[Intel-gfx] [PATCH v3 04/22] drm/i915: Split out functions for different kinds of workarounds
Oscar Mateo
oscar.mateo at intel.com
Fri Oct 13 20:53:58 UTC 2017
There are different kind of workarounds (those that modify registers that
live in the context image, those that modify global registers, those that
whitelist registers, etc...) and they have different requirements in terms
of where they are applied and how. Also, by splitting them apart, it should
be easier to decide where a new workaround should go.
v2:
- Add multiple MISSING_CASE
- Rebased
v3:
- Rename mmio_workarounds to gt_workarounds (Chris, Mika)
- Create empty placeholders for BDW and CHV GT WAs
- Rebased
Signed-off-by: Oscar Mateo <oscar.mateo at intel.com>
Cc: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala at linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala at linux.intel.com>
---
drivers/gpu/drm/i915/i915_gem.c | 3 +
drivers/gpu/drm/i915/i915_gem_context.c | 5 +
drivers/gpu/drm/i915/intel_lrc.c | 10 +-
drivers/gpu/drm/i915/intel_ringbuffer.c | 4 +-
drivers/gpu/drm/i915/intel_workarounds.c | 714 +++++++++++++++++++------------
drivers/gpu/drm/i915/intel_workarounds.h | 8 +-
6 files changed, 458 insertions(+), 286 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index f76890b..112a218 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -35,6 +35,7 @@
#include "intel_drv.h"
#include "intel_frontbuffer.h"
#include "intel_mocs.h"
+#include "intel_workarounds.h"
#include "i915_gemfs.h"
#include <linux/dma-fence-array.h>
#include <linux/kthread.h>
@@ -4804,6 +4805,8 @@ int i915_gem_init_hw(struct drm_i915_private *dev_priv)
}
}
+ intel_gt_workarounds_apply(dev_priv);
+
i915_gem_init_swizzling(dev_priv);
/*
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index 5bf96a2..cf104cf 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -90,6 +90,7 @@
#include <drm/i915_drm.h>
#include "i915_drv.h"
#include "i915_trace.h"
+#include "intel_workarounds.h"
#define ALL_L3_SLICES(dev) (1 << NUM_L3_SLICES(dev)) - 1
@@ -454,6 +455,10 @@ int i915_gem_contexts_init(struct drm_i915_private *dev_priv)
GEM_BUG_ON(dev_priv->kernel_context);
+ err = intel_ctx_workarounds_init(dev_priv);
+ if (err)
+ goto err;
+
INIT_LIST_HEAD(&dev_priv->contexts.list);
INIT_WORK(&dev_priv->contexts.free_work, contexts_free_worker);
init_llist_head(&dev_priv->contexts.free_list);
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 495ade6..e632a29 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -1488,7 +1488,7 @@ static int gen8_init_render_ring(struct intel_engine_cs *engine)
I915_WRITE(INSTPM, _MASKED_BIT_ENABLE(INSTPM_FORCE_ORDERING));
- return init_workarounds_ring(engine);
+ return 0;
}
static int gen9_init_render_ring(struct intel_engine_cs *engine)
@@ -1499,7 +1499,11 @@ static int gen9_init_render_ring(struct intel_engine_cs *engine)
if (ret)
return ret;
- return init_workarounds_ring(engine);
+ ret = intel_whitelist_workarounds_apply(engine);
+ if (ret)
+ return ret;
+
+ return 0;
}
static void reset_common_ring(struct intel_engine_cs *engine,
@@ -1827,7 +1831,7 @@ static int gen8_init_rcs_context(struct drm_i915_gem_request *req)
{
int ret;
- ret = intel_ring_workarounds_emit(req);
+ ret = intel_ctx_workarounds_emit(req);
if (ret)
return ret;
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 7e5d7d3..c3f8961 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -647,7 +647,7 @@ static int intel_rcs_ctx_init(struct drm_i915_gem_request *req)
{
int ret;
- ret = intel_ring_workarounds_emit(req);
+ ret = intel_ctx_workarounds_emit(req);
if (ret != 0)
return ret;
@@ -706,7 +706,7 @@ static int init_render_ring(struct intel_engine_cs *engine)
if (INTEL_INFO(dev_priv)->gen >= 6)
I915_WRITE_IMR(engine, ~engine->irq_keep_mask);
- return init_workarounds_ring(engine);
+ return 0;
}
static void render_ring_cleanup(struct intel_engine_cs *engine)
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 5f597d1..0a8f265 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -58,27 +58,8 @@ static int wa_add(struct drm_i915_private *dev_priv,
#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)
+static int gen8_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = engine->i915;
-
WA_SET_BIT_MASKED(INSTPM, INSTPM_FORCE_ORDERING);
/* WaDisableAsyncFlipPerfMode:bdw,chv */
@@ -126,12 +107,11 @@ static int gen8_init_workarounds(struct intel_engine_cs *engine)
return 0;
}
-static int bdw_init_workarounds(struct intel_engine_cs *engine)
+static int bdw_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = engine->i915;
int ret;
- ret = gen8_init_workarounds(engine);
+ ret = gen8_ctx_workarounds_init(dev_priv);
if (ret)
return ret;
@@ -158,12 +138,11 @@ static int bdw_init_workarounds(struct intel_engine_cs *engine)
return 0;
}
-static int chv_init_workarounds(struct intel_engine_cs *engine)
+static int chv_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = engine->i915;
int ret;
- ret = gen8_init_workarounds(engine);
+ ret = gen8_ctx_workarounds_init(dev_priv);
if (ret)
return ret;
@@ -176,23 +155,8 @@ static int chv_init_workarounds(struct intel_engine_cs *engine)
return 0;
}
-static int gen9_init_workarounds(struct intel_engine_cs *engine)
+static int gen9_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- 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
*
@@ -203,11 +167,6 @@ static int gen9_init_workarounds(struct intel_engine_cs *engine)
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 */
@@ -279,10 +238,6 @@ static int gen9_init_workarounds(struct intel_engine_cs *engine)
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) ||
@@ -294,10 +249,6 @@ static int gen9_init_workarounds(struct intel_engine_cs *engine)
/* WaDisableSTUnitPowerOptimization:skl,bxt,kbl,glk,cfl */
WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN2, GEN8_ST_PO_DISABLE);
- /* 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
@@ -316,29 +267,11 @@ static int gen9_init_workarounds(struct intel_engine_cs *engine)
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)
+static int skl_tune_iz_hashing(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = engine->i915;
u8 vals[3] = { 0, 0, 0 };
unsigned int i;
@@ -377,67 +310,29 @@ static int skl_tune_iz_hashing(struct intel_engine_cs *engine)
return 0;
}
-static int skl_init_workarounds(struct intel_engine_cs *engine)
+static int skl_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- 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);
+ ret = gen9_ctx_workarounds_init(dev_priv);
if (ret)
return ret;
- return skl_tune_iz_hashing(engine);
+ return skl_tune_iz_hashing(dev_priv);
}
-static int bxt_init_workarounds(struct intel_engine_cs *engine)
+static int bxt_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = engine->i915;
int ret;
- ret = gen9_init_workarounds(engine);
+ ret = gen9_ctx_workarounds_init(dev_priv);
if (ret)
return ret;
- /* WaStoreMultiplePTEenable:bxt */
- /* This is a requirement according to Hardware specification */
- if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
- I915_WRITE(TILECTL, I915_READ(TILECTL) | TILECTL_TLBPF);
-
- /* WaSetClckGatingDisableMedia:bxt */
- if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
- I915_WRITE(GEN7_MISCCPCTL, (I915_READ(GEN7_MISCCPCTL) &
- ~GEN8_DOP_CLOCK_GATE_MEDIA_ENABLE));
- }
-
/* WaDisableThreadStallDopClockGating:bxt */
WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN,
STALL_DOP_GATING_DISABLE);
- /* WaDisablePooledEuLoadBalancingFix:bxt */
- if (IS_BXT_REVID(dev_priv, BXT_REVID_B0, REVID_FOREVER)) {
- I915_WRITE(FF_SLICE_CS_CHICKEN2,
- _MASKED_BIT_ENABLE(GEN9_POOLED_EU_LOAD_BALANCING_FIX_DISABLE));
- }
-
/* WaDisableSbeCacheDispatchPortSharing:bxt */
if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_B0)) {
WA_SET_BIT_MASKED(
@@ -445,117 +340,22 @@ static int bxt_init_workarounds(struct intel_engine_cs *engine)
GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE);
}
- /* WaDisableObjectLevelPreemptionForTrifanOrPolygon:bxt */
- /* WaDisableObjectLevelPreemptionForInstancedDraw:bxt */
- /* WaDisableObjectLevelPreemtionForInstanceId:bxt */
- /* WaDisableLSQCROPERFforOCL:bxt */
- if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
- ret = wa_ring_whitelist_reg(engine, GEN9_CS_DEBUG_MODE1);
- if (ret)
- return ret;
-
- ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4);
- if (ret)
- return ret;
- }
-
- /* WaProgramL3SqcReg1DefaultForPerf:bxt */
- if (IS_BXT_REVID(dev_priv, BXT_REVID_B0, REVID_FOREVER)) {
- 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);
- }
-
/* WaToEnableHwFixForPushConstHWBug:bxt */
if (IS_BXT_REVID(dev_priv, BXT_REVID_C0, REVID_FOREVER))
WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION);
- /* WaInPlaceDecompressionHang:bxt */
- if (IS_BXT_REVID(dev_priv, BXT_REVID_C0, REVID_FOREVER))
- 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;
-
return 0;
}
-static int kbl_init_workarounds(struct intel_engine_cs *engine)
+static int kbl_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = engine->i915;
int ret;
- ret = gen9_init_workarounds(engine);
+ ret = gen9_ctx_workarounds_init(dev_priv);
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,
@@ -566,34 +366,19 @@ static int kbl_init_workarounds(struct intel_engine_cs *engine)
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)
+static int glk_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = engine->i915;
int ret;
- ret = gen9_init_workarounds(engine);
+ ret = gen9_ctx_workarounds_init(dev_priv);
if (ret)
return ret;
@@ -604,77 +389,98 @@ static int glk_init_workarounds(struct intel_engine_cs *engine)
return 0;
}
-static int cfl_init_workarounds(struct intel_engine_cs *engine)
+static int cfl_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = engine->i915;
int ret;
- ret = gen9_init_workarounds(engine);
+ ret = gen9_ctx_workarounds_init(dev_priv);
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)
+static int cnl_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = engine->i915;
- int err;
-
- WARN_ON(engine->id != RCS);
+ /* WaForceContextSaveRestoreNonCoherent:cnl */
+ WA_SET_BIT_MASKED(CNL_HDC_CHICKEN0,
+ HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT);
- dev_priv->workarounds.count = 0;
- dev_priv->workarounds.hw_whitelist_count[engine->id] = 0;
+ /* 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);
- 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;
+ /* WaDisableReplayBufferBankArbitrationOptimization:cnl */
+ WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
+ GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION);
- DRM_DEBUG_DRIVER("%s: Number of context specific w/a: %d\n",
- engine->name, dev_priv->workarounds.count);
- return 0;
-}
+ /* 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);
-int intel_ring_workarounds_emit(struct drm_i915_gem_request *req)
+ /* 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);
+
+ return 0;
+}
+
+int intel_ctx_workarounds_init(struct drm_i915_private *dev_priv)
+{
+ int err;
+
+ dev_priv->workarounds.count = 0;
+
+ if (INTEL_GEN(dev_priv) < 8)
+ err = 0;
+ else if (IS_BROADWELL(dev_priv))
+ err = bdw_ctx_workarounds_init(dev_priv);
+ else if (IS_CHERRYVIEW(dev_priv))
+ err = chv_ctx_workarounds_init(dev_priv);
+ else if (IS_SKYLAKE(dev_priv))
+ err = skl_ctx_workarounds_init(dev_priv);
+ else if (IS_BROXTON(dev_priv))
+ err = bxt_ctx_workarounds_init(dev_priv);
+ else if (IS_KABYLAKE(dev_priv))
+ err = kbl_ctx_workarounds_init(dev_priv);
+ else if (IS_GEMINILAKE(dev_priv))
+ err = glk_ctx_workarounds_init(dev_priv);
+ else if (IS_COFFEELAKE(dev_priv))
+ err = cfl_ctx_workarounds_init(dev_priv);
+ else if (IS_CANNONLAKE(dev_priv))
+ err = cnl_ctx_workarounds_init(dev_priv);
+ else {
+ MISSING_CASE(INTEL_GEN(dev_priv));
+ err = 0;
+ }
+ if (err)
+ return err;
+
+ DRM_DEBUG_DRIVER("Number of context specific w/a: %d\n",
+ dev_priv->workarounds.count);
+ return 0;
+}
+
+int intel_ctx_workarounds_emit(struct drm_i915_gem_request *req)
{
struct i915_workarounds *w = &req->i915->workarounds;
u32 *cs;
@@ -706,3 +512,353 @@ int intel_ring_workarounds_emit(struct drm_i915_gem_request *req)
return 0;
}
+
+static void bdw_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+}
+
+static void chv_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+}
+
+static void gen9_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ if (HAS_LLC(dev_priv)) {
+ /* WaCompressedResourceSamplerPbeMediaNewHashMode:skl,kbl
+ *
+ * Must match Display Engine. See
+ * WaCompressedResourceDisplayNewHashMode.
+ */
+ I915_WRITE(MMCD_MISC_CTRL,
+ I915_READ(MMCD_MISC_CTRL) |
+ MMCD_PCLA |
+ MMCD_HOTSPOT_EN);
+ }
+
+ /* WaContextSwitchWithConcurrentTLBInvalidate: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);
+
+ /* WaDisableHDCInvalidation:skl,bxt,kbl,cfl */
+ I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) |
+ BDW_DISABLE_HDC_INVALIDATION);
+
+ /* WaOCLCoherentLineFlush:skl,bxt,kbl,cfl */
+ I915_WRITE(GEN8_L3SQCREG4, (I915_READ(GEN8_L3SQCREG4) |
+ GEN8_LQSC_FLUSH_COHERENT_LINES));
+
+ /* WaEnablePreemptionGranularityControlByUMD:skl,bxt,kbl,cfl,[cnl] */
+ I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1,
+ _MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL));
+}
+
+static void skl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ gen9_gt_workarounds_apply(dev_priv);
+
+ /* 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));
+}
+
+static void bxt_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ gen9_gt_workarounds_apply(dev_priv);
+
+ /* WaStoreMultiplePTEenable:bxt */
+ /* This is a requirement according to Hardware specification */
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
+ I915_WRITE(TILECTL, I915_READ(TILECTL) | TILECTL_TLBPF);
+
+ /* WaSetClckGatingDisableMedia:bxt */
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
+ I915_WRITE(GEN7_MISCCPCTL, (I915_READ(GEN7_MISCCPCTL) &
+ ~GEN8_DOP_CLOCK_GATE_MEDIA_ENABLE));
+ }
+
+ /* WaDisablePooledEuLoadBalancingFix:bxt */
+ if (IS_BXT_REVID(dev_priv, BXT_REVID_B0, REVID_FOREVER)) {
+ I915_WRITE(FF_SLICE_CS_CHICKEN2,
+ _MASKED_BIT_ENABLE(GEN9_POOLED_EU_LOAD_BALANCING_FIX_DISABLE));
+ }
+
+ /* WaProgramL3SqcReg1DefaultForPerf:bxt */
+ if (IS_BXT_REVID(dev_priv, BXT_REVID_B0, REVID_FOREVER)) {
+ 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);
+ }
+
+ /* WaInPlaceDecompressionHang:bxt */
+ if (IS_BXT_REVID(dev_priv, BXT_REVID_C0, REVID_FOREVER))
+ I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA,
+ (I915_READ(GEN9_GAMT_ECO_REG_RW_IA) |
+ GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS));
+}
+
+static void kbl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ gen9_gt_workarounds_apply(dev_priv);
+
+ /* 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));
+
+ /* WaDisableGafsUnitClkGating:kbl */
+ I915_WRITE(GEN7_UCGCTL4, (I915_READ(GEN7_UCGCTL4) |
+ GEN8_EU_GAUNIT_CLOCK_GATE_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));
+}
+
+static void glk_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ gen9_gt_workarounds_apply(dev_priv);
+}
+
+static void cfl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ gen9_gt_workarounds_apply(dev_priv);
+
+ /* WaEnableGapsTsvCreditFix:cfl */
+ I915_WRITE(GEN8_GARBCNTL, (I915_READ(GEN8_GARBCNTL) |
+ GEN9_GAPS_TSV_CREDIT_DISABLE));
+
+ /* WaDisableGafsUnitClkGating:cfl */
+ I915_WRITE(GEN7_UCGCTL4, (I915_READ(GEN7_UCGCTL4) |
+ GEN8_EU_GAUNIT_CLOCK_GATE_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));
+}
+
+static void cnl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ /* 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));
+
+ /* WaInPlaceDecompressionHang:cnl */
+ I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA,
+ (I915_READ(GEN9_GAMT_ECO_REG_RW_IA) |
+ GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS));
+
+ /* WaEnablePreemptionGranularityControlByUMD:cnl */
+ I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1,
+ _MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL));
+}
+
+void intel_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ if (INTEL_GEN(dev_priv) < 8)
+ return;
+ else if (IS_BROADWELL(dev_priv))
+ bdw_gt_workarounds_apply(dev_priv);
+ else if (IS_CHERRYVIEW(dev_priv))
+ chv_gt_workarounds_apply(dev_priv);
+ else if (IS_SKYLAKE(dev_priv))
+ skl_gt_workarounds_apply(dev_priv);
+ else if (IS_BROXTON(dev_priv))
+ bxt_gt_workarounds_apply(dev_priv);
+ else if (IS_KABYLAKE(dev_priv))
+ kbl_gt_workarounds_apply(dev_priv);
+ else if (IS_GEMINILAKE(dev_priv))
+ glk_gt_workarounds_apply(dev_priv);
+ else if (IS_COFFEELAKE(dev_priv))
+ cfl_gt_workarounds_apply(dev_priv);
+ else if (IS_CANNONLAKE(dev_priv))
+ cnl_gt_workarounds_apply(dev_priv);
+ else
+ MISSING_CASE(INTEL_GEN(dev_priv));
+}
+
+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 gen9_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+{
+ int ret;
+
+ /* 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] */
+ 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_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+{
+ int ret = gen9_whitelist_workarounds_apply(engine);
+ if (ret)
+ return ret;
+
+ /* WaDisableLSQCROPERFforOCL:skl */
+ ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+static int bxt_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+{
+ struct drm_i915_private *dev_priv = engine->i915;
+
+ int ret = gen9_whitelist_workarounds_apply(engine);
+ if (ret)
+ return ret;
+
+ /* WaDisableObjectLevelPreemptionForTrifanOrPolygon:bxt */
+ /* WaDisableObjectLevelPreemptionForInstancedDraw:bxt */
+ /* WaDisableObjectLevelPreemtionForInstanceId:bxt */
+ /* WaDisableLSQCROPERFforOCL:bxt */
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
+ ret = wa_ring_whitelist_reg(engine, GEN9_CS_DEBUG_MODE1);
+ if (ret)
+ return ret;
+
+ ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4);
+ if (ret)
+ return ret;
+ }
+
+ return 0;
+}
+
+static int kbl_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+{
+ int ret = gen9_whitelist_workarounds_apply(engine);
+ if (ret)
+ return ret;
+
+ /* WaDisableLSQCROPERFforOCL:kbl */
+ ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+static int glk_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+{
+ int ret = gen9_whitelist_workarounds_apply(engine);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+static int cfl_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+{
+ int ret = gen9_whitelist_workarounds_apply(engine);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+static int cnl_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+{
+ int ret;
+
+ /* WaEnablePreemptionGranularityControlByUMD:cnl */
+ ret = wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+int intel_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+{
+ struct drm_i915_private *dev_priv = engine->i915;
+ int err;
+
+ WARN_ON(engine->id != RCS);
+
+ dev_priv->workarounds.hw_whitelist_count[engine->id] = 0;
+
+ if (INTEL_GEN(dev_priv) < 9) {
+ WARN(1, "No whitelisting in Gen%u\n", INTEL_GEN(dev_priv));
+ err = 0;
+ } else if (IS_SKYLAKE(dev_priv))
+ err = skl_whitelist_workarounds_apply(engine);
+ else if (IS_BROXTON(dev_priv))
+ err = bxt_whitelist_workarounds_apply(engine);
+ else if (IS_KABYLAKE(dev_priv))
+ err = kbl_whitelist_workarounds_apply(engine);
+ else if (IS_GEMINILAKE(dev_priv))
+ err = glk_whitelist_workarounds_apply(engine);
+ else if (IS_COFFEELAKE(dev_priv))
+ err = cfl_whitelist_workarounds_apply(engine);
+ else if (IS_CANNONLAKE(dev_priv))
+ err = cnl_whitelist_workarounds_apply(engine);
+ else {
+ MISSING_CASE(INTEL_GEN(dev_priv));
+ err = 0;
+ }
+ if (err)
+ return err;
+
+ DRM_DEBUG_DRIVER("%s: Number of whitelist w/a: %d\n", engine->name,
+ dev_priv->workarounds.hw_whitelist_count[engine->id]);
+ return 0;
+}
diff --git a/drivers/gpu/drm/i915/intel_workarounds.h b/drivers/gpu/drm/i915/intel_workarounds.h
index 27099fc..bba51bb 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.h
+++ b/drivers/gpu/drm/i915/intel_workarounds.h
@@ -25,7 +25,11 @@
#ifndef _I915_WORKAROUNDS_H_
#define _I915_WORKAROUNDS_H_
-int init_workarounds_ring(struct intel_engine_cs *engine);
-int intel_ring_workarounds_emit(struct drm_i915_gem_request *req);
+int intel_ctx_workarounds_init(struct drm_i915_private *dev_priv);
+int intel_ctx_workarounds_emit(struct drm_i915_gem_request *req);
+
+void intel_gt_workarounds_apply(struct drm_i915_private *dev_priv);
+
+int intel_whitelist_workarounds_apply(struct intel_engine_cs *engine);
#endif
--
1.9.1
More information about the Intel-gfx
mailing list