context)) + if (IS_GEN(req->engine->i915, 9) && is_inhibit_context(req->context)) intel_vgpu_restore_inhibit_context(vgpu, req); /* @@ -509,26 +509,18 @@ static int prepare_shadow_batch_buffer(struct intel_vgpu_workload *workload) bb->bb_start_cmd_va = workload->shadow_ring_buffer_va + bb->bb_offset; - if (bb->ppgtt) { - /* for non-priv bb, scan&shadow is only for - * debugging purpose, so the content of shadow bb - * is the same as original bb. Therefore, - * here, rather than switch to shadow bb's gma - * address, we directly use original batch buffer's - * gma address, and send original bb to hardware - * directly - */ - if (bb->clflush & CLFLUSH_AFTER) { - drm_clflush_virt_range(bb->va, - bb->obj->base.size); - bb->clflush &= ~CLFLUSH_AFTER; - } - i915_gem_object_finish_access(bb->obj); - bb->accessing = false; - - } else { + /* + * For non-priv bb, scan&shadow is only for + * debugging purpose, so the content of shadow bb + * is the same as original bb. Therefore, + * here, rather than switch to shadow bb's gma + * address, we directly use original batch buffer's + * gma address, and send original bb to hardware + * directly + */ + if (!bb->ppgtt) { bb->vma = i915_gem_object_ggtt_pin(bb->obj, - NULL, 0, 0, 0); + NULL, 0, 0, 0); if (IS_ERR(bb->vma)) { ret = PTR_ERR(bb->vma); goto err; @@ -539,27 +531,15 @@ static int prepare_shadow_batch_buffer(struct intel_vgpu_workload *workload) if (gmadr_bytes == 8) bb->bb_start_cmd_va[2] = 0; - /* No one is going to touch shadow bb from now on. */ - if (bb->clflush & CLFLUSH_AFTER) { - drm_clflush_virt_range(bb->va, - bb->obj->base.size); - bb->clflush &= ~CLFLUSH_AFTER; - } - - ret = i915_gem_object_set_to_gtt_domain(bb->obj, - false); - if (ret) - goto err; - ret = i915_vma_move_to_active(bb->vma, workload->req, 0); if (ret) goto err; - - i915_gem_object_finish_access(bb->obj); - bb->accessing = false; } + + /* No one is going to touch shadow bb from now on. */ + i915_gem_object_flush_map(bb->obj); } return 0; err: @@ -630,9 +610,6 @@ static void release_shadow_batch_buffer(struct intel_vgpu_workload *workload) list_for_each_entry_safe(bb, pos, &workload->shadow_bb, list) { if (bb->obj) { - if (bb->accessing) - i915_gem_object_finish_access(bb->obj); - if (bb->va && !IS_ERR(bb->va)) i915_gem_object_unpin_map(bb->obj); @@ -939,7 +916,7 @@ static void update_guest_context(struct intel_vgpu_workload *workload) context_page_num = rq->engine->context_size; context_page_num = context_page_num >> PAGE_SHIFT; - if (IS_BROADWELL(rq->i915) && rq->engine->id == RCS0) + if (IS_BROADWELL(rq->engine->i915) && rq->engine->id == RCS0) context_page_num = 19; context_base = (void *) ctx->lrc_reg_state - diff --git a/drivers/gpu/drm/i915/gvt/scheduler.h b/drivers/gpu/drm/i915/gvt/scheduler.h index 15d317f2a4a4..64e7a0b791c3 100644 --- a/drivers/gpu/drm/i915/gvt/scheduler.h +++ b/drivers/gpu/drm/i915/gvt/scheduler.h @@ -124,8 +124,6 @@ struct intel_vgpu_shadow_bb { struct i915_vma *vma; void *va; u32 *bb_start_cmd_va; - unsigned int clflush; - bool accessing; unsigned long bb_offset; bool ppgtt; }; diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c index bca036ac6621..8594a8ef08ce 100644 --- a/drivers/gpu/drm/i915/i915_debugfs.c +++ b/drivers/gpu/drm/i915/i915_debugfs.c @@ -64,7 +64,7 @@ static int i915_capabilities(struct seq_file *m, void *data) intel_driver_caps_print(&i915->caps, &p); kernel_param_lock(THIS_MODULE); - i915_params_dump(&i915_modparams, &p); + i915_params_dump(&i915->params, &p); kernel_param_unlock(THIS_MODULE); return 0; diff --git a/drivers/gpu/drm/i915/i915_debugfs_params.c b/drivers/gpu/drm/i915/i915_debugfs_params.c index 62b2c5f0495d..4e2b077692cb 100644 --- a/drivers/gpu/drm/i915/i915_debugfs_params.c +++ b/drivers/gpu/drm/i915/i915_debugfs_params.c @@ -138,9 +138,6 @@ static ssize_t i915_param_charp_write(struct file *file, char **s = m->private; char *new, *old; - /* FIXME: remove locking after params aren't the module params */ - kernel_param_lock(THIS_MODULE); - old = *s; new = strndup_user(ubuf, PAGE_SIZE); if (IS_ERR(new)) { @@ -152,8 +149,6 @@ static ssize_t i915_param_charp_write(struct file *file, kfree(old); out: - kernel_param_unlock(THIS_MODULE); - return len; } @@ -229,7 +224,7 @@ _i915_param_create_file(struct dentry *parent, const char *name, struct dentry *i915_debugfs_params(struct drm_i915_private *i915) { struct drm_minor *minor = i915->drm.primary; - struct i915_params *params = &i915_modparams; + struct i915_params *params = &i915->params; struct dentry *dir; dir = debugfs_create_dir("i915_params", minor->debugfs_root); diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index 34ee12f3f02d..67102dc26fce 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -501,6 +501,8 @@ static void i915_driver_late_release(struct drm_i915_private *dev_priv) cpu_latency_qos_remove_request(&dev_priv->sb_qos); mutex_destroy(&dev_priv->sb_lock); + + i915_params_free(&dev_priv->params); } /** @@ -915,6 +917,9 @@ i915_driver_create(struct pci_dev *pdev, const struct pci_device_id *ent) i915->drm.pdev = pdev; pci_set_drvdata(pdev, i915); + /* Device parameters start as a copy of module parameters. */ + i915_params_copy(&i915->params, &i915_modparams); + /* Setup the write-once "constant" device info */ device_info = mkwrite_device_info(i915); memcpy(device_info, match_info, sizeof(*device_info)); @@ -948,7 +953,7 @@ int i915_driver_probe(struct pci_dev *pdev, const struct pci_device_id *ent) return PTR_ERR(i915); /* Disable nuclear pageflip by default on pre-ILK */ - if (!i915_modparams.nuclear_pageflip && match_info->gen < 5) + if (!i915->params.nuclear_pageflip && match_info->gen < 5) i915->drm.driver_features &= ~DRIVER_ATOMIC; /* @@ -958,7 +963,7 @@ int i915_driver_probe(struct pci_dev *pdev, const struct pci_device_id *ent) #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST) if (IS_ENABLED(CONFIG_DRM_I915_UNSTABLE_FAKE_LMEM)) { if (INTEL_GEN(i915) >= 9 && i915_selftest.live < 0 && - i915_modparams.fake_lmem_start) { + i915->params.fake_lmem_start) { mkwrite_device_info(i915)->memory_regions = REGION_SMEM | REGION_LMEM | REGION_STOLEN; mkwrite_device_info(i915)->is_dgfx = true; diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index adb9bf34cf97..9aad3ec979bd 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -273,6 +273,7 @@ struct drm_i915_display_funcs { void (*set_cdclk)(struct drm_i915_private *dev_priv, const struct intel_cdclk_config *cdclk_config, enum pipe pipe); + int (*bw_calc_min_cdclk)(struct intel_atomic_state *state); int (*get_fifo_size)(struct drm_i915_private *dev_priv, enum i9xx_plane_id i9xx_plane); int (*compute_pipe_wm)(struct intel_crtc_state *crtc_state); @@ -829,6 +830,9 @@ struct drm_i915_private { /* FIXME: Device release actions should all be moved to drmm_ */ bool do_release; + /* i915 device parameters */ + struct i915_params params; + const struct intel_device_info __info; /* Use INTEL_INFO() to access. */ struct intel_runtime_info __runtime; /* Use RUNTIME_INFO() to access. */ struct intel_driver_caps caps; @@ -950,6 +954,13 @@ struct drm_i915_private { struct intel_global_obj obj; } cdclk; + struct { + /* The current hardware dbuf configuration */ + u8 enabled_slices; + + struct intel_global_obj obj; + } dbuf; + /** * wq - Driver workqueue for GEM. * @@ -1126,12 +1137,12 @@ struct drm_i915_private { * Set during HW readout of watermarks/DDB. Some platforms * need to know when we're still using BIOS-provided values * (which we don't fully trust). + * + * FIXME get rid of this. */ bool distrust_bios_wm; } wm; - u8 enabled_dbuf_slices_mask; /* GEN11 has configurable 2 slices */ - struct dram_info { bool valid; bool is_16gb_dimm; @@ -1256,6 +1267,11 @@ static inline struct drm_i915_private *pdev_to_i915(struct pci_dev *pdev) (engine__); \ (engine__) = rb_to_uabi_engine(rb_next(&(engine__)->uabi_node))) +#define for_each_uabi_class_engine(engine__, class__, i915__) \ + for ((engine__) = intel_engine_lookup_user((i915__), (class__), 0); \ + (engine__) && (engine__)->uabi_class == (class__); \ + (engine__) = rb_to_uabi_engine(rb_next(&(engine__)->uabi_node))) + #define I915_GTT_OFFSET_NONE ((u32)-1) /* @@ -1406,10 +1422,12 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915, #define IS_KABYLAKE(dev_priv) IS_PLATFORM(dev_priv, INTEL_KABYLAKE) #define IS_GEMINILAKE(dev_priv) IS_PLATFORM(dev_priv, INTEL_GEMINILAKE) #define IS_COFFEELAKE(dev_priv) IS_PLATFORM(dev_priv, INTEL_COFFEELAKE) +#define IS_COMETLAKE(dev_priv) IS_PLATFORM(dev_priv, INTEL_COMETLAKE) #define IS_CANNONLAKE(dev_priv) IS_PLATFORM(dev_priv, INTEL_CANNONLAKE) #define IS_ICELAKE(dev_priv) IS_PLATFORM(dev_priv, INTEL_ICELAKE) #define IS_ELKHARTLAKE(dev_priv) IS_PLATFORM(dev_priv, INTEL_ELKHARTLAKE) #define IS_TIGERLAKE(dev_priv) IS_PLATFORM(dev_priv, INTEL_TIGERLAKE) +#define IS_ROCKETLAKE(dev_priv) IS_PLATFORM(dev_priv, INTEL_ROCKETLAKE) #define IS_HSW_EARLY_SDV(dev_priv) (IS_HASWELL(dev_priv) && \ (INTEL_DEVID(dev_priv) & 0xFF00) == 0x0C00) #define IS_BDW_ULT(dev_priv) \ @@ -1453,6 +1471,14 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915, INTEL_INFO(dev_priv)->gt == 2) #define IS_CFL_GT3(dev_priv) (IS_COFFEELAKE(dev_priv) && \ INTEL_INFO(dev_priv)->gt == 3) + +#define IS_CML_ULT(dev_priv) \ + IS_SUBPLATFORM(dev_priv, INTEL_COMETLAKE, INTEL_SUBPLATFORM_ULT) +#define IS_CML_ULX(dev_priv) \ + IS_SUBPLATFORM(dev_priv, INTEL_COMETLAKE, INTEL_SUBPLATFORM_ULX) +#define IS_CML_GT2(dev_priv) (IS_COMETLAKE(dev_priv) && \ + INTEL_INFO(dev_priv)->gt == 2) + #define IS_CNL_WITH_PORT_F(dev_priv) \ IS_SUBPLATFORM(dev_priv, INTEL_CANNONLAKE, INTEL_SUBPLATFORM_PORTF) #define IS_ICL_WITH_PORT_F(dev_priv) \ @@ -1523,6 +1549,13 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915, #define IS_TGL_REVID(p, since, until) \ (IS_TIGERLAKE(p) && IS_REVID(p, since, until)) +#define RKL_REVID_A0 0x0 +#define RKL_REVID_B0 0x1 +#define RKL_REVID_C0 0x4 + +#define IS_RKL_REVID(p, since, until) \ + (IS_ROCKETLAKE(p) && IS_REVID(p, since, until)) + #define IS_LP(dev_priv) (INTEL_INFO(dev_priv)->is_lp) #define IS_GEN9_LP(dev_priv) (IS_GEN(dev_priv, 9) && IS_LP(dev_priv)) #define IS_GEN9_BC(dev_priv) (IS_GEN(dev_priv, 9) && !IS_LP(dev_priv)) @@ -1616,6 +1649,8 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915, #define HAS_DDI(dev_priv) (INTEL_INFO(dev_priv)->display.has_ddi) #define HAS_FPGA_DBG_UNCLAIMED(dev_priv) (INTEL_INFO(dev_priv)->has_fpga_dbg) #define HAS_PSR(dev_priv) (INTEL_INFO(dev_priv)->display.has_psr) +#define HAS_PSR_HW_TRACKING(dev_priv) \ + (INTEL_INFO(dev_priv)->display.has_psr_hw_tracking) #define HAS_TRANSCODER(dev_priv, trans) ((INTEL_INFO(dev_priv)->cpu_transcoder_mask & BIT(trans)) != 0) #define HAS_RC6(dev_priv) (INTEL_INFO(dev_priv)->has_rc6) @@ -1658,7 +1693,8 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915, #define HAS_DISPLAY(dev_priv) (INTEL_INFO(dev_priv)->pipe_mask != 0) /* Only valid when HAS_DISPLAY() is true */ -#define INTEL_DISPLAY_ENABLED(dev_priv) (WARN_ON(!HAS_DISPLAY(dev_priv)), !i915_modparams.disable_display) +#define INTEL_DISPLAY_ENABLED(dev_priv) \ + (drm_WARN_ON(&(dev_priv)->drm, !HAS_DISPLAY(dev_priv)), !(dev_priv)->params.disable_display) static inline bool intel_vtd_active(void) { diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 0cbcb9f54e7d..9aa3066cb75d 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -933,6 +933,18 @@ void i915_gem_runtime_suspend(struct drm_i915_private *i915) } } +static void discard_ggtt_vma(struct i915_vma *vma) +{ + struct drm_i915_gem_object *obj = vma->obj; + + spin_lock(&obj->vma.lock); + if (!RB_EMPTY_NODE(&vma->obj_node)) { + rb_erase(&vma->obj_node, &obj->vma.tree); + RB_CLEAR_NODE(&vma->obj_node); + } + spin_unlock(&obj->vma.lock); +} + struct i915_vma * i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj, const struct i915_ggtt_view *view, @@ -979,6 +991,7 @@ i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj, return ERR_PTR(-ENOSPC); } +new_vma: vma = i915_vma_instance(obj, &ggtt->vm, view); if (IS_ERR(vma)) return vma; @@ -993,6 +1006,11 @@ i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj, return ERR_PTR(-ENOSPC); } + if (i915_vma_is_pinned(vma) || i915_vma_is_active(vma)) { + discard_ggtt_vma(vma); + goto new_vma; + } + ret = i915_vma_unbind(vma); if (ret) return ERR_PTR(ret); diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.h b/drivers/gpu/drm/i915/i915_gem_gtt.h index f6226df9f972..c9b0ee5e1d23 100644 --- a/drivers/gpu/drm/i915/i915_gem_gtt.h +++ b/drivers/gpu/drm/i915/i915_gem_gtt.h @@ -42,7 +42,6 @@ int i915_gem_gtt_insert(struct i915_address_space *vm, #define PIN_OFFSET_BIAS BIT_ULL(6) #define PIN_OFFSET_FIXED BIT_ULL(7) -#define PIN_UPDATE BIT_ULL(9) #define PIN_GLOBAL BIT_ULL(10) /* I915_VMA_GLOBAL_BIND */ #define PIN_USER BIT_ULL(11) /* I915_VMA_LOCAL_BIND */ diff --git a/drivers/gpu/drm/i915/i915_getparam.c b/drivers/gpu/drm/i915/i915_getparam.c index d042644b9cd2..40390b2352b1 100644 --- a/drivers/gpu/drm/i915/i915_getparam.c +++ b/drivers/gpu/drm/i915/i915_getparam.c @@ -80,7 +80,7 @@ int i915_getparam_ioctl(struct drm_device *dev, void *data, return -ENODEV; break; case I915_PARAM_HAS_GPU_RESET: - value = i915_modparams.enable_hangcheck && + value = i915->params.enable_hangcheck && intel_has_gpu_reset(&i915->gt); if (value && intel_has_reset_engine(&i915->gt)) value = 2; diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c index eec292d06f11..866166ada10e 100644 --- a/drivers/gpu/drm/i915/i915_gpu_error.c +++ b/drivers/gpu/drm/i915/i915_gpu_error.c @@ -1698,7 +1698,7 @@ static void capture_gen(struct i915_gpu_coredump *error) error->reset_count = i915_reset_count(&i915->gpu_error); error->suspend_count = i915->suspend_count; - i915_params_copy(&error->params, &i915_modparams); + i915_params_copy(&error->params, &i915->params); memcpy(&error->device_info, INTEL_INFO(i915), sizeof(error->device_info)); @@ -1713,7 +1713,7 @@ i915_gpu_coredump_alloc(struct drm_i915_private *i915, gfp_t gfp) { struct i915_gpu_coredump *error; - if (!i915_modparams.error_capture) + if (!i915->params.error_capture) return NULL; error = kzalloc(sizeof(*error), gfp); diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c index 284cf078135a..710224d930c5 100644 --- a/drivers/gpu/drm/i915/i915_irq.c +++ b/drivers/gpu/drm/i915/i915_irq.c @@ -777,7 +777,7 @@ static int __intel_get_crtc_scanline(struct intel_crtc *crtc) vblank = &crtc->base.dev->vblank[drm_crtc_index(&crtc->base)]; mode = &vblank->hwmode; - if (mode->private_flags & I915_MODE_FLAG_GET_SCANLINE_FROM_TIMESTAMP) + if (crtc->mode_flags & I915_MODE_FLAG_GET_SCANLINE_FROM_TIMESTAMP) return __intel_get_crtc_scanline_from_timestamp(crtc); vtotal = mode->crtc_vtotal; @@ -836,7 +836,7 @@ static bool i915_get_crtc_scanoutpos(struct drm_crtc *_crtc, unsigned long irqflags; bool use_scanline_counter = INTEL_GEN(dev_priv) >= 5 || IS_G4X(dev_priv) || IS_GEN(dev_priv, 2) || - mode->private_flags & I915_MODE_FLAG_USE_SCANLINE_COUNTER; + crtc->mode_flags & I915_MODE_FLAG_USE_SCANLINE_COUNTER; if (drm_WARN_ON(&dev_priv->drm, !mode->crtc_clock)) { drm_dbg(&dev_priv->drm, @@ -2097,67 +2097,68 @@ static void ivb_display_irq_handler(struct drm_i915_private *dev_priv, */ static irqreturn_t ilk_irq_handler(int irq, void *arg) { - struct drm_i915_private *dev_priv = arg; + struct drm_i915_private *i915 = arg; + void __iomem * const regs = i915->uncore.regs; u32 de_iir, gt_iir, de_ier, sde_ier = 0; irqreturn_t ret = IRQ_NONE; - if (!intel_irqs_enabled(dev_priv)) + if (unlikely(!intel_irqs_enabled(i915))) return IRQ_NONE; /* IRQs are synced during runtime_suspend, we don't require a wakeref */ - disable_rpm_wakeref_asserts(&dev_priv->runtime_pm); + disable_rpm_wakeref_asserts(&i915->runtime_pm); /* disable master interrupt before clearing iir */ - de_ier = I915_READ(DEIER); - I915_WRITE(DEIER, de_ier & ~DE_MASTER_IRQ_CONTROL); + de_ier = raw_reg_read(regs, DEIER); + raw_reg_write(regs, DEIER, de_ier & ~DE_MASTER_IRQ_CONTROL); /* Disable south interrupts. We'll only write to SDEIIR once, so further * interrupts will will be stored on its back queue, and then we'll be * able to process them after we restore SDEIER (as soon as we restore * it, we'll get an interrupt if SDEIIR still has something to process * due to its back queue). */ - if (!HAS_PCH_NOP(dev_priv)) { - sde_ier = I915_READ(SDEIER); - I915_WRITE(SDEIER, 0); + if (!HAS_PCH_NOP(i915)) { + sde_ier = raw_reg_read(regs, SDEIER); + raw_reg_write(regs, SDEIER, 0); } /* Find, clear, then process each source of interrupt */ - gt_iir = I915_READ(GTIIR); + gt_iir = raw_reg_read(regs, GTIIR); if (gt_iir) { - I915_WRITE(GTIIR, gt_iir); - ret = IRQ_HANDLED; - if (INTEL_GEN(dev_priv) >= 6) - gen6_gt_irq_handler(&dev_priv->gt, gt_iir); + raw_reg_write(regs, GTIIR, gt_iir); + if (INTEL_GEN(i915) >= 6) + gen6_gt_irq_handler(&i915->gt, gt_iir); else - gen5_gt_irq_handler(&dev_priv->gt, gt_iir); + gen5_gt_irq_handler(&i915->gt, gt_iir); + ret = IRQ_HANDLED; } - de_iir = I915_READ(DEIIR); + de_iir = raw_reg_read(regs, DEIIR); if (de_iir) { - I915_WRITE(DEIIR, de_iir); - ret = IRQ_HANDLED; - if (INTEL_GEN(dev_priv) >= 7) - ivb_display_irq_handler(dev_priv, de_iir); + raw_reg_write(regs, DEIIR, de_iir); + if (INTEL_GEN(i915) >= 7) + ivb_display_irq_handler(i915, de_iir); else - ilk_display_irq_handler(dev_priv, de_iir); + ilk_display_irq_handler(i915, de_iir); + ret = IRQ_HANDLED; } - if (INTEL_GEN(dev_priv) >= 6) { - u32 pm_iir = I915_READ(GEN6_PMIIR); + if (INTEL_GEN(i915) >= 6) { + u32 pm_iir = raw_reg_read(regs, GEN6_PMIIR); if (pm_iir) { - I915_WRITE(GEN6_PMIIR, pm_iir); + raw_reg_write(regs, GEN6_PMIIR, pm_iir); + gen6_rps_irq_handler(&i915->gt.rps, pm_iir); ret = IRQ_HANDLED; - gen6_rps_irq_handler(&dev_priv->gt.rps, pm_iir); } } - I915_WRITE(DEIER, de_ier); - if (!HAS_PCH_NOP(dev_priv)) - I915_WRITE(SDEIER, sde_ier); + raw_reg_write(regs, DEIER, de_ier); + if (sde_ier) + raw_reg_write(regs, SDEIER, sde_ier); /* IRQs are synced during runtime_suspend, we don't require a wakeref */ - enable_rpm_wakeref_asserts(&dev_priv->runtime_pm); + enable_rpm_wakeref_asserts(&i915->runtime_pm); return ret; } @@ -2254,7 +2255,9 @@ static u32 gen8_de_port_aux_mask(struct drm_i915_private *dev_priv) static u32 gen8_de_pipe_fault_mask(struct drm_i915_private *dev_priv) { - if (INTEL_GEN(dev_priv) >= 11) + if (IS_ROCKETLAKE(dev_priv)) + return RKL_DE_PIPE_IRQ_FAULT_ERRORS; + else if (INTEL_GEN(dev_priv) >= 11) return GEN11_DE_PIPE_IRQ_FAULT_ERRORS; else if (INTEL_GEN(dev_priv) >= 9) return GEN9_DE_PIPE_IRQ_FAULT_ERRORS; @@ -2869,13 +2872,15 @@ static void gen11_display_irq_reset(struct drm_i915_private *dev_priv) { struct intel_uncore *uncore = &dev_priv->uncore; enum pipe pipe; + u32 trans_mask = BIT(TRANSCODER_A) | BIT(TRANSCODER_B) | + BIT(TRANSCODER_C) | BIT(TRANSCODER_D); intel_uncore_write(uncore, GEN11_DISPLAY_INT_CTL, 0); if (INTEL_GEN(dev_priv) >= 12) { enum transcoder trans; - for (trans = TRANSCODER_A; trans <= TRANSCODER_D; trans++) { + for_each_cpu_transcoder_masked(dev_priv, trans, trans_mask) { enum intel_display_power_domain domain; domain = POWER_DOMAIN_TRANSCODER(trans); @@ -2902,8 +2907,10 @@ static void gen11_display_irq_reset(struct drm_i915_private *dev_priv) if (INTEL_PCH_TYPE(dev_priv) >= PCH_ICP) GEN3_IRQ_RESET(uncore, SDE); - /* Wa_14010685332:icl */ - if (INTEL_PCH_TYPE(dev_priv) == PCH_ICP) { + /* Wa_14010685332:icl,jsl,ehl */ + if (INTEL_PCH_TYPE(dev_priv) == PCH_ICP || + INTEL_PCH_TYPE(dev_priv) == PCH_JSP || + INTEL_PCH_TYPE(dev_priv) == PCH_MCC) { intel_uncore_rmw(uncore, SOUTH_CHICKEN1, SBCLK_RUN_REFCLK_DIS, SBCLK_RUN_REFCLK_DIS); intel_uncore_rmw(uncore, SOUTH_CHICKEN1, @@ -3396,6 +3403,8 @@ static void gen8_de_irq_postinstall(struct drm_i915_private *dev_priv) u32 de_port_masked = gen8_de_port_aux_mask(dev_priv); u32 de_port_enables; u32 de_misc_masked = GEN8_DE_EDP_PSR; + u32 trans_mask = BIT(TRANSCODER_A) | BIT(TRANSCODER_B) | + BIT(TRANSCODER_C) | BIT(TRANSCODER_D); enum pipe pipe; if (INTEL_GEN(dev_priv) <= 10) @@ -3416,7 +3425,7 @@ static void gen8_de_irq_postinstall(struct drm_i915_private *dev_priv) if (INTEL_GEN(dev_priv) >= 12) { enum transcoder trans; - for (trans = TRANSCODER_A; trans <= TRANSCODER_D; trans++) { + for_each_cpu_transcoder_masked(dev_priv, trans, trans_mask) { enum intel_display_power_domain domain; domain = POWER_DOMAIN_TRANSCODER(trans); diff --git a/drivers/gpu/drm/i915/i915_params.c b/drivers/gpu/drm/i915/i915_params.c index 02559da61e6e..a7b61e6ec508 100644 --- a/drivers/gpu/drm/i915/i915_params.c +++ b/drivers/gpu/drm/i915/i915_params.c @@ -40,6 +40,15 @@ struct i915_params i915_modparams __read_mostly = { #undef MEMBER }; +/* + * Note: As a rule, keep module parameter sysfs permissions read-only + * 0400. Runtime changes are only supported through i915 debugfs. + * + * For any exceptions requiring write access and runtime changes through module + * parameter sysfs, prevent debugfs file creation by setting the parameter's + * debugfs mode to 0. + */ + i915_param_named(modeset, int, 0400, "Use kernel modesetting [KMS] (0=disable, " "1=on, -1=force vga console preference [default])"); @@ -49,7 +58,7 @@ i915_param_named_unsafe(enable_dc, int, 0400, "(-1=auto [default]; 0=disable; 1=up to DC5; 2=up to DC6; " "3=up to DC5 with DC3CO; 4=up to DC6 with DC3CO)"); -i915_param_named_unsafe(enable_fbc, int, 0600, +i915_param_named_unsafe(enable_fbc, int, 0400, "Enable frame buffer compression for power savings " "(default: -1 (use per-chip default))"); @@ -57,7 +66,7 @@ i915_param_named_unsafe(lvds_channel_mode, int, 0400, "Specify LVDS channel mode " "(0=probe BIOS [default], 1=single-channel, 2=dual-channel)"); -i915_param_named_unsafe(panel_use_ssc, int, 0600, +i915_param_named_unsafe(panel_use_ssc, int, 0400, "Use Spread Spectrum Clock with panels [LVDS/eDP] " "(default: auto from VBT)"); @@ -65,29 +74,34 @@ i915_param_named_unsafe(vbt_sdvo_panel_type, int, 0400, "Override/Ignore selection of SDVO panel mode in the VBT " "(-2=ignore, -1=auto [default], index in VBT BIOS table)"); -i915_param_named_unsafe(reset, uint, 0600, +i915_param_named_unsafe(reset, uint, 0400, "Attempt GPU resets (0=disabled, 1=full gpu reset, 2=engine reset [default])"); i915_param_named_unsafe(vbt_firmware, charp, 0400, "Load VBT from specified file under /lib/firmware"); #if IS_ENABLED(CONFIG_DRM_I915_CAPTURE_ERROR) -i915_param_named(error_capture, bool, 0600, +i915_param_named(error_capture, bool, 0400, "Record the GPU state following a hang. " "This information in /sys/class/drm/card<N>/error is vital for " "triaging and debugging hangs."); #endif -i915_param_named_unsafe(enable_hangcheck, bool, 0600, +i915_param_named_unsafe(enable_hangcheck, bool, 0400, "Periodically check GPU activity for detecting hangs. " "WARNING: Disabling this can cause system wide hangs. " "(default: true)"); -i915_param_named_unsafe(enable_psr, int, 0600, +i915_param_named_unsafe(enable_psr, int, 0400, "Enable PSR " "(0=disabled, 1=enabled) " "Default: -1 (use per-chip default)"); +i915_param_named(psr_safest_params, bool, 0400, + "Replace PSR VBT parameters by the safest and not optimal ones. This " + "is helpfull to detect if PSR issues are related to bad values set in " + " VBT. (0=use VBT paramters, 1=use safest parameters)"); + i915_param_named_unsafe(force_probe, charp, 0400, "Force probe the driver for specified devices. " "See CONFIG_DRM_I915_FORCE_PROBE for details."); @@ -96,22 +110,22 @@ i915_param_named_unsafe(disable_power_well, int, 0400, "Disable display power wells when possible " "(-1=auto [default], 0=power wells always on, 1=power wells disabled when possible)"); -i915_param_named_unsafe(enable_ips, int, 0600, "Enable IPS (default: true)"); +i915_param_named_unsafe(enable_ips, int, 0400, "Enable IPS (default: true)"); -i915_param_named(fastboot, int, 0600, +i915_param_named(fastboot, int, 0400, "Try to skip unnecessary mode sets at boot time " "(0=disabled, 1=enabled) " "Default: -1 (use per-chip default)"); -i915_param_named_unsafe(load_detect_test, bool, 0600, +i915_param_named_unsafe(load_detect_test, bool, 0400, "Force-enable the VGA load detect code for testing (default:false). " "For developers only."); -i915_param_named_unsafe(force_reset_modeset_test, bool, 0600, +i915_param_named_unsafe(force_reset_modeset_test, bool, 0400, "Force a modeset during gpu reset for testing (default:false). " "For developers only."); -i915_param_named_unsafe(invert_brightness, int, 0600, +i915_param_named_unsafe(invert_brightness, int, 0400, "Invert backlight brightness " "(-1 force normal, 0 machine defaults, 1 force inversion), please " "report PCI device ID, subsystem vendor and subsystem device ID " @@ -121,10 +135,11 @@ i915_param_named_unsafe(invert_brightness, int, 0600, i915_param_named(disable_display, bool, 0400, "Disable display (default: false)"); -i915_param_named(mmio_debug, int, 0600, +i915_param_named(mmio_debug, int, 0400, "Enable the MMIO debug code for the first N failures (default: off). " "This may negatively affect performance."); +/* Special case writable file */ i915_param_named(verbose_state_checks, bool, 0600, "Enable verbose logs (ie. WARN_ON()) in case of unexpected hw state conditions."); @@ -155,7 +170,7 @@ i915_param_named_unsafe(huc_firmware_path, charp, 0400, i915_param_named_unsafe(dmc_firmware_path, charp, 0400, "DMC firmware path to use instead of the default one"); -i915_param_named_unsafe(enable_dp_mst, bool, 0600, +i915_param_named_unsafe(enable_dp_mst, bool, 0400, "Enable multi-stream transport (MST) for new DisplayPort sinks. (default: true)"); #if IS_ENABLED(CONFIG_DRM_I915_DEBUG) @@ -163,7 +178,7 @@ i915_param_named_unsafe(inject_probe_failure, uint, 0400, "Force an error after a number of failure check points (0:disabled (default), N:force failure at the Nth failure check point)"); #endif -i915_param_named(enable_dpcd_backlight, int, 0600, +i915_param_named(enable_dpcd_backlight, int, 0400, "Enable support for DPCD backlight control" "(-1=use per-VBT LFP backlight type setting [default], 0=disabled, 1=enabled)"); diff --git a/drivers/gpu/drm/i915/i915_params.h b/drivers/gpu/drm/i915/i915_params.h index 4f21bfffbf0e..53fb5ba8fbed 100644 --- a/drivers/gpu/drm/i915/i915_params.h +++ b/drivers/gpu/drm/i915/i915_params.h @@ -53,6 +53,7 @@ struct drm_printer; param(int, enable_dc, -1, 0400) \ param(int, enable_fbc, -1, 0600) \ param(int, enable_psr, -1, 0600) \ + param(bool, psr_safest_params, false, 0600) \ param(int, disable_power_well, -1, 0400) \ param(int, enable_ips, 1, 0600) \ param(int, invert_brightness, 0, 0600) \ diff --git a/drivers/gpu/drm/i915/i915_pci.c b/drivers/gpu/drm/i915/i915_pci.c index eb0b5be7c35d..e5fdf17cd9cd 100644 --- a/drivers/gpu/drm/i915/i915_pci.c +++ b/drivers/gpu/drm/i915/i915_pci.c @@ -536,6 +536,7 @@ static const struct intel_device_info vlv_info = { .display.has_ddi = 1, \ .has_fpga_dbg = 1, \ .display.has_psr = 1, \ + .display.has_psr_hw_tracking = 1, \ .display.has_dp_mst = 1, \ .has_rc6p = 0 /* RC6p removed-by HSW */, \ HSW_PIPE_OFFSETS, \ @@ -690,6 +691,7 @@ static const struct intel_device_info skl_gt4_info = { .display.has_fbc = 1, \ .display.has_hdcp = 1, \ .display.has_psr = 1, \ + .display.has_psr_hw_tracking = 1, \ .has_runtime_pm = 1, \ .display.has_csr = 1, \ .has_rc6 = 1, \ @@ -766,6 +768,20 @@ static const struct intel_device_info cfl_gt3_info = { BIT(RCS0) | BIT(VCS0) | BIT(BCS0) | BIT(VECS0) | BIT(VCS1), }; +#define CML_PLATFORM \ + GEN9_FEATURES, \ + PLATFORM(INTEL_COMETLAKE) + +static const struct intel_device_info cml_gt1_info = { + CML_PLATFORM, + .gt = 1, +}; + +static const struct intel_device_info cml_gt2_info = { + CML_PLATFORM, + .gt = 2, +}; + #define GEN10_FEATURES \ GEN9_FEATURES, \ GEN(10), \ @@ -788,6 +804,7 @@ static const struct intel_device_info cnl_info = { #define GEN11_FEATURES \ GEN10_FEATURES, \ GEN11_DEFAULT_PAGE_SIZES, \ + .abox_mask = BIT(0), \ .cpu_transcoder_mask = BIT(TRANSCODER_A) | BIT(TRANSCODER_B) | \ BIT(TRANSCODER_C) | BIT(TRANSCODER_EDP) | \ BIT(TRANSCODER_DSI_0) | BIT(TRANSCODER_DSI_1), \ @@ -831,6 +848,7 @@ static const struct intel_device_info ehl_info = { #define GEN12_FEATURES \ GEN11_FEATURES, \ GEN(12), \ + .abox_mask = GENMASK(2, 1), \ .pipe_mask = BIT(PIPE_A) | BIT(PIPE_B) | BIT(PIPE_C) | BIT(PIPE_D), \ .cpu_transcoder_mask = BIT(TRANSCODER_A) | BIT(TRANSCODER_B) | \ BIT(TRANSCODER_C) | BIT(TRANSCODER_D) | \ @@ -863,6 +881,19 @@ static const struct intel_device_info tgl_info = { BIT(RCS0) | BIT(BCS0) | BIT(VECS0) | BIT(VCS0) | BIT(VCS2), }; +static const struct intel_device_info rkl_info = { + GEN12_FEATURES, + PLATFORM(INTEL_ROCKETLAKE), + .abox_mask = BIT(0), + .pipe_mask = BIT(PIPE_A) | BIT(PIPE_B) | BIT(PIPE_C), + .cpu_transcoder_mask = BIT(TRANSCODER_A) | BIT(TRANSCODER_B) | + BIT(TRANSCODER_C), + .require_force_probe = 1, + .display.has_psr_hw_tracking = 0, + .engine_mask = + BIT(RCS0) | BIT(BCS0) | BIT(VECS0) | BIT(VCS0), +}; + #define GEN12_DGFX_FEATURES \ GEN12_FEATURES, \ .is_dgfx = 1 @@ -933,14 +964,15 @@ static const struct pci_device_id pciidlist[] = { INTEL_WHL_U_GT2_IDS(&cfl_gt2_info), INTEL_AML_CFL_GT2_IDS(&cfl_gt2_info), INTEL_WHL_U_GT3_IDS(&cfl_gt3_info), - INTEL_CML_GT1_IDS(&cfl_gt1_info), - INTEL_CML_GT2_IDS(&cfl_gt2_info), - INTEL_CML_U_GT1_IDS(&cfl_gt1_info), - INTEL_CML_U_GT2_IDS(&cfl_gt2_info), + INTEL_CML_GT1_IDS(&cml_gt1_info), + INTEL_CML_GT2_IDS(&cml_gt2_info), + INTEL_CML_U_GT1_IDS(&cml_gt1_info), + INTEL_CML_U_GT2_IDS(&cml_gt2_info), INTEL_CNL_IDS(&cnl_info), INTEL_ICL_11_IDS(&icl_info), INTEL_EHL_IDS(&ehl_info), INTEL_TGL_12_IDS(&tgl_info), + INTEL_RKL_IDS(&rkl_info), {0, 0, 0} }; MODULE_DEVICE_TABLE(pci, pciidlist); diff --git a/drivers/gpu/drm/i915/i915_pmu.c b/drivers/gpu/drm/i915/i915_pmu.c index 962ded9ce73f..28bc5f13ae52 100644 --- a/drivers/gpu/drm/i915/i915_pmu.c +++ b/drivers/gpu/drm/i915/i915_pmu.c @@ -441,7 +441,11 @@ static u64 count_interrupts(struct drm_i915_private *i915) static void i915_pmu_event_destroy(struct perf_event *event) { - WARN_ON(event->parent); + struct drm_i915_private *i915 = + container_of(event->pmu, typeof(*i915), pmu.base); + + drm_WARN_ON(&i915->drm, event->parent); + module_put(THIS_MODULE); } @@ -561,7 +565,10 @@ static u64 __i915_pmu_event_read(struct perf_event *event) /* Do nothing */ } else if (sample == I915_SAMPLE_BUSY && intel_engine_supports_stats(engine)) { - val = ktime_to_ns(intel_engine_get_busy_time(engine)); + ktime_t unused; + + val = ktime_to_ns(intel_engine_get_busy_time(engine, + &unused)); } else { val = engine->pmu.sample[sample].cur; } @@ -1058,8 +1065,10 @@ static int i915_pmu_register_cpuhp_state(struct i915_pmu *pmu) static void i915_pmu_unregister_cpuhp_state(struct i915_pmu *pmu) { - WARN_ON(pmu->cpuhp.slot == CPUHP_INVALID); - WARN_ON(cpuhp_state_remove_instance(pmu->cpuhp.slot, &pmu->cpuhp.node)); + struct drm_i915_private *i915 = container_of(pmu, typeof(*i915), pmu); + + drm_WARN_ON(&i915->drm, pmu->cpuhp.slot == CPUHP_INVALID); + drm_WARN_ON(&i915->drm, cpuhp_state_remove_instance(pmu->cpuhp.slot, &pmu->cpuhp.node)); cpuhp_remove_multi_state(pmu->cpuhp.slot); pmu->cpuhp.slot = CPUHP_INVALID; } diff --git a/drivers/gpu/drm/i915/i915_query.c b/drivers/gpu/drm/i915/i915_query.c index e75c528ebbe0..c1ebda9b5627 100644 --- a/drivers/gpu/drm/i915/i915_query.c +++ b/drivers/gpu/drm/i915/i915_query.c @@ -109,8 +109,7 @@ query_engine_info(struct drm_i915_private *i915, for_each_uabi_engine(engine, i915) num_uabi_engines++; - len = sizeof(struct drm_i915_query_engine_info) + - num_uabi_engines * sizeof(struct drm_i915_engine_info); + len = struct_size(query_ptr, engines, num_uabi_engines); ret = copy_query_item(&query, sizeof(query), len, query_item); if (ret != 0) diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h index 06cd1d28a176..f09120cac89a 100644 --- a/drivers/gpu/drm/i915/i915_reg.h +++ b/drivers/gpu/drm/i915/i915_reg.h @@ -1869,9 +1869,11 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg) #define _ICL_COMBOPHY_A 0x162000 #define _ICL_COMBOPHY_B 0x6C000 #define _EHL_COMBOPHY_C 0x160000 +#define _RKL_COMBOPHY_D 0x161000 #define _ICL_COMBOPHY(phy) _PICK(phy, _ICL_COMBOPHY_A, \ _ICL_COMBOPHY_B, \ - _EHL_COMBOPHY_C) + _EHL_COMBOPHY_C, \ + _RKL_COMBOPHY_D) /* CNL/ICL Port CL_DW registers */ #define _ICL_PORT_CL_DW(dw, phy) (_ICL_COMBOPHY(phy) + \ @@ -2877,9 +2879,12 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg) #define LM_FIFO_WATERMARK 0x0000001F #define MI_ARB_STATE _MMIO(0x20e4) /* 915+ only */ -#define MBUS_ABOX_CTL _MMIO(0x45038) -#define MBUS_ABOX1_CTL _MMIO(0x45048) -#define MBUS_ABOX2_CTL _MMIO(0x4504C) +#define _MBUS_ABOX0_CTL 0x45038 +#define _MBUS_ABOX1_CTL 0x45048 +#define _MBUS_ABOX2_CTL 0x4504C +#define MBUS_ABOX_CTL(x) _MMIO(_PICK(x, _MBUS_ABOX0_CTL, \ + _MBUS_ABOX1_CTL, \ + _MBUS_ABOX2_CTL)) #define MBUS_ABOX_BW_CREDIT_MASK (3 << 20) #define MBUS_ABOX_BW_CREDIT(x) ((x) << 20) #define MBUS_ABOX_B_CREDIT_MASK (0xF << 16) @@ -3768,19 +3773,16 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg) /* Clocking configuration register */ #define CLKCFG _MMIO(MCHBAR_MIRROR_BASE + 0xc00) -#define CLKCFG_FSB_400 (5 << 0) /* hrawclk 100 */ +#define CLKCFG_FSB_400 (0 << 0) /* hrawclk 100 */ +#define CLKCFG_FSB_400_ALT (5 << 0) /* hrawclk 100 */ #define CLKCFG_FSB_533 (1 << 0) /* hrawclk 133 */ #define CLKCFG_FSB_667 (3 << 0) /* hrawclk 166 */ #define CLKCFG_FSB_800 (2 << 0) /* hrawclk 200 */ #define CLKCFG_FSB_1067 (6 << 0) /* hrawclk 266 */ #define CLKCFG_FSB_1067_ALT (0 << 0) /* hrawclk 266 */ #define CLKCFG_FSB_1333 (7 << 0) /* hrawclk 333 */ -/* - * Note that on at least on ELK the below value is reported for both - * 333 and 400 MHz BIOS FSB setting, but given that the gmch datasheet - * lists only 200/266/333 MHz FSB as supported let's decode it as 333 MHz. - */ #define CLKCFG_FSB_1333_ALT (4 << 0) /* hrawclk 333 */ +#define CLKCFG_FSB_1600_ALT (6 << 0) /* hrawclk 400 */ #define CLKCFG_FSB_MASK (7 << 0) #define CLKCFG_MEM_533 (1 << 4) #define CLKCFG_MEM_667 (2 << 4) @@ -4512,25 +4514,39 @@ enum { #define EDP_PSR_DEBUG_MASK_DISP_REG_WRITE (1 << 16) /* Reserved in ICL+ */ #define EDP_PSR_DEBUG_EXIT_ON_PIXEL_UNDERRUN (1 << 15) /* SKL+ */ -#define _PSR2_CTL_A 0x60900 -#define _PSR2_CTL_EDP 0x6f900 -#define EDP_PSR2_CTL(tran) _MMIO_TRANS2(tran, _PSR2_CTL_A) -#define EDP_PSR2_ENABLE (1 << 31) -#define EDP_SU_TRACK_ENABLE (1 << 30) -#define EDP_Y_COORDINATE_VALID (1 << 26) /* GLK and CNL+ */ -#define EDP_Y_COORDINATE_ENABLE (1 << 25) /* GLK and CNL+ */ -#define EDP_MAX_SU_DISABLE_TIME(t) ((t) << 20) -#define EDP_MAX_SU_DISABLE_TIME_MASK (0x1f << 20) -#define EDP_PSR2_TP2_TIME_500us (0 << 8) -#define EDP_PSR2_TP2_TIME_100us (1 << 8) -#define EDP_PSR2_TP2_TIME_2500us (2 << 8) -#define EDP_PSR2_TP2_TIME_50us (3 << 8) -#define EDP_PSR2_TP2_TIME_MASK (3 << 8) -#define EDP_PSR2_FRAME_BEFORE_SU_SHIFT 4 -#define EDP_PSR2_FRAME_BEFORE_SU_MASK (0xf << 4) -#define EDP_PSR2_FRAME_BEFORE_SU(a) ((a) << 4) -#define EDP_PSR2_IDLE_FRAME_MASK 0xf -#define EDP_PSR2_IDLE_FRAME_SHIFT 0 +#define _PSR2_CTL_A 0x60900 +#define _PSR2_CTL_EDP 0x6f900 +#define EDP_PSR2_CTL(tran) _MMIO_TRANS2(tran, _PSR2_CTL_A) +#define EDP_PSR2_ENABLE (1 << 31) +#define EDP_SU_TRACK_ENABLE (1 << 30) +#define TGL_EDP_PSR2_BLOCK_COUNT_NUM_2 (0 << 28) +#define TGL_EDP_PSR2_BLOCK_COUNT_NUM_3 (1 << 28) +#define EDP_Y_COORDINATE_VALID (1 << 26) /* GLK and CNL+ */ +#define EDP_Y_COORDINATE_ENABLE (1 << 25) /* GLK and CNL+ */ +#define EDP_MAX_SU_DISABLE_TIME(t) ((t) << 20) +#define EDP_MAX_SU_DISABLE_TIME_MASK (0x1f << 20) +#define EDP_PSR2_IO_BUFFER_WAKE_MAX_LINES 8 +#define EDP_PSR2_IO_BUFFER_WAKE(lines) ((EDP_PSR2_IO_BUFFER_WAKE_MAX_LINES - (lines)) << 13) +#define EDP_PSR2_IO_BUFFER_WAKE_MASK (3 << 13) +#define TGL_EDP_PSR2_IO_BUFFER_WAKE_MIN_LINES 5 +#define TGL_EDP_PSR2_IO_BUFFER_WAKE(lines) (((lines) - TGL_EDP_PSR2_IO_BUFFER_WAKE_MIN_LINES) << 13) +#define TGL_EDP_PSR2_IO_BUFFER_WAKE_MASK (7 << 13) +#define EDP_PSR2_FAST_WAKE_MAX_LINES 8 +#define EDP_PSR2_FAST_WAKE(lines) ((EDP_PSR2_FAST_WAKE_MAX_LINES - (lines)) << 11) +#define EDP_PSR2_FAST_WAKE_MASK (3 << 11) +#define TGL_EDP_PSR2_FAST_WAKE_MIN_LINES 5 +#define TGL_EDP_PSR2_FAST_WAKE(lines) (((lines) - TGL_EDP_PSR2_FAST_WAKE_MIN_LINES) << 10) +#define TGL_EDP_PSR2_FAST_WAKE_MASK (7 << 10) +#define EDP_PSR2_TP2_TIME_500us (0 << 8) +#define EDP_PSR2_TP2_TIME_100us (1 << 8) +#define EDP_PSR2_TP2_TIME_2500us (2 << 8) +#define EDP_PSR2_TP2_TIME_50us (3 << 8) +#define EDP_PSR2_TP2_TIME_MASK (3 << 8) +#define EDP_PSR2_FRAME_BEFORE_SU_SHIFT 4 +#define EDP_PSR2_FRAME_BEFORE_SU_MASK (0xf << 4) +#define EDP_PSR2_FRAME_BEFORE_SU(a) ((a) << 4) +#define EDP_PSR2_IDLE_FRAME_MASK 0xf +#define EDP_PSR2_IDLE_FRAME_SHIFT 0 #define _PSR_EVENT_TRANS_A 0x60848 #define _PSR_EVENT_TRANS_B 0x61848 @@ -6915,6 +6931,8 @@ enum { #define _PLANE_CUS_CTL_1_A 0x701c8 #define _PLANE_CUS_CTL_2_A 0x702c8 #define PLANE_CUS_ENABLE (1 << 31) +#define PLANE_CUS_PLANE_4_RKL (0 << 30) +#define PLANE_CUS_PLANE_5_RKL (1 << 30) #define PLANE_CUS_PLANE_6 (0 << 30) #define PLANE_CUS_PLANE_7 (1 << 30) #define PLANE_CUS_HPHASE_SIGN_NEGATIVE (1 << 19) @@ -6933,7 +6951,7 @@ enum { #define PLANE_COLOR_INPUT_CSC_ENABLE (1 << 20) /* ICL+ */ #define PLANE_COLOR_PIPE_CSC_ENABLE (1 << 23) /* Pre-ICL */ #define PLANE_COLOR_CSC_MODE_BYPASS (0 << 17) -#define PLANE_COLOR_CSC_MODE_YUV601_TO_RGB709 (1 << 17) +#define PLANE_COLOR_CSC_MODE_YUV601_TO_RGB601 (1 << 17) #define PLANE_COLOR_CSC_MODE_YUV709_TO_RGB709 (2 << 17) #define PLANE_COLOR_CSC_MODE_YUV2020_TO_RGB2020 (3 << 17) #define PLANE_COLOR_CSC_MODE_RGB709_TO_RGB2020 (4 << 17) @@ -7581,6 +7599,9 @@ enum { GEN11_PIPE_PLANE7_FAULT | \ GEN11_PIPE_PLANE6_FAULT | \ GEN11_PIPE_PLANE5_FAULT) +#define RKL_DE_PIPE_IRQ_FAULT_ERRORS \ + (GEN9_DE_PIPE_IRQ_FAULT_ERRORS | \ + GEN11_PIPE_PLANE5_FAULT) #define GEN8_DE_PORT_ISR _MMIO(0x44440) #define GEN8_DE_PORT_IMR _MMIO(0x44444) @@ -7835,13 +7856,20 @@ enum { #define WAIT_FOR_PCH_RESET_ACK (1 << 1) #define WAIT_FOR_PCH_FLR_ACK (1 << 0) -#define BW_BUDDY1_CTL _MMIO(0x45140) -#define BW_BUDDY2_CTL _MMIO(0x45150) +#define _BW_BUDDY0_CTL 0x45130 +#define _BW_BUDDY1_CTL 0x45140 +#define BW_BUDDY_CTL(x) _MMIO(_PICK_EVEN(x, \ + _BW_BUDDY0_CTL, \ + _BW_BUDDY1_CTL)) #define BW_BUDDY_DISABLE REG_BIT(31) #define BW_BUDDY_TLB_REQ_TIMER_MASK REG_GENMASK(21, 16) +#define BW_BUDDY_TLB_REQ_TIMER(x) REG_FIELD_PREP(BW_BUDDY_TLB_REQ_TIMER_MASK, x) -#define BW_BUDDY1_PAGE_MASK _MMIO(0x45144) -#define BW_BUDDY2_PAGE_MASK _MMIO(0x45154) +#define _BW_BUDDY0_PAGE_MASK 0x45134 +#define _BW_BUDDY1_PAGE_MASK 0x45144 +#define BW_BUDDY_PAGE_MASK(x) _MMIO(_PICK_EVEN(x, \ + _BW_BUDDY0_PAGE_MASK, \ + _BW_BUDDY1_PAGE_MASK)) #define HSW_NDE_RSTWRN_OPT _MMIO(0x46408) #define RESET_PCH_HANDSHAKE_ENABLE (1 << 4) @@ -8002,6 +8030,8 @@ enum { #define PER_PIXEL_ALPHA_BYPASS_EN (1 << 7) #define FF_MODE2 _MMIO(0x6604) +#define FF_MODE2_GS_TIMER_MASK REG_GENMASK(31, 24) +#define FF_MODE2_GS_TIMER_224 REG_FIELD_PREP(FF_MODE2_GS_TIMER_MASK, 224) #define FF_MODE2_TDS_TIMER_MASK REG_GENMASK(23, 16) #define FF_MODE2_TDS_TIMER_128 REG_FIELD_PREP(FF_MODE2_TDS_TIMER_MASK, 4) diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c index def62100e666..3bb7320249ae 100644 --- a/drivers/gpu/drm/i915/i915_request.c +++ b/drivers/gpu/drm/i915/i915_request.c @@ -42,7 +42,6 @@ #include "intel_pm.h" struct execute_cb { - struct list_head link; struct irq_work work; struct i915_sw_fence *fence; void (*hook)(struct i915_request *rq, struct dma_fence *signal); @@ -57,7 +56,7 @@ static struct i915_global_request { static const char *i915_fence_get_driver_name(struct dma_fence *fence) { - return dev_name(to_request(fence)->i915->drm.dev); + return dev_name(to_request(fence)->engine->i915->drm.dev); } static const char *i915_fence_get_timeline_name(struct dma_fence *fence) @@ -189,14 +188,15 @@ static void irq_execute_cb_hook(struct irq_work *wrk) static void __notify_execute_cb(struct i915_request *rq) { - struct execute_cb *cb; + struct execute_cb *cb, *cn; lockdep_assert_held(&rq->lock); - if (list_empty(&rq->execute_cb)) + GEM_BUG_ON(!i915_request_is_active(rq)); + if (llist_empty(&rq->execute_cb)) return; - list_for_each_entry(cb, &rq->execute_cb, link) + llist_for_each_entry_safe(cb, cn, rq->execute_cb.first, work.llnode) irq_work_queue(&cb->work); /* @@ -209,7 +209,7 @@ static void __notify_execute_cb(struct i915_request *rq) * preempt-to-idle cycle on the target engine, all the while the * master execute_cb may refire. */ - INIT_LIST_HEAD(&rq->execute_cb); + init_llist_head(&rq->execute_cb); } static inline void @@ -327,7 +327,7 @@ bool i915_request_retire(struct i915_request *rq) set_bit(I915_FENCE_FLAG_ACTIVE, &rq->fence.flags); __notify_execute_cb(rq); } - GEM_BUG_ON(!list_empty(&rq->execute_cb)); + GEM_BUG_ON(!llist_empty(&rq->execute_cb)); spin_unlock_irq(&rq->lock); remove_from_client(rq); @@ -357,6 +357,12 @@ void i915_request_retire_upto(struct i915_request *rq) } while (i915_request_retire(tmp) && tmp != rq); } +static void __llist_add(struct llist_node *node, struct llist_head *head) +{ + node->next = head->first; + head->first = node; +} + static struct i915_request * const * __engine_active(struct intel_engine_cs *engine) { @@ -442,7 +448,7 @@ __await_execution(struct i915_request *rq, i915_sw_fence_complete(cb->fence); kmem_cache_free(global.slab_execute_cbs, cb); } else { - list_add_tail(&cb->link, &signal->execute_cb); + __llist_add(&cb->work.llnode, &signal->execute_cb); } spin_unlock_irq(&signal->lock); @@ -560,15 +566,15 @@ xfer: /* We may be recursing from the signal callback of another i915 fence */ if (!test_and_set_bit(I915_FENCE_FLAG_ACTIVE, &request->fence.flags)) { list_move_tail(&request->sched.link, &engine->active.requests); clear_bit(I915_FENCE_FLAG_PQUEUE, &request->fence.flags); + __notify_execute_cb(request); } + GEM_BUG_ON(!llist_empty(&request->execute_cb)); if (test_bit(DMA_FENCE_FLAG_ENABLE_SIGNAL_BIT, &request->fence.flags) && !test_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &request->fence.flags) && !i915_request_enable_breadcrumb(request)) intel_engine_signal_breadcrumbs(engine); - __notify_execute_cb(request); - spin_unlock(&request->lock); return result; @@ -751,7 +757,7 @@ static void __i915_request_ctor(void *arg) rq->file_priv = NULL; rq->capture_list = NULL; - INIT_LIST_HEAD(&rq->execute_cb); + init_llist_head(&rq->execute_cb); } struct i915_request * @@ -806,7 +812,6 @@ __i915_request_create(struct intel_context *ce, gfp_t gfp) } } - rq->i915 = ce->engine->i915; rq->context = ce; rq->engine = ce->engine; rq->ring = ce->ring; @@ -841,7 +846,7 @@ __i915_request_create(struct intel_context *ce, gfp_t gfp) rq->batch = NULL; GEM_BUG_ON(rq->file_priv); GEM_BUG_ON(rq->capture_list); - GEM_BUG_ON(!list_empty(&rq->execute_cb)); + GEM_BUG_ON(!llist_empty(&rq->execute_cb)); /* * Reserve space in the ring buffer for all the commands required to @@ -1005,12 +1010,12 @@ __emit_semaphore_wait(struct i915_request *to, struct i915_request *from, u32 seqno) { - const int has_token = INTEL_GEN(to->i915) >= 12; + const int has_token = INTEL_GEN(to->engine->i915) >= 12; u32 hwsp_offset; int len, err; u32 *cs; - GEM_BUG_ON(INTEL_GEN(to->i915) < 8); + GEM_BUG_ON(INTEL_GEN(to->engine->i915) < 8); GEM_BUG_ON(i915_request_has_initial_breadcrumb(to)); /* We need to pin the signaler's HWSP until we are finished reading. */ @@ -1205,7 +1210,7 @@ __i915_request_await_external(struct i915_request *rq, struct dma_fence *fence) { mark_external(rq); return i915_sw_fence_await_dma_fence(&rq->submit, fence, - i915_fence_context_timeout(rq->i915, + i915_fence_context_timeout(rq->engine->i915, fence->context), I915_FENCE_GFP); } @@ -1776,7 +1781,8 @@ long i915_request_wait(struct i915_request *rq, * (bad for battery). */ if (flags & I915_WAIT_PRIORITY) { - if (!i915_request_started(rq) && INTEL_GEN(rq->i915) >= 6) + if (!i915_request_started(rq) && + INTEL_GEN(rq->engine->i915) >= 6) intel_rps_boost(rq); } diff --git a/drivers/gpu/drm/i915/i915_request.h b/drivers/gpu/drm/i915/i915_request.h index 8ec7ee4dbadc..590762820761 100644 --- a/drivers/gpu/drm/i915/i915_request.h +++ b/drivers/gpu/drm/i915/i915_request.h @@ -162,9 +162,6 @@ struct i915_request { struct dma_fence fence; spinlock_t lock; - /** On Which ring this request was generated */ - struct drm_i915_private *i915; - /** * Context and ring buffer related to this request * Contexts are refcounted, so when this request is associated with a @@ -214,7 +211,7 @@ struct i915_request { ktime_t emitted; } duration; }; - struct list_head execute_cb; + struct llist_head execute_cb; struct i915_sw_fence semaphore; /* @@ -564,7 +561,7 @@ static inline void i915_request_clear_hold(struct i915_request *rq) } static inline struct intel_timeline * -i915_request_timeline(struct i915_request *rq) +i915_request_timeline(const struct i915_request *rq) { /* Valid only while the request is being constructed (or retired). */ return rcu_dereference_protected(rq->timeline, @@ -572,14 +569,14 @@ i915_request_timeline(struct i915_request *rq) } static inline struct i915_gem_context * -i915_request_gem_context(struct i915_request *rq) +i915_request_gem_context(const struct i915_request *rq) { /* Valid only while the request is being constructed (or retired). */ return rcu_dereference_protected(rq->context->gem_context, true); } static inline struct intel_timeline * -i915_request_active_timeline(struct i915_request *rq) +i915_request_active_timeline(const struct i915_request *rq) { /* * When in use during submission, we are protected by a guarantee that diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h index bc854ad60954..a4addcc64978 100644 --- a/drivers/gpu/drm/i915/i915_trace.h +++ b/drivers/gpu/drm/i915/i915_trace.h @@ -735,7 +735,7 @@ TRACE_EVENT(i915_request_queue, ), TP_fast_assign( - __entry->dev = rq->i915->drm.primary->index; + __entry->dev = rq->engine->i915->drm.primary->index; __entry->class = rq->engine->uabi_class; __entry->instance = rq->engine->uabi_instance; __entry->ctx = rq->fence.context; @@ -761,7 +761,7 @@ DECLARE_EVENT_CLASS(i915_request, ), TP_fast_assign( - __entry->dev = rq->i915->drm.primary->index; + __entry->dev = rq->engine->i915->drm.primary->index; __entry->class = rq->engine->uabi_class; __entry->instance = rq->engine->uabi_instance; __entry->ctx = rq->fence.context; @@ -804,7 +804,7 @@ TRACE_EVENT(i915_request_in, ), TP_fast_assign( - __entry->dev = rq->i915->drm.primary->index; + __entry->dev = rq->engine->i915->drm.primary->index; __entry->class = rq->engine->uabi_class; __entry->instance = rq->engine->uabi_instance; __entry->ctx = rq->fence.context; @@ -833,7 +833,7 @@ TRACE_EVENT(i915_request_out, ), TP_fast_assign( - __entry->dev = rq->i915->drm.primary->index; + __entry->dev = rq->engine->i915->drm.primary->index; __entry->class = rq->engine->uabi_class; __entry->instance = rq->engine->uabi_instance; __entry->ctx = rq->fence.context; @@ -895,7 +895,7 @@ TRACE_EVENT(i915_request_wait_begin, * less desirable. */ TP_fast_assign( - __entry->dev = rq->i915->drm.primary->index; + __entry->dev = rq->engine->i915->drm.primary->index; __entry->class = rq->engine->uabi_class; __entry->instance = rq->engine->uabi_instance; __entry->ctx = rq->fence.context; diff --git a/drivers/gpu/drm/i915/i915_utils.c b/drivers/gpu/drm/i915/i915_utils.c index e28eae4a8f70..f42a9e9a0b4f 100644 --- a/drivers/gpu/drm/i915/i915_utils.c +++ b/drivers/gpu/drm/i915/i915_utils.c @@ -91,7 +91,7 @@ void set_timer_ms(struct timer_list *t, unsigned long timeout) return; } - timeout = msecs_to_jiffies_timeout(timeout); + timeout = msecs_to_jiffies(timeout); /* * Paranoia to make sure the compiler computes the timeout before diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c index fc14ebf9a0b7..1f63c4a1f055 100644 --- a/drivers/gpu/drm/i915/i915_vma.c +++ b/drivers/gpu/drm/i915/i915_vma.c @@ -397,17 +397,15 @@ int i915_vma_bind(struct i915_vma *vma, vma_flags = atomic_read(&vma->flags); vma_flags &= I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND; - if (flags & PIN_UPDATE) - bind_flags |= vma_flags; - else - bind_flags &= ~vma_flags; + + bind_flags &= ~vma_flags; if (bind_flags == 0) return 0; GEM_BUG_ON(!vma->pages); trace_i915_vma_bind(vma, bind_flags); - if (work && (bind_flags & ~vma_flags) & vma->vm->bind_async_flags) { + if (work && bind_flags & vma->vm->bind_async_flags) { struct dma_fence *prev; work->vma = vma; @@ -868,7 +866,6 @@ int i915_vma_pin(struct i915_vma *vma, u64 size, u64 alignment, u64 flags) BUILD_BUG_ON(PIN_GLOBAL != I915_VMA_GLOBAL_BIND); BUILD_BUG_ON(PIN_USER != I915_VMA_LOCAL_BIND); - GEM_BUG_ON(flags & PIN_UPDATE); GEM_BUG_ON(!(flags & (PIN_USER | PIN_GLOBAL))); /* First try and grab the pin without rebinding the vma */ @@ -1090,7 +1087,8 @@ void i915_vma_release(struct kref *ref) spin_lock(&obj->vma.lock); list_del(&vma->obj_link); - rb_erase(&vma->obj_node, &obj->vma.tree); + if (!RB_EMPTY_NODE(&vma->obj_node)) + rb_erase(&vma->obj_node, &obj->vma.tree); spin_unlock(&obj->vma.lock); } @@ -1232,31 +1230,9 @@ int i915_vma_move_to_active(struct i915_vma *vma, return 0; } -int __i915_vma_unbind(struct i915_vma *vma) +void __i915_vma_evict(struct i915_vma *vma) { - int ret; - - lockdep_assert_held(&vma->vm->mutex); - - if (i915_vma_is_pinned(vma)) { - vma_print_allocator(vma, "is pinned"); - return -EAGAIN; - } - - /* - * After confirming that no one else is pinning this vma, wait for - * any laggards who may have crept in during the wait (through - * a residual pin skipping the vm->mutex) to complete. - */ - ret = i915_vma_sync(vma); - if (ret) - return ret; - - if (!drm_mm_node_allocated(&vma->node)) - return 0; - GEM_BUG_ON(i915_vma_is_pinned(vma)); - GEM_BUG_ON(i915_vma_is_active(vma)); if (i915_vma_is_map_and_fenceable(vma)) { /* Force a pagefault for domain tracking on next user access */ @@ -1295,6 +1271,33 @@ int __i915_vma_unbind(struct i915_vma *vma) i915_vma_detach(vma); vma_unbind_pages(vma); +} + +int __i915_vma_unbind(struct i915_vma *vma) +{ + int ret; + + lockdep_assert_held(&vma->vm->mutex); + + if (!drm_mm_node_allocated(&vma->node)) + return 0; + + if (i915_vma_is_pinned(vma)) { + vma_print_allocator(vma, "is pinned"); + return -EAGAIN; + } + + /* + * After confirming that no one else is pinning this vma, wait for + * any laggards who may have crept in during the wait (through + * a residual pin skipping the vm->mutex) to complete. + */ + ret = i915_vma_sync(vma); + if (ret) + return ret; + + GEM_BUG_ON(i915_vma_is_active(vma)); + __i915_vma_evict(vma); drm_mm_remove_node(&vma->node); /* pairs with i915_vma_release() */ return 0; @@ -1306,13 +1309,13 @@ int i915_vma_unbind(struct i915_vma *vma) intel_wakeref_t wakeref = 0; int err; - if (!drm_mm_node_allocated(&vma->node)) - return 0; - /* Optimistic wait before taking the mutex */ err = i915_vma_sync(vma); if (err) - goto out_rpm; + return err; + + if (!drm_mm_node_allocated(&vma->node)) + return 0; if (i915_vma_is_pinned(vma)) { vma_print_allocator(vma, "is pinned"); diff --git a/drivers/gpu/drm/i915/i915_vma.h b/drivers/gpu/drm/i915/i915_vma.h index 8ad1daabcd58..d0d01f909548 100644 --- a/drivers/gpu/drm/i915/i915_vma.h +++ b/drivers/gpu/drm/i915/i915_vma.h @@ -203,6 +203,7 @@ bool i915_vma_misplaced(const struct i915_vma *vma, u64 size, u64 alignment, u64 flags); void __i915_vma_set_map_and_fenceable(struct i915_vma *vma); void i915_vma_revoke_mmap(struct i915_vma *vma); +void __i915_vma_evict(struct i915_vma *vma); int __i915_vma_unbind(struct i915_vma *vma); int __must_check i915_vma_unbind(struct i915_vma *vma); void i915_vma_unlink_ctx(struct i915_vma *vma); diff --git a/drivers/gpu/drm/i915/intel_device_info.c b/drivers/gpu/drm/i915/intel_device_info.c index 8a635bd4d5d8..544ac61fbc36 100644 --- a/drivers/gpu/drm/i915/intel_device_info.c +++ b/drivers/gpu/drm/i915/intel_device_info.c @@ -57,10 +57,12 @@ static const char * const platform_names[] = { PLATFORM_NAME(KABYLAKE), PLATFORM_NAME(GEMINILAKE), PLATFORM_NAME(COFFEELAKE), + PLATFORM_NAME(COMETLAKE), PLATFORM_NAME(CANNONLAKE), PLATFORM_NAME(ICELAKE), PLATFORM_NAME(ELKHARTLAKE), PLATFORM_NAME(TIGERLAKE), + PLATFORM_NAME(ROCKETLAKE), }; #undef PLATFORM_NAME @@ -933,7 +935,10 @@ void intel_device_info_runtime_init(struct drm_i915_private *dev_priv) BUILD_BUG_ON(BITS_PER_TYPE(intel_engine_mask_t) < I915_NUM_ENGINES); - if (INTEL_GEN(dev_priv) >= 11) + if (IS_ROCKETLAKE(dev_priv)) + for_each_pipe(dev_priv, pipe) + runtime->num_sprites[pipe] = 4; + else if (INTEL_GEN(dev_priv) >= 11) for_each_pipe(dev_priv, pipe) runtime->num_sprites[pipe] = 6; else if (IS_GEN(dev_priv, 10) || IS_GEMINILAKE(dev_priv)) diff --git a/drivers/gpu/drm/i915/intel_device_info.h b/drivers/gpu/drm/i915/intel_device_info.h index 62e03ffa377e..8d62b8538585 100644 --- a/drivers/gpu/drm/i915/intel_device_info.h +++ b/drivers/gpu/drm/i915/intel_device_info.h @@ -73,6 +73,7 @@ enum intel_platform { INTEL_KABYLAKE, INTEL_GEMINILAKE, INTEL_COFFEELAKE, + INTEL_COMETLAKE, /* gen10 */ INTEL_CANNONLAKE, /* gen11 */ @@ -80,6 +81,7 @@ enum intel_platform { INTEL_ELKHARTLAKE, /* gen12 */ INTEL_TIGERLAKE, + INTEL_ROCKETLAKE, INTEL_MAX_PLATFORMS }; @@ -146,6 +148,7 @@ enum intel_ppgtt_type { func(has_modular_fia); \ func(has_overlay); \ func(has_psr); \ + func(has_psr_hw_tracking); \ func(overlay_needs_physical); \ func(supports_tv); @@ -172,6 +175,8 @@ struct intel_device_info { u8 pipe_mask; u8 cpu_transcoder_mask; + u8 abox_mask; + #define DEFINE_FLAG(name) u8 name:1 DEV_INFO_FOR_EACH_FLAG(DEFINE_FLAG); #undef DEFINE_FLAG diff --git a/drivers/gpu/drm/i915/intel_gvt.c b/drivers/gpu/drm/i915/intel_gvt.c index 21b91313cc5d..99fe8aef1c67 100644 --- a/drivers/gpu/drm/i915/intel_gvt.c +++ b/drivers/gpu/drm/i915/intel_gvt.c @@ -52,6 +52,8 @@ static bool is_supported_device(struct drm_i915_private *dev_priv) return true; if (IS_COFFEELAKE(dev_priv)) return true; + if (IS_COMETLAKE(dev_priv)) + return true; return false; } @@ -64,7 +66,7 @@ static bool is_supported_device(struct drm_i915_private *dev_priv) */ void intel_gvt_sanitize_options(struct drm_i915_private *dev_priv) { - if (!i915_modparams.enable_gvt) + if (!dev_priv->params.enable_gvt) return; if (intel_vgpu_active(dev_priv)) { @@ -80,7 +82,7 @@ void intel_gvt_sanitize_options(struct drm_i915_private *dev_priv) return; bail: - i915_modparams.enable_gvt = 0; + dev_priv->params.enable_gvt = 0; } /** @@ -100,7 +102,7 @@ int intel_gvt_init(struct drm_i915_private *dev_priv) if (i915_inject_probe_failure(dev_priv)) return -ENODEV; - if (!i915_modparams.enable_gvt) { + if (!dev_priv->params.enable_gvt) { drm_dbg(&dev_priv->drm, "GVT-g is disabled by kernel params\n"); return 0; @@ -121,7 +123,7 @@ int intel_gvt_init(struct drm_i915_private *dev_priv) return 0; bail: - i915_modparams.enable_gvt = 0; + dev_priv->params.enable_gvt = 0; return 0; } diff --git a/drivers/gpu/drm/i915/intel_pch.c b/drivers/gpu/drm/i915/intel_pch.c index 20ab9a5023b5..c668e99eb2e4 100644 --- a/drivers/gpu/drm/i915/intel_pch.c +++ b/drivers/gpu/drm/i915/intel_pch.c @@ -64,36 +64,49 @@ intel_pch_type(const struct drm_i915_private *dev_priv, unsigned short id) case INTEL_PCH_SPT_LP_DEVICE_ID_TYPE: drm_dbg_kms(&dev_priv->drm, "Found SunrisePoint LP PCH\n"); drm_WARN_ON(&dev_priv->drm, - !IS_SKYLAKE(dev_priv) && !IS_KABYLAKE(dev_priv) && - !IS_COFFEELAKE(dev_priv)); + !IS_SKYLAKE(dev_priv) && + !IS_KABYLAKE(dev_priv) && + !IS_COFFEELAKE(dev_priv) && + !IS_COMETLAKE(dev_priv)); return PCH_SPT; case INTEL_PCH_KBP_DEVICE_ID_TYPE: drm_dbg_kms(&dev_priv->drm, "Found Kaby Lake PCH (KBP)\n"); drm_WARN_ON(&dev_priv->drm, - !IS_SKYLAKE(dev_priv) && !IS_KABYLAKE(dev_priv) && - !IS_COFFEELAKE(dev_priv)); + !IS_SKYLAKE(dev_priv) && + !IS_KABYLAKE(dev_priv) && + !IS_COFFEELAKE(dev_priv) && + !IS_COMETLAKE(dev_priv)); /* KBP is SPT compatible */ return PCH_SPT; case INTEL_PCH_CNP_DEVICE_ID_TYPE: drm_dbg_kms(&dev_priv->drm, "Found Cannon Lake PCH (CNP)\n"); - drm_WARN_ON(&dev_priv->drm, !IS_CANNONLAKE(dev_priv) && - !IS_COFFEELAKE(dev_priv)); + drm_WARN_ON(&dev_priv->drm, + !IS_CANNONLAKE(dev_priv) && + !IS_COFFEELAKE(dev_priv) && + !IS_COMETLAKE(dev_priv)); return PCH_CNP; case INTEL_PCH_CNP_LP_DEVICE_ID_TYPE: drm_dbg_kms(&dev_priv->drm, "Found Cannon Lake LP PCH (CNP-LP)\n"); - drm_WARN_ON(&dev_priv->drm, !IS_CANNONLAKE(dev_priv) && - !IS_COFFEELAKE(dev_priv)); + drm_WARN_ON(&dev_priv->drm, + !IS_CANNONLAKE(dev_priv) && + !IS_COFFEELAKE(dev_priv) && + !IS_COMETLAKE(dev_priv)); return PCH_CNP; case INTEL_PCH_CMP_DEVICE_ID_TYPE: case INTEL_PCH_CMP2_DEVICE_ID_TYPE: drm_dbg_kms(&dev_priv->drm, "Found Comet Lake PCH (CMP)\n"); - drm_WARN_ON(&dev_priv->drm, !IS_COFFEELAKE(dev_priv)); + drm_WARN_ON(&dev_priv->drm, + !IS_COFFEELAKE(dev_priv) && + !IS_COMETLAKE(dev_priv) && + !IS_ROCKETLAKE(dev_priv)); /* CometPoint is CNP Compatible */ return PCH_CNP; case INTEL_PCH_CMP_V_DEVICE_ID_TYPE: drm_dbg_kms(&dev_priv->drm, "Found Comet Lake V PCH (CMP-V)\n"); - drm_WARN_ON(&dev_priv->drm, !IS_COFFEELAKE(dev_priv)); + drm_WARN_ON(&dev_priv->drm, + !IS_COFFEELAKE(dev_priv) && + !IS_COMETLAKE(dev_priv)); /* Comet Lake V PCH is based on KBP, which is SPT compatible */ return PCH_SPT; case INTEL_PCH_ICP_DEVICE_ID_TYPE: @@ -107,7 +120,8 @@ intel_pch_type(const struct drm_i915_private *dev_priv, unsigned short id) case INTEL_PCH_TGP_DEVICE_ID_TYPE: case INTEL_PCH_TGP2_DEVICE_ID_TYPE: drm_dbg_kms(&dev_priv->drm, "Found Tiger Lake LP PCH\n"); - drm_WARN_ON(&dev_priv->drm, !IS_TIGERLAKE(dev_priv)); + drm_WARN_ON(&dev_priv->drm, !IS_TIGERLAKE(dev_priv) && + !IS_ROCKETLAKE(dev_priv)); return PCH_TGP; case INTEL_PCH_JSP_DEVICE_ID_TYPE: case INTEL_PCH_JSP2_DEVICE_ID_TYPE: @@ -141,13 +155,15 @@ intel_virt_detect_pch(const struct drm_i915_private *dev_priv) * make an educated guess as to which PCH is really there. */ - if (IS_TIGERLAKE(dev_priv)) + if (IS_TIGERLAKE(dev_priv) || IS_ROCKETLAKE(dev_priv)) id = INTEL_PCH_TGP_DEVICE_ID_TYPE; else if (IS_ELKHARTLAKE(dev_priv)) id = INTEL_PCH_MCC_DEVICE_ID_TYPE; else if (IS_ICELAKE(dev_priv)) id = INTEL_PCH_ICP_DEVICE_ID_TYPE; - else if (IS_CANNONLAKE(dev_priv) || IS_COFFEELAKE(dev_priv)) + else if (IS_CANNONLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) id = INTEL_PCH_CNP_DEVICE_ID_TYPE; else if (IS_KABYLAKE(dev_priv) || IS_SKYLAKE(dev_priv)) id = INTEL_PCH_SPT_DEVICE_ID_TYPE; diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 07f663cd2d1c..2a32d6230795 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -33,6 +33,7 @@ #include <drm/drm_plane_helper.h> #include "display/intel_atomic.h" +#include "display/intel_bw.h" #include "display/intel_display_types.h" #include "display/intel_fbc.h" #include "display/intel_sprite.h" @@ -43,7 +44,6 @@ #include "i915_fixed.h" #include "i915_irq.h" #include "i915_trace.h" -#include "display/intel_bw.h" #include "intel_pm.h" #include "intel_sideband.h" #include "../../../platform/x86/intel_ips.h" @@ -1437,6 +1437,7 @@ static int g4x_compute_pipe_wm(struct intel_crtc_state *crtc_state) static int g4x_compute_intermediate_wm(struct intel_crtc_state *new_crtc_state) { struct intel_crtc *crtc = to_intel_crtc(new_crtc_state->uapi.crtc); + struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); struct g4x_wm_state *intermediate = &new_crtc_state->wm.g4x.intermediate; const struct g4x_wm_state *optimal = &new_crtc_state->wm.g4x.optimal; struct intel_atomic_state *intel_state = @@ -1465,8 +1466,8 @@ static int g4x_compute_intermediate_wm(struct intel_crtc_state *new_crtc_state) max(optimal->wm.plane[plane_id], active->wm.plane[plane_id]); - WARN_ON(intermediate->wm.plane[plane_id] > - g4x_plane_fifo_size(plane_id, G4X_WM_LEVEL_NORMAL)); + drm_WARN_ON(&dev_priv->drm, intermediate->wm.plane[plane_id] > + g4x_plane_fifo_size(plane_id, G4X_WM_LEVEL_NORMAL)); } intermediate->sr.plane = max(optimal->sr.plane, @@ -1483,21 +1484,25 @@ static int g4x_compute_intermediate_wm(struct intel_crtc_state *new_crtc_state) intermediate->hpll.fbc = max(optimal->hpll.fbc, active->hpll.fbc); - WARN_ON((intermediate->sr.plane > - g4x_plane_fifo_size(PLANE_PRIMARY, G4X_WM_LEVEL_SR) || - intermediate->sr.cursor > - g4x_plane_fifo_size(PLANE_CURSOR, G4X_WM_LEVEL_SR)) && - intermediate->cxsr); - WARN_ON((intermediate->sr.plane > - g4x_plane_fifo_size(PLANE_PRIMARY, G4X_WM_LEVEL_HPLL) || - intermediate->sr.cursor > - g4x_plane_fifo_size(PLANE_CURSOR, G4X_WM_LEVEL_HPLL)) && - intermediate->hpll_en); - - WARN_ON(intermediate->sr.fbc > g4x_fbc_fifo_size(1) && - intermediate->fbc_en && intermediate->cxsr); - WARN_ON(intermediate->hpll.fbc > g4x_fbc_fifo_size(2) && - intermediate->fbc_en && intermediate->hpll_en); + drm_WARN_ON(&dev_priv->drm, + (intermediate->sr.plane > + g4x_plane_fifo_size(PLANE_PRIMARY, G4X_WM_LEVEL_SR) || + intermediate->sr.cursor > + g4x_plane_fifo_size(PLANE_CURSOR, G4X_WM_LEVEL_SR)) && + intermediate->cxsr); + drm_WARN_ON(&dev_priv->drm, + (intermediate->sr.plane > + g4x_plane_fifo_size(PLANE_PRIMARY, G4X_WM_LEVEL_HPLL) || + intermediate->sr.cursor > + g4x_plane_fifo_size(PLANE_CURSOR, G4X_WM_LEVEL_HPLL)) && + intermediate->hpll_en); + + drm_WARN_ON(&dev_priv->drm, + intermediate->sr.fbc > g4x_fbc_fifo_size(1) && + intermediate->fbc_en && intermediate->cxsr); + drm_WARN_ON(&dev_priv->drm, + intermediate->hpll.fbc > g4x_fbc_fifo_size(2) && + intermediate->fbc_en && intermediate->hpll_en); out: /* @@ -1681,6 +1686,7 @@ static bool vlv_need_sprite0_fifo_workaround(unsigned int active_planes) static int vlv_compute_fifo(struct intel_crtc_state *crtc_state) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); + struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); const struct g4x_pipe_wm *raw = &crtc_state->wm.vlv.raw[VLV_WM_LEVEL_PM2]; struct vlv_fifo_state *fifo_state = &crtc_state->wm.vlv.fifo_state; @@ -1749,11 +1755,11 @@ static int vlv_compute_fifo(struct intel_crtc_state *crtc_state) fifo_left -= plane_extra; } - WARN_ON(active_planes != 0 && fifo_left != 0); + drm_WARN_ON(&dev_priv->drm, active_planes != 0 && fifo_left != 0); /* give it all to the first plane if none are active */ if (active_planes == 0) { - WARN_ON(fifo_left != fifo_size); + drm_WARN_ON(&dev_priv->drm, fifo_left != fifo_size); fifo_state->plane[PLANE_PRIMARY] = fifo_left; } @@ -4025,10 +4031,9 @@ icl_get_first_dbuf_slice_offset(u32 dbuf_slice_mask, return offset; } -static u16 intel_get_ddb_size(struct drm_i915_private *dev_priv) +u16 intel_get_ddb_size(struct drm_i915_private *dev_priv) { u16 ddb_size = INTEL_INFO(dev_priv)->ddb_size; - drm_WARN_ON(&dev_priv->drm, ddb_size == 0); if (INTEL_GEN(dev_priv) < 11) @@ -4037,10 +4042,38 @@ static u16 intel_get_ddb_size(struct drm_i915_private *dev_priv) return ddb_size; } +u32 skl_ddb_dbuf_slice_mask(struct drm_i915_private *dev_priv, + const struct skl_ddb_entry *entry) +{ + u32 slice_mask = 0; + u16 ddb_size = intel_get_ddb_size(dev_priv); + u16 num_supported_slices = INTEL_INFO(dev_priv)->num_supported_dbuf_slices; + u16 slice_size = ddb_size / num_supported_slices; + u16 start_slice; + u16 end_slice; + + if (!skl_ddb_entry_size(entry)) + return 0; + + start_slice = entry->start / slice_size; + end_slice = (entry->end - 1) / slice_size; + + /* + * Per plane DDB entry can in a really worst case be on multiple slices + * but single entry is anyway contigious. + */ + while (start_slice <= end_slice) { + slice_mask |= BIT(start_slice); + start_slice++; + } + + return slice_mask; +} + static u8 skl_compute_dbuf_slices(const struct intel_crtc_state *crtc_state, u8 active_pipes); -static void +static int skl_ddb_get_pipe_allocation_limits(struct drm_i915_private *dev_priv, const struct intel_crtc_state *crtc_state, const u64 total_data_rate, @@ -4053,30 +4086,29 @@ skl_ddb_get_pipe_allocation_limits(struct drm_i915_private *dev_priv, const struct intel_crtc *crtc; u32 pipe_width = 0, total_width_in_range = 0, width_before_pipe_in_range = 0; enum pipe for_pipe = to_intel_crtc(for_crtc)->pipe; + struct intel_dbuf_state *new_dbuf_state = + intel_atomic_get_new_dbuf_state(intel_state); + const struct intel_dbuf_state *old_dbuf_state = + intel_atomic_get_old_dbuf_state(intel_state); + u8 active_pipes = new_dbuf_state->active_pipes; u16 ddb_size; u32 ddb_range_size; u32 i; u32 dbuf_slice_mask; - u32 active_pipes; u32 offset; u32 slice_size; u32 total_slice_mask; u32 start, end; + int ret; + + *num_active = hweight8(active_pipes); - if (drm_WARN_ON(&dev_priv->drm, !state) || !crtc_state->hw.active) { + if (!crtc_state->hw.active) { alloc->start = 0; alloc->end = 0; - *num_active = hweight8(dev_priv->active_pipes); - return; + return 0; } - if (intel_state->active_pipe_changes) - active_pipes = intel_state->active_pipes; - else - active_pipes = dev_priv->active_pipes; - - *num_active = hweight8(active_pipes); - ddb_size = intel_get_ddb_size(dev_priv); slice_size = ddb_size / INTEL_INFO(dev_priv)->num_supported_dbuf_slices; @@ -4089,13 +4121,16 @@ skl_ddb_get_pipe_allocation_limits(struct drm_i915_private *dev_priv, * that changes the active CRTC list or do modeset would need to * grab _all_ crtc locks, including the one we currently hold. */ - if (!intel_state->active_pipe_changes && !intel_state->modeset) { + if (old_dbuf_state->active_pipes == new_dbuf_state->active_pipes && + !dev_priv->wm.distrust_bios_wm) { /* * alloc may be cleared by clear_intel_crtc_state, * copy from old state to be sure + * + * FIXME get rid of this mess */ *alloc = to_intel_crtc_state(for_crtc->state)->wm.skl.ddb; - return; + return 0; } /* @@ -4103,10 +4138,6 @@ skl_ddb_get_pipe_allocation_limits(struct drm_i915_private *dev_priv, */ dbuf_slice_mask = skl_compute_dbuf_slices(crtc_state, active_pipes); - DRM_DEBUG_KMS("DBuf slice mask %x pipe %c active pipes %x\n", - dbuf_slice_mask, - pipe_name(for_pipe), active_pipes); - /* * Figure out at which DBuf slice we start, i.e if we start at Dbuf S2 * and slice size is 1024, the offset would be 1024 @@ -4174,7 +4205,13 @@ skl_ddb_get_pipe_allocation_limits(struct drm_i915_private *dev_priv, * FIXME: For now we always enable slice S1 as per * the Bspec display initialization sequence. */ - intel_state->enabled_dbuf_slices_mask = total_slice_mask | BIT(DBUF_S1); + new_dbuf_state->enabled_slices = total_slice_mask | BIT(DBUF_S1); + + if (old_dbuf_state->enabled_slices != new_dbuf_state->enabled_slices) { + ret = intel_atomic_serialize_global_state(&new_dbuf_state->base); + if (ret) + return ret; + } start = ddb_range_size * width_before_pipe_in_range / total_width_in_range; end = ddb_range_size * @@ -4183,11 +4220,12 @@ skl_ddb_get_pipe_allocation_limits(struct drm_i915_private *dev_priv, alloc->start = offset + start; alloc->end = offset + end; - DRM_DEBUG_KMS("Pipe %d ddb %d-%d\n", for_pipe, - alloc->start, alloc->end); - DRM_DEBUG_KMS("Enabled ddb slices mask %x num supported %d\n", - intel_state->enabled_dbuf_slices_mask, - INTEL_INFO(dev_priv)->num_supported_dbuf_slices); + drm_dbg_kms(&dev_priv->drm, + "[CRTC:%d:%s] dbuf slices 0x%x, ddb (%d - %d), active pipes 0x%x\n", + for_crtc->base.id, for_crtc->name, + dbuf_slice_mask, alloc->start, alloc->end, active_pipes); + + return 0; } static int skl_compute_wm_params(const struct intel_crtc_state *crtc_state, @@ -4308,12 +4346,6 @@ void skl_pipe_ddb_get_hw_state(struct intel_crtc *crtc, intel_display_power_put(dev_priv, power_domain, wakeref); } -void skl_ddb_get_hw_state(struct drm_i915_private *dev_priv) -{ - dev_priv->enabled_dbuf_slices_mask = - intel_enabled_dbuf_slices_mask(dev_priv); -} - /* * Determines the downscale amount of a plane for the purposes of watermark calculations. * The bspec defines downscale amount as: @@ -4334,11 +4366,13 @@ static uint_fixed_16_16_t skl_plane_downscale_amount(const struct intel_crtc_state *crtc_state, const struct intel_plane_state *plane_state) { + struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); u32 src_w, src_h, dst_w, dst_h; uint_fixed_16_16_t fp_w_ratio, fp_h_ratio; uint_fixed_16_16_t downscale_h, downscale_w; - if (WARN_ON(!intel_wm_plane_visible(crtc_state, plane_state))) + if (drm_WARN_ON(&dev_priv->drm, + !intel_wm_plane_visible(crtc_state, plane_state))) return u32_to_fixed16(0); /* @@ -4606,7 +4640,7 @@ static u8 skl_compute_dbuf_slices(const struct intel_crtc_state *crtc_state, * For anything else just return one slice yet. * Should be extended for other platforms. */ - return BIT(DBUF_S1); + return active_pipes & BIT(pipe) ? BIT(DBUF_S1) : 0; } static u64 @@ -4758,12 +4792,37 @@ skl_allocate_pipe_ddb(struct intel_crtc_state *crtc_state) u64 uv_plane_data_rate[I915_MAX_PLANES] = {}; u32 blocks; int level; + int ret; /* Clear the partitioning for disabled planes. */ memset(crtc_state->wm.skl.plane_ddb_y, 0, sizeof(crtc_state->wm.skl.plane_ddb_y)); memset(crtc_state->wm.skl.plane_ddb_uv, 0, sizeof(crtc_state->wm.skl.plane_ddb_uv)); if (!crtc_state->hw.active) { + struct intel_atomic_state *state = + to_intel_atomic_state(crtc_state->uapi.state); + struct intel_dbuf_state *new_dbuf_state = + intel_atomic_get_new_dbuf_state(state); + const struct intel_dbuf_state *old_dbuf_state = + intel_atomic_get_old_dbuf_state(state); + + /* + * FIXME hack to make sure we compute this sensibly when + * turning off all the pipes. Otherwise we leave it at + * whatever we had previously, and then runtime PM will + * mess it up by turning off all but S1. Remove this + * once the dbuf state computation flow becomes sane. + */ + if (new_dbuf_state->active_pipes == 0) { + new_dbuf_state->enabled_slices = BIT(DBUF_S1); + + if (old_dbuf_state->enabled_slices != new_dbuf_state->enabled_slices) { + ret = intel_atomic_serialize_global_state(&new_dbuf_state->base); + if (ret) + return ret; + } + } + alloc->start = alloc->end = 0; return 0; } @@ -4778,8 +4837,12 @@ skl_allocate_pipe_ddb(struct intel_crtc_state *crtc_state) plane_data_rate, uv_plane_data_rate); - skl_ddb_get_pipe_allocation_limits(dev_priv, crtc_state, total_data_rate, - alloc, &num_active); + ret = skl_ddb_get_pipe_allocation_limits(dev_priv, crtc_state, + total_data_rate, + alloc, &num_active); + if (ret) + return ret; + alloc_size = skl_ddb_entry_size(alloc); if (alloc_size == 0) return 0; @@ -5003,6 +5066,7 @@ skl_wm_method2(u32 pixel_rate, u32 pipe_htotal, u32 latency, static uint_fixed_16_16_t intel_get_linetime_us(const struct intel_crtc_state *crtc_state) { + struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); u32 pixel_rate; u32 crtc_htotal; uint_fixed_16_16_t linetime_us; @@ -5012,7 +5076,7 @@ intel_get_linetime_us(const struct intel_crtc_state *crtc_state) pixel_rate = crtc_state->pixel_rate; - if (WARN_ON(pixel_rate == 0)) + if (drm_WARN_ON(&dev_priv->drm, pixel_rate == 0)) return u32_to_fixed16(0); crtc_htotal = crtc_state->hw.adjusted_mode.crtc_htotal; @@ -5025,11 +5089,13 @@ static u32 skl_adjusted_plane_pixel_rate(const struct intel_crtc_state *crtc_state, const struct intel_plane_state *plane_state) { + struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); u64 adjusted_pixel_rate; uint_fixed_16_16_t downscale_amount; /* Shouldn't reach here on disabled planes... */ - if (WARN_ON(!intel_wm_plane_visible(crtc_state, plane_state))) + if (drm_WARN_ON(&dev_priv->drm, + !intel_wm_plane_visible(crtc_state, plane_state))) return 0; /* @@ -5190,7 +5256,9 @@ static void skl_compute_plane_wm(const struct intel_crtc_state *crtc_state, * WaIncreaseLatencyIPCEnabled: kbl,cfl * Display WA #1141: kbl,cfl */ - if ((IS_KABYLAKE(dev_priv) || IS_COFFEELAKE(dev_priv)) && + if ((IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) && dev_priv->ipc_enabled) latency += 4; @@ -5465,6 +5533,7 @@ static int skl_build_plane_wm(struct intel_crtc_state *crtc_state, static int icl_build_plane_wm(struct intel_crtc_state *crtc_state, const struct intel_plane_state *plane_state) { + struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); enum plane_id plane_id = to_intel_plane(plane_state->uapi.plane)->id; int ret; @@ -5476,9 +5545,10 @@ static int icl_build_plane_wm(struct intel_crtc_state *crtc_state, const struct drm_framebuffer *fb = plane_state->hw.fb; enum plane_id y_plane_id = plane_state->planar_linked_plane->id; - WARN_ON(!intel_wm_plane_visible(crtc_state, plane_state)); - WARN_ON(!fb->format->is_yuv || - fb->format->num_planes == 1); + drm_WARN_ON(&dev_priv->drm, + !intel_wm_plane_visible(crtc_state, plane_state)); + drm_WARN_ON(&dev_priv->drm, !fb->format->is_yuv || + fb->format->num_planes == 1); ret = skl_build_plane_wm_single(crtc_state, plane_state, y_plane_id, 0); @@ -5701,13 +5771,13 @@ static int skl_compute_ddb(struct intel_atomic_state *state) { struct drm_i915_private *dev_priv = to_i915(state->base.dev); - struct intel_crtc_state *old_crtc_state; + const struct intel_dbuf_state *old_dbuf_state; + const struct intel_dbuf_state *new_dbuf_state; + const struct intel_crtc_state *old_crtc_state; struct intel_crtc_state *new_crtc_state; struct intel_crtc *crtc; int ret, i; - state->enabled_dbuf_slices_mask = dev_priv->enabled_dbuf_slices_mask; - for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) { ret = skl_allocate_pipe_ddb(new_crtc_state); @@ -5720,6 +5790,17 @@ skl_compute_ddb(struct intel_atomic_state *state) return ret; } + old_dbuf_state = intel_atomic_get_old_dbuf_state(state); + new_dbuf_state = intel_atomic_get_new_dbuf_state(state); + + if (new_dbuf_state && + new_dbuf_state->enabled_slices != old_dbuf_state->enabled_slices) + drm_dbg_kms(&dev_priv->drm, + "Enabled dbuf slices 0x%x -> 0x%x (out of %d dbuf slices)\n", + old_dbuf_state->enabled_slices, + new_dbuf_state->enabled_slices, + INTEL_INFO(dev_priv)->num_supported_dbuf_slices); + return 0; } @@ -5855,7 +5936,8 @@ skl_print_wm_changes(struct intel_atomic_state *state) } } -static int intel_add_all_pipes(struct intel_atomic_state *state) +static int intel_add_affected_pipes(struct intel_atomic_state *state, + u8 pipe_mask) { struct drm_i915_private *dev_priv = to_i915(state->base.dev); struct intel_crtc *crtc; @@ -5863,6 +5945,9 @@ static int intel_add_all_pipes(struct intel_atomic_state *state) for_each_intel_crtc(&dev_priv->drm, crtc) { struct intel_crtc_state *crtc_state; + if ((pipe_mask & BIT(crtc->pipe)) == 0) + continue; + crtc_state = intel_atomic_get_crtc_state(&state->base, crtc); if (IS_ERR(crtc_state)) return PTR_ERR(crtc_state); @@ -5875,49 +5960,54 @@ static int skl_ddb_add_affected_pipes(struct intel_atomic_state *state) { struct drm_i915_private *dev_priv = to_i915(state->base.dev); - int ret; + struct intel_crtc_state *crtc_state; + struct intel_crtc *crtc; + int i, ret; - /* - * If this is our first atomic update following hardware readout, - * we can't trust the DDB that the BIOS programmed for us. Let's - * pretend that all pipes switched active status so that we'll - * ensure a full DDB recompute. - */ if (dev_priv->wm.distrust_bios_wm) { - ret = drm_modeset_lock(&dev_priv->drm.mode_config.connection_mutex, - state->base.acquire_ctx); + /* + * skl_ddb_get_pipe_allocation_limits() currently requires + * all active pipes to be included in the state so that + * it can redistribute the dbuf among them, and it really + * wants to recompute things when distrust_bios_wm is set + * so we add all the pipes to the state. + */ + ret = intel_add_affected_pipes(state, ~0); if (ret) return ret; + } + + for_each_new_intel_crtc_in_state(state, crtc, crtc_state, i) { + struct intel_dbuf_state *new_dbuf_state; + const struct intel_dbuf_state *old_dbuf_state; + + new_dbuf_state = intel_atomic_get_dbuf_state(state); + if (IS_ERR(new_dbuf_state)) + return PTR_ERR(new_dbuf_state); + + old_dbuf_state = intel_atomic_get_old_dbuf_state(state); - state->active_pipe_changes = INTEL_INFO(dev_priv)->pipe_mask; + new_dbuf_state->active_pipes = + intel_calc_active_pipes(state, old_dbuf_state->active_pipes); + + if (old_dbuf_state->active_pipes == new_dbuf_state->active_pipes) + break; + + ret = intel_atomic_lock_global_state(&new_dbuf_state->base); + if (ret) + return ret; /* - * We usually only initialize state->active_pipes if we - * we're doing a modeset; make sure this field is always - * initialized during the sanitization process that happens - * on the first commit too. + * skl_ddb_get_pipe_allocation_limits() currently requires + * all active pipes to be included in the state so that + * it can redistribute the dbuf among them. */ - if (!state->modeset) - state->active_pipes = dev_priv->active_pipes; - } - - /* - * If the modeset changes which CRTC's are active, we need to - * recompute the DDB allocation for *all* active pipes, even - * those that weren't otherwise being modified in any way by this - * atomic commit. Due to the shrinking of the per-pipe allocations - * when new active CRTC's are added, it's possible for a pipe that - * we were already using and aren't changing at all here to suddenly - * become invalid if its DDB needs exceeds its new allocation. - * - * Note that if we wind up doing a full DDB recompute, we can't let - * any other display updates race with this transaction, so we need - * to grab the lock on *all* CRTC's. - */ - if (state->active_pipe_changes || state->modeset) { - ret = intel_add_all_pipes(state); + ret = intel_add_affected_pipes(state, + new_dbuf_state->active_pipes); if (ret) return ret; + + break; } return 0; @@ -6163,7 +6253,6 @@ void skl_wm_get_hw_state(struct drm_i915_private *dev_priv) struct intel_crtc *crtc; struct intel_crtc_state *crtc_state; - skl_ddb_get_hw_state(dev_priv); for_each_intel_crtc(&dev_priv->drm, crtc) { crtc_state = to_intel_crtc_state(crtc->base.state); @@ -6735,7 +6824,9 @@ static bool intel_can_enable_ipc(struct drm_i915_private *dev_priv) return false; /* Display WA #1141: SKL:all KBL:all CFL */ - if (IS_KABYLAKE(dev_priv) || IS_COFFEELAKE(dev_priv)) + if (IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv) || + IS_COMETLAKE(dev_priv)) return dev_priv->dram_info.symmetric_memory; return true; @@ -7414,7 +7505,7 @@ void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv) dev_priv->display.init_clock_gating = icl_init_clock_gating; else if (IS_CANNONLAKE(dev_priv)) dev_priv->display.init_clock_gating = cnl_init_clock_gating; - else if (IS_COFFEELAKE(dev_priv)) + else if (IS_COFFEELAKE(dev_priv) || IS_COMETLAKE(dev_priv)) dev_priv->display.init_clock_gating = cfl_init_clock_gating; else if (IS_SKYLAKE(dev_priv)) dev_priv->display.init_clock_gating = skl_init_clock_gating; @@ -7544,3 +7635,89 @@ void intel_pm_setup(struct drm_i915_private *dev_priv) dev_priv->runtime_pm.suspended = false; atomic_set(&dev_priv->runtime_pm.wakeref_count, 0); } + +static struct intel_global_state *intel_dbuf_duplicate_state(struct intel_global_obj *obj) +{ + struct intel_dbuf_state *dbuf_state; + + dbuf_state = kmemdup(obj->state, sizeof(*dbuf_state), GFP_KERNEL); + if (!dbuf_state) + return NULL; + + return &dbuf_state->base; +} + +static void intel_dbuf_destroy_state(struct intel_global_obj *obj, + struct intel_global_state *state) +{ + kfree(state); +} + +static const struct intel_global_state_funcs intel_dbuf_funcs = { + .atomic_duplicate_state = intel_dbuf_duplicate_state, + .atomic_destroy_state = intel_dbuf_destroy_state, +}; + +struct intel_dbuf_state * +intel_atomic_get_dbuf_state(struct intel_atomic_state *state) +{ + struct drm_i915_private *dev_priv = to_i915(state->base.dev); + struct intel_global_state *dbuf_state; + + dbuf_state = intel_atomic_get_global_obj_state(state, &dev_priv->dbuf.obj); + if (IS_ERR(dbuf_state)) + return ERR_CAST(dbuf_state); + + return to_intel_dbuf_state(dbuf_state); +} + +int intel_dbuf_init(struct drm_i915_private *dev_priv) +{ + struct intel_dbuf_state *dbuf_state; + + dbuf_state = kzalloc(sizeof(*dbuf_state), GFP_KERNEL); + if (!dbuf_state) + return -ENOMEM; + + intel_atomic_global_obj_init(dev_priv, &dev_priv->dbuf.obj, + &dbuf_state->base, &intel_dbuf_funcs); + + return 0; +} + +void intel_dbuf_pre_plane_update(struct intel_atomic_state *state) +{ + struct drm_i915_private *dev_priv = to_i915(state->base.dev); + const struct intel_dbuf_state *new_dbuf_state = + intel_atomic_get_new_dbuf_state(state); + const struct intel_dbuf_state *old_dbuf_state = + intel_atomic_get_old_dbuf_state(state); + + if (!new_dbuf_state || + new_dbuf_state->enabled_slices == old_dbuf_state->enabled_slices) + return; + + WARN_ON(!new_dbuf_state->base.changed); + + gen9_dbuf_slices_update(dev_priv, + old_dbuf_state->enabled_slices | + new_dbuf_state->enabled_slices); +} + +void intel_dbuf_post_plane_update(struct intel_atomic_state *state) +{ + struct drm_i915_private *dev_priv = to_i915(state->base.dev); + const struct intel_dbuf_state *new_dbuf_state = + intel_atomic_get_new_dbuf_state(state); + const struct intel_dbuf_state *old_dbuf_state = + intel_atomic_get_old_dbuf_state(state); + + if (!new_dbuf_state || + new_dbuf_state->enabled_slices == old_dbuf_state->enabled_slices) + return; + + WARN_ON(!new_dbuf_state->base.changed); + + gen9_dbuf_slices_update(dev_priv, + new_dbuf_state->enabled_slices); +} diff --git a/drivers/gpu/drm/i915/intel_pm.h b/drivers/gpu/drm/i915/intel_pm.h index 614ac7f8d4cc..a2473594c2db 100644 --- a/drivers/gpu/drm/i915/intel_pm.h +++ b/drivers/gpu/drm/i915/intel_pm.h @@ -8,8 +8,10 @@ #include <linux/types.h> -#include "i915_reg.h" #include "display/intel_bw.h" +#include "display/intel_global_state.h" + +#include "i915_reg.h" struct drm_device; struct drm_i915_private; @@ -38,6 +40,9 @@ void skl_pipe_ddb_get_hw_state(struct intel_crtc *crtc, struct skl_ddb_entry *ddb_y, struct skl_ddb_entry *ddb_uv); void skl_ddb_get_hw_state(struct drm_i915_private *dev_priv); +u16 intel_get_ddb_size(struct drm_i915_private *dev_priv); +u32 skl_ddb_dbuf_slice_mask(struct drm_i915_private *dev_priv, + const struct skl_ddb_entry *entry); void skl_pipe_wm_get_hw_state(struct intel_crtc *crtc, struct skl_pipe_wm *out); void g4x_wm_sanitize(struct drm_i915_private *dev_priv); @@ -63,4 +68,26 @@ void intel_enable_ipc(struct drm_i915_private *dev_priv); bool intel_set_memory_cxsr(struct drm_i915_private *dev_priv, bool enable); +struct intel_dbuf_state { + struct intel_global_state base; + + u8 enabled_slices; + u8 active_pipes; +}; + +int intel_dbuf_init(struct drm_i915_private *dev_priv); + +struct intel_dbuf_state * +intel_atomic_get_dbuf_state(struct intel_atomic_state *state); + +#define to_intel_dbuf_state(x) container_of((x), struct intel_dbuf_state, base) +#define intel_atomic_get_old_dbuf_state(state) \ + to_intel_dbuf_state(intel_atomic_get_old_global_obj_state(state, &to_i915(state->base.dev)->dbuf.obj)) +#define intel_atomic_get_new_dbuf_state(state) \ + to_intel_dbuf_state(intel_atomic_get_new_global_obj_state(state, &to_i915(state->base.dev)->dbuf.obj)) + +int intel_dbuf_init(struct drm_i915_private *dev_priv); +void intel_dbuf_pre_plane_update(struct intel_atomic_state *state); +void intel_dbuf_post_plane_update(struct intel_atomic_state *state); + #endif /* __INTEL_PM_H__ */ diff --git a/drivers/gpu/drm/i915/intel_region_lmem.c b/drivers/gpu/drm/i915/intel_region_lmem.c index 14b59b899c9b..40d8f1a95df6 100644 --- a/drivers/gpu/drm/i915/intel_region_lmem.c +++ b/drivers/gpu/drm/i915/intel_region_lmem.c @@ -76,7 +76,7 @@ region_lmem_init(struct intel_memory_region *mem) { int ret; - if (i915_modparams.fake_lmem_start) { + if (mem->i915->params.fake_lmem_start) { ret = init_fake_lmem_bar(mem); GEM_BUG_ON(ret); } @@ -111,12 +111,12 @@ intel_setup_fake_lmem(struct drm_i915_private *i915) resource_size_t start; GEM_BUG_ON(i915_ggtt_has_aperture(&i915->ggtt)); - GEM_BUG_ON(!i915_modparams.fake_lmem_start); + GEM_BUG_ON(!i915->params.fake_lmem_start); /* Your mappable aperture belongs to me now! */ mappable_end = pci_resource_len(pdev, 2); io_start = pci_resource_start(pdev, 2), - start = i915_modparams.fake_lmem_start; + start = i915->params.fake_lmem_start; mem = intel_memory_region_create(i915, start, diff --git a/drivers/gpu/drm/i915/intel_runtime_pm.c b/drivers/gpu/drm/i915/intel_runtime_pm.c index 9cb2d7548daa..153ca9e65382 100644 --- a/drivers/gpu/drm/i915/intel_runtime_pm.c +++ b/drivers/gpu/drm/i915/intel_runtime_pm.c @@ -116,6 +116,9 @@ track_intel_runtime_pm_wakeref(struct intel_runtime_pm *rpm) static void untrack_intel_runtime_pm_wakeref(struct intel_runtime_pm *rpm, depot_stack_handle_t stack) { + struct drm_i915_private *i915 = container_of(rpm, + struct drm_i915_private, + runtime_pm); unsigned long flags, n; bool found = false; @@ -134,9 +137,9 @@ static void untrack_intel_runtime_pm_wakeref(struct intel_runtime_pm *rpm, } spin_unlock_irqrestore(&rpm->debug.lock, flags); - if (WARN(!found, - "Unmatched wakeref (tracking %lu), count %u\n", - rpm->debug.count, atomic_read(&rpm->wakeref_count))) { + if (drm_WARN(&i915->drm, !found, + "Unmatched wakeref (tracking %lu), count %u\n", + rpm->debug.count, atomic_read(&rpm->wakeref_count))) { char *buf; buf = kmalloc(PAGE_SIZE, GFP_NOWAIT | __GFP_NOWARN); @@ -355,10 +358,14 @@ intel_runtime_pm_release(struct intel_runtime_pm *rpm, int wakelock) static intel_wakeref_t __intel_runtime_pm_get(struct intel_runtime_pm *rpm, bool wakelock) { + struct drm_i915_private *i915 = container_of(rpm, + struct drm_i915_private, + runtime_pm); int ret; ret = pm_runtime_get_sync(rpm->kdev); - WARN_ONCE(ret < 0, "pm_runtime_get_sync() failed: %d\n", ret); + drm_WARN_ONCE(&i915->drm, ret < 0, + "pm_runtime_get_sync() failed: %d\n", ret); intel_runtime_pm_acquire(rpm, wakelock); @@ -539,6 +546,9 @@ void intel_runtime_pm_put(struct intel_runtime_pm *rpm, intel_wakeref_t wref) */ void intel_runtime_pm_enable(struct intel_runtime_pm *rpm) { + struct drm_i915_private *i915 = container_of(rpm, + struct drm_i915_private, + runtime_pm); struct device *kdev = rpm->kdev; /* @@ -565,7 +575,8 @@ void intel_runtime_pm_enable(struct intel_runtime_pm *rpm) pm_runtime_dont_use_autosuspend(kdev); ret = pm_runtime_get_sync(kdev); - WARN(ret < 0, "pm_runtime_get_sync() failed: %d\n", ret); + drm_WARN(&i915->drm, ret < 0, + "pm_runtime_get_sync() failed: %d\n", ret); } else { pm_runtime_use_autosuspend(kdev); } @@ -580,11 +591,14 @@ void intel_runtime_pm_enable(struct intel_runtime_pm *rpm) void intel_runtime_pm_disable(struct intel_runtime_pm *rpm) { + struct drm_i915_private *i915 = container_of(rpm, + struct drm_i915_private, + runtime_pm); struct device *kdev = rpm->kdev; /* Transfer rpm ownership back to core */ - WARN(pm_runtime_get_sync(kdev) < 0, - "Failed to pass rpm ownership back to core\n"); + drm_WARN(&i915->drm, pm_runtime_get_sync(kdev) < 0, + "Failed to pass rpm ownership back to core\n"); pm_runtime_dont_use_autosuspend(kdev); @@ -594,12 +608,15 @@ void intel_runtime_pm_disable(struct intel_runtime_pm *rpm) void intel_runtime_pm_driver_release(struct intel_runtime_pm *rpm) { + struct drm_i915_private *i915 = container_of(rpm, + struct drm_i915_private, + runtime_pm); int count = atomic_read(&rpm->wakeref_count); - WARN(count, - "i915 raw-wakerefs=%d wakelocks=%d on cleanup\n", - intel_rpm_raw_wakeref_count(count), - intel_rpm_wakelock_count(count)); + drm_WARN(&i915->drm, count, + "i915 raw-wakerefs=%d wakelocks=%d on cleanup\n", + intel_rpm_raw_wakeref_count(count), + intel_rpm_wakelock_count(count)); untrack_all_intel_runtime_pm_wakerefs(rpm); } diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c index a61cb8ca4d50..592364aed2da 100644 --- a/drivers/gpu/drm/i915/intel_uncore.c +++ b/drivers/gpu/drm/i915/intel_uncore.c @@ -709,7 +709,7 @@ static void __intel_uncore_forcewake_put(struct intel_uncore *uncore, continue; } - fw_domain_arm_timer(domain); + uncore->funcs.force_wake_put(uncore, domain->mask); } } @@ -1185,7 +1185,7 @@ __unclaimed_reg_debug(struct intel_uncore *uncore, read ? "read from" : "write to", i915_mmio_reg_offset(reg))) /* Only report the first N failures */ - i915_modparams.mmio_debug--; + uncore->i915->params.mmio_debug--; } static inline void @@ -1194,7 +1194,7 @@ unclaimed_reg_debug(struct intel_uncore *uncore, const bool read, const bool before) { - if (likely(!i915_modparams.mmio_debug)) + if (likely(!uncore->i915->params.mmio_debug)) return; /* interrupts are disabled and re-enabled around uncore->lock usage */ @@ -2093,12 +2093,12 @@ intel_uncore_arm_unclaimed_mmio_detection(struct intel_uncore *uncore) goto out; if (unlikely(check_for_unclaimed_mmio(uncore))) { - if (!i915_modparams.mmio_debug) { + if (!uncore->i915->params.mmio_debug) { drm_dbg(&uncore->i915->drm, "Unclaimed register detected, " "enabling oneshot unclaimed register reporting. " "Please use i915.mmio_debug=N for more information.\n"); - i915_modparams.mmio_debug++; + uncore->i915->params.mmio_debug++; } uncore->debug->unclaimed_mmio_check--; ret = true; diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c index 2e471500a646..0016ffc7d914 100644 --- a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c @@ -97,6 +97,7 @@ static void fake_put_pages(struct drm_i915_gem_object *obj, } static const struct drm_i915_gem_object_ops fake_ops = { + .name = "fake-gem", .flags = I915_GEM_OBJECT_IS_SHRINKABLE, .get_pages = fake_get_pages, .put_pages = fake_put_pages, diff --git a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h index 5dd5d81646c4..a92c0e9b7e6b 100644 --- a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h +++ b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h @@ -11,9 +11,9 @@ * a module parameter. It must be unique and legal for a C identifier. * * The function should be of type int function(void). It may be conditionally - * compiled using #if IS_ENABLED(DRM_I915_SELFTEST). + * compiled using #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST). * - * Tests are executed in order by igt/drv_selftest + * Tests are executed in order by igt/i915_selftest */ selftest(sanitycheck, i915_live_sanitycheck) /* keep first (igt selfcheck) */ selftest(uncore, intel_uncore_live_selftests) diff --git a/drivers/gpu/drm/i915/selftests/i915_mock_selftests.h b/drivers/gpu/drm/i915/selftests/i915_mock_selftests.h index 6090ce35226b..3db34d3eea58 100644 --- a/drivers/gpu/drm/i915/selftests/i915_mock_selftests.h +++ b/drivers/gpu/drm/i915/selftests/i915_mock_selftests.h @@ -11,9 +11,9 @@ * a module parameter. It must be unique and legal for a C identifier. * * The function should be of type int function(void). It may be conditionally - * compiled using #if IS_ENABLED(DRM_I915_SELFTEST). + * compiled using #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST). * - * Tests are executed in order by igt/drv_selftest + * Tests are executed in order by igt/i915_selftest */ selftest(sanitycheck, i915_mock_sanitycheck) /* keep first (igt selfcheck) */ selftest(shmem, shmem_utils_mock_selftests) diff --git a/drivers/gpu/drm/i915/selftests/i915_perf.c b/drivers/gpu/drm/i915/selftests/i915_perf.c index 8eb3108f1767..be54570c407c 100644 --- a/drivers/gpu/drm/i915/selftests/i915_perf.c +++ b/drivers/gpu/drm/i915/selftests/i915_perf.c @@ -162,7 +162,7 @@ static int write_timestamp(struct i915_request *rq, int slot) return PTR_ERR(cs); len = 5; - if (INTEL_GEN(rq->i915) >= 8) + if (INTEL_GEN(rq->engine->i915) >= 8) len++; *cs++ = GFX_OP_PIPE_CONTROL(len); diff --git a/drivers/gpu/drm/i915/selftests/i915_perf_selftests.h b/drivers/gpu/drm/i915/selftests/i915_perf_selftests.h index d8da142985eb..c2389f8a257d 100644 --- a/drivers/gpu/drm/i915/selftests/i915_perf_selftests.h +++ b/drivers/gpu/drm/i915/selftests/i915_perf_selftests.h @@ -11,7 +11,7 @@ * a module parameter. It must be unique and legal for a C identifier. * * The function should be of type int function(void). It may be conditionally - * compiled using #if IS_ENABLED(DRM_I915_SELFTEST). + * compiled using #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST). * * Tests are executed in order by igt/i915_selftest */ diff --git a/drivers/gpu/drm/i915/selftests/i915_request.c b/drivers/gpu/drm/i915/selftests/i915_request.c index 6014e8dfcbb1..9271aad7f779 100644 --- a/drivers/gpu/drm/i915/selftests/i915_request.c +++ b/drivers/gpu/drm/i915/selftests/i915_request.c @@ -24,16 +24,21 @@ #include <linux/prime_numbers.h> #include <linux/pm_qos.h> +#include <linux/sort.h> #include "gem/i915_gem_pm.h" #include "gem/selftests/mock_context.h" +#include "gt/intel_engine_heartbeat.h" #include "gt/intel_engine_pm.h" #include "gt/intel_engine_user.h" #include "gt/intel_gt.h" +#include "gt/intel_gt_requests.h" +#include "gt/selftest_engine_heartbeat.h" #include "i915_random.h" #include "i915_selftest.h" +#include "igt_flush_test.h" #include "igt_live_test.h" #include "igt_spinner.h" #include "lib_sw_fence.h" @@ -1524,6 +1529,808 @@ struct perf_series { struct intel_context *ce[]; }; +static int cmp_u32(const void *A, const void *B) +{ + const u32 *a = A, *b = B; + + return *a - *b; +} + +static u32 trifilter(u32 *a) +{ + u64 sum; + +#define TF_COUNT 5 + sort(a, TF_COUNT, sizeof(*a), cmp_u32, NULL); + + sum = mul_u32_u32(a[2], 2); + sum += a[1]; + sum += a[3]; + + GEM_BUG_ON(sum > U32_MAX); + return sum; +#define TF_BIAS 2 +} + +static u64 cycles_to_ns(struct intel_engine_cs *engine, u32 cycles) +{ + u64 ns = i915_cs_timestamp_ticks_to_ns(engine->i915, cycles); + + return DIV_ROUND_CLOSEST(ns, 1 << TF_BIAS); +} + +static u32 *emit_timestamp_store(u32 *cs, struct intel_context *ce, u32 offset) +{ + *cs++ = MI_STORE_REGISTER_MEM_GEN8 | MI_USE_GGTT; + *cs++ = i915_mmio_reg_offset(RING_TIMESTAMP((ce->engine->mmio_base))); + *cs++ = offset; + *cs++ = 0; + + return cs; +} + +static u32 *emit_store_dw(u32 *cs, u32 offset, u32 value) +{ + *cs++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT; + *cs++ = offset; + *cs++ = 0; + *cs++ = value; + + return cs; +} + +static u32 *emit_semaphore_poll(u32 *cs, u32 mode, u32 value, u32 offset) +{ + *cs++ = MI_SEMAPHORE_WAIT | + MI_SEMAPHORE_GLOBAL_GTT | + MI_SEMAPHORE_POLL | + mode; + *cs++ = value; + *cs++ = offset; + *cs++ = 0; + + return cs; +} + +static u32 *emit_semaphore_poll_until(u32 *cs, u32 offset, u32 value) +{ + return emit_semaphore_poll(cs, MI_SEMAPHORE_SAD_EQ_SDD, value, offset); +} + +static void semaphore_set(u32 *sema, u32 value) +{ + WRITE_ONCE(*sema, value); + wmb(); /* flush the update to the cache, and beyond */ +} + +static u32 *hwsp_scratch(const struct intel_context *ce) +{ + return memset32(ce->engine->status_page.addr + 1000, 0, 21); +} + +static u32 hwsp_offset(const struct intel_context *ce, u32 *dw) +{ + return (i915_ggtt_offset(ce->engine->status_page.vma) + + offset_in_page(dw)); +} + +static int measure_semaphore_response(struct intel_context *ce) +{ + u32 *sema = hwsp_scratch(ce); + const u32 offset = hwsp_offset(ce, sema); + u32 elapsed[TF_COUNT], cycles; + struct i915_request *rq; + u32 *cs; + int err; + int i; + + /* + * Measure how many cycles it takes for the HW to detect the change + * in a semaphore value. + * + * A: read CS_TIMESTAMP from CPU + * poke semaphore + * B: read CS_TIMESTAMP on GPU + * + * Semaphore latency: B - A + */ + + semaphore_set(sema, -1); + + rq = i915_request_create(ce); + if (IS_ERR(rq)) + return PTR_ERR(rq); + + cs = intel_ring_begin(rq, 4 + 12 * ARRAY_SIZE(elapsed)); + if (IS_ERR(cs)) { + i915_request_add(rq); + err = PTR_ERR(cs); + goto err; + } + + cs = emit_store_dw(cs, offset, 0); + for (i = 1; i <= ARRAY_SIZE(elapsed); i++) { + cs = emit_semaphore_poll_until(cs, offset, i); + cs = emit_timestamp_store(cs, ce, offset + i * sizeof(u32)); + cs = emit_store_dw(cs, offset, 0); + } + + intel_ring_advance(rq, cs); + i915_request_add(rq); + + if (wait_for(READ_ONCE(*sema) == 0, 50)) { + err = -EIO; + goto err; + } + + for (i = 1; i <= ARRAY_SIZE(elapsed); i++) { + preempt_disable(); + cycles = ENGINE_READ_FW(ce->engine, RING_TIMESTAMP); + semaphore_set(sema, i); + preempt_enable(); + + if (wait_for(READ_ONCE(*sema) == 0, 50)) { + err = -EIO; + goto err; + } + + elapsed[i - 1] = sema[i] - cycles; + } + + cycles = trifilter(elapsed); + pr_info("%s: semaphore response %d cycles, %lluns\n", + ce->engine->name, cycles >> TF_BIAS, + cycles_to_ns(ce->engine, cycles)); + + return intel_gt_wait_for_idle(ce->engine->gt, HZ); + +err: + intel_gt_set_wedged(ce->engine->gt); + return err; +} + +static int measure_idle_dispatch(struct intel_context *ce) +{ + u32 *sema = hwsp_scratch(ce); + const u32 offset = hwsp_offset(ce, sema); + u32 elapsed[TF_COUNT], cycles; + u32 *cs; + int err; + int i; + + /* + * Measure how long it takes for us to submit a request while the + * engine is idle, but is resting in our context. + * + * A: read CS_TIMESTAMP from CPU + * submit request + * B: read CS_TIMESTAMP on GPU + * + * Submission latency: B - A + */ + + for (i = 0; i < ARRAY_SIZE(elapsed); i++) { + struct i915_request *rq; + + err = intel_gt_wait_for_idle(ce->engine->gt, HZ / 2); + if (err) + return err; + + rq = i915_request_create(ce); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err; + } + + cs = intel_ring_begin(rq, 4); + if (IS_ERR(cs)) { + i915_request_add(rq); + err = PTR_ERR(cs); + goto err; + } + + cs = emit_timestamp_store(cs, ce, offset + i * sizeof(u32)); + + intel_ring_advance(rq, cs); + + preempt_disable(); + local_bh_disable(); + elapsed[i] = ENGINE_READ_FW(ce->engine, RING_TIMESTAMP); + i915_request_add(rq); + local_bh_enable(); + preempt_enable(); + } + + err = intel_gt_wait_for_idle(ce->engine->gt, HZ / 2); + if (err) + goto err; + + for (i = 0; i < ARRAY_SIZE(elapsed); i++) + elapsed[i] = sema[i] - elapsed[i]; + + cycles = trifilter(elapsed); + pr_info("%s: idle dispatch latency %d cycles, %lluns\n", + ce->engine->name, cycles >> TF_BIAS, + cycles_to_ns(ce->engine, cycles)); + + return intel_gt_wait_for_idle(ce->engine->gt, HZ); + +err: + intel_gt_set_wedged(ce->engine->gt); + return err; +} + +static int measure_busy_dispatch(struct intel_context *ce) +{ + u32 *sema = hwsp_scratch(ce); + const u32 offset = hwsp_offset(ce, sema); + u32 elapsed[TF_COUNT + 1], cycles; + u32 *cs; + int err; + int i; + + /* + * Measure how long it takes for us to submit a request while the + * engine is busy, polling on a semaphore in our context. With + * direct submission, this will include the cost of a lite restore. + * + * A: read CS_TIMESTAMP from CPU + * submit request + * B: read CS_TIMESTAMP on GPU + * + * Submission latency: B - A + */ + + for (i = 1; i <= ARRAY_SIZE(elapsed); i++) { + struct i915_request *rq; + + rq = i915_request_create(ce); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err; + } + + cs = intel_ring_begin(rq, 12); + if (IS_ERR(cs)) { + i915_request_add(rq); + err = PTR_ERR(cs); + goto err; + } + + cs = emit_store_dw(cs, offset + i * sizeof(u32), -1); + cs = emit_semaphore_poll_until(cs, offset, i); + cs = emit_timestamp_store(cs, ce, offset + i * sizeof(u32)); + + intel_ring_advance(rq, cs); + + if (i > 1 && wait_for(READ_ONCE(sema[i - 1]), 500)) { + err = -EIO; + goto err; + } + + preempt_disable(); + local_bh_disable(); + elapsed[i - 1] = ENGINE_READ_FW(ce->engine, RING_TIMESTAMP); + i915_request_add(rq); + local_bh_enable(); + semaphore_set(sema, i - 1); + preempt_enable(); + } + + wait_for(READ_ONCE(sema[i - 1]), 500); + semaphore_set(sema, i - 1); + + for (i = 1; i <= TF_COUNT; i++) { + GEM_BUG_ON(sema[i] == -1); + elapsed[i - 1] = sema[i] - elapsed[i]; + } + + cycles = trifilter(elapsed); + pr_info("%s: busy dispatch latency %d cycles, %lluns\n", + ce->engine->name, cycles >> TF_BIAS, + cycles_to_ns(ce->engine, cycles)); + + return intel_gt_wait_for_idle(ce->engine->gt, HZ); + +err: + intel_gt_set_wedged(ce->engine->gt); + return err; +} + +static int plug(struct intel_engine_cs *engine, u32 *sema, u32 mode, int value) +{ + const u32 offset = + i915_ggtt_offset(engine->status_page.vma) + + offset_in_page(sema); + struct i915_request *rq; + u32 *cs; + + rq = i915_request_create(engine->kernel_context); + if (IS_ERR(rq)) + return PTR_ERR(rq); + + cs = intel_ring_begin(rq, 4); + if (IS_ERR(cs)) { + i915_request_add(rq); + return PTR_ERR(cs); + } + + cs = emit_semaphore_poll(cs, mode, value, offset); + + intel_ring_advance(rq, cs); + i915_request_add(rq); + + return 0; +} + +static int measure_inter_request(struct intel_context *ce) +{ + u32 *sema = hwsp_scratch(ce); + const u32 offset = hwsp_offset(ce, sema); + u32 elapsed[TF_COUNT + 1], cycles; + struct i915_sw_fence *submit; + int i, err; + + /* + * Measure how long it takes to advance from one request into the + * next. Between each request we flush the GPU caches to memory, + * update the breadcrumbs, and then invalidate those caches. + * We queue up all the requests to be submitted in one batch so + * it should be one set of contiguous measurements. + * + * A: read CS_TIMESTAMP on GPU + * advance request + * B: read CS_TIMESTAMP on GPU + * + * Request latency: B - A + */ + + err = plug(ce->engine, sema, MI_SEMAPHORE_SAD_NEQ_SDD, 0); + if (err) + return err; + + submit = heap_fence_create(GFP_KERNEL); + if (!submit) { + semaphore_set(sema, 1); + return -ENOMEM; + } + + intel_engine_flush_submission(ce->engine); + for (i = 1; i <= ARRAY_SIZE(elapsed); i++) { + struct i915_request *rq; + u32 *cs; + + rq = i915_request_create(ce); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err_submit; + } + + err = i915_sw_fence_await_sw_fence_gfp(&rq->submit, + submit, + GFP_KERNEL); + if (err < 0) { + i915_request_add(rq); + goto err_submit; + } + + cs = intel_ring_begin(rq, 4); + if (IS_ERR(cs)) { + i915_request_add(rq); + err = PTR_ERR(cs); + goto err_submit; + } + + cs = emit_timestamp_store(cs, ce, offset + i * sizeof(u32)); + + intel_ring_advance(rq, cs); + i915_request_add(rq); + } + local_bh_disable(); + i915_sw_fence_commit(submit); + local_bh_enable(); + intel_engine_flush_submission(ce->engine); + heap_fence_put(submit); + + semaphore_set(sema, 1); + err = intel_gt_wait_for_idle(ce->engine->gt, HZ / 2); + if (err) + goto err; + + for (i = 1; i <= TF_COUNT; i++) + elapsed[i - 1] = sema[i + 1] - sema[i]; + + cycles = trifilter(elapsed); + pr_info("%s: inter-request latency %d cycles, %lluns\n", + ce->engine->name, cycles >> TF_BIAS, + cycles_to_ns(ce->engine, cycles)); + + return intel_gt_wait_for_idle(ce->engine->gt, HZ); + +err_submit: + i915_sw_fence_commit(submit); + heap_fence_put(submit); + semaphore_set(sema, 1); +err: + intel_gt_set_wedged(ce->engine->gt); + return err; +} + +static int measure_context_switch(struct intel_context *ce) +{ + u32 *sema = hwsp_scratch(ce); + const u32 offset = hwsp_offset(ce, sema); + struct i915_request *fence = NULL; + u32 elapsed[TF_COUNT + 1], cycles; + int i, j, err; + u32 *cs; + + /* + * Measure how long it takes to advance from one request in one + * context to a request in another context. This allows us to + * measure how long the context save/restore take, along with all + * the inter-context setup we require. + * + * A: read CS_TIMESTAMP on GPU + * switch context + * B: read CS_TIMESTAMP on GPU + * + * Context switch latency: B - A + */ + + err = plug(ce->engine, sema, MI_SEMAPHORE_SAD_NEQ_SDD, 0); + if (err) + return err; + + for (i = 1; i <= ARRAY_SIZE(elapsed); i++) { + struct intel_context *arr[] = { + ce, ce->engine->kernel_context + }; + u32 addr = offset + ARRAY_SIZE(arr) * i * sizeof(u32); + + for (j = 0; j < ARRAY_SIZE(arr); j++) { + struct i915_request *rq; + + rq = i915_request_create(arr[j]); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err_fence; + } + + if (fence) { + err = i915_request_await_dma_fence(rq, + &fence->fence); + if (err) { + i915_request_add(rq); + goto err_fence; + } + } + + cs = intel_ring_begin(rq, 4); + if (IS_ERR(cs)) { + i915_request_add(rq); + err = PTR_ERR(cs); + goto err_fence; + } + + cs = emit_timestamp_store(cs, ce, addr); + addr += sizeof(u32); + + intel_ring_advance(rq, cs); + + i915_request_put(fence); + fence = i915_request_get(rq); + + i915_request_add(rq); + } + } + i915_request_put(fence); + intel_engine_flush_submission(ce->engine); + + semaphore_set(sema, 1); + err = intel_gt_wait_for_idle(ce->engine->gt, HZ / 2); + if (err) + goto err; + + for (i = 1; i <= TF_COUNT; i++) + elapsed[i - 1] = sema[2 * i + 2] - sema[2 * i + 1]; + + cycles = trifilter(elapsed); + pr_info("%s: context switch latency %d cycles, %lluns\n", + ce->engine->name, cycles >> TF_BIAS, + cycles_to_ns(ce->engine, cycles)); + + return intel_gt_wait_for_idle(ce->engine->gt, HZ); + +err_fence: + i915_request_put(fence); + semaphore_set(sema, 1); +err: + intel_gt_set_wedged(ce->engine->gt); + return err; +} + +static int measure_preemption(struct intel_context *ce) +{ + u32 *sema = hwsp_scratch(ce); + const u32 offset = hwsp_offset(ce, sema); + u32 elapsed[TF_COUNT], cycles; + u32 *cs; + int err; + int i; + + /* + * We measure two latencies while triggering preemption. The first + * latency is how long it takes for us to submit a preempting request. + * The second latency is how it takes for us to return from the + * preemption back to the original context. + * + * A: read CS_TIMESTAMP from CPU + * submit preemption + * B: read CS_TIMESTAMP on GPU (in preempting context) + * context switch + * C: read CS_TIMESTAMP on GPU (in original context) + * + * Preemption dispatch latency: B - A + * Preemption switch latency: C - B + */ + + if (!intel_engine_has_preemption(ce->engine)) + return 0; + + for (i = 1; i <= ARRAY_SIZE(elapsed); i++) { + u32 addr = offset + 2 * i * sizeof(u32); + struct i915_request *rq; + + rq = i915_request_create(ce); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err; + } + + cs = intel_ring_begin(rq, 12); + if (IS_ERR(cs)) { + i915_request_add(rq); + err = PTR_ERR(cs); + goto err; + } + + cs = emit_store_dw(cs, addr, -1); + cs = emit_semaphore_poll_until(cs, offset, i); + cs = emit_timestamp_store(cs, ce, addr + sizeof(u32)); + + intel_ring_advance(rq, cs); + i915_request_add(rq); + + if (wait_for(READ_ONCE(sema[2 * i]) == -1, 500)) { + err = -EIO; + goto err; + } + + rq = i915_request_create(ce->engine->kernel_context); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err; + } + + cs = intel_ring_begin(rq, 8); + if (IS_ERR(cs)) { + i915_request_add(rq); + err = PTR_ERR(cs); + goto err; + } + + cs = emit_timestamp_store(cs, ce, addr); + cs = emit_store_dw(cs, offset, i); + + intel_ring_advance(rq, cs); + rq->sched.attr.priority = I915_PRIORITY_BARRIER; + + elapsed[i - 1] = ENGINE_READ_FW(ce->engine, RING_TIMESTAMP); + i915_request_add(rq); + } + + if (wait_for(READ_ONCE(sema[2 * i - 2]) != -1, 500)) { + err = -EIO; + goto err; + } + + for (i = 1; i <= TF_COUNT; i++) + elapsed[i - 1] = sema[2 * i + 0] - elapsed[i - 1]; + + cycles = trifilter(elapsed); + pr_info("%s: preemption dispatch latency %d cycles, %lluns\n", + ce->engine->name, cycles >> TF_BIAS, + cycles_to_ns(ce->engine, cycles)); + + for (i = 1; i <= TF_COUNT; i++) + elapsed[i - 1] = sema[2 * i + 1] - sema[2 * i + 0]; + + cycles = trifilter(elapsed); + pr_info("%s: preemption switch latency %d cycles, %lluns\n", + ce->engine->name, cycles >> TF_BIAS, + cycles_to_ns(ce->engine, cycles)); + + return intel_gt_wait_for_idle(ce->engine->gt, HZ); + +err: + intel_gt_set_wedged(ce->engine->gt); + return err; +} + +struct signal_cb { + struct dma_fence_cb base; + bool seen; +}; + +static void signal_cb(struct dma_fence *fence, struct dma_fence_cb *cb) +{ + struct signal_cb *s = container_of(cb, typeof(*s), base); + + smp_store_mb(s->seen, true); /* be safe, be strong */ +} + +static int measure_completion(struct intel_context *ce) +{ + u32 *sema = hwsp_scratch(ce); + const u32 offset = hwsp_offset(ce, sema); + u32 elapsed[TF_COUNT], cycles; + u32 *cs; + int err; + int i; + + /* + * Measure how long it takes for the signal (interrupt) to be + * sent from the GPU to be processed by the CPU. + * + * A: read CS_TIMESTAMP on GPU + * signal + * B: read CS_TIMESTAMP from CPU + * + * Completion latency: B - A + */ + + for (i = 1; i <= ARRAY_SIZE(elapsed); i++) { + struct signal_cb cb = { .seen = false }; + struct i915_request *rq; + + rq = i915_request_create(ce); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err; + } + + cs = intel_ring_begin(rq, 12); + if (IS_ERR(cs)) { + i915_request_add(rq); + err = PTR_ERR(cs); + goto err; + } + + cs = emit_store_dw(cs, offset + i * sizeof(u32), -1); + cs = emit_semaphore_poll_until(cs, offset, i); + cs = emit_timestamp_store(cs, ce, offset + i * sizeof(u32)); + + intel_ring_advance(rq, cs); + + dma_fence_add_callback(&rq->fence, &cb.base, signal_cb); + + local_bh_disable(); + i915_request_add(rq); + local_bh_enable(); + + if (wait_for(READ_ONCE(sema[i]) == -1, 50)) { + err = -EIO; + goto err; + } + + preempt_disable(); + semaphore_set(sema, i); + while (!READ_ONCE(cb.seen)) + cpu_relax(); + + elapsed[i - 1] = ENGINE_READ_FW(ce->engine, RING_TIMESTAMP); + preempt_enable(); + } + + err = intel_gt_wait_for_idle(ce->engine->gt, HZ / 2); + if (err) + goto err; + + for (i = 0; i < ARRAY_SIZE(elapsed); i++) { + GEM_BUG_ON(sema[i + 1] == -1); + elapsed[i] = elapsed[i] - sema[i + 1]; + } + + cycles = trifilter(elapsed); + pr_info("%s: completion latency %d cycles, %lluns\n", + ce->engine->name, cycles >> TF_BIAS, + cycles_to_ns(ce->engine, cycles)); + + return intel_gt_wait_for_idle(ce->engine->gt, HZ); + +err: + intel_gt_set_wedged(ce->engine->gt); + return err; +} + +static void rps_pin(struct intel_gt *gt) +{ + /* Pin the frequency to max */ + atomic_inc(>->rps.num_waiters); + intel_uncore_forcewake_get(gt->uncore, FORCEWAKE_ALL); + + mutex_lock(>->rps.lock); + intel_rps_set(>->rps, gt->rps.max_freq); + mutex_unlock(>->rps.lock); +} + +static void rps_unpin(struct intel_gt *gt) +{ + intel_uncore_forcewake_put(gt->uncore, FORCEWAKE_ALL); + atomic_dec(>->rps.num_waiters); +} + +static int perf_request_latency(void *arg) +{ + struct drm_i915_private *i915 = arg; + struct intel_engine_cs *engine; + struct pm_qos_request qos; + int err = 0; + + if (INTEL_GEN(i915) < 8) /* per-engine CS timestamp, semaphores */ + return 0; + + cpu_latency_qos_add_request(&qos, 0); /* disable cstates */ + + for_each_uabi_engine(engine, i915) { + struct intel_context *ce; + + ce = intel_context_create(engine); + if (IS_ERR(ce)) + goto out; + + err = intel_context_pin(ce); + if (err) { + intel_context_put(ce); + goto out; + } + + st_engine_heartbeat_disable(engine); + rps_pin(engine->gt); + + if (err == 0) + err = measure_semaphore_response(ce); + if (err == 0) + err = measure_idle_dispatch(ce); + if (err == 0) + err = measure_busy_dispatch(ce); + if (err == 0) + err = measure_inter_request(ce); + if (err == 0) + err = measure_context_switch(ce); + if (err == 0) + err = measure_preemption(ce); + if (err == 0) + err = measure_completion(ce); + + rps_unpin(engine->gt); + st_engine_heartbeat_enable(engine); + + intel_context_unpin(ce); + intel_context_put(ce); + if (err) + goto out; + } + +out: + if (igt_flush_test(i915)) + err = -EIO; + + cpu_latency_qos_remove_request(&qos); + return err; +} + static int s_sync0(void *arg) { struct perf_series *ps = arg; @@ -1685,9 +2492,11 @@ static int perf_series_engines(void *arg) intel_engine_pm_get(p->engine); if (intel_engine_supports_stats(p->engine)) - p->busy = intel_engine_get_busy_time(p->engine) + 1; + p->busy = intel_engine_get_busy_time(p->engine, + &p->time) + 1; + else + p->time = ktime_get(); p->runtime = -intel_context_get_total_runtime_ns(ce); - p->time = ktime_get(); } err = (*fn)(ps); @@ -1698,13 +2507,15 @@ static int perf_series_engines(void *arg) struct perf_stats *p = &stats[idx]; struct intel_context *ce = ps->ce[idx]; int integer, decimal; - u64 busy, dt; + u64 busy, dt, now; - p->time = ktime_sub(ktime_get(), p->time); - if (p->busy) { - p->busy = ktime_sub(intel_engine_get_busy_time(p->engine), + if (p->busy) + p->busy = ktime_sub(intel_engine_get_busy_time(p->engine, + &now), p->busy - 1); - } + else + now = ktime_get(); + p->time = ktime_sub(now, p->time); err = switch_to_kernel_sync(ce, err); p->runtime += intel_context_get_total_runtime_ns(ce); @@ -1764,13 +2575,14 @@ static int p_sync0(void *arg) return err; } - busy = false; if (intel_engine_supports_stats(engine)) { - p->busy = intel_engine_get_busy_time(engine); + p->busy = intel_engine_get_busy_time(engine, &p->time); busy = true; + } else { + p->time = ktime_get(); + busy = false; } - p->time = ktime_get(); count = 0; do { struct i915_request *rq; @@ -1793,11 +2605,15 @@ static int p_sync0(void *arg) count++; } while (!__igt_timeout(end_time, NULL)); - p->time = ktime_sub(ktime_get(), p->time); if (busy) { - p->busy = ktime_sub(intel_engine_get_busy_time(engine), + ktime_t now; + + p->busy = ktime_sub(intel_engine_get_busy_time(engine, &now), p->busy); + p->time = ktime_sub(now, p->time); + } else { + p->time = ktime_sub(ktime_get(), p->time); } err = switch_to_kernel_sync(ce, err); @@ -1830,13 +2646,14 @@ static int p_sync1(void *arg) return err; } - busy = false; if (intel_engine_supports_stats(engine)) { - p->busy = intel_engine_get_busy_time(engine); + p->busy = intel_engine_get_busy_time(engine, &p->time); busy = true; + } else { + p->time = ktime_get(); + busy = false; } - p->time = ktime_get(); count = 0; do { struct i915_request *rq; @@ -1861,11 +2678,15 @@ static int p_sync1(void *arg) count++; } while (!__igt_timeout(end_time, NULL)); i915_request_put(prev); - p->time = ktime_sub(ktime_get(), p->time); if (busy) { - p->busy = ktime_sub(intel_engine_get_busy_time(engine), + ktime_t now; + + p->busy = ktime_sub(intel_engine_get_busy_time(engine, &now), p->busy); + p->time = ktime_sub(now, p->time); + } else { + p->time = ktime_sub(ktime_get(), p->time); } err = switch_to_kernel_sync(ce, err); @@ -1897,14 +2718,15 @@ static int p_many(void *arg) return err; } - busy = false; if (intel_engine_supports_stats(engine)) { - p->busy = intel_engine_get_busy_time(engine); + p->busy = intel_engine_get_busy_time(engine, &p->time); busy = true; + } else { + p->time = ktime_get(); + busy = false; } count = 0; - p->time = ktime_get(); do { struct i915_request *rq; @@ -1917,11 +2739,15 @@ static int p_many(void *arg) i915_request_add(rq); count++; } while (!__igt_timeout(end_time, NULL)); - p->time = ktime_sub(ktime_get(), p->time); if (busy) { - p->busy = ktime_sub(intel_engine_get_busy_time(engine), + ktime_t now; + + p->busy = ktime_sub(intel_engine_get_busy_time(engine, &now), p->busy); + p->time = ktime_sub(now, p->time); + } else { + p->time = ktime_sub(ktime_get(), p->time); } err = switch_to_kernel_sync(ce, err); @@ -2042,6 +2868,7 @@ static int perf_parallel_engines(void *arg) int i915_request_perf_selftests(struct drm_i915_private *i915) { static const struct i915_subtest tests[] = { + SUBTEST(perf_request_latency), SUBTEST(perf_series_engines), SUBTEST(perf_parallel_engines), }; diff --git a/drivers/gpu/drm/i915/selftests/igt_spinner.c b/drivers/gpu/drm/i915/selftests/igt_spinner.c index e35ba5f9e73f..ec0ecb4e4ca6 100644 --- a/drivers/gpu/drm/i915/selftests/igt_spinner.c +++ b/drivers/gpu/drm/i915/selftests/igt_spinner.c @@ -134,15 +134,15 @@ igt_spinner_create_request(struct igt_spinner *spin, batch = spin->batch; - if (INTEL_GEN(rq->i915) >= 8) { + if (INTEL_GEN(rq->engine->i915) >= 8) { *batch++ = MI_STORE_DWORD_IMM_GEN4; *batch++ = lower_32_bits(hws_address(hws, rq)); *batch++ = upper_32_bits(hws_address(hws, rq)); - } else if (INTEL_GEN(rq->i915) >= 6) { + } else if (INTEL_GEN(rq->engine->i915) >= 6) { *batch++ = MI_STORE_DWORD_IMM_GEN4; *batch++ = 0; *batch++ = hws_address(hws, rq); - } else if (INTEL_GEN(rq->i915) >= 4) { + } else if (INTEL_GEN(rq->engine->i915) >= 4) { *batch++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT; *batch++ = 0; *batch++ = hws_address(hws, rq); @@ -154,11 +154,11 @@ igt_spinner_create_request(struct igt_spinner *spin, *batch++ = arbitration_command; - if (INTEL_GEN(rq->i915) >= 8) + if (INTEL_GEN(rq->engine->i915) >= 8) *batch++ = MI_BATCH_BUFFER_START | BIT(8) | 1; - else if (IS_HASWELL(rq->i915)) + else if (IS_HASWELL(rq->engine->i915)) *batch++ = MI_BATCH_BUFFER_START | MI_BATCH_PPGTT_HSW; - else if (INTEL_GEN(rq->i915) >= 6) + else if (INTEL_GEN(rq->engine->i915) >= 6) *batch++ = MI_BATCH_BUFFER_START; else *batch++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT; @@ -176,7 +176,7 @@ igt_spinner_create_request(struct igt_spinner *spin, } flags = 0; - if (INTEL_GEN(rq->i915) <= 5) + if (INTEL_GEN(rq->engine->i915) <= 5) flags |= I915_DISPATCH_SECURE; err = engine->emit_bb_start(rq, vma->node.start, PAGE_SIZE, flags); @@ -221,8 +221,8 @@ bool igt_wait_for_spinner(struct igt_spinner *spin, struct i915_request *rq) { return !(wait_for_us(i915_seqno_passed(hws_seqno(spin, rq), rq->fence.seqno), - 10) && + 100) && wait_for(i915_seqno_passed(hws_seqno(spin, rq), rq->fence.seqno), - 1000)); + 50)); } diff --git a/drivers/gpu/drm/i915/selftests/mock_region.c b/drivers/gpu/drm/i915/selftests/mock_region.c index b2ad41c27e67..09660f5a0a4c 100644 --- a/drivers/gpu/drm/i915/selftests/mock_region.c +++ b/drivers/gpu/drm/i915/selftests/mock_region.c @@ -9,6 +9,7 @@ #include "mock_region.h" static const struct drm_i915_gem_object_ops mock_region_obj_ops = { + .name = "mock-region", .get_pages = i915_gem_object_get_pages_buddy, .put_pages = i915_gem_object_put_pages_buddy, .release = i915_gem_object_release_memory_region, diff --git a/include/drm/i915_pciids.h b/include/drm/i915_pciids.h index 662d8351c87a..bc989de2aac2 100644 --- a/include/drm/i915_pciids.h +++ b/include/drm/i915_pciids.h @@ -605,4 +605,13 @@ INTEL_VGA_DEVICE(0x9AD9, info), \ INTEL_VGA_DEVICE(0x9AF8, info) +/* RKL */ +#define INTEL_RKL_IDS(info) \ + INTEL_VGA_DEVICE(0x4C80, info), \ + INTEL_VGA_DEVICE(0x4C8A, info), \ + INTEL_VGA_DEVICE(0x4C8B, info), \ + INTEL_VGA_DEVICE(0x4C8C, info), \ + INTEL_VGA_DEVICE(0x4C90, info), \ + INTEL_VGA_DEVICE(0x4C9A, info) + #endif /* _I915_PCIIDS_H */ |