diff options
Diffstat (limited to 'drivers/gpu/drm/i915/gt/intel_workarounds.c')
| -rw-r--r-- | drivers/gpu/drm/i915/gt/intel_workarounds.c | 290 |
1 files changed, 260 insertions, 30 deletions
diff --git a/drivers/gpu/drm/i915/gt/intel_workarounds.c b/drivers/gpu/drm/i915/gt/intel_workarounds.c index 5176ad1a3976..85d2bef51524 100644 --- a/drivers/gpu/drm/i915/gt/intel_workarounds.c +++ b/drivers/gpu/drm/i915/gt/intel_workarounds.c @@ -179,6 +179,12 @@ wa_write_or(struct i915_wa_list *wal, i915_reg_t reg, u32 set) } static void +wa_write_clr(struct i915_wa_list *wal, i915_reg_t reg, u32 clr) +{ + wa_write_masked_or(wal, reg, clr, 0); +} + +static void wa_masked_en(struct i915_wa_list *wal, i915_reg_t reg, u32 val) { wa_add(wal, reg, 0, _MASKED_BIT_ENABLE(val), val); @@ -485,25 +491,14 @@ static void cfl_ctx_workarounds_init(struct intel_engine_cs *engine, static void cnl_ctx_workarounds_init(struct intel_engine_cs *engine, struct i915_wa_list *wal) { - struct drm_i915_private *i915 = engine->i915; - /* WaForceContextSaveRestoreNonCoherent:cnl */ WA_SET_BIT_MASKED(CNL_HDC_CHICKEN0, HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT); - /* WaThrottleEUPerfToAvoidTDBackPressure:cnl(pre-prod) */ - if (IS_CNL_REVID(i915, CNL_REVID_B0, CNL_REVID_B0)) - WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, THROTTLE_12_5); - /* WaDisableReplayBufferBankArbitrationOptimization:cnl */ WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION); - /* WaDisableEnhancedSBEVertexCaching:cnl (pre-prod) */ - if (IS_CNL_REVID(i915, 0, CNL_REVID_B0)) - WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, - GEN8_CSC2_SBE_VUE_CACHE_CONSERVATIVE); - /* WaPushConstantDereferenceHoldDisable:cnl */ WA_SET_BIT_MASKED(GEN7_ROW_CHICKEN2, PUSH_CONSTANT_DEREF_DISABLE); @@ -698,6 +693,227 @@ int intel_engine_emit_ctx_wa(struct i915_request *rq) } static void +gen4_gt_workarounds_init(struct drm_i915_private *i915, + struct i915_wa_list *wal) +{ + /* WaDisable_RenderCache_OperationalFlush:gen4,ilk */ + wa_masked_dis(wal, CACHE_MODE_0, RC_OP_FLUSH_ENABLE); +} + +static void +g4x_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) +{ + gen4_gt_workarounds_init(i915, wal); + + /* WaDisableRenderCachePipelinedFlush:g4x,ilk */ + wa_masked_en(wal, CACHE_MODE_0, CM0_PIPELINED_RENDER_FLUSH_DISABLE); +} + +static void +ilk_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) +{ + g4x_gt_workarounds_init(i915, wal); + + wa_masked_en(wal, _3D_CHICKEN2, _3D_CHICKEN2_WM_READ_PIPELINED); +} + +static void +snb_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) +{ + /* WaDisableHiZPlanesWhenMSAAEnabled:snb */ + wa_masked_en(wal, + _3D_CHICKEN, + _3D_CHICKEN_HIZ_PLANE_DISABLE_MSAA_4X_SNB); + + /* WaDisable_RenderCache_OperationalFlush:snb */ + wa_masked_dis(wal, CACHE_MODE_0, RC_OP_FLUSH_ENABLE); + + /* + * BSpec recommends 8x4 when MSAA is used, + * however in practice 16x4 seems fastest. + * + * Note that PS/WM thread counts depend on the WIZ hashing + * disable bit, which we don't touch here, but it's good + * to keep in mind (see 3DSTATE_PS and 3DSTATE_WM). + */ + wa_add(wal, + GEN6_GT_MODE, 0, + _MASKED_FIELD(GEN6_WIZ_HASHING_MASK, GEN6_WIZ_HASHING_16x4), + GEN6_WIZ_HASHING_16x4); + + wa_masked_dis(wal, CACHE_MODE_0, CM0_STC_EVICT_DISABLE_LRA_SNB); + + wa_masked_en(wal, + _3D_CHICKEN3, + /* WaStripsFansDisableFastClipPerformanceFix:snb */ + _3D_CHICKEN3_SF_DISABLE_FASTCLIP_CULL | + /* + * Bspec says: + * "This bit must be set if 3DSTATE_CLIP clip mode is set + * to normal and 3DSTATE_SF number of SF output attributes + * is more than 16." + */ + _3D_CHICKEN3_SF_DISABLE_PIPELINED_ATTR_FETCH); +} + +static void +ivb_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) +{ + /* WaDisableEarlyCull:ivb */ + wa_masked_en(wal, _3D_CHICKEN3, _3D_CHICKEN_SF_DISABLE_OBJEND_CULL); + + /* WaDisablePSDDualDispatchEnable:ivb */ + if (IS_IVB_GT1(i915)) + wa_masked_en(wal, + GEN7_HALF_SLICE_CHICKEN1, + GEN7_PSD_SINGLE_PORT_DISPATCH_ENABLE); + + /* WaDisable_RenderCache_OperationalFlush:ivb */ + wa_masked_dis(wal, CACHE_MODE_0_GEN7, RC_OP_FLUSH_ENABLE); + + /* Apply the WaDisableRHWOOptimizationForRenderHang:ivb workaround. */ + wa_masked_dis(wal, + GEN7_COMMON_SLICE_CHICKEN1, + GEN7_CSC1_RHWO_OPT_DISABLE_IN_RCC); + + /* WaApplyL3ControlAndL3ChickenMode:ivb */ + wa_write(wal, GEN7_L3CNTLREG1, GEN7_WA_FOR_GEN7_L3_CONTROL); + wa_write(wal, GEN7_L3_CHICKEN_MODE_REGISTER, GEN7_WA_L3_CHICKEN_MODE); + + /* WaForceL3Serialization:ivb */ + wa_write_clr(wal, GEN7_L3SQCREG4, L3SQ_URB_READ_CAM_MATCH_DISABLE); + + /* + * WaVSThreadDispatchOverride:ivb,vlv + * + * This actually overrides the dispatch + * mode for all thread types. + */ + wa_write_masked_or(wal, GEN7_FF_THREAD_MODE, + GEN7_FF_SCHED_MASK, + GEN7_FF_TS_SCHED_HW | + GEN7_FF_VS_SCHED_HW | + GEN7_FF_DS_SCHED_HW); + + if (0) { /* causes HiZ corruption on ivb:gt1 */ + /* enable HiZ Raw Stall Optimization */ + wa_masked_dis(wal, CACHE_MODE_0_GEN7, HIZ_RAW_STALL_OPT_DISABLE); + } + + /* WaDisable4x2SubspanOptimization:ivb */ + wa_masked_en(wal, CACHE_MODE_1, PIXEL_SUBSPAN_COLLECT_OPT_DISABLE); + + /* + * BSpec recommends 8x4 when MSAA is used, + * however in practice 16x4 seems fastest. + * + * Note that PS/WM thread counts depend on the WIZ hashing + * disable bit, which we don't touch here, but it's good + * to keep in mind (see 3DSTATE_PS and 3DSTATE_WM). + */ + wa_add(wal, GEN7_GT_MODE, 0, + _MASKED_FIELD(GEN6_WIZ_HASHING_MASK, GEN6_WIZ_HASHING_16x4), + GEN6_WIZ_HASHING_16x4); +} + +static void +vlv_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) +{ + /* WaDisableEarlyCull:vlv */ + wa_masked_en(wal, _3D_CHICKEN3, _3D_CHICKEN_SF_DISABLE_OBJEND_CULL); + + /* WaPsdDispatchEnable:vlv */ + /* WaDisablePSDDualDispatchEnable:vlv */ + wa_masked_en(wal, + GEN7_HALF_SLICE_CHICKEN1, + GEN7_MAX_PS_THREAD_DEP | + GEN7_PSD_SINGLE_PORT_DISPATCH_ENABLE); + + /* WaDisable_RenderCache_OperationalFlush:vlv */ + wa_masked_dis(wal, CACHE_MODE_0_GEN7, RC_OP_FLUSH_ENABLE); + + /* WaForceL3Serialization:vlv */ + wa_write_clr(wal, GEN7_L3SQCREG4, L3SQ_URB_READ_CAM_MATCH_DISABLE); + + /* + * WaVSThreadDispatchOverride:ivb,vlv + * + * This actually overrides the dispatch + * mode for all thread types. + */ + wa_write_masked_or(wal, + GEN7_FF_THREAD_MODE, + GEN7_FF_SCHED_MASK, + GEN7_FF_TS_SCHED_HW | + GEN7_FF_VS_SCHED_HW | + GEN7_FF_DS_SCHED_HW); + + /* + * BSpec says this must be set, even though + * WaDisable4x2SubspanOptimization isn't listed for VLV. + */ + wa_masked_en(wal, CACHE_MODE_1, PIXEL_SUBSPAN_COLLECT_OPT_DISABLE); + + /* + * BSpec recommends 8x4 when MSAA is used, + * however in practice 16x4 seems fastest. + * + * Note that PS/WM thread counts depend on the WIZ hashing + * disable bit, which we don't touch here, but it's good + * to keep in mind (see 3DSTATE_PS and 3DSTATE_WM). + */ + wa_add(wal, GEN7_GT_MODE, 0, + _MASKED_FIELD(GEN6_WIZ_HASHING_MASK, GEN6_WIZ_HASHING_16x4), + GEN6_WIZ_HASHING_16x4); + + /* + * WaIncreaseL3CreditsForVLVB0:vlv + * This is the hardware default actually. + */ + wa_write(wal, GEN7_L3SQCREG1, VLV_B0_WA_L3SQCREG1_VALUE); +} + +static void +hsw_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) +{ + /* L3 caching of data atomics doesn't work -- disable it. */ + wa_write(wal, HSW_SCRATCH1, HSW_SCRATCH1_L3_DATA_ATOMICS_DISABLE); + + wa_add(wal, + HSW_ROW_CHICKEN3, 0, + _MASKED_BIT_ENABLE(HSW_ROW_CHICKEN3_L3_GLOBAL_ATOMICS_DISABLE), + 0 /* XXX does this reg exist? */); + + /* WaVSRefCountFullforceMissDisable:hsw */ + wa_write_clr(wal, GEN7_FF_THREAD_MODE, GEN7_FF_VS_REF_CNT_FFME); + + wa_masked_dis(wal, + CACHE_MODE_0_GEN7, + /* WaDisable_RenderCache_OperationalFlush:hsw */ + RC_OP_FLUSH_ENABLE | + /* enable HiZ Raw Stall Optimization */ + HIZ_RAW_STALL_OPT_DISABLE); + + /* WaDisable4x2SubspanOptimization:hsw */ + wa_masked_en(wal, CACHE_MODE_1, PIXEL_SUBSPAN_COLLECT_OPT_DISABLE); + + /* + * BSpec recommends 8x4 when MSAA is used, + * however in practice 16x4 seems fastest. + * + * Note that PS/WM thread counts depend on the WIZ hashing + * disable bit, which we don't touch here, but it's good + * to keep in mind (see 3DSTATE_PS and 3DSTATE_WM). + */ + wa_add(wal, GEN7_GT_MODE, 0, + _MASKED_FIELD(GEN6_WIZ_HASHING_MASK, GEN6_WIZ_HASHING_16x4), + GEN6_WIZ_HASHING_16x4); + + /* WaSampleCChickenBitEnable:hsw */ + wa_masked_en(wal, HALF_SLICE_CHICKEN3, HSW_SAMPLE_C_PERFORMANCE); +} + +static void gen9_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) { /* WaDisableKillLogic:bxt,skl,kbl */ @@ -837,7 +1053,7 @@ wa_init_mcr(struct drm_i915_private *i915, struct i915_wa_list *wal) intel_uncore_read(&i915->uncore, GEN10_MIRROR_FUSE3) & GEN10_L3BANK_MASK; - DRM_DEBUG_DRIVER("L3 fuse = %x\n", l3_fuse); + drm_dbg(&i915->drm, "L3 fuse = %x\n", l3_fuse); l3_en = ~(l3_fuse << GEN10_L3BANK_PAIR_COUNT | l3_fuse); } else { l3_en = ~0; @@ -846,7 +1062,8 @@ wa_init_mcr(struct drm_i915_private *i915, struct i915_wa_list *wal) slice = fls(sseu->slice_mask) - 1; subslice = fls(l3_en & intel_sseu_get_subslices(sseu, slice)); if (!subslice) { - DRM_WARN("No common index found between subslice mask %x and L3 bank mask %x!\n", + drm_warn(&i915->drm, + "No common index found between subslice mask %x and L3 bank mask %x!\n", intel_sseu_get_subslices(sseu, slice), l3_en); subslice = fls(l3_en); drm_WARN_ON(&i915->drm, !subslice); @@ -861,7 +1078,7 @@ wa_init_mcr(struct drm_i915_private *i915, struct i915_wa_list *wal) mcr_mask = GEN8_MCR_SLICE_MASK | GEN8_MCR_SUBSLICE_MASK; } - DRM_DEBUG_DRIVER("MCR slice/subslice = %x\n", mcr); + drm_dbg(&i915->drm, "MCR slice/subslice = %x\n", mcr); wa_write_masked_or(wal, GEN8_MCR_SELECTOR, mcr_mask, mcr); } @@ -871,12 +1088,6 @@ cnl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) { wa_init_mcr(i915, wal); - /* WaDisableI2mCycleOnWRPort:cnl (pre-prod) */ - if (IS_CNL_REVID(i915, CNL_REVID_B0, CNL_REVID_B0)) - wa_write_or(wal, - GAMT_CHKN_BIT_REG, - GAMT_CHKN_DISABLE_I2M_CYCLE_ON_WR_PORT); - /* WaInPlaceDecompressionHang:cnl */ wa_write_or(wal, GEN9_GAMT_ECO_REG_RW_IA, @@ -933,15 +1144,20 @@ icl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) GAMT_CHKN_BIT_REG, GAMT_CHKN_DISABLE_L3_COH_PIPE); - /* Wa_1607087056:icl */ - wa_write_or(wal, - SLICE_UNIT_LEVEL_CLKGATE, - L3_CLKGATE_DIS | L3_CR2X_CLKGATE_DIS); + /* Wa_1607087056:icl,ehl,jsl */ + if (IS_ICELAKE(i915) || + IS_EHL_REVID(i915, EHL_REVID_A0, EHL_REVID_A0)) { + wa_write_or(wal, + SLICE_UNIT_LEVEL_CLKGATE, + L3_CLKGATE_DIS | L3_CR2X_CLKGATE_DIS); + } } static void tgl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) { + wa_init_mcr(i915, wal); + /* Wa_1409420604:tgl */ if (IS_TGL_REVID(i915, TGL_REVID_A0, TGL_REVID_A0)) wa_write_or(wal, @@ -974,6 +1190,20 @@ gt_init_workarounds(struct drm_i915_private *i915, struct i915_wa_list *wal) bxt_gt_workarounds_init(i915, wal); else if (IS_SKYLAKE(i915)) skl_gt_workarounds_init(i915, wal); + else if (IS_HASWELL(i915)) + hsw_gt_workarounds_init(i915, wal); + else if (IS_VALLEYVIEW(i915)) + vlv_gt_workarounds_init(i915, wal); + else if (IS_IVYBRIDGE(i915)) + ivb_gt_workarounds_init(i915, wal); + else if (IS_GEN(i915, 6)) + snb_gt_workarounds_init(i915, wal); + else if (IS_GEN(i915, 5)) + ilk_gt_workarounds_init(i915, wal); + else if (IS_G4X(i915)) + g4x_gt_workarounds_init(i915, wal); + else if (IS_GEN(i915, 4)) + gen4_gt_workarounds_init(i915, wal); else if (INTEL_GEN(i915) <= 8) return; else @@ -1379,12 +1609,6 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal) GEN7_FF_THREAD_MODE, GEN12_FF_TESSELATION_DOP_GATE_DISABLE); - /* - * Wa_1409085225:tgl - * Wa_14010229206:tgl - */ - wa_masked_en(wal, GEN9_ROW_CHICKEN4, GEN12_DISABLE_TDL_PUSH); - /* Wa_1408615072:tgl */ wa_write_or(wal, UNSLICE_UNIT_LEVEL_CLKGATE2, VSUNIT_CLKGATE_DIS_TGL); @@ -1402,6 +1626,12 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal) wa_masked_en(wal, GEN9_CS_DEBUG_MODE1, FF_DOP_CLOCK_GATE_DISABLE); + + /* + * Wa_1409085225:tgl + * Wa_14010229206:tgl + */ + wa_masked_en(wal, GEN9_ROW_CHICKEN4, GEN12_DISABLE_TDL_PUSH); } if (IS_GEN(i915, 11)) { |