From: Oscar Mateo <oscar.mateo@xxxxxxxxx> This has grown to be a sizable amount of code, so move it to its own file before we try to refactor anything. For the moment, we are leaving behind the WA BB code and the WAs that get applied (incorrectly) in init_clock_gating, but we will deal with it later. v2: Use intel_ prefix for code that deals with the hardware (Chris) v3: Rebased v4: - Rebased - New license header v5: - Rebased - Added some organisational notes to the file (Chris) v6: Include DOC section in the documentation build (Jani) Signed-off-by: Oscar Mateo <oscar.mateo@xxxxxxxxx> Cc: Mika Kuoppala <mika.kuoppala@xxxxxxxxxxxxxxx> Cc: Jani Nikula <jani.nikula@xxxxxxxxx> Reviewed-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> [ickle: appease checkpatch, mostly] Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> Link: https://patchwork.freedesktop.org/patch/msgid/1523376767-18480-1-git-send-email-oscar.mateo@xxxxxxxxx --- Documentation/gpu/i915.rst | 6 + drivers/gpu/drm/i915/Makefile | 3 +- drivers/gpu/drm/i915/intel_engine_cs.c | 634 --------------------- drivers/gpu/drm/i915/intel_lrc.c | 1 + drivers/gpu/drm/i915/intel_ringbuffer.c | 1 + drivers/gpu/drm/i915/intel_ringbuffer.h | 3 - drivers/gpu/drm/i915/intel_workarounds.c | 686 +++++++++++++++++++++++ drivers/gpu/drm/i915/intel_workarounds.h | 13 + 8 files changed, 709 insertions(+), 638 deletions(-) create mode 100644 drivers/gpu/drm/i915/intel_workarounds.c create mode 100644 drivers/gpu/drm/i915/intel_workarounds.h diff --git a/Documentation/gpu/i915.rst b/Documentation/gpu/i915.rst index 34d22f275708..055df45596c1 100644 --- a/Documentation/gpu/i915.rst +++ b/Documentation/gpu/i915.rst @@ -58,6 +58,12 @@ Intel GVT-g Host Support(vGPU device model) .. kernel-doc:: drivers/gpu/drm/i915/intel_gvt.c :internal: +Workarounds +----------- + +.. kernel-doc:: drivers/gpu/drm/i915/intel_workarounds.c + :doc: Hardware workarounds + Display Hardware Handling ========================= diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile index 0c79c19223af..9bee52a949a9 100644 --- a/drivers/gpu/drm/i915/Makefile +++ b/drivers/gpu/drm/i915/Makefile @@ -43,7 +43,8 @@ i915-y := i915_drv.o \ intel_csr.o \ intel_device_info.o \ intel_pm.o \ - intel_runtime_pm.o + intel_runtime_pm.o \ + intel_workarounds.o i915-$(CONFIG_COMPAT) += i915_ioc32.o i915-$(CONFIG_DEBUG_FS) += i915_debugfs.o intel_pipe_crc.o diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c index a217b3fe5f0b..68898d58dd1e 100644 --- a/drivers/gpu/drm/i915/intel_engine_cs.c +++ b/drivers/gpu/drm/i915/intel_engine_cs.c @@ -903,640 +903,6 @@ void intel_engine_get_instdone(struct intel_engine_cs *engine, } } -static int wa_add(struct drm_i915_private *dev_priv, - i915_reg_t addr, - const u32 mask, const u32 val) -{ - const u32 idx = dev_priv->workarounds.count; - - if (WARN_ON(idx >= I915_MAX_WA_REGS)) - return -ENOSPC; - - dev_priv->workarounds.reg[idx].addr = addr; - dev_priv->workarounds.reg[idx].value = val; - dev_priv->workarounds.reg[idx].mask = mask; - - dev_priv->workarounds.count++; - - return 0; -} - -#define WA_REG(addr, mask, val) do { \ - const int r = wa_add(dev_priv, (addr), (mask), (val)); \ - if (r) \ - return r; \ - } while (0) - -#define WA_SET_BIT_MASKED(addr, mask) \ - WA_REG(addr, (mask), _MASKED_BIT_ENABLE(mask)) - -#define WA_CLR_BIT_MASKED(addr, mask) \ - WA_REG(addr, (mask), _MASKED_BIT_DISABLE(mask)) - -#define WA_SET_FIELD_MASKED(addr, mask, value) \ - WA_REG(addr, mask, _MASKED_FIELD(mask, value)) - -static int wa_ring_whitelist_reg(struct intel_engine_cs *engine, - i915_reg_t reg) -{ - struct drm_i915_private *dev_priv = engine->i915; - struct i915_workarounds *wa = &dev_priv->workarounds; - const uint32_t index = wa->hw_whitelist_count[engine->id]; - - if (WARN_ON(index >= RING_MAX_NONPRIV_SLOTS)) - return -EINVAL; - - I915_WRITE(RING_FORCE_TO_NONPRIV(engine->mmio_base, index), - i915_mmio_reg_offset(reg)); - wa->hw_whitelist_count[engine->id]++; - - return 0; -} - -static int gen8_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - - WA_SET_BIT_MASKED(INSTPM, INSTPM_FORCE_ORDERING); - - /* WaDisableAsyncFlipPerfMode:bdw,chv */ - WA_SET_BIT_MASKED(MI_MODE, ASYNC_FLIP_PERF_DISABLE); - - /* WaDisablePartialInstShootdown:bdw,chv */ - WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, - PARTIAL_INSTRUCTION_SHOOTDOWN_DISABLE); - - /* Use Force Non-Coherent whenever executing a 3D context. This is a - * workaround for for a possible hang in the unlikely event a TLB - * invalidation occurs during a PSD flush. - */ - /* WaForceEnableNonCoherent:bdw,chv */ - /* WaHdcDisableFetchWhenMasked:bdw,chv */ - WA_SET_BIT_MASKED(HDC_CHICKEN0, - HDC_DONOT_FETCH_MEM_WHEN_MASKED | - HDC_FORCE_NON_COHERENT); - - /* From the Haswell PRM, Command Reference: Registers, CACHE_MODE_0: - * "The Hierarchical Z RAW Stall Optimization allows non-overlapping - * polygons in the same 8x4 pixel/sample area to be processed without - * stalling waiting for the earlier ones to write to Hierarchical Z - * buffer." - * - * This optimization is off by default for BDW and CHV; turn it on. - */ - WA_CLR_BIT_MASKED(CACHE_MODE_0_GEN7, HIZ_RAW_STALL_OPT_DISABLE); - - /* Wa4x4STCOptimizationDisable:bdw,chv */ - WA_SET_BIT_MASKED(CACHE_MODE_1, GEN8_4x4_STC_OPTIMIZATION_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_SET_FIELD_MASKED(GEN7_GT_MODE, - GEN6_WIZ_HASHING_MASK, - GEN6_WIZ_HASHING_16x4); - - return 0; -} - -static int bdw_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int ret; - - ret = gen8_init_workarounds(engine); - if (ret) - return ret; - - /* WaDisableThreadStallDopClockGating:bdw (pre-production) */ - WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, STALL_DOP_GATING_DISABLE); - - /* WaDisableDopClockGating:bdw - * - * Also see the related UCGTCL1 write in broadwell_init_clock_gating() - * to disable EUTC clock gating. - */ - WA_SET_BIT_MASKED(GEN7_ROW_CHICKEN2, - DOP_CLOCK_GATING_DISABLE); - - WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, - GEN8_SAMPLER_POWER_BYPASS_DIS); - - WA_SET_BIT_MASKED(HDC_CHICKEN0, - /* WaForceContextSaveRestoreNonCoherent:bdw */ - HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT | - /* WaDisableFenceDestinationToSLM:bdw (pre-prod) */ - (IS_BDW_GT3(dev_priv) ? HDC_FENCE_DEST_SLM_DISABLE : 0)); - - return 0; -} - -static int chv_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int ret; - - ret = gen8_init_workarounds(engine); - if (ret) - return ret; - - /* WaDisableThreadStallDopClockGating:chv */ - WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, STALL_DOP_GATING_DISABLE); - - /* Improve HiZ throughput on CHV. */ - WA_SET_BIT_MASKED(HIZ_CHICKEN, CHV_HZ_8X8_MODE_IN_1X); - - return 0; -} - -static int gen9_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int ret; - - /* WaConextSwitchWithConcurrentTLBInvalidate:skl,bxt,kbl,glk,cfl */ - I915_WRITE(GEN9_CSFE_CHICKEN1_RCS, _MASKED_BIT_ENABLE(GEN9_PREEMPT_GPGPU_SYNC_SWITCH_DISABLE)); - - /* WaEnableLbsSlaRetryTimerDecrement:skl,bxt,kbl,glk,cfl */ - I915_WRITE(BDW_SCRATCH1, I915_READ(BDW_SCRATCH1) | - GEN9_LBS_SLA_RETRY_TIMER_DECREMENT_ENABLE); - - /* WaDisableKillLogic:bxt,skl,kbl */ - if (!IS_COFFEELAKE(dev_priv)) - I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) | - ECOCHK_DIS_TLB); - - if (HAS_LLC(dev_priv)) { - /* WaCompressedResourceSamplerPbeMediaNewHashMode:skl,kbl - * - * Must match Display Engine. See - * WaCompressedResourceDisplayNewHashMode. - */ - WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, - GEN9_PBE_COMPRESSED_HASH_SELECTION); - WA_SET_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN7, - GEN9_SAMPLER_HASH_COMPRESSED_READ_ADDR); - - I915_WRITE(MMCD_MISC_CTRL, - I915_READ(MMCD_MISC_CTRL) | - MMCD_PCLA | - MMCD_HOTSPOT_EN); - } - - /* WaClearFlowControlGpgpuContextSave:skl,bxt,kbl,glk,cfl */ - /* WaDisablePartialInstShootdown:skl,bxt,kbl,glk,cfl */ - WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, - FLOW_CONTROL_ENABLE | - PARTIAL_INSTRUCTION_SHOOTDOWN_DISABLE); - - /* Syncing dependencies between camera and graphics:skl,bxt,kbl */ - if (!IS_COFFEELAKE(dev_priv)) - WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, - GEN9_DISABLE_OCL_OOB_SUPPRESS_LOGIC); - - /* WaEnableYV12BugFixInHalfSliceChicken7:skl,bxt,kbl,glk,cfl */ - /* WaEnableSamplerGPGPUPreemptionSupport:skl,bxt,kbl,cfl */ - WA_SET_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN7, - GEN9_ENABLE_YV12_BUGFIX | - GEN9_ENABLE_GPGPU_PREEMPTION); - - /* Wa4x4STCOptimizationDisable:skl,bxt,kbl,glk,cfl */ - /* WaDisablePartialResolveInVc:skl,bxt,kbl,cfl */ - WA_SET_BIT_MASKED(CACHE_MODE_1, (GEN8_4x4_STC_OPTIMIZATION_DISABLE | - GEN9_PARTIAL_RESOLVE_IN_VC_DISABLE)); - - /* WaCcsTlbPrefetchDisable:skl,bxt,kbl,glk,cfl */ - WA_CLR_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN5, - GEN9_CCS_TLB_PREFETCH_ENABLE); - - /* WaForceContextSaveRestoreNonCoherent:skl,bxt,kbl,cfl */ - WA_SET_BIT_MASKED(HDC_CHICKEN0, - HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT | - HDC_FORCE_CSR_NON_COHERENT_OVR_DISABLE); - - /* WaForceEnableNonCoherent and WaDisableHDCInvalidation are - * both tied to WaForceContextSaveRestoreNonCoherent - * in some hsds for skl. We keep the tie for all gen9. The - * documentation is a bit hazy and so we want to get common behaviour, - * even though there is no clear evidence we would need both on kbl/bxt. - * This area has been source of system hangs so we play it safe - * and mimic the skl regardless of what bspec says. - * - * Use Force Non-Coherent whenever executing a 3D context. This - * is a workaround for a possible hang in the unlikely event - * a TLB invalidation occurs during a PSD flush. - */ - - /* WaForceEnableNonCoherent:skl,bxt,kbl,cfl */ - WA_SET_BIT_MASKED(HDC_CHICKEN0, - HDC_FORCE_NON_COHERENT); - - /* WaDisableHDCInvalidation:skl,bxt,kbl,cfl */ - I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) | - BDW_DISABLE_HDC_INVALIDATION); - - /* WaDisableSamplerPowerBypassForSOPingPong:skl,bxt,kbl,cfl */ - if (IS_SKYLAKE(dev_priv) || - IS_KABYLAKE(dev_priv) || - IS_COFFEELAKE(dev_priv)) - WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, - GEN8_SAMPLER_POWER_BYPASS_DIS); - - /* WaDisableSTUnitPowerOptimization:skl,bxt,kbl,glk,cfl */ - WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN2, GEN8_ST_PO_DISABLE); - - /* WaProgramL3SqcReg1DefaultForPerf:bxt,glk */ - if (IS_GEN9_LP(dev_priv)) { - u32 val = I915_READ(GEN8_L3SQCREG1); - - val &= ~L3_PRIO_CREDITS_MASK; - val |= L3_GENERAL_PRIO_CREDITS(62) | L3_HIGH_PRIO_CREDITS(2); - I915_WRITE(GEN8_L3SQCREG1, val); - } - - /* WaOCLCoherentLineFlush:skl,bxt,kbl,cfl */ - I915_WRITE(GEN8_L3SQCREG4, (I915_READ(GEN8_L3SQCREG4) | - GEN8_LQSC_FLUSH_COHERENT_LINES)); - - /* - * Supporting preemption with fine-granularity requires changes in the - * batch buffer programming. Since we can't break old userspace, we - * need to set our default preemption level to safe value. Userspace is - * still able to use more fine-grained preemption levels, since in - * WaEnablePreemptionGranularityControlByUMD we're whitelisting the - * per-ctx register. As such, WaDisable{3D,GPGPU}MidCmdPreemption are - * not real HW workarounds, but merely a way to start using preemption - * while maintaining old contract with userspace. - */ - - /* WaDisable3DMidCmdPreemption:skl,bxt,glk,cfl,[cnl] */ - WA_CLR_BIT_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_3D_OBJECT_LEVEL); - - /* WaDisableGPGPUMidCmdPreemption:skl,bxt,blk,cfl,[cnl] */ - WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_GPGPU_LEVEL_MASK, - GEN9_PREEMPT_GPGPU_COMMAND_LEVEL); - - /* WaVFEStateAfterPipeControlwithMediaStateClear:skl,bxt,glk,cfl */ - ret = wa_ring_whitelist_reg(engine, GEN9_CTX_PREEMPT_REG); - if (ret) - return ret; - - /* WaEnablePreemptionGranularityControlByUMD:skl,bxt,kbl,cfl,[cnl] */ - I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1, - _MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL)); - ret = wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1); - if (ret) - return ret; - - /* WaAllowUMDToModifyHDCChicken1:skl,bxt,kbl,glk,cfl */ - ret = wa_ring_whitelist_reg(engine, GEN8_HDC_CHICKEN1); - if (ret) - return ret; - - return 0; -} - -static int skl_tune_iz_hashing(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - u8 vals[3] = { 0, 0, 0 }; - unsigned int i; - - for (i = 0; i < 3; i++) { - u8 ss; - - /* - * Only consider slices where one, and only one, subslice has 7 - * EUs - */ - if (!is_power_of_2(INTEL_INFO(dev_priv)->sseu.subslice_7eu[i])) - continue; - - /* - * subslice_7eu[i] != 0 (because of the check above) and - * ss_max == 4 (maximum number of subslices possible per slice) - * - * -> 0 <= ss <= 3; - */ - ss = ffs(INTEL_INFO(dev_priv)->sseu.subslice_7eu[i]) - 1; - vals[i] = 3 - ss; - } - - if (vals[0] == 0 && vals[1] == 0 && vals[2] == 0) - return 0; - - /* Tune IZ hashing. See intel_device_info_runtime_init() */ - WA_SET_FIELD_MASKED(GEN7_GT_MODE, - GEN9_IZ_HASHING_MASK(2) | - GEN9_IZ_HASHING_MASK(1) | - GEN9_IZ_HASHING_MASK(0), - GEN9_IZ_HASHING(2, vals[2]) | - GEN9_IZ_HASHING(1, vals[1]) | - GEN9_IZ_HASHING(0, vals[0])); - - return 0; -} - -static int skl_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int ret; - - ret = gen9_init_workarounds(engine); - if (ret) - return ret; - - /* WaEnableGapsTsvCreditFix:skl */ - I915_WRITE(GEN8_GARBCNTL, (I915_READ(GEN8_GARBCNTL) | - GEN9_GAPS_TSV_CREDIT_DISABLE)); - - /* WaDisableGafsUnitClkGating:skl */ - I915_WRITE(GEN7_UCGCTL4, (I915_READ(GEN7_UCGCTL4) | - GEN8_EU_GAUNIT_CLOCK_GATE_DISABLE)); - - /* WaInPlaceDecompressionHang:skl */ - if (IS_SKL_REVID(dev_priv, SKL_REVID_H0, REVID_FOREVER)) - I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, - (I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | - GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS)); - - /* WaDisableLSQCROPERFforOCL:skl */ - ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4); - if (ret) - return ret; - - return skl_tune_iz_hashing(engine); -} - -static int bxt_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int ret; - - ret = gen9_init_workarounds(engine); - if (ret) - return ret; - - /* WaDisableThreadStallDopClockGating:bxt */ - WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, - STALL_DOP_GATING_DISABLE); - - /* WaDisablePooledEuLoadBalancingFix:bxt */ - I915_WRITE(FF_SLICE_CS_CHICKEN2, - _MASKED_BIT_ENABLE(GEN9_POOLED_EU_LOAD_BALANCING_FIX_DISABLE)); - - /* WaToEnableHwFixForPushConstHWBug:bxt */ - WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, - GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION); - - /* WaInPlaceDecompressionHang:bxt */ - I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, - (I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | - GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS)); - - return 0; -} - -static int cnl_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int ret; - - /* WaDisableI2mCycleOnWRPort:cnl (pre-prod) */ - if (IS_CNL_REVID(dev_priv, CNL_REVID_B0, CNL_REVID_B0)) - I915_WRITE(GAMT_CHKN_BIT_REG, - (I915_READ(GAMT_CHKN_BIT_REG) | - GAMT_CHKN_DISABLE_I2M_CYCLE_ON_WR_PORT)); - - /* WaForceContextSaveRestoreNonCoherent:cnl */ - WA_SET_BIT_MASKED(CNL_HDC_CHICKEN0, - HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT); - - /* WaThrottleEUPerfToAvoidTDBackPressure:cnl(pre-prod) */ - if (IS_CNL_REVID(dev_priv, 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(dev_priv, 0, CNL_REVID_B0)) - WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, - GEN8_CSC2_SBE_VUE_CACHE_CONSERVATIVE); - - /* WaInPlaceDecompressionHang:cnl */ - I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, - (I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | - GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS)); - - /* WaPushConstantDereferenceHoldDisable:cnl */ - WA_SET_BIT_MASKED(GEN7_ROW_CHICKEN2, PUSH_CONSTANT_DEREF_DISABLE); - - /* FtrEnableFastAnisoL1BankingFix: cnl */ - WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, CNL_FAST_ANISO_L1_BANKING_FIX); - - /* WaDisable3DMidCmdPreemption:cnl */ - WA_CLR_BIT_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_3D_OBJECT_LEVEL); - - /* WaDisableGPGPUMidCmdPreemption:cnl */ - WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_GPGPU_LEVEL_MASK, - GEN9_PREEMPT_GPGPU_COMMAND_LEVEL); - - /* WaEnablePreemptionGranularityControlByUMD:cnl */ - I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1, - _MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL)); - ret= wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1); - if (ret) - return ret; - - /* WaDisableEarlyEOT:cnl */ - WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, DISABLE_EARLY_EOT); - - return 0; -} - -static int kbl_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int ret; - - ret = gen9_init_workarounds(engine); - if (ret) - return ret; - - /* WaEnableGapsTsvCreditFix:kbl */ - I915_WRITE(GEN8_GARBCNTL, (I915_READ(GEN8_GARBCNTL) | - GEN9_GAPS_TSV_CREDIT_DISABLE)); - - /* WaDisableDynamicCreditSharing:kbl */ - if (IS_KBL_REVID(dev_priv, 0, KBL_REVID_B0)) - I915_WRITE(GAMT_CHKN_BIT_REG, - (I915_READ(GAMT_CHKN_BIT_REG) | - GAMT_CHKN_DISABLE_DYNAMIC_CREDIT_SHARING)); - - /* WaDisableFenceDestinationToSLM:kbl (pre-prod) */ - if (IS_KBL_REVID(dev_priv, KBL_REVID_A0, KBL_REVID_A0)) - WA_SET_BIT_MASKED(HDC_CHICKEN0, - HDC_FENCE_DEST_SLM_DISABLE); - - /* WaToEnableHwFixForPushConstHWBug:kbl */ - if (IS_KBL_REVID(dev_priv, KBL_REVID_C0, REVID_FOREVER)) - WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, - GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION); - - /* WaDisableGafsUnitClkGating:kbl */ - I915_WRITE(GEN7_UCGCTL4, (I915_READ(GEN7_UCGCTL4) | - GEN8_EU_GAUNIT_CLOCK_GATE_DISABLE)); - - /* WaDisableSbeCacheDispatchPortSharing:kbl */ - WA_SET_BIT_MASKED( - GEN7_HALF_SLICE_CHICKEN1, - GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE); - - /* WaInPlaceDecompressionHang:kbl */ - I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, - (I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | - GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS)); - - /* WaDisableLSQCROPERFforOCL:kbl */ - ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4); - if (ret) - return ret; - - return 0; -} - -static int glk_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int ret; - - ret = gen9_init_workarounds(engine); - if (ret) - return ret; - - /* WA #0862: Userspace has to set "Barrier Mode" to avoid hangs. */ - ret = wa_ring_whitelist_reg(engine, GEN9_SLICE_COMMON_ECO_CHICKEN1); - if (ret) - return ret; - - /* WaToEnableHwFixForPushConstHWBug:glk */ - WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, - GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION); - - return 0; -} - -static int cfl_init_workarounds(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int ret; - - ret = gen9_init_workarounds(engine); - if (ret) - return ret; - - /* WaEnableGapsTsvCreditFix:cfl */ - I915_WRITE(GEN8_GARBCNTL, (I915_READ(GEN8_GARBCNTL) | - GEN9_GAPS_TSV_CREDIT_DISABLE)); - - /* WaToEnableHwFixForPushConstHWBug:cfl */ - WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, - GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION); - - /* WaDisableGafsUnitClkGating:cfl */ - I915_WRITE(GEN7_UCGCTL4, (I915_READ(GEN7_UCGCTL4) | - GEN8_EU_GAUNIT_CLOCK_GATE_DISABLE)); - - /* WaDisableSbeCacheDispatchPortSharing:cfl */ - WA_SET_BIT_MASKED( - GEN7_HALF_SLICE_CHICKEN1, - GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE); - - /* WaInPlaceDecompressionHang:cfl */ - I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, - (I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | - GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS)); - - return 0; -} - -int init_workarounds_ring(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - int err; - - if (GEM_WARN_ON(engine->id != RCS)) - return -EINVAL; - - dev_priv->workarounds.count = 0; - dev_priv->workarounds.hw_whitelist_count[engine->id] = 0; - - if (IS_BROADWELL(dev_priv)) - err = bdw_init_workarounds(engine); - else if (IS_CHERRYVIEW(dev_priv)) - err = chv_init_workarounds(engine); - else if (IS_SKYLAKE(dev_priv)) - err = skl_init_workarounds(engine); - else if (IS_BROXTON(dev_priv)) - err = bxt_init_workarounds(engine); - else if (IS_KABYLAKE(dev_priv)) - err = kbl_init_workarounds(engine); - else if (IS_GEMINILAKE(dev_priv)) - err = glk_init_workarounds(engine); - else if (IS_COFFEELAKE(dev_priv)) - err = cfl_init_workarounds(engine); - else if (IS_CANNONLAKE(dev_priv)) - err = cnl_init_workarounds(engine); - else - err = 0; - if (err) - return err; - - DRM_DEBUG_DRIVER("%s: Number of context specific w/a: %d\n", - engine->name, dev_priv->workarounds.count); - return 0; -} - -int intel_ring_workarounds_emit(struct i915_request *rq) -{ - struct i915_workarounds *w = &rq->i915->workarounds; - u32 *cs; - int ret, i; - - if (w->count == 0) - return 0; - - ret = rq->engine->emit_flush(rq, EMIT_BARRIER); - if (ret) - return ret; - - cs = intel_ring_begin(rq, w->count * 2 + 2); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = MI_LOAD_REGISTER_IMM(w->count); - for (i = 0; i < w->count; i++) { - *cs++ = i915_mmio_reg_offset(w->reg[i].addr); - *cs++ = w->reg[i].value; - } - *cs++ = MI_NOOP; - - intel_ring_advance(rq, cs); - - ret = rq->engine->emit_flush(rq, EMIT_BARRIER); - if (ret) - return ret; - - return 0; -} - static bool ring_is_idle(struct intel_engine_cs *engine) { struct drm_i915_private *dev_priv = engine->i915; diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c index 665d9e82e954..03b9d5ae883a 100644 --- a/drivers/gpu/drm/i915/intel_lrc.c +++ b/drivers/gpu/drm/i915/intel_lrc.c @@ -139,6 +139,7 @@ #include "i915_gem_render_state.h" #include "intel_lrc_reg.h" #include "intel_mocs.h" +#include "intel_workarounds.h" #define RING_EXECLIST_QFULL (1 << 0x2) #define RING_EXECLIST1_VALID (1 << 0x3) diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index 04d9d9a946a7..36acc32374e4 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -36,6 +36,7 @@ #include "i915_gem_render_state.h" #include "i915_trace.h" #include "intel_drv.h" +#include "intel_workarounds.h" /* Rough estimate of the typical request size, performing a flush, * set-context and then emitting the batch. diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h index 256d58487559..717041640135 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.h +++ b/drivers/gpu/drm/i915/intel_ringbuffer.h @@ -885,9 +885,6 @@ static inline u32 intel_engine_last_submit(struct intel_engine_cs *engine) return READ_ONCE(engine->timeline->seqno); } -int init_workarounds_ring(struct intel_engine_cs *engine); -int intel_ring_workarounds_emit(struct i915_request *rq); - void intel_engine_get_instdone(struct intel_engine_cs *engine, struct intel_instdone *instdone); diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c new file mode 100644 index 000000000000..d60a37700f84 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_workarounds.c @@ -0,0 +1,686 @@ +/* + * SPDX-License-Identifier: MIT + * + * Copyright © 2014-2018 Intel Corporation + */ + +#include "i915_drv.h" +#include "intel_workarounds.h" + +/** + * DOC: Hardware workarounds + * + * This file is intended as a central place to implement most [1]_ of the + * required workarounds for hardware to work as originally intended. They fall + * in five basic categories depending on how/when they are applied: + * + * - Workarounds that touch registers that are saved/restored to/from the HW + * context image. The list is emitted (via Load Register Immediate commands) + * everytime a new context is created. + * - GT workarounds. The list of these WAs is applied whenever these registers + * revert to default values (on GPU reset, suspend/resume [2]_, etc..). + * - Display workarounds. The list is applied during display clock-gating + * initialization. + * - Workarounds that whitelist a privileged register, so that UMDs can manage + * them directly. This is just a special case of a MMMIO workaround (as we + * write the list of these to/be-whitelisted registers to some special HW + * registers). + * - Workaround batchbuffers, that get executed automatically by the hardware + * on every HW context restore. + * + * .. [1] Please notice that there are other WAs that, due to their nature, + * cannot be applied from a central place. Those are peppered around the rest + * of the code, as needed. + * + * .. [2] Technically, some registers are powercontext saved & restored, so they + * survive a suspend/resume. In practice, writing them again is not too + * costly and simplifies things. We can revisit this in the future. + * + * Layout + * '''''' + * + * Keep things in this file ordered by WA type, as per the above (context, GT, + * display, register whitelist, batchbuffer). Then, inside each type, keep the + * following order: + * + * - Infrastructure functions and macros + * - WAs per platform in standard gen/chrono order + * - Public functions to init or apply the given workaround type. + */ + +static int wa_add(struct drm_i915_private *dev_priv, + i915_reg_t addr, + const u32 mask, const u32 val) +{ + const unsigned int idx = dev_priv->workarounds.count; + + if (WARN_ON(idx >= I915_MAX_WA_REGS)) + return -ENOSPC; + + dev_priv->workarounds.reg[idx].addr = addr; + dev_priv->workarounds.reg[idx].value = val; + dev_priv->workarounds.reg[idx].mask = mask; + + dev_priv->workarounds.count++; + + return 0; +} + +#define WA_REG(addr, mask, val) do { \ + const int r = wa_add(dev_priv, (addr), (mask), (val)); \ + if (r) \ + return r; \ + } while (0) + +#define WA_SET_BIT_MASKED(addr, mask) \ + WA_REG(addr, (mask), _MASKED_BIT_ENABLE(mask)) + +#define WA_CLR_BIT_MASKED(addr, mask) \ + WA_REG(addr, (mask), _MASKED_BIT_DISABLE(mask)) + +#define WA_SET_FIELD_MASKED(addr, mask, value) \ + WA_REG(addr, (mask), _MASKED_FIELD(mask, value)) + +static int wa_ring_whitelist_reg(struct intel_engine_cs *engine, + i915_reg_t reg) +{ + struct drm_i915_private *dev_priv = engine->i915; + struct i915_workarounds *wa = &dev_priv->workarounds; + const unsigned int index = wa->hw_whitelist_count[engine->id]; + + if (WARN_ON(index >= RING_MAX_NONPRIV_SLOTS)) + return -EINVAL; + + I915_WRITE(RING_FORCE_TO_NONPRIV(engine->mmio_base, index), + i915_mmio_reg_offset(reg)); + wa->hw_whitelist_count[engine->id]++; + + return 0; +} + +static int gen8_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + + WA_SET_BIT_MASKED(INSTPM, INSTPM_FORCE_ORDERING); + + /* WaDisableAsyncFlipPerfMode:bdw,chv */ + WA_SET_BIT_MASKED(MI_MODE, ASYNC_FLIP_PERF_DISABLE); + + /* WaDisablePartialInstShootdown:bdw,chv */ + WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, + PARTIAL_INSTRUCTION_SHOOTDOWN_DISABLE); + + /* Use Force Non-Coherent whenever executing a 3D context. This is a + * workaround for for a possible hang in the unlikely event a TLB + * invalidation occurs during a PSD flush. + */ + /* WaForceEnableNonCoherent:bdw,chv */ + /* WaHdcDisableFetchWhenMasked:bdw,chv */ + WA_SET_BIT_MASKED(HDC_CHICKEN0, + HDC_DONOT_FETCH_MEM_WHEN_MASKED | + HDC_FORCE_NON_COHERENT); + + /* From the Haswell PRM, Command Reference: Registers, CACHE_MODE_0: + * "The Hierarchical Z RAW Stall Optimization allows non-overlapping + * polygons in the same 8x4 pixel/sample area to be processed without + * stalling waiting for the earlier ones to write to Hierarchical Z + * buffer." + * + * This optimization is off by default for BDW and CHV; turn it on. + */ + WA_CLR_BIT_MASKED(CACHE_MODE_0_GEN7, HIZ_RAW_STALL_OPT_DISABLE); + + /* Wa4x4STCOptimizationDisable:bdw,chv */ + WA_SET_BIT_MASKED(CACHE_MODE_1, GEN8_4x4_STC_OPTIMIZATION_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_SET_FIELD_MASKED(GEN7_GT_MODE, + GEN6_WIZ_HASHING_MASK, + GEN6_WIZ_HASHING_16x4); + + return 0; +} + +static int bdw_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int ret; + + ret = gen8_init_workarounds(engine); + if (ret) + return ret; + + /* WaDisableThreadStallDopClockGating:bdw (pre-production) */ + WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, STALL_DOP_GATING_DISABLE); + + /* WaDisableDopClockGating:bdw + * + * Also see the related UCGTCL1 write in broadwell_init_clock_gating() + * to disable EUTC clock gating. + */ + WA_SET_BIT_MASKED(GEN7_ROW_CHICKEN2, + DOP_CLOCK_GATING_DISABLE); + + WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, + GEN8_SAMPLER_POWER_BYPASS_DIS); + + WA_SET_BIT_MASKED(HDC_CHICKEN0, + /* WaForceContextSaveRestoreNonCoherent:bdw */ + HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT | + /* WaDisableFenceDestinationToSLM:bdw (pre-prod) */ + (IS_BDW_GT3(dev_priv) ? HDC_FENCE_DEST_SLM_DISABLE : 0)); + + return 0; +} + +static int chv_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int ret; + + ret = gen8_init_workarounds(engine); + if (ret) + return ret; + + /* WaDisableThreadStallDopClockGating:chv */ + WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, STALL_DOP_GATING_DISABLE); + + /* Improve HiZ throughput on CHV. */ + WA_SET_BIT_MASKED(HIZ_CHICKEN, CHV_HZ_8X8_MODE_IN_1X); + + return 0; +} + +static int gen9_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int ret; + + /* WaConextSwitchWithConcurrentTLBInvalidate:skl,bxt,kbl,glk,cfl */ + I915_WRITE(GEN9_CSFE_CHICKEN1_RCS, + _MASKED_BIT_ENABLE(GEN9_PREEMPT_GPGPU_SYNC_SWITCH_DISABLE)); + + /* WaEnableLbsSlaRetryTimerDecrement:skl,bxt,kbl,glk,cfl */ + I915_WRITE(BDW_SCRATCH1, + I915_READ(BDW_SCRATCH1) | + GEN9_LBS_SLA_RETRY_TIMER_DECREMENT_ENABLE); + + /* WaDisableKillLogic:bxt,skl,kbl */ + if (!IS_COFFEELAKE(dev_priv)) + I915_WRITE(GAM_ECOCHK, + I915_READ(GAM_ECOCHK) | ECOCHK_DIS_TLB); + + if (HAS_LLC(dev_priv)) { + /* WaCompressedResourceSamplerPbeMediaNewHashMode:skl,kbl + * + * Must match Display Engine. See + * WaCompressedResourceDisplayNewHashMode. + */ + WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, + GEN9_PBE_COMPRESSED_HASH_SELECTION); + WA_SET_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN7, + GEN9_SAMPLER_HASH_COMPRESSED_READ_ADDR); + + I915_WRITE(MMCD_MISC_CTRL, + I915_READ(MMCD_MISC_CTRL) | + MMCD_PCLA | + MMCD_HOTSPOT_EN); + } + + /* WaClearFlowControlGpgpuContextSave:skl,bxt,kbl,glk,cfl */ + /* WaDisablePartialInstShootdown:skl,bxt,kbl,glk,cfl */ + WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, + FLOW_CONTROL_ENABLE | + PARTIAL_INSTRUCTION_SHOOTDOWN_DISABLE); + + /* Syncing dependencies between camera and graphics:skl,bxt,kbl */ + if (!IS_COFFEELAKE(dev_priv)) + WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, + GEN9_DISABLE_OCL_OOB_SUPPRESS_LOGIC); + + /* WaEnableYV12BugFixInHalfSliceChicken7:skl,bxt,kbl,glk,cfl */ + /* WaEnableSamplerGPGPUPreemptionSupport:skl,bxt,kbl,cfl */ + WA_SET_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN7, + GEN9_ENABLE_YV12_BUGFIX | + GEN9_ENABLE_GPGPU_PREEMPTION); + + /* Wa4x4STCOptimizationDisable:skl,bxt,kbl,glk,cfl */ + /* WaDisablePartialResolveInVc:skl,bxt,kbl,cfl */ + WA_SET_BIT_MASKED(CACHE_MODE_1, + GEN8_4x4_STC_OPTIMIZATION_DISABLE | + GEN9_PARTIAL_RESOLVE_IN_VC_DISABLE); + + /* WaCcsTlbPrefetchDisable:skl,bxt,kbl,glk,cfl */ + WA_CLR_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN5, + GEN9_CCS_TLB_PREFETCH_ENABLE); + + /* WaForceContextSaveRestoreNonCoherent:skl,bxt,kbl,cfl */ + WA_SET_BIT_MASKED(HDC_CHICKEN0, + HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT | + HDC_FORCE_CSR_NON_COHERENT_OVR_DISABLE); + + /* WaForceEnableNonCoherent and WaDisableHDCInvalidation are + * both tied to WaForceContextSaveRestoreNonCoherent + * in some hsds for skl. We keep the tie for all gen9. The + * documentation is a bit hazy and so we want to get common behaviour, + * even though there is no clear evidence we would need both on kbl/bxt. + * This area has been source of system hangs so we play it safe + * and mimic the skl regardless of what bspec says. + * + * Use Force Non-Coherent whenever executing a 3D context. This + * is a workaround for a possible hang in the unlikely event + * a TLB invalidation occurs during a PSD flush. + */ + + /* WaForceEnableNonCoherent:skl,bxt,kbl,cfl */ + WA_SET_BIT_MASKED(HDC_CHICKEN0, + HDC_FORCE_NON_COHERENT); + + /* WaDisableHDCInvalidation:skl,bxt,kbl,cfl */ + I915_WRITE(GAM_ECOCHK, + I915_READ(GAM_ECOCHK) | BDW_DISABLE_HDC_INVALIDATION); + + /* WaDisableSamplerPowerBypassForSOPingPong:skl,bxt,kbl,cfl */ + if (IS_SKYLAKE(dev_priv) || + IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv)) + WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, + GEN8_SAMPLER_POWER_BYPASS_DIS); + + /* WaDisableSTUnitPowerOptimization:skl,bxt,kbl,glk,cfl */ + WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN2, GEN8_ST_PO_DISABLE); + + /* WaProgramL3SqcReg1DefaultForPerf:bxt,glk */ + if (IS_GEN9_LP(dev_priv)) { + u32 val = I915_READ(GEN8_L3SQCREG1); + + val &= ~L3_PRIO_CREDITS_MASK; + val |= L3_GENERAL_PRIO_CREDITS(62) | L3_HIGH_PRIO_CREDITS(2); + I915_WRITE(GEN8_L3SQCREG1, val); + } + + /* WaOCLCoherentLineFlush:skl,bxt,kbl,cfl */ + I915_WRITE(GEN8_L3SQCREG4, + I915_READ(GEN8_L3SQCREG4) | GEN8_LQSC_FLUSH_COHERENT_LINES); + + /* + * Supporting preemption with fine-granularity requires changes in the + * batch buffer programming. Since we can't break old userspace, we + * need to set our default preemption level to safe value. Userspace is + * still able to use more fine-grained preemption levels, since in + * WaEnablePreemptionGranularityControlByUMD we're whitelisting the + * per-ctx register. As such, WaDisable{3D,GPGPU}MidCmdPreemption are + * not real HW workarounds, but merely a way to start using preemption + * while maintaining old contract with userspace. + */ + + /* WaDisable3DMidCmdPreemption:skl,bxt,glk,cfl,[cnl] */ + WA_CLR_BIT_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_3D_OBJECT_LEVEL); + + /* WaDisableGPGPUMidCmdPreemption:skl,bxt,blk,cfl,[cnl] */ + WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1, + GEN9_PREEMPT_GPGPU_LEVEL_MASK, + GEN9_PREEMPT_GPGPU_COMMAND_LEVEL); + + /* WaVFEStateAfterPipeControlwithMediaStateClear:skl,bxt,glk,cfl */ + ret = wa_ring_whitelist_reg(engine, GEN9_CTX_PREEMPT_REG); + if (ret) + return ret; + + /* WaEnablePreemptionGranularityControlByUMD:skl,bxt,kbl,cfl,[cnl] */ + I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1, + _MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL)); + ret = wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1); + if (ret) + return ret; + + /* WaAllowUMDToModifyHDCChicken1:skl,bxt,kbl,glk,cfl */ + ret = wa_ring_whitelist_reg(engine, GEN8_HDC_CHICKEN1); + if (ret) + return ret; + + return 0; +} + +static int skl_tune_iz_hashing(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + u8 vals[3] = { 0, 0, 0 }; + unsigned int i; + + for (i = 0; i < 3; i++) { + u8 ss; + + /* + * Only consider slices where one, and only one, subslice has 7 + * EUs + */ + if (!is_power_of_2(INTEL_INFO(dev_priv)->sseu.subslice_7eu[i])) + continue; + + /* + * subslice_7eu[i] != 0 (because of the check above) and + * ss_max == 4 (maximum number of subslices possible per slice) + * + * -> 0 <= ss <= 3; + */ + ss = ffs(INTEL_INFO(dev_priv)->sseu.subslice_7eu[i]) - 1; + vals[i] = 3 - ss; + } + + if (vals[0] == 0 && vals[1] == 0 && vals[2] == 0) + return 0; + + /* Tune IZ hashing. See intel_device_info_runtime_init() */ + WA_SET_FIELD_MASKED(GEN7_GT_MODE, + GEN9_IZ_HASHING_MASK(2) | + GEN9_IZ_HASHING_MASK(1) | + GEN9_IZ_HASHING_MASK(0), + GEN9_IZ_HASHING(2, vals[2]) | + GEN9_IZ_HASHING(1, vals[1]) | + GEN9_IZ_HASHING(0, vals[0])); + + return 0; +} + +static int skl_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int ret; + + ret = gen9_init_workarounds(engine); + if (ret) + return ret; + + /* WaEnableGapsTsvCreditFix:skl */ + I915_WRITE(GEN8_GARBCNTL, + I915_READ(GEN8_GARBCNTL) | GEN9_GAPS_TSV_CREDIT_DISABLE); + + /* WaDisableGafsUnitClkGating:skl */ + I915_WRITE(GEN7_UCGCTL4, + I915_READ(GEN7_UCGCTL4) | GEN8_EU_GAUNIT_CLOCK_GATE_DISABLE); + + /* WaInPlaceDecompressionHang:skl */ + if (IS_SKL_REVID(dev_priv, SKL_REVID_H0, REVID_FOREVER)) + I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, + I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | + GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS); + + /* WaDisableLSQCROPERFforOCL:skl */ + ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4); + if (ret) + return ret; + + return skl_tune_iz_hashing(engine); +} + +static int bxt_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int ret; + + ret = gen9_init_workarounds(engine); + if (ret) + return ret; + + /* WaDisableThreadStallDopClockGating:bxt */ + WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, + STALL_DOP_GATING_DISABLE); + + /* WaDisablePooledEuLoadBalancingFix:bxt */ + I915_WRITE(FF_SLICE_CS_CHICKEN2, + _MASKED_BIT_ENABLE(GEN9_POOLED_EU_LOAD_BALANCING_FIX_DISABLE)); + + /* WaToEnableHwFixForPushConstHWBug:bxt */ + WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, + GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION); + + /* WaInPlaceDecompressionHang:bxt */ + I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, + I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | + GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS); + + return 0; +} + +static int cnl_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int ret; + + /* WaDisableI2mCycleOnWRPort:cnl (pre-prod) */ + if (IS_CNL_REVID(dev_priv, CNL_REVID_B0, CNL_REVID_B0)) + I915_WRITE(GAMT_CHKN_BIT_REG, + I915_READ(GAMT_CHKN_BIT_REG) | + GAMT_CHKN_DISABLE_I2M_CYCLE_ON_WR_PORT); + + /* WaForceContextSaveRestoreNonCoherent:cnl */ + WA_SET_BIT_MASKED(CNL_HDC_CHICKEN0, + HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT); + + /* WaThrottleEUPerfToAvoidTDBackPressure:cnl(pre-prod) */ + if (IS_CNL_REVID(dev_priv, 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(dev_priv, 0, CNL_REVID_B0)) + WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, + GEN8_CSC2_SBE_VUE_CACHE_CONSERVATIVE); + + /* WaInPlaceDecompressionHang:cnl */ + I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, + I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | + GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS); + + /* WaPushConstantDereferenceHoldDisable:cnl */ + WA_SET_BIT_MASKED(GEN7_ROW_CHICKEN2, PUSH_CONSTANT_DEREF_DISABLE); + + /* FtrEnableFastAnisoL1BankingFix: cnl */ + WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, CNL_FAST_ANISO_L1_BANKING_FIX); + + /* WaDisable3DMidCmdPreemption:cnl */ + WA_CLR_BIT_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_3D_OBJECT_LEVEL); + + /* WaDisableGPGPUMidCmdPreemption:cnl */ + WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1, + GEN9_PREEMPT_GPGPU_LEVEL_MASK, + GEN9_PREEMPT_GPGPU_COMMAND_LEVEL); + + /* WaEnablePreemptionGranularityControlByUMD:cnl */ + I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1, + _MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL)); + ret = wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1); + if (ret) + return ret; + + /* WaDisableEarlyEOT:cnl */ + WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, DISABLE_EARLY_EOT); + + return 0; +} + +static int kbl_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int ret; + + ret = gen9_init_workarounds(engine); + if (ret) + return ret; + + /* WaEnableGapsTsvCreditFix:kbl */ + I915_WRITE(GEN8_GARBCNTL, + I915_READ(GEN8_GARBCNTL) | GEN9_GAPS_TSV_CREDIT_DISABLE); + + /* WaDisableDynamicCreditSharing:kbl */ + if (IS_KBL_REVID(dev_priv, 0, KBL_REVID_B0)) + I915_WRITE(GAMT_CHKN_BIT_REG, + I915_READ(GAMT_CHKN_BIT_REG) | + GAMT_CHKN_DISABLE_DYNAMIC_CREDIT_SHARING); + + /* WaDisableFenceDestinationToSLM:kbl (pre-prod) */ + if (IS_KBL_REVID(dev_priv, KBL_REVID_A0, KBL_REVID_A0)) + WA_SET_BIT_MASKED(HDC_CHICKEN0, + HDC_FENCE_DEST_SLM_DISABLE); + + /* WaToEnableHwFixForPushConstHWBug:kbl */ + if (IS_KBL_REVID(dev_priv, KBL_REVID_C0, REVID_FOREVER)) + WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, + GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION); + + /* WaDisableGafsUnitClkGating:kbl */ + I915_WRITE(GEN7_UCGCTL4, + I915_READ(GEN7_UCGCTL4) | GEN8_EU_GAUNIT_CLOCK_GATE_DISABLE); + + /* WaDisableSbeCacheDispatchPortSharing:kbl */ + WA_SET_BIT_MASKED(GEN7_HALF_SLICE_CHICKEN1, + GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE); + + /* WaInPlaceDecompressionHang:kbl */ + I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, + I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | + GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS); + + /* WaDisableLSQCROPERFforOCL:kbl */ + ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4); + if (ret) + return ret; + + return 0; +} + +static int glk_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int ret; + + ret = gen9_init_workarounds(engine); + if (ret) + return ret; + + /* WA #0862: Userspace has to set "Barrier Mode" to avoid hangs. */ + ret = wa_ring_whitelist_reg(engine, GEN9_SLICE_COMMON_ECO_CHICKEN1); + if (ret) + return ret; + + /* WaToEnableHwFixForPushConstHWBug:glk */ + WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, + GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION); + + return 0; +} + +static int cfl_init_workarounds(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int ret; + + ret = gen9_init_workarounds(engine); + if (ret) + return ret; + + /* WaEnableGapsTsvCreditFix:cfl */ + I915_WRITE(GEN8_GARBCNTL, + I915_READ(GEN8_GARBCNTL) | GEN9_GAPS_TSV_CREDIT_DISABLE); + + /* WaToEnableHwFixForPushConstHWBug:cfl */ + WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2, + GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION); + + /* WaDisableGafsUnitClkGating:cfl */ + I915_WRITE(GEN7_UCGCTL4, + I915_READ(GEN7_UCGCTL4) | GEN8_EU_GAUNIT_CLOCK_GATE_DISABLE); + + /* WaDisableSbeCacheDispatchPortSharing:cfl */ + WA_SET_BIT_MASKED(GEN7_HALF_SLICE_CHICKEN1, + GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE); + + /* WaInPlaceDecompressionHang:cfl */ + I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, + I915_READ(GEN9_GAMT_ECO_REG_RW_IA) | + GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS); + + return 0; +} + +int init_workarounds_ring(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + int err; + + if (GEM_WARN_ON(engine->id != RCS)) + return -EINVAL; + + dev_priv->workarounds.count = 0; + dev_priv->workarounds.hw_whitelist_count[engine->id] = 0; + + if (IS_BROADWELL(dev_priv)) + err = bdw_init_workarounds(engine); + else if (IS_CHERRYVIEW(dev_priv)) + err = chv_init_workarounds(engine); + else if (IS_SKYLAKE(dev_priv)) + err = skl_init_workarounds(engine); + else if (IS_BROXTON(dev_priv)) + err = bxt_init_workarounds(engine); + else if (IS_KABYLAKE(dev_priv)) + err = kbl_init_workarounds(engine); + else if (IS_GEMINILAKE(dev_priv)) + err = glk_init_workarounds(engine); + else if (IS_COFFEELAKE(dev_priv)) + err = cfl_init_workarounds(engine); + else if (IS_CANNONLAKE(dev_priv)) + err = cnl_init_workarounds(engine); + else + err = 0; + if (err) + return err; + + DRM_DEBUG_DRIVER("%s: Number of context specific w/a: %d\n", + engine->name, dev_priv->workarounds.count); + return 0; +} + +int intel_ring_workarounds_emit(struct i915_request *rq) +{ + struct i915_workarounds *w = &rq->i915->workarounds; + u32 *cs; + int ret, i; + + if (w->count == 0) + return 0; + + ret = rq->engine->emit_flush(rq, EMIT_BARRIER); + if (ret) + return ret; + + cs = intel_ring_begin(rq, w->count * 2 + 2); + if (IS_ERR(cs)) + return PTR_ERR(cs); + + *cs++ = MI_LOAD_REGISTER_IMM(w->count); + for (i = 0; i < w->count; i++) { + *cs++ = i915_mmio_reg_offset(w->reg[i].addr); + *cs++ = w->reg[i].value; + } + *cs++ = MI_NOOP; + + intel_ring_advance(rq, cs); + + ret = rq->engine->emit_flush(rq, EMIT_BARRIER); + if (ret) + return ret; + + return 0; +} diff --git a/drivers/gpu/drm/i915/intel_workarounds.h b/drivers/gpu/drm/i915/intel_workarounds.h new file mode 100644 index 000000000000..2afea73aeeae --- /dev/null +++ b/drivers/gpu/drm/i915/intel_workarounds.h @@ -0,0 +1,13 @@ +/* + * SPDX-License-Identifier: MIT + * + * Copyright © 2014-2018 Intel Corporation + */ + +#ifndef _I915_WORKAROUNDS_H_ +#define _I915_WORKAROUNDS_H_ + +int init_workarounds_ring(struct intel_engine_cs *engine); +int intel_ring_workarounds_emit(struct i915_request *rq); + +#endif -- 2.17.0 _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
