* [PATCH v4 00/20] Refactor HW workaround code
@ 2017-10-30 20:17 Oscar Mateo
2017-10-30 20:17 ` [PATCH 01/20] drm/i915: Remove Gen9 WAs with no effect Oscar Mateo
` (20 more replies)
0 siblings, 21 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
This v4 implements review comments from Chris and Ville and removes the
early bug fixes that have already been upstreamed.
Currently, deciding how/where to apply new workarounds is challenging. Often,
workarounds end up applied incorrectly and get lost under certain circumstances
(e.g. a context switch or a GPU reset). This is a proposal to attempt to
eliminate some of this pain, by clarifying the current classification of
workarounds (context saved/restored, global registers, whitelisting, BB),
putting them together on the same file, and improving the existing validation
infrastructure (debugfs/i-g-t).
Oscar Mateo (20):
drm/i915: Remove Gen9 WAs with no effect
drm/i915: Move a bunch of workaround-related code to its own file
drm/i915: Split out functions for different kinds of workarounds
drm/i915: Rename saved workarounds to make it explicit that they are
context WAs
drm/i915: Save all GT WAs and apply them at a later time
drm/i915: Save all Whitelist WAs and apply them at a later time
drm/i915: Create a new category of display WAs
drm/i915: Print all workaround types correctly in debugfs
drm/i915: Move WA BB stuff to the workarounds file as well
drm/i915/cnl: Move GT and Display workarounds from init_clock_gating
drm/i915/gen9: Move GT and Display workarounds from init_clock_gating
drm/i915/cfl: Move GT and Display workarounds from init_clock_gating
drm/i915/glk: Move GT and Display workarounds from init_clock_gating
drm/i915/kbl: Move GT and Display workarounds from init_clock_gating
drm/i915/bxt: Move GT and Display workarounds from init_clock_gating
drm/i915/skl: Move GT and Display workarounds from init_clock_gating
drm/i915/chv: Move GT and Display workarounds from init_clock_gating
drm/i915/bdw: Move GT and Display workarounds from init_clock_gating
drm/i915: Move WaProgramL3SqcReg1Default to the workarounds file
drm/i915: Document the i915_workarounds file
drivers/gpu/drm/i915/Makefile | 3 +-
drivers/gpu/drm/i915/i915_debugfs.c | 63 +-
drivers/gpu/drm/i915/i915_drv.c | 5 +
drivers/gpu/drm/i915/i915_drv.h | 27 +-
drivers/gpu/drm/i915/i915_gem.c | 3 +
drivers/gpu/drm/i915/i915_gem_context.c | 5 +
drivers/gpu/drm/i915/i915_reg.h | 3 -
drivers/gpu/drm/i915/intel_engine_cs.c | 682 -------------
drivers/gpu/drm/i915/intel_lrc.c | 266 +----
drivers/gpu/drm/i915/intel_pm.c | 312 +-----
drivers/gpu/drm/i915/intel_ringbuffer.c | 5 +-
drivers/gpu/drm/i915/intel_ringbuffer.h | 3 -
drivers/gpu/drm/i915/intel_workarounds.c | 1622 ++++++++++++++++++++++++++++++
drivers/gpu/drm/i915/intel_workarounds.h | 41 +
14 files changed, 1771 insertions(+), 1269 deletions(-)
create mode 100644 drivers/gpu/drm/i915/intel_workarounds.c
create mode 100644 drivers/gpu/drm/i915/intel_workarounds.h
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply [flat|nested] 26+ messages in thread
* [PATCH 01/20] drm/i915: Remove Gen9 WAs with no effect
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 02/20] drm/i915: Move a bunch of workaround-related code to its own file Oscar Mateo
` (19 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
GEN8_CONFIG0 (0xD00) is a protected by a lock (bit 31) which is set by
the BIOS, so there is no way we can enable the three chicken bits
mandated by the WA (the BIOS should be doing it instead).
v2: Rebased
v3: Standalone patch
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
---
drivers/gpu/drm/i915/i915_reg.h | 3 ---
drivers/gpu/drm/i915/intel_pm.c | 3 ---
2 files changed, 6 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 8c775e9..9c57e0c 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -355,9 +355,6 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
#define ECOCHK_PPGTT_WT_HSW (0x2<<3)
#define ECOCHK_PPGTT_WB_HSW (0x3<<3)
-#define GEN8_CONFIG0 _MMIO(0xD00)
-#define GEN9_DEFAULT_FIXES (1 << 3 | 1 << 2 | 1 << 1)
-
#define GAC_ECO_BITS _MMIO(0x14090)
#define ECOBITS_SNB_BIT (1<<13)
#define ECOBITS_PPGTT_CACHE64B (3<<8)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 07118c0..acd0cbb 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -75,9 +75,6 @@ static void gen9_init_clock_gating(struct drm_i915_private *dev_priv)
I915_WRITE(CHICKEN_PAR1_1,
I915_READ(CHICKEN_PAR1_1) | SKL_EDP_PSR_FIX_RDWRAP);
- I915_WRITE(GEN8_CONFIG0,
- I915_READ(GEN8_CONFIG0) | GEN9_DEFAULT_FIXES);
-
/* WaEnableChickenDCPR:skl,bxt,kbl,glk,cfl */
I915_WRITE(GEN8_CHICKEN_DCPR_1,
I915_READ(GEN8_CHICKEN_DCPR_1) | MASK_WAKEMEM);
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 02/20] drm/i915: Move a bunch of workaround-related code to its own file
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
2017-10-30 20:17 ` [PATCH 01/20] drm/i915: Remove Gen9 WAs with no effect Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 03/20] drm/i915: Split out functions for different kinds of workarounds Oscar Mateo
` (18 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=a, Size: 48863 bytes --]
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
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
---
drivers/gpu/drm/i915/Makefile | 3 +-
drivers/gpu/drm/i915/intel_engine_cs.c | 682 -----------------------------
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 | 708 +++++++++++++++++++++++++++++++
drivers/gpu/drm/i915/intel_workarounds.h | 31 ++
7 files changed, 743 insertions(+), 686 deletions(-)
create mode 100644 drivers/gpu/drm/i915/intel_workarounds.c
create mode 100644 drivers/gpu/drm/i915/intel_workarounds.h
diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile
index 7750be8..03cd0df 100644
--- a/drivers/gpu/drm/i915/Makefile
+++ b/drivers/gpu/drm/i915/Makefile
@@ -41,7 +41,8 @@ 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 f31f2d6..6e4440f 100644
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -818,688 +818,6 @@ 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);
-
- /* WaDisableDgMirrorFixInHalfSliceChicken5:bxt */
- if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
- WA_CLR_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN5,
- GEN9_DG_MIRROR_FIX_ENABLE);
-
- /* WaSetDisablePixMaskCammingAndRhwoInCommonSliceChicken:bxt */
- if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
- WA_SET_BIT_MASKED(GEN7_COMMON_SLICE_CHICKEN1,
- GEN9_RHWO_OPTIMIZATION_DISABLE);
- /*
- * WA also requires GEN9_SLICE_COMMON_ECO_CHICKEN0[14:14] to be set
- * but we do that in per ctx batchbuffer as there is an issue
- * with this register not getting restored on ctx restore
- */
- }
-
- /* 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);
-
- /* WaDisableMaskBasedCammingInRCC:bxt */
- if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
- WA_SET_BIT_MASKED(SLICE_ECO_CHICKEN0,
- PIXEL_MASK_CAMMING_DISABLE);
-
- /* 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) ||
- IS_BXT_REVID(dev_priv, 0, BXT_REVID_B0))
- 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);
-
- /* 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;
-
- /* 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(
- GEN7_HALF_SLICE_CHICKEN1,
- 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)
-{
- 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;
-
- /* 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;
-
- WARN_ON(engine->id != RCS);
-
- 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 drm_i915_gem_request *req)
-{
- struct i915_workarounds *w = &req->i915->workarounds;
- u32 *cs;
- int ret, i;
-
- if (w->count == 0)
- return 0;
-
- ret = req->engine->emit_flush(req, EMIT_BARRIER);
- if (ret)
- return ret;
-
- cs = intel_ring_begin(req, (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(req, cs);
-
- ret = req->engine->emit_flush(req, 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 6840ec8..911df0c 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -137,6 +137,7 @@
#include <drm/i915_drm.h>
#include "i915_drv.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 47fadf8..1c721b2 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -33,6 +33,7 @@
#include <drm/i915_drm.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 69ad875..8613e4a 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -766,9 +766,6 @@ 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 drm_i915_gem_request *req);
-
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 0000000..5f597d1
--- /dev/null
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -0,0 +1,708 @@
+/*
+ * Copyright © 2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#include "i915_drv.h"
+#include "intel_workarounds.h"
+
+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);
+
+ /* WaDisableDgMirrorFixInHalfSliceChicken5:bxt */
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
+ WA_CLR_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN5,
+ GEN9_DG_MIRROR_FIX_ENABLE);
+
+ /* WaSetDisablePixMaskCammingAndRhwoInCommonSliceChicken:bxt */
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
+ WA_SET_BIT_MASKED(GEN7_COMMON_SLICE_CHICKEN1,
+ GEN9_RHWO_OPTIMIZATION_DISABLE);
+ /*
+ * WA also requires GEN9_SLICE_COMMON_ECO_CHICKEN0[14:14] to be set
+ * but we do that in per ctx batchbuffer as there is an issue
+ * with this register not getting restored on ctx restore
+ */
+ }
+
+ /* 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);
+
+ /* WaDisableMaskBasedCammingInRCC:bxt */
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
+ WA_SET_BIT_MASKED(SLICE_ECO_CHICKEN0,
+ PIXEL_MASK_CAMMING_DISABLE);
+
+ /* 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) ||
+ IS_BXT_REVID(dev_priv, 0, BXT_REVID_B0))
+ 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);
+
+ /* 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;
+
+ /* 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(
+ GEN7_HALF_SLICE_CHICKEN1,
+ 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)
+{
+ 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;
+
+ /* 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;
+
+ WARN_ON(engine->id != RCS);
+
+ 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 drm_i915_gem_request *req)
+{
+ struct i915_workarounds *w = &req->i915->workarounds;
+ u32 *cs;
+ int ret, i;
+
+ if (w->count == 0)
+ return 0;
+
+ ret = req->engine->emit_flush(req, EMIT_BARRIER);
+ if (ret)
+ return ret;
+
+ cs = intel_ring_begin(req, (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(req, cs);
+
+ ret = req->engine->emit_flush(req, 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 0000000..27099fc
--- /dev/null
+++ b/drivers/gpu/drm/i915/intel_workarounds.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright © 2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#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);
+
+#endif
--
1.9.1
[-- Attachment #2: Type: text/plain, Size: 160 bytes --]
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 03/20] drm/i915: Split out functions for different kinds of workarounds
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
2017-10-30 20:17 ` [PATCH 01/20] drm/i915: Remove Gen9 WAs with no effect Oscar Mateo
2017-10-30 20:17 ` [PATCH 02/20] drm/i915: Move a bunch of workaround-related code to its own file Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 04/20] drm/i915: Rename saved workarounds to make it explicit that they are context WAs Oscar Mateo
` (17 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
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@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@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 e43688f..750e014 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>
@@ -4916,6 +4917,8 @@ 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 10affb3..8548e571 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
@@ -456,6 +457,10 @@ 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 911df0c..f0b4d2f 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -1503,7 +1503,7 @@ 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)
@@ -1514,7 +1514,11 @@ 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,
@@ -1830,7 +1834,7 @@ 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 1c721b2..b053fed 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -649,7 +649,7 @@ 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;
@@ -708,7 +708,7 @@ 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
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 04/20] drm/i915: Rename saved workarounds to make it explicit that they are context WAs
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (2 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 03/20] drm/i915: Split out functions for different kinds of workarounds Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 05/20] drm/i915: Save all GT WAs and apply them at a later time Oscar Mateo
` (16 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
Some WAs touch registers that get saved/restored together with the logical context.
Make this very explicit by renaming a few things in the code.
v2:
- Improved naming
- Rebased
v3: Also rename I915_MAX_CTX_WA_REGS
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
---
drivers/gpu/drm/i915/i915_debugfs.c | 10 +-
drivers/gpu/drm/i915/i915_drv.h | 7 +-
drivers/gpu/drm/i915/intel_workarounds.c | 232 +++++++++++++++----------------
3 files changed, 125 insertions(+), 124 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index c1e5bba..e31aba9 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -3371,18 +3371,18 @@ static int i915_wa_registers(struct seq_file *m, void *unused)
intel_runtime_pm_get(dev_priv);
- seq_printf(m, "Workarounds applied: %d\n", workarounds->count);
+ seq_printf(m, "Context workarounds applied: %d\n", workarounds->ctx_wa_count);
for_each_engine(engine, dev_priv, id)
seq_printf(m, "HW whitelist count for %s: %d\n",
engine->name, workarounds->hw_whitelist_count[id]);
- for (i = 0; i < workarounds->count; ++i) {
+ for (i = 0; i < workarounds->ctx_wa_count; ++i) {
i915_reg_t addr;
u32 mask, value, read;
bool ok;
- addr = workarounds->reg[i].addr;
- mask = workarounds->reg[i].mask;
- value = workarounds->reg[i].value;
+ addr = workarounds->ctx_wa_reg[i].addr;
+ mask = workarounds->ctx_wa_reg[i].mask;
+ value = workarounds->ctx_wa_reg[i].value;
read = I915_READ(addr);
ok = (value & mask) == (read & mask);
seq_printf(m, "0x%X: 0x%08X, mask: 0x%08X, read: 0x%08x, status: %s\n",
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 61c155c..e6daf53 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1975,11 +1975,12 @@ struct i915_wa_reg {
u32 mask;
};
-#define I915_MAX_WA_REGS 16
+#define I915_MAX_CTX_WA_REGS 16
struct i915_workarounds {
- struct i915_wa_reg reg[I915_MAX_WA_REGS];
- u32 count;
+ struct i915_wa_reg ctx_wa_reg[I915_MAX_CTX_WA_REGS];
+ u32 ctx_wa_count;
+
u32 hw_whitelist_count[I915_NUM_ENGINES];
};
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 0a8f265..f4bc2d3 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -25,49 +25,49 @@
#include "i915_drv.h"
#include "intel_workarounds.h"
-static int wa_add(struct drm_i915_private *dev_priv,
- i915_reg_t addr,
- const u32 mask, const u32 val)
+static int ctx_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;
+ const u32 idx = dev_priv->workarounds.ctx_wa_count;
- if (WARN_ON(idx >= I915_MAX_WA_REGS))
+ if (WARN_ON(idx >= I915_MAX_CTX_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.ctx_wa_reg[idx].addr = addr;
+ dev_priv->workarounds.ctx_wa_reg[idx].value = val;
+ dev_priv->workarounds.ctx_wa_reg[idx].mask = mask;
- dev_priv->workarounds.count++;
+ dev_priv->workarounds.ctx_wa_count++;
return 0;
}
-#define WA_REG(addr, mask, val) do { \
- const int r = wa_add(dev_priv, (addr), (mask), (val)); \
+#define CTX_WA_REG(addr, mask, val) do { \
+ const int r = ctx_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 CTX_WA_SET_BIT_MASKED(addr, mask) \
+ CTX_WA_REG(addr, (mask), _MASKED_BIT_ENABLE(mask))
-#define WA_CLR_BIT_MASKED(addr, mask) \
- WA_REG(addr, (mask), _MASKED_BIT_DISABLE(mask))
+#define CTX_WA_CLR_BIT_MASKED(addr, mask) \
+ CTX_WA_REG(addr, (mask), _MASKED_BIT_DISABLE(mask))
-#define WA_SET_FIELD_MASKED(addr, mask, value) \
- WA_REG(addr, mask, _MASKED_FIELD(mask, value))
+#define CTX_WA_SET_FIELD_MASKED(addr, mask, value) \
+ CTX_WA_REG(addr, mask, _MASKED_FIELD(mask, value))
static int gen8_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
- WA_SET_BIT_MASKED(INSTPM, INSTPM_FORCE_ORDERING);
+ CTX_WA_SET_BIT_MASKED(INSTPM, INSTPM_FORCE_ORDERING);
/* WaDisableAsyncFlipPerfMode:bdw,chv */
- WA_SET_BIT_MASKED(MI_MODE, ASYNC_FLIP_PERF_DISABLE);
+ CTX_WA_SET_BIT_MASKED(MI_MODE, ASYNC_FLIP_PERF_DISABLE);
/* WaDisablePartialInstShootdown:bdw,chv */
- WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN,
- PARTIAL_INSTRUCTION_SHOOTDOWN_DISABLE);
+ CTX_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
@@ -75,9 +75,9 @@ static int gen8_ctx_workarounds_init(struct drm_i915_private *dev_priv)
*/
/* WaForceEnableNonCoherent:bdw,chv */
/* WaHdcDisableFetchWhenMasked:bdw,chv */
- WA_SET_BIT_MASKED(HDC_CHICKEN0,
- HDC_DONOT_FETCH_MEM_WHEN_MASKED |
- HDC_FORCE_NON_COHERENT);
+ CTX_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
@@ -87,10 +87,10 @@ static int gen8_ctx_workarounds_init(struct drm_i915_private *dev_priv)
*
* 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);
+ CTX_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);
+ CTX_WA_SET_BIT_MASKED(CACHE_MODE_1, GEN8_4x4_STC_OPTIMIZATION_DISABLE);
/*
* BSpec recommends 8x4 when MSAA is used,
@@ -100,9 +100,9 @@ static int gen8_ctx_workarounds_init(struct drm_i915_private *dev_priv)
* 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);
+ CTX_WA_SET_FIELD_MASKED(GEN7_GT_MODE,
+ GEN6_WIZ_HASHING_MASK,
+ GEN6_WIZ_HASHING_16x4);
return 0;
}
@@ -116,24 +116,24 @@ static int bdw_ctx_workarounds_init(struct drm_i915_private *dev_priv)
return ret;
/* WaDisableThreadStallDopClockGating:bdw (pre-production) */
- WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, STALL_DOP_GATING_DISABLE);
+ CTX_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);
+ CTX_WA_SET_BIT_MASKED(GEN7_ROW_CHICKEN2,
+ DOP_CLOCK_GATING_DISABLE);
- WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3,
- GEN8_SAMPLER_POWER_BYPASS_DIS);
+ CTX_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));
+ CTX_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;
}
@@ -147,10 +147,10 @@ static int chv_ctx_workarounds_init(struct drm_i915_private *dev_priv)
return ret;
/* WaDisableThreadStallDopClockGating:chv */
- WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN, STALL_DOP_GATING_DISABLE);
+ CTX_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);
+ CTX_WA_SET_BIT_MASKED(HIZ_CHICKEN, CHV_HZ_8X8_MODE_IN_1X);
return 0;
}
@@ -163,32 +163,32 @@ static int gen9_ctx_workarounds_init(struct drm_i915_private *dev_priv)
* 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);
+ CTX_WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
+ GEN9_PBE_COMPRESSED_HASH_SELECTION);
+ CTX_WA_SET_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN7,
+ GEN9_SAMPLER_HASH_COMPRESSED_READ_ADDR);
}
/* 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);
+ CTX_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);
+ CTX_WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3,
+ GEN9_DISABLE_OCL_OOB_SUPPRESS_LOGIC);
/* WaDisableDgMirrorFixInHalfSliceChicken5:bxt */
if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
- WA_CLR_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN5,
- GEN9_DG_MIRROR_FIX_ENABLE);
+ CTX_WA_CLR_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN5,
+ GEN9_DG_MIRROR_FIX_ENABLE);
/* WaSetDisablePixMaskCammingAndRhwoInCommonSliceChicken:bxt */
if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
- WA_SET_BIT_MASKED(GEN7_COMMON_SLICE_CHICKEN1,
- GEN9_RHWO_OPTIMIZATION_DISABLE);
+ CTX_WA_SET_BIT_MASKED(GEN7_COMMON_SLICE_CHICKEN1,
+ GEN9_RHWO_OPTIMIZATION_DISABLE);
/*
* WA also requires GEN9_SLICE_COMMON_ECO_CHICKEN0[14:14] to be set
* but we do that in per ctx batchbuffer as there is an issue
@@ -198,28 +198,28 @@ static int gen9_ctx_workarounds_init(struct drm_i915_private *dev_priv)
/* 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);
+ CTX_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));
+ CTX_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);
+ CTX_WA_CLR_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN5,
+ GEN9_CCS_TLB_PREFETCH_ENABLE);
/* WaDisableMaskBasedCammingInRCC:bxt */
if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
- WA_SET_BIT_MASKED(SLICE_ECO_CHICKEN0,
- PIXEL_MASK_CAMMING_DISABLE);
+ CTX_WA_SET_BIT_MASKED(SLICE_ECO_CHICKEN0,
+ PIXEL_MASK_CAMMING_DISABLE);
/* 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);
+ CTX_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
@@ -235,19 +235,19 @@ static int gen9_ctx_workarounds_init(struct drm_i915_private *dev_priv)
*/
/* WaForceEnableNonCoherent:skl,bxt,kbl,cfl */
- WA_SET_BIT_MASKED(HDC_CHICKEN0,
- HDC_FORCE_NON_COHERENT);
+ CTX_WA_SET_BIT_MASKED(HDC_CHICKEN0,
+ HDC_FORCE_NON_COHERENT);
/* WaDisableSamplerPowerBypassForSOPingPong:skl,bxt,kbl,cfl */
if (IS_SKYLAKE(dev_priv) ||
IS_KABYLAKE(dev_priv) ||
IS_COFFEELAKE(dev_priv) ||
IS_BXT_REVID(dev_priv, 0, BXT_REVID_B0))
- WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3,
- GEN8_SAMPLER_POWER_BYPASS_DIS);
+ CTX_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);
+ CTX_WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN2, GEN8_ST_PO_DISABLE);
/*
* Supporting preemption with fine-granularity requires changes in the
@@ -261,11 +261,11 @@ static int gen9_ctx_workarounds_init(struct drm_i915_private *dev_priv)
*/
/* WaDisable3DMidCmdPreemption:skl,bxt,glk,cfl,[cnl] */
- WA_CLR_BIT_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_3D_OBJECT_LEVEL);
+ CTX_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);
+ CTX_WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_GPGPU_LEVEL_MASK,
+ GEN9_PREEMPT_GPGPU_COMMAND_LEVEL);
return 0;
}
@@ -299,13 +299,13 @@ static int skl_tune_iz_hashing(struct drm_i915_private *dev_priv)
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]));
+ CTX_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;
}
@@ -330,20 +330,20 @@ static int bxt_ctx_workarounds_init(struct drm_i915_private *dev_priv)
return ret;
/* WaDisableThreadStallDopClockGating:bxt */
- WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN,
- STALL_DOP_GATING_DISABLE);
+ CTX_WA_SET_BIT_MASKED(GEN8_ROW_CHICKEN,
+ STALL_DOP_GATING_DISABLE);
/* WaDisableSbeCacheDispatchPortSharing:bxt */
if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_B0)) {
- WA_SET_BIT_MASKED(
+ CTX_WA_SET_BIT_MASKED(
GEN7_HALF_SLICE_CHICKEN1,
GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE);
}
/* 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);
+ CTX_WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
+ GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION);
return 0;
}
@@ -358,16 +358,16 @@ static int kbl_ctx_workarounds_init(struct drm_i915_private *dev_priv)
/* 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);
+ CTX_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);
+ CTX_WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
+ GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION);
/* WaDisableSbeCacheDispatchPortSharing:kbl */
- WA_SET_BIT_MASKED(
+ CTX_WA_SET_BIT_MASKED(
GEN7_HALF_SLICE_CHICKEN1,
GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE);
@@ -383,8 +383,8 @@ static int glk_ctx_workarounds_init(struct drm_i915_private *dev_priv)
return ret;
/* WaToEnableHwFixForPushConstHWBug:glk */
- WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
- GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION);
+ CTX_WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
+ GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION);
return 0;
}
@@ -398,11 +398,11 @@ static int cfl_ctx_workarounds_init(struct drm_i915_private *dev_priv)
return ret;
/* WaToEnableHwFixForPushConstHWBug:cfl */
- WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
- GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION);
+ CTX_WA_SET_BIT_MASKED(COMMON_SLICE_CHICKEN2,
+ GEN8_SBE_DISABLE_REPLAY_BUF_OPTIMIZATION);
/* WaDisableSbeCacheDispatchPortSharing:cfl */
- WA_SET_BIT_MASKED(
+ CTX_WA_SET_BIT_MASKED(
GEN7_HALF_SLICE_CHICKEN1,
GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE);
@@ -412,34 +412,34 @@ static int cfl_ctx_workarounds_init(struct drm_i915_private *dev_priv)
static int cnl_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
/* WaForceContextSaveRestoreNonCoherent:cnl */
- WA_SET_BIT_MASKED(CNL_HDC_CHICKEN0,
- HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT);
+ CTX_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);
+ CTX_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);
+ CTX_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);
+ CTX_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);
+ CTX_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);
+ CTX_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);
+ CTX_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);
+ CTX_WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_GPGPU_LEVEL_MASK,
+ GEN9_PREEMPT_GPGPU_COMMAND_LEVEL);
return 0;
}
@@ -448,7 +448,7 @@ int intel_ctx_workarounds_init(struct drm_i915_private *dev_priv)
{
int err;
- dev_priv->workarounds.count = 0;
+ dev_priv->workarounds.ctx_wa_count = 0;
if (INTEL_GEN(dev_priv) < 8)
err = 0;
@@ -476,7 +476,7 @@ int intel_ctx_workarounds_init(struct drm_i915_private *dev_priv)
return err;
DRM_DEBUG_DRIVER("Number of context specific w/a: %d\n",
- dev_priv->workarounds.count);
+ dev_priv->workarounds.ctx_wa_count);
return 0;
}
@@ -486,21 +486,21 @@ int intel_ctx_workarounds_emit(struct drm_i915_gem_request *req)
u32 *cs;
int ret, i;
- if (w->count == 0)
+ if (w->ctx_wa_count == 0)
return 0;
ret = req->engine->emit_flush(req, EMIT_BARRIER);
if (ret)
return ret;
- cs = intel_ring_begin(req, (w->count * 2 + 2));
+ cs = intel_ring_begin(req, (w->ctx_wa_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_LOAD_REGISTER_IMM(w->ctx_wa_count);
+ for (i = 0; i < w->ctx_wa_count; i++) {
+ *cs++ = i915_mmio_reg_offset(w->ctx_wa_reg[i].addr);
+ *cs++ = w->ctx_wa_reg[i].value;
}
*cs++ = MI_NOOP;
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 05/20] drm/i915: Save all GT WAs and apply them at a later time
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (3 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 04/20] drm/i915: Rename saved workarounds to make it explicit that they are context WAs Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-31 14:14 ` Joonas Lahtinen
2017-10-30 20:17 ` [PATCH 06/20] drm/i915: Save all Whitelist " Oscar Mateo
` (15 subsequent siblings)
20 siblings, 1 reply; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
By doing this, we can dump these workarounds in debugfs for validation (which,
at the moment, we are only able to do for the contexts WAs).
v2:
- Wrong macro used for MMIO set bit masked
- Improved naming
- Rebased
v3:
- GT instead of MMIO (Chris, Mika)
- Leave L3_PRIO_CREDITS_MASK for a separate patch
- Rebased
v4: Carry the init_early nomenclature over (Chris)
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
---
drivers/gpu/drm/i915/i915_drv.c | 5 +
drivers/gpu/drm/i915/i915_drv.h | 4 +
drivers/gpu/drm/i915/intel_workarounds.c | 247 ++++++++++++++++++++-----------
drivers/gpu/drm/i915/intel_workarounds.h | 1 +
4 files changed, 170 insertions(+), 87 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index af74574..361d4a3 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -49,6 +49,7 @@
#include "i915_drv.h"
#include "i915_trace.h"
#include "i915_vgpu.h"
+#include "intel_workarounds.h"
#include "intel_drv.h"
#include "intel_uc.h"
@@ -886,6 +887,10 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv,
BUG_ON(device_info->gen > sizeof(device_info->gen_mask) * BITS_PER_BYTE);
device_info->gen_mask = BIT(device_info->gen - 1);
+ ret = intel_gt_workarounds_init_early(dev_priv);
+ if (ret < 0)
+ return ret;
+
spin_lock_init(&dev_priv->irq_lock);
spin_lock_init(&dev_priv->gpu_error.lock);
mutex_init(&dev_priv->backlight_lock);
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index e6daf53..ca88f31 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1976,11 +1976,15 @@ struct i915_wa_reg {
};
#define I915_MAX_CTX_WA_REGS 16
+#define I915_MAX_MMIO_WA_REGS 32
struct i915_workarounds {
struct i915_wa_reg ctx_wa_reg[I915_MAX_CTX_WA_REGS];
u32 ctx_wa_count;
+ struct i915_wa_reg gt_wa_reg[I915_MAX_MMIO_WA_REGS];
+ u32 gt_wa_count;
+
u32 hw_whitelist_count[I915_NUM_ENGINES];
};
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index f4bc2d3..51868fe 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -513,15 +513,53 @@ int intel_ctx_workarounds_emit(struct drm_i915_gem_request *req)
return 0;
}
-static void bdw_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+static int gt_wa_add(struct drm_i915_private *dev_priv,
+ i915_reg_t addr,
+ const u32 mask, const u32 val)
{
+ const u32 idx = dev_priv->workarounds.gt_wa_count;
+
+ if (WARN_ON(idx >= I915_MAX_MMIO_WA_REGS))
+ return -ENOSPC;
+
+ dev_priv->workarounds.gt_wa_reg[idx].addr = addr;
+ dev_priv->workarounds.gt_wa_reg[idx].value = val;
+ dev_priv->workarounds.gt_wa_reg[idx].mask = mask;
+
+ dev_priv->workarounds.gt_wa_count++;
+
+ return 0;
+}
+
+#define GT_WA_REG(addr, mask, val) do { \
+ const int r = gt_wa_add(dev_priv, (addr), (mask), (val)); \
+ if (r) \
+ return r; \
+ } while (0)
+
+#define GT_WA_SET_BIT(addr, mask) \
+ GT_WA_REG(addr, (mask), (mask))
+
+#define GT_WA_SET_BIT_MASKED(addr, mask) \
+ GT_WA_REG(addr, (mask), _MASKED_BIT_ENABLE(mask))
+
+#define GT_WA_CLR_BIT(addr, mask) \
+ GT_WA_REG(addr, (mask), 0)
+
+#define GT_WA_SET_FIELD(addr, mask, value) \
+ GT_WA_REG(addr, (mask), (value))
+
+static int bdw_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ return 0;
}
-static void chv_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+static int chv_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ return 0;
}
-static void gen9_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+static int gen9_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
if (HAS_LLC(dev_priv)) {
/* WaCompressedResourceSamplerPbeMediaNewHashMode:skl,kbl
@@ -529,180 +567,215 @@ static void gen9_gt_workarounds_apply(struct drm_i915_private *dev_priv)
* Must match Display Engine. See
* WaCompressedResourceDisplayNewHashMode.
*/
- I915_WRITE(MMCD_MISC_CTRL,
- I915_READ(MMCD_MISC_CTRL) |
- MMCD_PCLA |
- MMCD_HOTSPOT_EN);
+ GT_WA_SET_BIT(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));
+ GT_WA_SET_BIT_MASKED(GEN9_CSFE_CHICKEN1_RCS,
+ 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);
+ GT_WA_SET_BIT(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);
+ GT_WA_SET_BIT(GAM_ECOCHK, ECOCHK_DIS_TLB);
/* WaDisableHDCInvalidation:skl,bxt,kbl,cfl */
- I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) |
- BDW_DISABLE_HDC_INVALIDATION);
+ GT_WA_SET_BIT(GAM_ECOCHK, BDW_DISABLE_HDC_INVALIDATION);
/* WaOCLCoherentLineFlush:skl,bxt,kbl,cfl */
- I915_WRITE(GEN8_L3SQCREG4, (I915_READ(GEN8_L3SQCREG4) |
- GEN8_LQSC_FLUSH_COHERENT_LINES));
+ GT_WA_SET_BIT(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));
+ GT_WA_SET_BIT_MASKED(GEN7_FF_SLICE_CS_CHICKEN1,
+ GEN9_FFSC_PERCTX_PREEMPT_CTRL);
+
+ return 0;
}
-static void skl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+static int skl_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
- gen9_gt_workarounds_apply(dev_priv);
+ int ret;
+
+ ret = gen9_gt_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
/* WaEnableGapsTsvCreditFix:skl */
- I915_WRITE(GEN8_GARBCNTL, (I915_READ(GEN8_GARBCNTL) |
- GEN9_GAPS_TSV_CREDIT_DISABLE));
+ GT_WA_SET_BIT(GEN8_GARBCNTL, GEN9_GAPS_TSV_CREDIT_DISABLE);
/* WaDisableGafsUnitClkGating:skl */
- I915_WRITE(GEN7_UCGCTL4, (I915_READ(GEN7_UCGCTL4) |
- GEN8_EU_GAUNIT_CLOCK_GATE_DISABLE));
+ GT_WA_SET_BIT(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));
+ GT_WA_SET_BIT(GEN9_GAMT_ECO_REG_RW_IA,
+ GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
+
+ return 0;
}
-static void bxt_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+static int bxt_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
- gen9_gt_workarounds_apply(dev_priv);
+ int ret;
+
+ ret = gen9_gt_workarounds_init_early(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);
+ GT_WA_SET_BIT(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));
+ GT_WA_CLR_BIT(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));
+ GT_WA_SET_BIT_MASKED(FF_SLICE_CS_CHICKEN2,
+ 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);
- }
+ if (IS_BXT_REVID(dev_priv, BXT_REVID_B0, REVID_FOREVER))
+ GT_WA_SET_FIELD(GEN8_L3SQCREG1, L3_PRIO_CREDITS_MASK,
+ L3_GENERAL_PRIO_CREDITS(62) |
+ L3_HIGH_PRIO_CREDITS(2));
/* 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));
+ GT_WA_SET_BIT(GEN9_GAMT_ECO_REG_RW_IA,
+ GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
+
+ return 0;
}
-static void kbl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+static int kbl_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
- gen9_gt_workarounds_apply(dev_priv);
+ int ret;
+
+ ret = gen9_gt_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
/* WaEnableGapsTsvCreditFix:kbl */
- I915_WRITE(GEN8_GARBCNTL, (I915_READ(GEN8_GARBCNTL) |
- GEN9_GAPS_TSV_CREDIT_DISABLE));
+ GT_WA_SET_BIT(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));
+ GT_WA_SET_BIT(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));
+ GT_WA_SET_BIT(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));
+ GT_WA_SET_BIT(GEN9_GAMT_ECO_REG_RW_IA,
+ GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
+
+ return 0;
}
-static void glk_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+static int glk_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
- gen9_gt_workarounds_apply(dev_priv);
+ int ret = gen9_gt_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
+
+ return 0;
}
-static void cfl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+static int cfl_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
- gen9_gt_workarounds_apply(dev_priv);
+ int ret;
+
+ ret = gen9_gt_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
/* WaEnableGapsTsvCreditFix:cfl */
- I915_WRITE(GEN8_GARBCNTL, (I915_READ(GEN8_GARBCNTL) |
- GEN9_GAPS_TSV_CREDIT_DISABLE));
+ GT_WA_SET_BIT(GEN8_GARBCNTL, GEN9_GAPS_TSV_CREDIT_DISABLE);
/* WaDisableGafsUnitClkGating:cfl */
- I915_WRITE(GEN7_UCGCTL4, (I915_READ(GEN7_UCGCTL4) |
- GEN8_EU_GAUNIT_CLOCK_GATE_DISABLE));
+ GT_WA_SET_BIT(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));
+ GT_WA_SET_BIT(GEN9_GAMT_ECO_REG_RW_IA,
+ GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
+
+ return 0;
}
-static void cnl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+static int cnl_gt_workarounds_init_early(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));
+ GT_WA_SET_BIT(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));
+ GT_WA_SET_BIT(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));
+ GT_WA_SET_BIT_MASKED(GEN7_FF_SLICE_CS_CHICKEN1,
+ GEN9_FFSC_PERCTX_PREEMPT_CTRL);
+
+ return 0;
}
-void intel_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+int intel_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ int err;
+
+ dev_priv->workarounds.gt_wa_count = 0;
+
if (INTEL_GEN(dev_priv) < 8)
- return;
+ err = 0;
else if (IS_BROADWELL(dev_priv))
- bdw_gt_workarounds_apply(dev_priv);
+ err = bdw_gt_workarounds_init_early(dev_priv);
else if (IS_CHERRYVIEW(dev_priv))
- chv_gt_workarounds_apply(dev_priv);
+ err = chv_gt_workarounds_init_early(dev_priv);
else if (IS_SKYLAKE(dev_priv))
- skl_gt_workarounds_apply(dev_priv);
+ err = skl_gt_workarounds_init_early(dev_priv);
else if (IS_BROXTON(dev_priv))
- bxt_gt_workarounds_apply(dev_priv);
+ err = bxt_gt_workarounds_init_early(dev_priv);
else if (IS_KABYLAKE(dev_priv))
- kbl_gt_workarounds_apply(dev_priv);
+ err = kbl_gt_workarounds_init_early(dev_priv);
else if (IS_GEMINILAKE(dev_priv))
- glk_gt_workarounds_apply(dev_priv);
+ err = glk_gt_workarounds_init_early(dev_priv);
else if (IS_COFFEELAKE(dev_priv))
- cfl_gt_workarounds_apply(dev_priv);
+ err = cfl_gt_workarounds_init_early(dev_priv);
else if (IS_CANNONLAKE(dev_priv))
- cnl_gt_workarounds_apply(dev_priv);
- else
+ err = cnl_gt_workarounds_init_early(dev_priv);
+ else {
MISSING_CASE(INTEL_GEN(dev_priv));
+ err = 0;
+ }
+ if (err)
+ return err;
+
+ DRM_DEBUG_DRIVER("Number of GT w/a: %d\n",
+ dev_priv->workarounds.gt_wa_count);
+ return 0;
+}
+
+void intel_gt_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ struct i915_workarounds *w = &dev_priv->workarounds;
+ int i;
+
+ for (i = 0; i < w->gt_wa_count; i++) {
+ i915_reg_t addr = w->gt_wa_reg[i].addr;
+ u32 value = w->gt_wa_reg[i].value;
+ u32 mask = w->gt_wa_reg[i].mask;
+
+ I915_WRITE(addr, (I915_READ(addr) & ~mask) | value);
+ }
}
static int wa_ring_whitelist_reg(struct intel_engine_cs *engine,
diff --git a/drivers/gpu/drm/i915/intel_workarounds.h b/drivers/gpu/drm/i915/intel_workarounds.h
index bba51bb..19cd81e 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.h
+++ b/drivers/gpu/drm/i915/intel_workarounds.h
@@ -28,6 +28,7 @@
int intel_ctx_workarounds_init(struct drm_i915_private *dev_priv);
int intel_ctx_workarounds_emit(struct drm_i915_gem_request *req);
+int intel_gt_workarounds_init_early(struct drm_i915_private *dev_priv);
void intel_gt_workarounds_apply(struct drm_i915_private *dev_priv);
int intel_whitelist_workarounds_apply(struct intel_engine_cs *engine);
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 06/20] drm/i915: Save all Whitelist WAs and apply them at a later time
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (4 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 05/20] drm/i915: Save all GT WAs and apply them at a later time Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 07/20] drm/i915: Create a new category of display WAs Oscar Mateo
` (14 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
Same as we have been doing for other types, this allow us to dump
the whole list of workarounds to debugs, for validation purposes.
v2:
- Improved naming
- Rebased
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
---
drivers/gpu/drm/i915/i915_debugfs.c | 2 +-
drivers/gpu/drm/i915/i915_drv.h | 3 +-
drivers/gpu/drm/i915/intel_lrc.c | 8 ++-
drivers/gpu/drm/i915/intel_workarounds.c | 113 ++++++++++++++++---------------
drivers/gpu/drm/i915/intel_workarounds.h | 3 +-
5 files changed, 67 insertions(+), 62 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index e31aba9..2890c02 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -3374,7 +3374,7 @@ static int i915_wa_registers(struct seq_file *m, void *unused)
seq_printf(m, "Context workarounds applied: %d\n", workarounds->ctx_wa_count);
for_each_engine(engine, dev_priv, id)
seq_printf(m, "HW whitelist count for %s: %d\n",
- engine->name, workarounds->hw_whitelist_count[id]);
+ engine->name, workarounds->whitelist_wa_count[id]);
for (i = 0; i < workarounds->ctx_wa_count; ++i) {
i915_reg_t addr;
u32 mask, value, read;
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index ca88f31..bd18f0e 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1985,7 +1985,8 @@ struct i915_workarounds {
struct i915_wa_reg gt_wa_reg[I915_MAX_MMIO_WA_REGS];
u32 gt_wa_count;
- u32 hw_whitelist_count[I915_NUM_ENGINES];
+ struct i915_wa_reg whitelist_wa_reg[I915_NUM_ENGINES][RING_MAX_NONPRIV_SLOTS];
+ u32 whitelist_wa_count[I915_NUM_ENGINES];
};
struct i915_virtual_gpu {
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index f0b4d2f..a489e82 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -1514,9 +1514,7 @@ static int gen9_init_render_ring(struct intel_engine_cs *engine)
if (ret)
return ret;
- ret = intel_whitelist_workarounds_apply(engine);
- if (ret)
- return ret;
+ intel_whitelist_workarounds_apply(engine);
return 0;
}
@@ -1994,6 +1992,10 @@ int logical_render_ring_init(struct intel_engine_cs *engine)
if (ret)
return ret;
+ ret = intel_whitelist_workarounds_init(engine);
+ if (ret)
+ return ret;
+
ret = intel_init_workaround_bb(engine);
if (ret) {
/*
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 51868fe..bb64fde 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -778,64 +778,64 @@ void intel_gt_workarounds_apply(struct drm_i915_private *dev_priv)
}
}
-static int wa_ring_whitelist_reg(struct intel_engine_cs *engine,
- i915_reg_t reg)
+static int whitelist_wa_add(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];
+ const uint32_t index = wa->whitelist_wa_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]++;
+ wa->whitelist_wa_reg[engine->id][index].addr =
+ RING_FORCE_TO_NONPRIV(engine->mmio_base, index);
+ wa->whitelist_wa_reg[engine->id][index].value = i915_mmio_reg_offset(reg);
+ wa->whitelist_wa_reg[engine->id][index].mask = 0xffffffff;
+
+ wa->whitelist_wa_count[engine->id]++;
return 0;
}
-static int gen9_whitelist_workarounds_apply(struct intel_engine_cs *engine)
-{
- int ret;
+#define WHITELIST_WA_REG(engine, reg) do { \
+ const int r = whitelist_wa_add(engine, reg); \
+ if (r) \
+ return r; \
+} while (0)
+static int gen9_whitelist_workarounds_init(struct intel_engine_cs *engine)
+{
/* WaVFEStateAfterPipeControlwithMediaStateClear:skl,bxt,glk,cfl */
- ret = wa_ring_whitelist_reg(engine, GEN9_CTX_PREEMPT_REG);
- if (ret)
- return ret;
+ WHITELIST_WA_REG(engine, GEN9_CTX_PREEMPT_REG);
/* WaEnablePreemptionGranularityControlByUMD:skl,bxt,kbl,cfl,[cnl] */
- ret = wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1);
- if (ret)
- return ret;
+ WHITELIST_WA_REG(engine, GEN8_CS_CHICKEN1);
/* WaAllowUMDToModifyHDCChicken1:skl,bxt,kbl,glk,cfl */
- ret = wa_ring_whitelist_reg(engine, GEN8_HDC_CHICKEN1);
- if (ret)
- return ret;
+ WHITELIST_WA_REG(engine, GEN8_HDC_CHICKEN1);
return 0;
}
-static int skl_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+static int skl_whitelist_workarounds_init(struct intel_engine_cs *engine)
{
- int ret = gen9_whitelist_workarounds_apply(engine);
+ int ret = gen9_whitelist_workarounds_init(engine);
if (ret)
return ret;
/* WaDisableLSQCROPERFforOCL:skl */
- ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4);
- if (ret)
- return ret;
+ WHITELIST_WA_REG(engine, GEN8_L3SQCREG4);
return 0;
}
-static int bxt_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+static int bxt_whitelist_workarounds_init(struct intel_engine_cs *engine)
{
struct drm_i915_private *dev_priv = engine->i915;
+ int ret;
- int ret = gen9_whitelist_workarounds_apply(engine);
+ ret = gen9_whitelist_workarounds_init(engine);
if (ret)
return ret;
@@ -844,86 +844,75 @@ static int bxt_whitelist_workarounds_apply(struct intel_engine_cs *engine)
/* 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;
+ WHITELIST_WA_REG(engine, GEN9_CS_DEBUG_MODE1);
+ WHITELIST_WA_REG(engine, GEN8_L3SQCREG4);
}
return 0;
}
-static int kbl_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+static int kbl_whitelist_workarounds_init(struct intel_engine_cs *engine)
{
- int ret = gen9_whitelist_workarounds_apply(engine);
+ int ret = gen9_whitelist_workarounds_init(engine);
if (ret)
return ret;
/* WaDisableLSQCROPERFforOCL:kbl */
- ret = wa_ring_whitelist_reg(engine, GEN8_L3SQCREG4);
- if (ret)
- return ret;
+ WHITELIST_WA_REG(engine, GEN8_L3SQCREG4);
return 0;
}
-static int glk_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+static int glk_whitelist_workarounds_init(struct intel_engine_cs *engine)
{
- int ret = gen9_whitelist_workarounds_apply(engine);
+ int ret = gen9_whitelist_workarounds_init(engine);
if (ret)
return ret;
return 0;
}
-static int cfl_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+static int cfl_whitelist_workarounds_init(struct intel_engine_cs *engine)
{
- int ret = gen9_whitelist_workarounds_apply(engine);
+ int ret = gen9_whitelist_workarounds_init(engine);
if (ret)
return ret;
return 0;
}
-static int cnl_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+static int cnl_whitelist_workarounds_init(struct intel_engine_cs *engine)
{
- int ret;
-
/* WaEnablePreemptionGranularityControlByUMD:cnl */
- ret = wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1);
- if (ret)
- return ret;
+ WHITELIST_WA_REG(engine, GEN8_CS_CHICKEN1);
return 0;
}
-int intel_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+int intel_whitelist_workarounds_init(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;
+ dev_priv->workarounds.whitelist_wa_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);
+ err = skl_whitelist_workarounds_init(engine);
else if (IS_BROXTON(dev_priv))
- err = bxt_whitelist_workarounds_apply(engine);
+ err = bxt_whitelist_workarounds_init(engine);
else if (IS_KABYLAKE(dev_priv))
- err = kbl_whitelist_workarounds_apply(engine);
+ err = kbl_whitelist_workarounds_init(engine);
else if (IS_GEMINILAKE(dev_priv))
- err = glk_whitelist_workarounds_apply(engine);
+ err = glk_whitelist_workarounds_init(engine);
else if (IS_COFFEELAKE(dev_priv))
- err = cfl_whitelist_workarounds_apply(engine);
+ err = cfl_whitelist_workarounds_init(engine);
else if (IS_CANNONLAKE(dev_priv))
- err = cnl_whitelist_workarounds_apply(engine);
+ err = cnl_whitelist_workarounds_init(engine);
else {
MISSING_CASE(INTEL_GEN(dev_priv));
err = 0;
@@ -932,6 +921,18 @@ int intel_whitelist_workarounds_apply(struct intel_engine_cs *engine)
return err;
DRM_DEBUG_DRIVER("%s: Number of whitelist w/a: %d\n", engine->name,
- dev_priv->workarounds.hw_whitelist_count[engine->id]);
+ dev_priv->workarounds.whitelist_wa_count[engine->id]);
return 0;
}
+
+void intel_whitelist_workarounds_apply(struct intel_engine_cs *engine)
+{
+ struct drm_i915_private *dev_priv = engine->i915;
+ struct i915_workarounds *w = &dev_priv->workarounds;
+ int i;
+
+ for (i = 0; i < w->whitelist_wa_count[engine->id]; i++) {
+ I915_WRITE(w->whitelist_wa_reg[engine->id][i].addr,
+ w->whitelist_wa_reg[engine->id][i].value);
+ }
+}
diff --git a/drivers/gpu/drm/i915/intel_workarounds.h b/drivers/gpu/drm/i915/intel_workarounds.h
index 19cd81e..425d48e 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.h
+++ b/drivers/gpu/drm/i915/intel_workarounds.h
@@ -31,6 +31,7 @@
int intel_gt_workarounds_init_early(struct drm_i915_private *dev_priv);
void intel_gt_workarounds_apply(struct drm_i915_private *dev_priv);
-int intel_whitelist_workarounds_apply(struct intel_engine_cs *engine);
+int intel_whitelist_workarounds_init(struct intel_engine_cs *engine);
+void intel_whitelist_workarounds_apply(struct intel_engine_cs *engine);
#endif
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 07/20] drm/i915: Create a new category of display WAs
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (5 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 06/20] drm/i915: Save all Whitelist " Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 08/20] drm/i915: Print all workaround types correctly in debugfs Oscar Mateo
` (13 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
Display workarounds do not need to be re-applied on a GPU reset
(this is, in Ville's words: "at the very least wasted effort [...]
and could even be actively harmful in case we end up clobbering
something the current display configuration depends on"). Therefore,
they have to be applied in a different place that GT ones so they
deserve their own category.
Actually populating this is left for future patches: we have to
start moving WAs from init_clock_gating into either GT or
Display functions, and this requires a good deal of careful code
reviewing.
v2: Rebased to carry the init_early nomenclature over (Chris)
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
---
drivers/gpu/drm/i915/i915_drv.c | 2 +-
drivers/gpu/drm/i915/i915_drv.h | 3 +
drivers/gpu/drm/i915/intel_pm.c | 2 +
drivers/gpu/drm/i915/intel_workarounds.c | 154 +++++++++++++++++++++++++++++--
drivers/gpu/drm/i915/intel_workarounds.h | 3 +-
5 files changed, 155 insertions(+), 9 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 361d4a3..e6c1f17 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -887,7 +887,7 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv,
BUG_ON(device_info->gen > sizeof(device_info->gen_mask) * BITS_PER_BYTE);
device_info->gen_mask = BIT(device_info->gen - 1);
- ret = intel_gt_workarounds_init_early(dev_priv);
+ ret = intel_mmio_workarounds_init_early(dev_priv);
if (ret < 0)
return ret;
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index bd18f0e..f34e318 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1985,6 +1985,9 @@ struct i915_workarounds {
struct i915_wa_reg gt_wa_reg[I915_MAX_MMIO_WA_REGS];
u32 gt_wa_count;
+ struct i915_wa_reg display_wa_reg[I915_MAX_MMIO_WA_REGS];
+ u32 display_wa_count;
+
struct i915_wa_reg whitelist_wa_reg[I915_NUM_ENGINES][RING_MAX_NONPRIV_SLOTS];
u32 whitelist_wa_count[I915_NUM_ENGINES];
};
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index acd0cbb..0d0e84b 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -29,6 +29,7 @@
#include <drm/drm_plane_helper.h>
#include "i915_drv.h"
#include "intel_drv.h"
+#include "intel_workarounds.h"
#include "../../../platform/x86/intel_ips.h"
#include <linux/module.h>
#include <drm/drm_atomic_helper.h>
@@ -9013,6 +9014,7 @@ static void i830_init_clock_gating(struct drm_i915_private *dev_priv)
void intel_init_clock_gating(struct drm_i915_private *dev_priv)
{
dev_priv->display.init_clock_gating(dev_priv);
+ intel_display_workarounds_apply(dev_priv);
}
void intel_suspend_hw(struct drm_i915_private *dev_priv)
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index bb64fde..3ee56eb 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -513,6 +513,18 @@ int intel_ctx_workarounds_emit(struct drm_i915_gem_request *req)
return 0;
}
+static void mmio_workarounds_apply(struct drm_i915_private *dev_priv,
+ struct i915_wa_reg *wa,
+ u32 count)
+{
+ int i;
+
+ for (i = 0; i < count; i++) {
+ u32 value = I915_READ(wa[i].addr);
+ I915_WRITE(wa[i].addr, (value & ~wa[i].mask) | wa[i].value);
+ }
+}
+
static int gt_wa_add(struct drm_i915_private *dev_priv,
i915_reg_t addr,
const u32 mask, const u32 val)
@@ -728,7 +740,7 @@ static int cnl_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
return 0;
}
-int intel_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
+static int gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
int err;
@@ -767,15 +779,143 @@ int intel_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
void intel_gt_workarounds_apply(struct drm_i915_private *dev_priv)
{
struct i915_workarounds *w = &dev_priv->workarounds;
- int i;
- for (i = 0; i < w->gt_wa_count; i++) {
- i915_reg_t addr = w->gt_wa_reg[i].addr;
- u32 value = w->gt_wa_reg[i].value;
- u32 mask = w->gt_wa_reg[i].mask;
+ mmio_workarounds_apply(dev_priv, w->gt_wa_reg, w->gt_wa_count);
+}
+
+static int display_wa_add(struct drm_i915_private *dev_priv,
+ i915_reg_t addr,
+ const u32 mask, const u32 val)
+{
+ const u32 idx = dev_priv->workarounds.display_wa_count;
+
+ if (WARN_ON(idx >= I915_MAX_MMIO_WA_REGS))
+ return -ENOSPC;
+
+ dev_priv->workarounds.display_wa_reg[idx].addr = addr;
+ dev_priv->workarounds.display_wa_reg[idx].value = val;
+ dev_priv->workarounds.display_wa_reg[idx].mask = mask;
+
+ dev_priv->workarounds.display_wa_count++;
+
+ return 0;
+}
+
+#define DISPLAY_WA_REG(addr, mask, val) do { \
+ const int r = display_wa_add(dev_priv, (addr), (mask), (val)); \
+ if (r) \
+ return r; \
+ } while (0)
+
+#define DISPLAY_WA_SET_BIT(addr, mask) \
+ DISPLAY_WA_REG(addr, (mask), (mask))
+
+#define DISPLAY_WA_SET_BIT_MASKED(addr, mask) \
+ DISPLAY_WA_REG(addr, (mask), _MASKED_BIT_ENABLE(mask))
+
+#define DISPLAY_WA_CLR_BIT(addr, mask) \
+ DISPLAY_WA_REG(addr, (mask), 0)
+
+#define DISPLAY_WA_SET_FIELD(addr, mask, value) \
+ DISPLAY_WA_REG(addr, (mask), (value))
+
+static int bdw_display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ return 0;
+}
+
+static int chv_display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ return 0;
+}
+
+static int skl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ return 0;
+}
+
+static int bxt_display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ return 0;
+}
+
+static int kbl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ return 0;
+}
+
+static int glk_display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ return 0;
+}
+
+static int cfl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ return 0;
+}
+
+static int cnl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ return 0;
+}
+
+static int display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ int err;
+
+ DISPLAY_WA_SET_BIT(_MMIO(0), 0); /* shut up gcc temporarily */
+ dev_priv->workarounds.display_wa_count = 0;
- I915_WRITE(addr, (I915_READ(addr) & ~mask) | value);
+ if (INTEL_GEN(dev_priv) < 8)
+ err = 0;
+ else if (IS_BROADWELL(dev_priv))
+ err = bdw_display_workarounds_init_early(dev_priv);
+ else if (IS_CHERRYVIEW(dev_priv))
+ err = chv_display_workarounds_init_early(dev_priv);
+ else if (IS_SKYLAKE(dev_priv))
+ err = skl_display_workarounds_init_early(dev_priv);
+ else if (IS_BROXTON(dev_priv))
+ err = bxt_display_workarounds_init_early(dev_priv);
+ else if (IS_KABYLAKE(dev_priv))
+ err = kbl_display_workarounds_init_early(dev_priv);
+ else if (IS_GEMINILAKE(dev_priv))
+ err = glk_display_workarounds_init_early(dev_priv);
+ else if (IS_COFFEELAKE(dev_priv))
+ err = cfl_display_workarounds_init_early(dev_priv);
+ else if (IS_CANNONLAKE(dev_priv))
+ err = cnl_display_workarounds_init_early(dev_priv);
+ else {
+ MISSING_CASE(INTEL_GEN(dev_priv));
+ err = 0;
}
+ if (err)
+ return err;
+
+ DRM_DEBUG_DRIVER("Number of display w/a: %d\n",
+ dev_priv->workarounds.display_wa_count);
+ return 0;
+}
+
+void intel_display_workarounds_apply(struct drm_i915_private *dev_priv)
+{
+ struct i915_workarounds *w = &dev_priv->workarounds;
+
+ mmio_workarounds_apply(dev_priv, w->display_wa_reg, w->display_wa_count);
+}
+
+int intel_mmio_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ int ret;
+
+ ret = gt_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
+
+ ret = display_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
+
+ return 0;
}
static int whitelist_wa_add(struct intel_engine_cs *engine,
diff --git a/drivers/gpu/drm/i915/intel_workarounds.h b/drivers/gpu/drm/i915/intel_workarounds.h
index 425d48e..f1045f1 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.h
+++ b/drivers/gpu/drm/i915/intel_workarounds.h
@@ -28,8 +28,9 @@
int intel_ctx_workarounds_init(struct drm_i915_private *dev_priv);
int intel_ctx_workarounds_emit(struct drm_i915_gem_request *req);
-int intel_gt_workarounds_init_early(struct drm_i915_private *dev_priv);
+int intel_mmio_workarounds_init_early(struct drm_i915_private *dev_priv);
void intel_gt_workarounds_apply(struct drm_i915_private *dev_priv);
+void intel_display_workarounds_apply(struct drm_i915_private *dev_priv);
int intel_whitelist_workarounds_init(struct intel_engine_cs *engine);
void intel_whitelist_workarounds_apply(struct intel_engine_cs *engine);
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 08/20] drm/i915: Print all workaround types correctly in debugfs
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (6 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 07/20] drm/i915: Create a new category of display WAs Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 09/20] drm/i915: Move WA BB stuff to the workarounds file as well Oscar Mateo
` (12 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
Let's try to make sure that all WAs are applied correctly and survive
resumes, resets, etc... (with some help from a companion i-g-t patch).
v2:
- Rebased
- Print display WAs as well (Ville)
v3:
- Grab the forcewake once for everyone, so that all reads are from
the same powercontext (Chris)
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
---
drivers/gpu/drm/i915/i915_debugfs.c | 59 ++++++++++++++++++++++++++++---------
1 file changed, 45 insertions(+), 14 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 2890c02..03ae750 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -3355,6 +3355,22 @@ static int i915_shared_dplls_info(struct seq_file *m, void *unused)
return 0;
}
+static void check_wa_register(struct seq_file *m, struct i915_wa_reg *wa_reg)
+{
+ struct drm_i915_private *dev_priv = node_to_i915(m->private);
+ u32 read;
+ bool ok;
+
+ assert_forcewakes_active(dev_priv, FORCEWAKE_ALL);
+
+ read = I915_READ_FW(wa_reg->addr);
+ ok = (wa_reg->value & wa_reg->mask) == (read & wa_reg->mask);
+ seq_printf(m, "0x%X: 0x%08x, mask: 0x%08x, read: 0x%08x, status: %s\n",
+ i915_mmio_reg_offset(wa_reg->addr),
+ wa_reg->value, wa_reg->mask, read,
+ ok ? "OK" : "FAIL");
+}
+
static int i915_wa_registers(struct seq_file *m, void *unused)
{
int i;
@@ -3364,6 +3380,7 @@ static int i915_wa_registers(struct seq_file *m, void *unused)
struct drm_device *dev = &dev_priv->drm;
struct i915_workarounds *workarounds = &dev_priv->workarounds;
enum intel_engine_id id;
+ u32 whitelist_wa_count = 0;
ret = mutex_lock_interruptible(&dev->struct_mutex);
if (ret)
@@ -3372,22 +3389,36 @@ static int i915_wa_registers(struct seq_file *m, void *unused)
intel_runtime_pm_get(dev_priv);
seq_printf(m, "Context workarounds applied: %d\n", workarounds->ctx_wa_count);
- for_each_engine(engine, dev_priv, id)
- seq_printf(m, "HW whitelist count for %s: %d\n",
- engine->name, workarounds->whitelist_wa_count[id]);
for (i = 0; i < workarounds->ctx_wa_count; ++i) {
- i915_reg_t addr;
- u32 mask, value, read;
- bool ok;
-
- addr = workarounds->ctx_wa_reg[i].addr;
- mask = workarounds->ctx_wa_reg[i].mask;
- value = workarounds->ctx_wa_reg[i].value;
- read = I915_READ(addr);
- ok = (value & mask) == (read & mask);
- seq_printf(m, "0x%X: 0x%08X, mask: 0x%08X, read: 0x%08x, status: %s\n",
- i915_mmio_reg_offset(addr), value, mask, read, ok ? "OK" : "FAIL");
+ struct i915_wa_reg *wa_reg = &workarounds->ctx_wa_reg[i];
+
+ seq_printf(m, "0x%X: 0x%08x, mask: 0x%08x\n",
+ i915_mmio_reg_offset(wa_reg->addr),
+ wa_reg->value, wa_reg->mask);
}
+ seq_putc(m, '\n');
+
+ intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);
+
+ seq_printf(m, "GT workarounds applied: %d\n", workarounds->gt_wa_count);
+ for (i = 0; i < workarounds->gt_wa_count; ++i)
+ check_wa_register(m, &workarounds->gt_wa_reg[i]);
+ seq_putc(m, '\n');
+
+ seq_printf(m, "Display workarounds applied: %d\n", workarounds->display_wa_count);
+ for (i = 0; i < workarounds->display_wa_count; ++i)
+ check_wa_register(m, &workarounds->display_wa_reg[i]);
+ seq_putc(m, '\n');
+
+ for_each_engine(engine, dev_priv, id)
+ whitelist_wa_count += workarounds->whitelist_wa_count[id];
+ seq_printf(m, "Whitelist workarounds applied: %d\n", whitelist_wa_count);
+ for_each_engine(engine, dev_priv, id)
+ for (i = 0; i < workarounds->whitelist_wa_count[id]; ++i)
+ check_wa_register(m, &workarounds->whitelist_wa_reg[id][i]);
+ seq_putc(m, '\n');
+
+ intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
intel_runtime_pm_put(dev_priv);
mutex_unlock(&dev->struct_mutex);
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 09/20] drm/i915: Move WA BB stuff to the workarounds file as well
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (7 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 08/20] drm/i915: Print all workaround types correctly in debugfs Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 10/20] drm/i915/cnl: Move GT and Display workarounds from init_clock_gating Oscar Mateo
` (11 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
Since we are trying to put all WA stuff together, do not forget about the BB WAs.
v2: s/intel_bb_workarounds_init/intel_engine_init_bb_workarounds (Chris)
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
---
drivers/gpu/drm/i915/intel_lrc.c | 253 +-----------------------------
drivers/gpu/drm/i915/intel_workarounds.c | 254 +++++++++++++++++++++++++++++++
drivers/gpu/drm/i915/intel_workarounds.h | 3 +
3 files changed, 259 insertions(+), 251 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index a489e82..d148d89d 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -1181,255 +1181,6 @@ static int execlists_request_alloc(struct drm_i915_gem_request *request)
return 0;
}
-/*
- * In this WA we need to set GEN8_L3SQCREG4[21:21] and reset it after
- * PIPE_CONTROL instruction. This is required for the flush to happen correctly
- * but there is a slight complication as this is applied in WA batch where the
- * values are only initialized once so we cannot take register value at the
- * beginning and reuse it further; hence we save its value to memory, upload a
- * constant value with bit21 set and then we restore it back with the saved value.
- * To simplify the WA, a constant value is formed by using the default value
- * of this register. This shouldn't be a problem because we are only modifying
- * it for a short period and this batch in non-premptible. We can ofcourse
- * use additional instructions that read the actual value of the register
- * at that time and set our bit of interest but it makes the WA complicated.
- *
- * This WA is also required for Gen9 so extracting as a function avoids
- * code duplication.
- */
-static u32 *
-gen8_emit_flush_coherentl3_wa(struct intel_engine_cs *engine, u32 *batch)
-{
- *batch++ = MI_STORE_REGISTER_MEM_GEN8 | MI_SRM_LRM_GLOBAL_GTT;
- *batch++ = i915_mmio_reg_offset(GEN8_L3SQCREG4);
- *batch++ = i915_ggtt_offset(engine->scratch) + 256;
- *batch++ = 0;
-
- *batch++ = MI_LOAD_REGISTER_IMM(1);
- *batch++ = i915_mmio_reg_offset(GEN8_L3SQCREG4);
- *batch++ = 0x40400000 | GEN8_LQSC_FLUSH_COHERENT_LINES;
-
- batch = gen8_emit_pipe_control(batch,
- PIPE_CONTROL_CS_STALL |
- PIPE_CONTROL_DC_FLUSH_ENABLE,
- 0);
-
- *batch++ = MI_LOAD_REGISTER_MEM_GEN8 | MI_SRM_LRM_GLOBAL_GTT;
- *batch++ = i915_mmio_reg_offset(GEN8_L3SQCREG4);
- *batch++ = i915_ggtt_offset(engine->scratch) + 256;
- *batch++ = 0;
-
- return batch;
-}
-
-/*
- * Typically we only have one indirect_ctx and per_ctx batch buffer which are
- * initialized at the beginning and shared across all contexts but this field
- * helps us to have multiple batches at different offsets and select them based
- * on a criteria. At the moment this batch always start at the beginning of the page
- * and at this point we don't have multiple wa_ctx batch buffers.
- *
- * The number of WA applied are not known at the beginning; we use this field
- * to return the no of DWORDS written.
- *
- * It is to be noted that this batch does not contain MI_BATCH_BUFFER_END
- * so it adds NOOPs as padding to make it cacheline aligned.
- * MI_BATCH_BUFFER_END will be added to perctx batch and both of them together
- * makes a complete batch buffer.
- */
-static u32 *gen8_init_indirectctx_bb(struct intel_engine_cs *engine, u32 *batch)
-{
- /* WaDisableCtxRestoreArbitration:bdw,chv */
- *batch++ = MI_ARB_ON_OFF | MI_ARB_DISABLE;
-
- /* WaFlushCoherentL3CacheLinesAtContextSwitch:bdw */
- if (IS_BROADWELL(engine->i915))
- batch = gen8_emit_flush_coherentl3_wa(engine, batch);
-
- /* WaClearSlmSpaceAtContextSwitch:bdw,chv */
- /* Actual scratch location is at 128 bytes offset */
- batch = gen8_emit_pipe_control(batch,
- PIPE_CONTROL_FLUSH_L3 |
- PIPE_CONTROL_GLOBAL_GTT_IVB |
- PIPE_CONTROL_CS_STALL |
- PIPE_CONTROL_QW_WRITE,
- i915_ggtt_offset(engine->scratch) +
- 2 * CACHELINE_BYTES);
-
- *batch++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
-
- /* Pad to end of cacheline */
- while ((unsigned long)batch % CACHELINE_BYTES)
- *batch++ = MI_NOOP;
-
- /*
- * MI_BATCH_BUFFER_END is not required in Indirect ctx BB because
- * execution depends on the length specified in terms of cache lines
- * in the register CTX_RCS_INDIRECT_CTX
- */
-
- return batch;
-}
-
-static u32 *gen9_init_indirectctx_bb(struct intel_engine_cs *engine, u32 *batch)
-{
- *batch++ = MI_ARB_ON_OFF | MI_ARB_DISABLE;
-
- /* WaFlushCoherentL3CacheLinesAtContextSwitch:skl,bxt,glk */
- batch = gen8_emit_flush_coherentl3_wa(engine, batch);
-
- /* WaDisableGatherAtSetShaderCommonSlice:skl,bxt,kbl,glk */
- *batch++ = MI_LOAD_REGISTER_IMM(1);
- *batch++ = i915_mmio_reg_offset(COMMON_SLICE_CHICKEN2);
- *batch++ = _MASKED_BIT_DISABLE(
- GEN9_DISABLE_GATHER_AT_SET_SHADER_COMMON_SLICE);
- *batch++ = MI_NOOP;
-
- /* WaClearSlmSpaceAtContextSwitch:kbl */
- /* Actual scratch location is at 128 bytes offset */
- if (IS_KBL_REVID(engine->i915, 0, KBL_REVID_A0)) {
- batch = gen8_emit_pipe_control(batch,
- PIPE_CONTROL_FLUSH_L3 |
- PIPE_CONTROL_GLOBAL_GTT_IVB |
- PIPE_CONTROL_CS_STALL |
- PIPE_CONTROL_QW_WRITE,
- i915_ggtt_offset(engine->scratch)
- + 2 * CACHELINE_BYTES);
- }
-
- /* WaMediaPoolStateCmdInWABB:bxt,glk */
- if (HAS_POOLED_EU(engine->i915)) {
- /*
- * EU pool configuration is setup along with golden context
- * during context initialization. This value depends on
- * device type (2x6 or 3x6) and needs to be updated based
- * on which subslice is disabled especially for 2x6
- * devices, however it is safe to load default
- * configuration of 3x6 device instead of masking off
- * corresponding bits because HW ignores bits of a disabled
- * subslice and drops down to appropriate config. Please
- * see render_state_setup() in i915_gem_render_state.c for
- * possible configurations, to avoid duplication they are
- * not shown here again.
- */
- *batch++ = GEN9_MEDIA_POOL_STATE;
- *batch++ = GEN9_MEDIA_POOL_ENABLE;
- *batch++ = 0x00777000;
- *batch++ = 0;
- *batch++ = 0;
- *batch++ = 0;
- }
-
- *batch++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
-
- /* Pad to end of cacheline */
- while ((unsigned long)batch % CACHELINE_BYTES)
- *batch++ = MI_NOOP;
-
- return batch;
-}
-
-#define CTX_WA_BB_OBJ_SIZE (PAGE_SIZE)
-
-static int lrc_setup_wa_ctx(struct intel_engine_cs *engine)
-{
- struct drm_i915_gem_object *obj;
- struct i915_vma *vma;
- int err;
-
- obj = i915_gem_object_create(engine->i915, CTX_WA_BB_OBJ_SIZE);
- if (IS_ERR(obj))
- return PTR_ERR(obj);
-
- vma = i915_vma_instance(obj, &engine->i915->ggtt.base, NULL);
- if (IS_ERR(vma)) {
- err = PTR_ERR(vma);
- goto err;
- }
-
- err = i915_vma_pin(vma, 0, PAGE_SIZE, PIN_GLOBAL | PIN_HIGH);
- if (err)
- goto err;
-
- engine->wa_ctx.vma = vma;
- return 0;
-
-err:
- i915_gem_object_put(obj);
- return err;
-}
-
-static void lrc_destroy_wa_ctx(struct intel_engine_cs *engine)
-{
- i915_vma_unpin_and_release(&engine->wa_ctx.vma);
-}
-
-typedef u32 *(*wa_bb_func_t)(struct intel_engine_cs *engine, u32 *batch);
-
-static int intel_init_workaround_bb(struct intel_engine_cs *engine)
-{
- struct i915_ctx_workarounds *wa_ctx = &engine->wa_ctx;
- struct i915_wa_ctx_bb *wa_bb[2] = { &wa_ctx->indirect_ctx,
- &wa_ctx->per_ctx };
- wa_bb_func_t wa_bb_fn[2];
- struct page *page;
- void *batch, *batch_ptr;
- unsigned int i;
- int ret;
-
- if (WARN_ON(engine->id != RCS || !engine->scratch))
- return -EINVAL;
-
- switch (INTEL_GEN(engine->i915)) {
- case 10:
- return 0;
- case 9:
- wa_bb_fn[0] = gen9_init_indirectctx_bb;
- wa_bb_fn[1] = NULL;
- break;
- case 8:
- wa_bb_fn[0] = gen8_init_indirectctx_bb;
- wa_bb_fn[1] = NULL;
- break;
- default:
- MISSING_CASE(INTEL_GEN(engine->i915));
- return 0;
- }
-
- ret = lrc_setup_wa_ctx(engine);
- if (ret) {
- DRM_DEBUG_DRIVER("Failed to setup context WA page: %d\n", ret);
- return ret;
- }
-
- page = i915_gem_object_get_dirty_page(wa_ctx->vma->obj, 0);
- batch = batch_ptr = kmap_atomic(page);
-
- /*
- * Emit the two workaround batch buffers, recording the offset from the
- * start of the workaround batch buffer object for each and their
- * respective sizes.
- */
- for (i = 0; i < ARRAY_SIZE(wa_bb_fn); i++) {
- wa_bb[i]->offset = batch_ptr - batch;
- if (WARN_ON(!IS_ALIGNED(wa_bb[i]->offset, CACHELINE_BYTES))) {
- ret = -EINVAL;
- break;
- }
- if (wa_bb_fn[i])
- batch_ptr = wa_bb_fn[i](engine, batch_ptr);
- wa_bb[i]->size = batch_ptr - (batch + wa_bb[i]->offset);
- }
-
- BUG_ON(batch_ptr - batch > CTX_WA_BB_OBJ_SIZE);
-
- kunmap_atomic(batch);
- if (ret)
- lrc_destroy_wa_ctx(engine);
-
- return ret;
-}
-
static u8 gtiir[] = {
[RCS] = 0,
[BCS] = 0,
@@ -1873,7 +1624,7 @@ void intel_logical_ring_cleanup(struct intel_engine_cs *engine)
intel_engine_cleanup_common(engine);
- lrc_destroy_wa_ctx(engine);
+ intel_engine_fini_bb_workarounds(engine);
engine->i915 = NULL;
dev_priv->engine[engine->id] = NULL;
kfree(engine);
@@ -1996,7 +1747,7 @@ int logical_render_ring_init(struct intel_engine_cs *engine)
if (ret)
return ret;
- ret = intel_init_workaround_bb(engine);
+ ret = intel_engine_init_bb_workarounds(engine);
if (ret) {
/*
* We continue even if we fail to initialize WA batch
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 3ee56eb..3e643b6 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -1076,3 +1076,257 @@ void intel_whitelist_workarounds_apply(struct intel_engine_cs *engine)
w->whitelist_wa_reg[engine->id][i].value);
}
}
+
+/*
+ * In this WA we need to set GEN8_L3SQCREG4[21:21] and reset it after
+ * PIPE_CONTROL instruction. This is required for the flush to happen correctly
+ * but there is a slight complication as this is applied in WA batch where the
+ * values are only initialized once so we cannot take register value at the
+ * beginning and reuse it further; hence we save its value to memory, upload a
+ * constant value with bit21 set and then we restore it back with the saved value.
+ * To simplify the WA, a constant value is formed by using the default value
+ * of this register. This shouldn't be a problem because we are only modifying
+ * it for a short period and this batch in non-premptible. We can ofcourse
+ * use additional instructions that read the actual value of the register
+ * at that time and set our bit of interest but it makes the WA complicated.
+ *
+ * This WA is also required for Gen9 so extracting as a function avoids
+ * code duplication.
+ */
+static u32 *
+gen8_emit_flush_coherentl3_wa(struct intel_engine_cs *engine, u32 *batch)
+{
+ *batch++ = MI_STORE_REGISTER_MEM_GEN8 | MI_SRM_LRM_GLOBAL_GTT;
+ *batch++ = i915_mmio_reg_offset(GEN8_L3SQCREG4);
+ *batch++ = i915_ggtt_offset(engine->scratch) + 256;
+ *batch++ = 0;
+
+ *batch++ = MI_LOAD_REGISTER_IMM(1);
+ *batch++ = i915_mmio_reg_offset(GEN8_L3SQCREG4);
+ *batch++ = 0x40400000 | GEN8_LQSC_FLUSH_COHERENT_LINES;
+
+ batch = gen8_emit_pipe_control(batch,
+ PIPE_CONTROL_CS_STALL |
+ PIPE_CONTROL_DC_FLUSH_ENABLE,
+ 0);
+
+ *batch++ = MI_LOAD_REGISTER_MEM_GEN8 | MI_SRM_LRM_GLOBAL_GTT;
+ *batch++ = i915_mmio_reg_offset(GEN8_L3SQCREG4);
+ *batch++ = i915_ggtt_offset(engine->scratch) + 256;
+ *batch++ = 0;
+
+ return batch;
+}
+
+/*
+ * Typically we only have one indirect_ctx and per_ctx batch buffer which are
+ * initialized at the beginning and shared across all contexts but this field
+ * helps us to have multiple batches at different offsets and select them based
+ * on a criteria. At the moment this batch always start at the beginning of the page
+ * and at this point we don't have multiple wa_ctx batch buffers.
+ *
+ * The number of WA applied are not known at the beginning; we use this field
+ * to return the no of DWORDS written.
+ *
+ * It is to be noted that this batch does not contain MI_BATCH_BUFFER_END
+ * so it adds NOOPs as padding to make it cacheline aligned.
+ * MI_BATCH_BUFFER_END will be added to perctx batch and both of them together
+ * makes a complete batch buffer.
+ */
+static u32 *gen8_init_indirectctx_bb(struct intel_engine_cs *engine, u32 *batch)
+{
+ /* WaDisableCtxRestoreArbitration:bdw,chv */
+ *batch++ = MI_ARB_ON_OFF | MI_ARB_DISABLE;
+
+ /* WaFlushCoherentL3CacheLinesAtContextSwitch:bdw */
+ if (IS_BROADWELL(engine->i915))
+ batch = gen8_emit_flush_coherentl3_wa(engine, batch);
+
+ /* WaClearSlmSpaceAtContextSwitch:bdw,chv */
+ /* Actual scratch location is at 128 bytes offset */
+ batch = gen8_emit_pipe_control(batch,
+ PIPE_CONTROL_FLUSH_L3 |
+ PIPE_CONTROL_GLOBAL_GTT_IVB |
+ PIPE_CONTROL_CS_STALL |
+ PIPE_CONTROL_QW_WRITE,
+ i915_ggtt_offset(engine->scratch) +
+ 2 * CACHELINE_BYTES);
+
+ *batch++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
+
+ /* Pad to end of cacheline */
+ while ((unsigned long)batch % CACHELINE_BYTES)
+ *batch++ = MI_NOOP;
+
+ /*
+ * MI_BATCH_BUFFER_END is not required in Indirect ctx BB because
+ * execution depends on the length specified in terms of cache lines
+ * in the register CTX_RCS_INDIRECT_CTX
+ */
+
+ return batch;
+}
+
+static u32 *gen9_init_indirectctx_bb(struct intel_engine_cs *engine, u32 *batch)
+{
+ *batch++ = MI_ARB_ON_OFF | MI_ARB_DISABLE;
+
+ /* WaFlushCoherentL3CacheLinesAtContextSwitch:skl,bxt,glk */
+ batch = gen8_emit_flush_coherentl3_wa(engine, batch);
+
+ /* WaDisableGatherAtSetShaderCommonSlice:skl,bxt,kbl,glk */
+ *batch++ = MI_LOAD_REGISTER_IMM(1);
+ *batch++ = i915_mmio_reg_offset(COMMON_SLICE_CHICKEN2);
+ *batch++ = _MASKED_BIT_DISABLE(
+ GEN9_DISABLE_GATHER_AT_SET_SHADER_COMMON_SLICE);
+ *batch++ = MI_NOOP;
+
+ /* WaClearSlmSpaceAtContextSwitch:kbl */
+ /* Actual scratch location is at 128 bytes offset */
+ if (IS_KBL_REVID(engine->i915, 0, KBL_REVID_A0)) {
+ batch = gen8_emit_pipe_control(batch,
+ PIPE_CONTROL_FLUSH_L3 |
+ PIPE_CONTROL_GLOBAL_GTT_IVB |
+ PIPE_CONTROL_CS_STALL |
+ PIPE_CONTROL_QW_WRITE,
+ i915_ggtt_offset(engine->scratch)
+ + 2 * CACHELINE_BYTES);
+ }
+
+ /* WaMediaPoolStateCmdInWABB:bxt,glk */
+ if (HAS_POOLED_EU(engine->i915)) {
+ /*
+ * EU pool configuration is setup along with golden context
+ * during context initialization. This value depends on
+ * device type (2x6 or 3x6) and needs to be updated based
+ * on which subslice is disabled especially for 2x6
+ * devices, however it is safe to load default
+ * configuration of 3x6 device instead of masking off
+ * corresponding bits because HW ignores bits of a disabled
+ * subslice and drops down to appropriate config. Please
+ * see render_state_setup() in i915_gem_render_state.c for
+ * possible configurations, to avoid duplication they are
+ * not shown here again.
+ */
+ *batch++ = GEN9_MEDIA_POOL_STATE;
+ *batch++ = GEN9_MEDIA_POOL_ENABLE;
+ *batch++ = 0x00777000;
+ *batch++ = 0;
+ *batch++ = 0;
+ *batch++ = 0;
+ }
+
+ *batch++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
+
+ /* Pad to end of cacheline */
+ while ((unsigned long)batch % CACHELINE_BYTES)
+ *batch++ = MI_NOOP;
+
+ return batch;
+}
+
+#define CTX_WA_BB_OBJ_SIZE (PAGE_SIZE)
+
+static int lrc_setup_wa_ctx(struct intel_engine_cs *engine)
+{
+ struct drm_i915_gem_object *obj;
+ struct i915_vma *vma;
+ int err;
+
+ obj = i915_gem_object_create(engine->i915, CTX_WA_BB_OBJ_SIZE);
+ if (IS_ERR(obj))
+ return PTR_ERR(obj);
+
+ vma = i915_vma_instance(obj, &engine->i915->ggtt.base, NULL);
+ if (IS_ERR(vma)) {
+ err = PTR_ERR(vma);
+ goto err;
+ }
+
+ err = i915_vma_pin(vma, 0, PAGE_SIZE, PIN_GLOBAL | PIN_HIGH);
+ if (err)
+ goto err;
+
+ engine->wa_ctx.vma = vma;
+ return 0;
+
+err:
+ i915_gem_object_put(obj);
+ return err;
+}
+
+static void lrc_destroy_wa_ctx(struct intel_engine_cs *engine)
+{
+ i915_vma_unpin_and_release(&engine->wa_ctx.vma);
+}
+
+typedef u32 *(*wa_bb_func_t)(struct intel_engine_cs *engine, u32 *batch);
+
+int intel_engine_init_bb_workarounds(struct intel_engine_cs *engine)
+{
+ struct i915_ctx_workarounds *wa_ctx = &engine->wa_ctx;
+ struct i915_wa_ctx_bb *wa_bb[2] = { &wa_ctx->indirect_ctx,
+ &wa_ctx->per_ctx };
+ wa_bb_func_t wa_bb_fn[2];
+ struct page *page;
+ void *batch, *batch_ptr;
+ unsigned int i;
+ int ret;
+
+ if (WARN_ON(engine->id != RCS || !engine->scratch))
+ return -EINVAL;
+
+ switch (INTEL_GEN(engine->i915)) {
+ case 10:
+ return 0;
+ case 9:
+ wa_bb_fn[0] = gen9_init_indirectctx_bb;
+ wa_bb_fn[1] = NULL;
+ break;
+ case 8:
+ wa_bb_fn[0] = gen8_init_indirectctx_bb;
+ wa_bb_fn[1] = NULL;
+ break;
+ default:
+ MISSING_CASE(INTEL_GEN(engine->i915));
+ return 0;
+ }
+
+ ret = lrc_setup_wa_ctx(engine);
+ if (ret) {
+ DRM_DEBUG_DRIVER("Failed to setup context WA page: %d\n", ret);
+ return ret;
+ }
+
+ page = i915_gem_object_get_dirty_page(wa_ctx->vma->obj, 0);
+ batch = batch_ptr = kmap_atomic(page);
+
+ /*
+ * Emit the two workaround batch buffers, recording the offset from the
+ * start of the workaround batch buffer object for each and their
+ * respective sizes.
+ */
+ for (i = 0; i < ARRAY_SIZE(wa_bb_fn); i++) {
+ wa_bb[i]->offset = batch_ptr - batch;
+ if (WARN_ON(!IS_ALIGNED(wa_bb[i]->offset, CACHELINE_BYTES))) {
+ ret = -EINVAL;
+ break;
+ }
+ if (wa_bb_fn[i])
+ batch_ptr = wa_bb_fn[i](engine, batch_ptr);
+ wa_bb[i]->size = batch_ptr - (batch + wa_bb[i]->offset);
+ }
+
+ BUG_ON(batch_ptr - batch > CTX_WA_BB_OBJ_SIZE);
+
+ kunmap_atomic(batch);
+ if (ret)
+ lrc_destroy_wa_ctx(engine);
+
+ return ret;
+}
+
+void intel_engine_fini_bb_workarounds(struct intel_engine_cs *engine)
+{
+ lrc_destroy_wa_ctx(engine);
+}
diff --git a/drivers/gpu/drm/i915/intel_workarounds.h b/drivers/gpu/drm/i915/intel_workarounds.h
index f1045f1..9cc8e41 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.h
+++ b/drivers/gpu/drm/i915/intel_workarounds.h
@@ -35,4 +35,7 @@
int intel_whitelist_workarounds_init(struct intel_engine_cs *engine);
void intel_whitelist_workarounds_apply(struct intel_engine_cs *engine);
+int intel_engine_init_bb_workarounds(struct intel_engine_cs *engine);
+void intel_engine_fini_bb_workarounds(struct intel_engine_cs *engine);
+
#endif
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 10/20] drm/i915/cnl: Move GT and Display workarounds from init_clock_gating
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (8 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 09/20] drm/i915: Move WA BB stuff to the workarounds file as well Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 11/20] drm/i915/gen9: " Oscar Mateo
` (10 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
To their rightful place inside intel_workarounds.c
v2: classify WaSarbUnitClockGatingDisable as GT WA (Ville)
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v1)
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
---
drivers/gpu/drm/i915/intel_pm.c | 32 +-------------------------------
drivers/gpu/drm/i915/intel_workarounds.c | 26 +++++++++++++++++++++++++-
2 files changed, 26 insertions(+), 32 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 0d0e84b..ff3ac6c 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -8519,36 +8519,6 @@ static void cnp_init_clock_gating(struct drm_i915_private *dev_priv)
CNP_PWM_CGE_GATING_DISABLE);
}
-static void cnl_init_clock_gating(struct drm_i915_private *dev_priv)
-{
- u32 val;
- cnp_init_clock_gating(dev_priv);
-
- /* This is not an Wa. Enable for better image quality */
- I915_WRITE(_3D_CHICKEN3,
- _MASKED_BIT_ENABLE(_3D_CHICKEN3_AA_LINE_QUALITY_FIX_ENABLE));
-
- /* WaEnableChickenDCPR:cnl */
- I915_WRITE(GEN8_CHICKEN_DCPR_1,
- I915_READ(GEN8_CHICKEN_DCPR_1) | MASK_WAKEMEM);
-
- /* WaFbcWakeMemOn:cnl */
- I915_WRITE(DISP_ARB_CTL, I915_READ(DISP_ARB_CTL) |
- DISP_FBC_MEMORY_WAKE);
-
- /* WaSarbUnitClockGatingDisable:cnl (pre-prod) */
- if (IS_CNL_REVID(dev_priv, CNL_REVID_A0, CNL_REVID_B0))
- I915_WRITE(SLICE_UNIT_LEVEL_CLKGATE,
- I915_READ(SLICE_UNIT_LEVEL_CLKGATE) |
- SARBUNIT_CLKGATE_DIS);
-
- /* Display WA #1133: WaFbcSkipSegments:cnl */
- val = I915_READ(ILK_DPFC_CHICKEN);
- val &= ~GLK_SKIP_SEG_COUNT_MASK;
- val |= GLK_SKIP_SEG_EN | GLK_SKIP_SEG_COUNT(1);
- I915_WRITE(ILK_DPFC_CHICKEN, val);
-}
-
static void cfl_init_clock_gating(struct drm_i915_private *dev_priv)
{
cnp_init_clock_gating(dev_priv);
@@ -9040,7 +9010,7 @@ static void nop_init_clock_gating(struct drm_i915_private *dev_priv)
void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
{
if (IS_CANNONLAKE(dev_priv))
- dev_priv->display.init_clock_gating = cnl_init_clock_gating;
+ dev_priv->display.init_clock_gating = nop_init_clock_gating;
else if (IS_COFFEELAKE(dev_priv))
dev_priv->display.init_clock_gating = cfl_init_clock_gating;
else if (IS_SKYLAKE(dev_priv))
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 3e643b6..82defb1 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -724,6 +724,15 @@ static int cfl_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
static int cnl_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ /* This is not an Wa. Enable for better image quality */
+ GT_WA_SET_BIT_MASKED(_3D_CHICKEN3,
+ _3D_CHICKEN3_AA_LINE_QUALITY_FIX_ENABLE);
+
+ /* WaSarbUnitClockGatingDisable:cnl (pre-prod) */
+ if (IS_CNL_REVID(dev_priv, CNL_REVID_A0, CNL_REVID_B0))
+ GT_WA_SET_BIT(SLICE_UNIT_LEVEL_CLKGATE,
+ SARBUNIT_CLKGATE_DIS);
+
/* WaDisableI2mCycleOnWRPort:cnl (pre-prod) */
if (IS_CNL_REVID(dev_priv, CNL_REVID_B0, CNL_REVID_B0))
GT_WA_SET_BIT(GAMT_CHKN_BIT_REG,
@@ -856,6 +865,22 @@ static int cfl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
static int cnl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ if (HAS_PCH_CNP(dev_priv)) {
+ /* Wa #1181 */
+ DISPLAY_WA_SET_BIT(SOUTH_DSPCLK_GATE_D,
+ CNP_PWM_CGE_GATING_DISABLE);
+ }
+
+ /* WaEnableChickenDCPR:cnl */
+ DISPLAY_WA_SET_BIT(GEN8_CHICKEN_DCPR_1, MASK_WAKEMEM);
+
+ /* WaFbcWakeMemOn:cnl */
+ DISPLAY_WA_SET_BIT(DISP_ARB_CTL, DISP_FBC_MEMORY_WAKE);
+
+ /* Display WA #1133: WaFbcSkipSegments:cnl */
+ DISPLAY_WA_SET_FIELD(ILK_DPFC_CHICKEN, GLK_SKIP_SEG_COUNT_MASK,
+ GLK_SKIP_SEG_EN | GLK_SKIP_SEG_COUNT(1));
+
return 0;
}
@@ -863,7 +888,6 @@ static int display_workarounds_init_early(struct drm_i915_private *dev_priv)
{
int err;
- DISPLAY_WA_SET_BIT(_MMIO(0), 0); /* shut up gcc temporarily */
dev_priv->workarounds.display_wa_count = 0;
if (INTEL_GEN(dev_priv) < 8)
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 11/20] drm/i915/gen9: Move GT and Display workarounds from init_clock_gating
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (9 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 10/20] drm/i915/cnl: Move GT and Display workarounds from init_clock_gating Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 12/20] drm/i915/cfl: " Oscar Mateo
` (9 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
To their rightful place inside intel_workarounds.c
v2:
- Rebase on WA removed
- Rebased to carry the init_early nomenclature over (Chris)
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v1)
---
drivers/gpu/drm/i915/intel_pm.c | 48 -----------------------------
drivers/gpu/drm/i915/intel_workarounds.c | 52 ++++++++++++++++++++++++++++++++
2 files changed, 52 insertions(+), 48 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index ff3ac6c..f712b02 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -57,50 +57,8 @@
#define INTEL_RC6p_ENABLE (1<<1)
#define INTEL_RC6pp_ENABLE (1<<2)
-static void gen9_init_clock_gating(struct drm_i915_private *dev_priv)
-{
- if (HAS_LLC(dev_priv)) {
- /*
- * WaCompressedResourceDisplayNewHashMode:skl,kbl
- * Display WA#0390: skl,kbl
- *
- * Must match Sampler, Pixel Back End, and Media. See
- * WaCompressedResourceSamplerPbeMediaNewHashMode.
- */
- I915_WRITE(CHICKEN_PAR1_1,
- I915_READ(CHICKEN_PAR1_1) |
- SKL_DE_COMPRESSED_HASH_MODE);
- }
-
- /* See Bspec note for PSR2_CTL bit 31, Wa#828:skl,bxt,kbl,cfl */
- I915_WRITE(CHICKEN_PAR1_1,
- I915_READ(CHICKEN_PAR1_1) | SKL_EDP_PSR_FIX_RDWRAP);
-
- /* WaEnableChickenDCPR:skl,bxt,kbl,glk,cfl */
- I915_WRITE(GEN8_CHICKEN_DCPR_1,
- I915_READ(GEN8_CHICKEN_DCPR_1) | MASK_WAKEMEM);
-
- /* WaFbcTurnOffFbcWatermark:skl,bxt,kbl,cfl */
- /* WaFbcWakeMemOn:skl,bxt,kbl,glk,cfl */
- I915_WRITE(DISP_ARB_CTL, I915_READ(DISP_ARB_CTL) |
- DISP_FBC_WM_DIS |
- DISP_FBC_MEMORY_WAKE);
-
- /* WaFbcHighMemBwCorruptionAvoidance:skl,bxt,kbl,cfl */
- I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
- ILK_DPFC_DISABLE_DUMMY0);
-
- if (IS_SKYLAKE(dev_priv)) {
- /* WaDisableDopClockGating */
- I915_WRITE(GEN7_MISCCPCTL, I915_READ(GEN7_MISCCPCTL)
- & ~GEN7_DOP_CLOCK_GATE_ENABLE);
- }
-}
-
static void bxt_init_clock_gating(struct drm_i915_private *dev_priv)
{
- gen9_init_clock_gating(dev_priv);
-
/* WaDisableSDEUnitClockGating:bxt */
I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
@@ -123,7 +81,6 @@ static void bxt_init_clock_gating(struct drm_i915_private *dev_priv)
static void glk_init_clock_gating(struct drm_i915_private *dev_priv)
{
u32 val;
- gen9_init_clock_gating(dev_priv);
/*
* WaDisablePWMClockGating:glk
@@ -8522,7 +8479,6 @@ static void cnp_init_clock_gating(struct drm_i915_private *dev_priv)
static void cfl_init_clock_gating(struct drm_i915_private *dev_priv)
{
cnp_init_clock_gating(dev_priv);
- gen9_init_clock_gating(dev_priv);
/* WaFbcNukeOnHostModify:cfl */
I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
@@ -8531,8 +8487,6 @@ static void cfl_init_clock_gating(struct drm_i915_private *dev_priv)
static void kbl_init_clock_gating(struct drm_i915_private *dev_priv)
{
- gen9_init_clock_gating(dev_priv);
-
/* WaDisableSDEUnitClockGating:kbl */
if (IS_KBL_REVID(dev_priv, 0, KBL_REVID_B0))
I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
@@ -8550,8 +8504,6 @@ static void kbl_init_clock_gating(struct drm_i915_private *dev_priv)
static void skl_init_clock_gating(struct drm_i915_private *dev_priv)
{
- gen9_init_clock_gating(dev_priv);
-
/* WAC6entrylatency:skl */
I915_WRITE(FBC_LLC_READ_CTRL, I915_READ(FBC_LLC_READ_CTRL) |
FBC_LLC_FULLY_OPEN);
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 82defb1..08b2e8c 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -838,28 +838,80 @@ static int chv_display_workarounds_init_early(struct drm_i915_private *dev_priv)
return 0;
}
+static int gen9_display_workarounds_init_early(struct drm_i915_private *dev_priv)
+{
+ if (HAS_LLC(dev_priv)) {
+ /*
+ * WaCompressedResourceDisplayNewHashMode:skl,kbl
+ * Display WA#0390: skl,kbl
+ *
+ * Must match Sampler, Pixel Back End, and Media. See
+ * WaCompressedResourceSamplerPbeMediaNewHashMode.
+ */
+ DISPLAY_WA_SET_BIT(CHICKEN_PAR1_1, SKL_DE_COMPRESSED_HASH_MODE);
+ }
+
+ /* See Bspec note for PSR2_CTL bit 31, Wa#828:skl,bxt,kbl,cfl */
+ DISPLAY_WA_SET_BIT(CHICKEN_PAR1_1, SKL_EDP_PSR_FIX_RDWRAP);
+
+ /* WaEnableChickenDCPR:skl,bxt,kbl,glk,cfl */
+ DISPLAY_WA_SET_BIT(GEN8_CHICKEN_DCPR_1, MASK_WAKEMEM);
+
+ /* WaFbcTurnOffFbcWatermark:skl,bxt,kbl,cfl */
+ /* WaFbcWakeMemOn:skl,bxt,kbl,glk,cfl */
+ DISPLAY_WA_SET_BIT(DISP_ARB_CTL, DISP_FBC_WM_DIS | DISP_FBC_MEMORY_WAKE);
+
+ /* WaFbcHighMemBwCorruptionAvoidance:skl,bxt,kbl,cfl */
+ DISPLAY_WA_SET_BIT(ILK_DPFC_CHICKEN, ILK_DPFC_DISABLE_DUMMY0);
+
+ return 0;
+}
+
static int skl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ int ret = gen9_display_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
+
+ /* WaDisableDopClockGating */
+ DISPLAY_WA_CLR_BIT(GEN7_MISCCPCTL, GEN7_DOP_CLOCK_GATE_ENABLE);
+
return 0;
}
static int bxt_display_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ int ret = gen9_display_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
+
return 0;
}
static int kbl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ int ret = gen9_display_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
+
return 0;
}
static int glk_display_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ int ret = gen9_display_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
+
return 0;
}
static int cfl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ int ret = gen9_display_workarounds_init_early(dev_priv);
+ if (ret)
+ return ret;
+
return 0;
}
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 12/20] drm/i915/cfl: Move GT and Display workarounds from init_clock_gating
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (10 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 11/20] drm/i915/gen9: " Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 13/20] drm/i915/glk: " Oscar Mateo
` (8 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
To their rightful place inside intel_workarounds.c
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
---
drivers/gpu/drm/i915/intel_pm.c | 23 +----------------------
drivers/gpu/drm/i915/intel_workarounds.c | 9 +++++++++
2 files changed, 10 insertions(+), 22 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index f712b02..a85a001 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -8466,25 +8466,6 @@ static void gen8_set_l3sqc_credits(struct drm_i915_private *dev_priv,
I915_WRITE(GEN7_MISCCPCTL, misccpctl);
}
-static void cnp_init_clock_gating(struct drm_i915_private *dev_priv)
-{
- if (!HAS_PCH_CNP(dev_priv))
- return;
-
- /* Wa #1181 */
- I915_WRITE(SOUTH_DSPCLK_GATE_D, I915_READ(SOUTH_DSPCLK_GATE_D) |
- CNP_PWM_CGE_GATING_DISABLE);
-}
-
-static void cfl_init_clock_gating(struct drm_i915_private *dev_priv)
-{
- cnp_init_clock_gating(dev_priv);
-
- /* WaFbcNukeOnHostModify:cfl */
- I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
- ILK_DPFC_NUKE_ON_ANY_MODIFICATION);
-}
-
static void kbl_init_clock_gating(struct drm_i915_private *dev_priv)
{
/* WaDisableSDEUnitClockGating:kbl */
@@ -8961,10 +8942,8 @@ static void nop_init_clock_gating(struct drm_i915_private *dev_priv)
*/
void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
{
- if (IS_CANNONLAKE(dev_priv))
+ if (IS_CANNONLAKE(dev_priv) || IS_COFFEELAKE(dev_priv))
dev_priv->display.init_clock_gating = nop_init_clock_gating;
- else if (IS_COFFEELAKE(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;
else if (IS_KABYLAKE(dev_priv))
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 08b2e8c..6c40627 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -912,6 +912,15 @@ static int cfl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
if (ret)
return ret;
+ if (HAS_PCH_CNP(dev_priv)) {
+ /* Wa #1181 */
+ DISPLAY_WA_SET_BIT(SOUTH_DSPCLK_GATE_D,
+ CNP_PWM_CGE_GATING_DISABLE);
+ }
+
+ /* WaFbcNukeOnHostModify:cfl */
+ DISPLAY_WA_SET_BIT(ILK_DPFC_CHICKEN, ILK_DPFC_NUKE_ON_ANY_MODIFICATION);
+
return 0;
}
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 13/20] drm/i915/glk: Move GT and Display workarounds from init_clock_gating
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (11 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 12/20] drm/i915/cfl: " Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 14/20] drm/i915/kbl: " Oscar Mateo
` (7 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
To their rightful place inside intel_workarounds.c
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
---
drivers/gpu/drm/i915/intel_pm.c | 33 ++------------------------------
drivers/gpu/drm/i915/intel_workarounds.c | 17 ++++++++++++++++
2 files changed, 19 insertions(+), 31 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index a85a001..b5e7432 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -78,34 +78,6 @@ static void bxt_init_clock_gating(struct drm_i915_private *dev_priv)
PWM1_GATING_DIS | PWM2_GATING_DIS);
}
-static void glk_init_clock_gating(struct drm_i915_private *dev_priv)
-{
- u32 val;
-
- /*
- * WaDisablePWMClockGating:glk
- * Backlight PWM may stop in the asserted state, causing backlight
- * to stay fully on.
- */
- I915_WRITE(GEN9_CLKGATE_DIS_0, I915_READ(GEN9_CLKGATE_DIS_0) |
- PWM1_GATING_DIS | PWM2_GATING_DIS);
-
- /* WaDDIIOTimeout:glk */
- if (IS_GLK_REVID(dev_priv, 0, GLK_REVID_A1)) {
- u32 val = I915_READ(CHICKEN_MISC_2);
- val &= ~(GLK_CL0_PWR_DOWN |
- GLK_CL1_PWR_DOWN |
- GLK_CL2_PWR_DOWN);
- I915_WRITE(CHICKEN_MISC_2, val);
- }
-
- /* Display WA #1133: WaFbcSkipSegments:glk */
- val = I915_READ(ILK_DPFC_CHICKEN);
- val &= ~GLK_SKIP_SEG_COUNT_MASK;
- val |= GLK_SKIP_SEG_EN | GLK_SKIP_SEG_COUNT(1);
- I915_WRITE(ILK_DPFC_CHICKEN, val);
-}
-
static void i915_pineview_get_mem_freq(struct drm_i915_private *dev_priv)
{
u32 tmp;
@@ -8942,7 +8914,8 @@ static void nop_init_clock_gating(struct drm_i915_private *dev_priv)
*/
void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
{
- if (IS_CANNONLAKE(dev_priv) || IS_COFFEELAKE(dev_priv))
+ if (IS_CANNONLAKE(dev_priv) || IS_COFFEELAKE(dev_priv) ||
+ IS_GEMINILAKE(dev_priv))
dev_priv->display.init_clock_gating = nop_init_clock_gating;
else if (IS_SKYLAKE(dev_priv))
dev_priv->display.init_clock_gating = skl_init_clock_gating;
@@ -8950,8 +8923,6 @@ void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
dev_priv->display.init_clock_gating = kbl_init_clock_gating;
else if (IS_BROXTON(dev_priv))
dev_priv->display.init_clock_gating = bxt_init_clock_gating;
- else if (IS_GEMINILAKE(dev_priv))
- dev_priv->display.init_clock_gating = glk_init_clock_gating;
else if (IS_BROADWELL(dev_priv))
dev_priv->display.init_clock_gating = bdw_init_clock_gating;
else if (IS_CHERRYVIEW(dev_priv))
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 6c40627..e69753d 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -903,6 +903,23 @@ static int glk_display_workarounds_init_early(struct drm_i915_private *dev_priv)
if (ret)
return ret;
+ /*
+ * WaDisablePWMClockGating:glk
+ * Backlight PWM may stop in the asserted state, causing backlight
+ * to stay fully on.
+ */
+ DISPLAY_WA_SET_BIT(GEN9_CLKGATE_DIS_0, PWM1_GATING_DIS | PWM2_GATING_DIS);
+
+ /* WaDDIIOTimeout:glk */
+ if (IS_GLK_REVID(dev_priv, 0, GLK_REVID_A1))
+ DISPLAY_WA_CLR_BIT(CHICKEN_MISC_2, GLK_CL0_PWR_DOWN |
+ GLK_CL1_PWR_DOWN |
+ GLK_CL2_PWR_DOWN);
+
+ /* Display WA #1133: WaFbcSkipSegments:glk */
+ DISPLAY_WA_SET_FIELD(ILK_DPFC_CHICKEN, GLK_SKIP_SEG_COUNT_MASK,
+ GLK_SKIP_SEG_EN | GLK_SKIP_SEG_COUNT(1));
+
return 0;
}
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 14/20] drm/i915/kbl: Move GT and Display workarounds from init_clock_gating
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (12 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 13/20] drm/i915/glk: " Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 15/20] drm/i915/bxt: " Oscar Mateo
` (6 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
To their rightful place inside intel_workarounds.c
v2: Classify WaDisableSDEUnitClockGating and WaDisableGamClockGating
as GT WAs
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v1)
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
---
drivers/gpu/drm/i915/intel_pm.c | 21 +--------------------
drivers/gpu/drm/i915/intel_workarounds.c | 11 +++++++++++
2 files changed, 12 insertions(+), 20 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index b5e7432..046553b 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -8438,23 +8438,6 @@ static void gen8_set_l3sqc_credits(struct drm_i915_private *dev_priv,
I915_WRITE(GEN7_MISCCPCTL, misccpctl);
}
-static void kbl_init_clock_gating(struct drm_i915_private *dev_priv)
-{
- /* WaDisableSDEUnitClockGating:kbl */
- if (IS_KBL_REVID(dev_priv, 0, KBL_REVID_B0))
- I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
- GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
-
- /* WaDisableGamClockGating:kbl */
- if (IS_KBL_REVID(dev_priv, 0, KBL_REVID_B0))
- I915_WRITE(GEN6_UCGCTL1, I915_READ(GEN6_UCGCTL1) |
- GEN6_GAMUNIT_CLOCK_GATE_DISABLE);
-
- /* WaFbcNukeOnHostModify:kbl */
- I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
- ILK_DPFC_NUKE_ON_ANY_MODIFICATION);
-}
-
static void skl_init_clock_gating(struct drm_i915_private *dev_priv)
{
/* WAC6entrylatency:skl */
@@ -8915,12 +8898,10 @@ static void nop_init_clock_gating(struct drm_i915_private *dev_priv)
void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
{
if (IS_CANNONLAKE(dev_priv) || IS_COFFEELAKE(dev_priv) ||
- IS_GEMINILAKE(dev_priv))
+ IS_GEMINILAKE(dev_priv) || IS_KABYLAKE(dev_priv))
dev_priv->display.init_clock_gating = nop_init_clock_gating;
else if (IS_SKYLAKE(dev_priv))
dev_priv->display.init_clock_gating = skl_init_clock_gating;
- else if (IS_KABYLAKE(dev_priv))
- dev_priv->display.init_clock_gating = kbl_init_clock_gating;
else if (IS_BROXTON(dev_priv))
dev_priv->display.init_clock_gating = bxt_init_clock_gating;
else if (IS_BROADWELL(dev_priv))
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index e69753d..cb38c84 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -674,6 +674,14 @@ static int kbl_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
if (ret)
return ret;
+ /* WaDisableSDEUnitClockGating:kbl */
+ if (IS_KBL_REVID(dev_priv, 0, KBL_REVID_B0))
+ GT_WA_SET_BIT(GEN8_UCGCTL6, GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
+
+ /* WaDisableGamClockGating:kbl */
+ if (IS_KBL_REVID(dev_priv, 0, KBL_REVID_B0))
+ GT_WA_SET_BIT(GEN6_UCGCTL1, GEN6_GAMUNIT_CLOCK_GATE_DISABLE);
+
/* WaEnableGapsTsvCreditFix:kbl */
GT_WA_SET_BIT(GEN8_GARBCNTL, GEN9_GAPS_TSV_CREDIT_DISABLE);
@@ -894,6 +902,9 @@ static int kbl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
if (ret)
return ret;
+ /* WaFbcNukeOnHostModify:kbl */
+ DISPLAY_WA_SET_BIT(ILK_DPFC_CHICKEN, ILK_DPFC_NUKE_ON_ANY_MODIFICATION);
+
return 0;
}
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 15/20] drm/i915/bxt: Move GT and Display workarounds from init_clock_gating
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (13 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 14/20] drm/i915/kbl: " Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 16/20] drm/i915/skl: " Oscar Mateo
` (5 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
To their rightful place inside intel_workarounds.c
v2: Classify WaDisableSDEUnitClockGating as GT WA
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v1)
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
---
drivers/gpu/drm/i915/intel_pm.c | 26 ++------------------------
drivers/gpu/drm/i915/intel_workarounds.c | 15 +++++++++++++++
2 files changed, 17 insertions(+), 24 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 046553b..8b5a83c 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -57,27 +57,6 @@
#define INTEL_RC6p_ENABLE (1<<1)
#define INTEL_RC6pp_ENABLE (1<<2)
-static void bxt_init_clock_gating(struct drm_i915_private *dev_priv)
-{
- /* WaDisableSDEUnitClockGating:bxt */
- I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
- GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
-
- /*
- * FIXME:
- * GEN8_HDCUNIT_CLOCK_GATE_DISABLE_HDCREQ applies on 3x6 GT SKUs only.
- */
- I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
- GEN8_HDCUNIT_CLOCK_GATE_DISABLE_HDCREQ);
-
- /*
- * Wa: Backlight PWM may stop in the asserted state, causing backlight
- * to stay fully on.
- */
- I915_WRITE(GEN9_CLKGATE_DIS_0, I915_READ(GEN9_CLKGATE_DIS_0) |
- PWM1_GATING_DIS | PWM2_GATING_DIS);
-}
-
static void i915_pineview_get_mem_freq(struct drm_i915_private *dev_priv)
{
u32 tmp;
@@ -8898,12 +8877,11 @@ static void nop_init_clock_gating(struct drm_i915_private *dev_priv)
void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
{
if (IS_CANNONLAKE(dev_priv) || IS_COFFEELAKE(dev_priv) ||
- IS_GEMINILAKE(dev_priv) || IS_KABYLAKE(dev_priv))
+ IS_GEMINILAKE(dev_priv) || IS_KABYLAKE(dev_priv) ||
+ IS_BROXTON(dev_priv))
dev_priv->display.init_clock_gating = nop_init_clock_gating;
else if (IS_SKYLAKE(dev_priv))
dev_priv->display.init_clock_gating = skl_init_clock_gating;
- else if (IS_BROXTON(dev_priv))
- dev_priv->display.init_clock_gating = bxt_init_clock_gating;
else if (IS_BROADWELL(dev_priv))
dev_priv->display.init_clock_gating = bdw_init_clock_gating;
else if (IS_CHERRYVIEW(dev_priv))
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index cb38c84..da82562 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -636,6 +636,9 @@ static int bxt_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
if (ret)
return ret;
+ /* WaDisableSDEUnitClockGating:bxt */
+ GT_WA_SET_BIT(GEN8_UCGCTL6, GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
+
/* WaStoreMultiplePTEenable:bxt */
/* This is a requirement according to Hardware specification */
if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
@@ -893,6 +896,18 @@ static int bxt_display_workarounds_init_early(struct drm_i915_private *dev_priv)
if (ret)
return ret;
+ /*
+ * FIXME:
+ * GEN8_HDCUNIT_CLOCK_GATE_DISABLE_HDCREQ applies on 3x6 GT SKUs only.
+ */
+ DISPLAY_WA_SET_BIT(GEN8_UCGCTL6, GEN8_HDCUNIT_CLOCK_GATE_DISABLE_HDCREQ);
+
+ /*
+ * Wa: Backlight PWM may stop in the asserted state, causing backlight
+ * to stay fully on.
+ */
+ DISPLAY_WA_SET_BIT(GEN9_CLKGATE_DIS_0, PWM1_GATING_DIS | PWM2_GATING_DIS);
+
return 0;
}
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 16/20] drm/i915/skl: Move GT and Display workarounds from init_clock_gating
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (14 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 15/20] drm/i915/bxt: " Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 17/20] drm/i915/chv: " Oscar Mateo
` (4 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
To their rightful place inside intel_workarounds.c
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
---
drivers/gpu/drm/i915/intel_pm.c | 15 +--------------
drivers/gpu/drm/i915/intel_workarounds.c | 6 ++++++
2 files changed, 7 insertions(+), 14 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 8b5a83c..ee0e784 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -8417,17 +8417,6 @@ static void gen8_set_l3sqc_credits(struct drm_i915_private *dev_priv,
I915_WRITE(GEN7_MISCCPCTL, misccpctl);
}
-static void skl_init_clock_gating(struct drm_i915_private *dev_priv)
-{
- /* WAC6entrylatency:skl */
- I915_WRITE(FBC_LLC_READ_CTRL, I915_READ(FBC_LLC_READ_CTRL) |
- FBC_LLC_FULLY_OPEN);
-
- /* WaFbcNukeOnHostModify:skl */
- I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
- ILK_DPFC_NUKE_ON_ANY_MODIFICATION);
-}
-
static void bdw_init_clock_gating(struct drm_i915_private *dev_priv)
{
/* The GTT cache must be disabled if the system is using 2M pages. */
@@ -8878,10 +8867,8 @@ void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
{
if (IS_CANNONLAKE(dev_priv) || IS_COFFEELAKE(dev_priv) ||
IS_GEMINILAKE(dev_priv) || IS_KABYLAKE(dev_priv) ||
- IS_BROXTON(dev_priv))
+ IS_BROXTON(dev_priv) || IS_SKYLAKE(dev_priv))
dev_priv->display.init_clock_gating = nop_init_clock_gating;
- else if (IS_SKYLAKE(dev_priv))
- dev_priv->display.init_clock_gating = skl_init_clock_gating;
else if (IS_BROADWELL(dev_priv))
dev_priv->display.init_clock_gating = bdw_init_clock_gating;
else if (IS_CHERRYVIEW(dev_priv))
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index da82562..71e1858 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -887,6 +887,12 @@ static int skl_display_workarounds_init_early(struct drm_i915_private *dev_priv)
/* WaDisableDopClockGating */
DISPLAY_WA_CLR_BIT(GEN7_MISCCPCTL, GEN7_DOP_CLOCK_GATE_ENABLE);
+ /* WAC6entrylatency:skl */
+ DISPLAY_WA_SET_BIT(FBC_LLC_READ_CTRL, FBC_LLC_FULLY_OPEN);
+
+ /* WaFbcNukeOnHostModify:skl */
+ DISPLAY_WA_SET_BIT(ILK_DPFC_CHICKEN, ILK_DPFC_NUKE_ON_ANY_MODIFICATION);
+
return 0;
}
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 17/20] drm/i915/chv: Move GT and Display workarounds from init_clock_gating
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (15 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 16/20] drm/i915/skl: " Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 18/20] drm/i915/bdw: " Oscar Mateo
` (3 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
To their rightful place inside intel_workarounds.c
TODO: Notice that we are leaving WaProgramL3SqcReg1Default (and the
associated WaTempDisableDOPClkGating) behind because it requires extra
careful reviewing. We'll deal with it in a separate patch.
v2: Classify WaDisableCSUnitClockGating and WaDisableSDEUnitClockGating
as GT WAs
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v1)
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
---
drivers/gpu/drm/i915/intel_pm.c | 24 ------------------------
drivers/gpu/drm/i915/intel_workarounds.c | 21 +++++++++++++++++++++
2 files changed, 21 insertions(+), 24 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index ee0e784..b15892b 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -8707,36 +8707,12 @@ static void vlv_init_clock_gating(struct drm_i915_private *dev_priv)
static void chv_init_clock_gating(struct drm_i915_private *dev_priv)
{
- /* WaVSRefCountFullforceMissDisable:chv */
- /* WaDSRefCountFullforceMissDisable:chv */
- I915_WRITE(GEN7_FF_THREAD_MODE,
- I915_READ(GEN7_FF_THREAD_MODE) &
- ~(GEN8_FF_DS_REF_CNT_FFME | GEN7_FF_VS_REF_CNT_FFME));
-
- /* WaDisableSemaphoreAndSyncFlipWait:chv */
- I915_WRITE(GEN6_RC_SLEEP_PSMI_CONTROL,
- _MASKED_BIT_ENABLE(GEN8_RC_SEMA_IDLE_MSG_DISABLE));
-
- /* WaDisableCSUnitClockGating:chv */
- I915_WRITE(GEN6_UCGCTL1, I915_READ(GEN6_UCGCTL1) |
- GEN6_CSUNIT_CLOCK_GATE_DISABLE);
-
- /* WaDisableSDEUnitClockGating:chv */
- I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
- GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
-
/*
* WaProgramL3SqcReg1Default:chv
* See gfxspecs/Related Documents/Performance Guide/
* LSQC Setting Recommendations.
*/
gen8_set_l3sqc_credits(dev_priv, 38, 2);
-
- /*
- * GTT cache may not work with big pages, so if those
- * are ever enabled GTT cache may need to be disabled.
- */
- I915_WRITE(HSW_GTT_CACHE_EN, GTT_CACHE_EN_ALL);
}
static void g4x_init_clock_gating(struct drm_i915_private *dev_priv)
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 71e1858..c0f4513 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -568,6 +568,27 @@ static int bdw_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
static int chv_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ /* WaVSRefCountFullforceMissDisable:chv */
+ /* WaDSRefCountFullforceMissDisable:chv */
+ GT_WA_CLR_BIT(GEN7_FF_THREAD_MODE, GEN8_FF_DS_REF_CNT_FFME |
+ GEN7_FF_VS_REF_CNT_FFME);
+
+ /* WaDisableSemaphoreAndSyncFlipWait:chv */
+ GT_WA_SET_BIT_MASKED(GEN6_RC_SLEEP_PSMI_CONTROL,
+ GEN8_RC_SEMA_IDLE_MSG_DISABLE);
+
+ /*
+ * GTT cache may not work with big pages, so if those
+ * are ever enabled GTT cache may need to be disabled.
+ */
+ GT_WA_SET_FIELD(HSW_GTT_CACHE_EN, 0xFFFFFFFF, GTT_CACHE_EN_ALL);
+
+ /* WaDisableCSUnitClockGating:chv */
+ GT_WA_SET_BIT(GEN6_UCGCTL1, GEN6_CSUNIT_CLOCK_GATE_DISABLE);
+
+ /* WaDisableSDEUnitClockGating:chv */
+ GT_WA_SET_BIT(GEN8_UCGCTL6, GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
+
return 0;
}
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 18/20] drm/i915/bdw: Move GT and Display workarounds from init_clock_gating
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (16 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 17/20] drm/i915/chv: " Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 19/20] drm/i915: Move WaProgramL3SqcReg1Default to the workarounds file Oscar Mateo
` (2 subsequent siblings)
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Paulo Zanoni, Rodrigo Vivi
To their rightful place inside intel_workarounds.c
TODO: Notice that we are leaving WaProgramL3SqcReg1Default (and the
associated WaTempDisableDOPClkGating) behind because it requires
extra careful reviewing. We'll deal with it in a separate patch.
TODO2: Decide what to do with lpt_init_clock_gating (shouldn't
WADPOClockGatingDisable be marked as "bdw"? shouldn't it be
protected by HAS_PCH_LPT_LP? do we want to move the whole thing
to the workarounds file or not?).
v2: Classify WaDisableSDEUnitClockGating as GT WA
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
---
drivers/gpu/drm/i915/intel_pm.c | 47 --------------------------------
drivers/gpu/drm/i915/intel_workarounds.c | 43 +++++++++++++++++++++++++++++
2 files changed, 43 insertions(+), 47 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index b15892b..f1e3a04 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -8419,59 +8419,12 @@ static void gen8_set_l3sqc_credits(struct drm_i915_private *dev_priv,
static void bdw_init_clock_gating(struct drm_i915_private *dev_priv)
{
- /* The GTT cache must be disabled if the system is using 2M pages. */
- bool can_use_gtt_cache = !HAS_PAGE_SIZES(dev_priv,
- I915_GTT_PAGE_SIZE_2M);
- enum pipe pipe;
-
ilk_init_lp_watermarks(dev_priv);
- /* WaSwitchSolVfFArbitrationPriority:bdw */
- I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) | HSW_ECOCHK_ARB_PRIO_SOL);
-
- /* WaPsrDPAMaskVBlankInSRD:bdw */
- I915_WRITE(CHICKEN_PAR1_1,
- I915_READ(CHICKEN_PAR1_1) | DPA_MASK_VBLANK_SRD);
-
- /* WaPsrDPRSUnmaskVBlankInSRD:bdw */
- for_each_pipe(dev_priv, pipe) {
- I915_WRITE(CHICKEN_PIPESL_1(pipe),
- I915_READ(CHICKEN_PIPESL_1(pipe)) |
- BDW_DPRS_MASK_VBLANK_SRD);
- }
-
- /* WaVSRefCountFullforceMissDisable:bdw */
- /* WaDSRefCountFullforceMissDisable:bdw */
- I915_WRITE(GEN7_FF_THREAD_MODE,
- I915_READ(GEN7_FF_THREAD_MODE) &
- ~(GEN8_FF_DS_REF_CNT_FFME | GEN7_FF_VS_REF_CNT_FFME));
-
- I915_WRITE(GEN6_RC_SLEEP_PSMI_CONTROL,
- _MASKED_BIT_ENABLE(GEN8_RC_SEMA_IDLE_MSG_DISABLE));
-
- /* WaDisableSDEUnitClockGating:bdw */
- I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
- GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
-
/* WaProgramL3SqcReg1Default:bdw */
gen8_set_l3sqc_credits(dev_priv, 30, 2);
- /* WaGttCachingOffByDefault:bdw */
- I915_WRITE(HSW_GTT_CACHE_EN, can_use_gtt_cache ? GTT_CACHE_EN_ALL : 0);
-
- /* WaKVMNotificationOnConfigChange:bdw */
- I915_WRITE(CHICKEN_PAR2_1, I915_READ(CHICKEN_PAR2_1)
- | KVM_CONFIG_CHANGE_NOTIFICATION_SELECT);
-
lpt_init_clock_gating(dev_priv);
-
- /* WaDisableDopClockGating:bdw
- *
- * Also see the CHICKEN2 write in bdw_init_workarounds() to disable DOP
- * clock gating.
- */
- I915_WRITE(GEN6_UCGCTL1,
- I915_READ(GEN6_UCGCTL1) | GEN6_EU_TCUNIT_CLOCK_GATE_DISABLE);
}
static void hsw_init_clock_gating(struct drm_i915_private *dev_priv)
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index c0f4513..0a88c92 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -563,6 +563,28 @@ static int gt_wa_add(struct drm_i915_private *dev_priv,
static int bdw_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ /* The GTT cache must be disabled if the system is using 2M pages. */
+ bool can_use_gtt_cache = !HAS_PAGE_SIZES(dev_priv, I915_GTT_PAGE_SIZE_2M);
+
+ /* WaSwitchSolVfFArbitrationPriority:bdw */
+ GT_WA_SET_BIT(GAM_ECOCHK, HSW_ECOCHK_ARB_PRIO_SOL);
+
+ /* WaVSRefCountFullforceMissDisable:bdw */
+ /* WaDSRefCountFullforceMissDisable:bdw */
+ GT_WA_CLR_BIT(GEN7_FF_THREAD_MODE, GEN8_FF_DS_REF_CNT_FFME |
+ GEN7_FF_VS_REF_CNT_FFME);
+
+ /* WaDisableSemaphoreAndSyncFlipWait:bdw */
+ GT_WA_SET_BIT_MASKED(GEN6_RC_SLEEP_PSMI_CONTROL,
+ GEN8_RC_SEMA_IDLE_MSG_DISABLE);
+
+ /* WaDisableSDEUnitClockGating:bdw */
+ GT_WA_SET_BIT(GEN8_UCGCTL6, GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
+
+ /* WaGttCachingOffByDefault:bdw */
+ GT_WA_SET_FIELD(HSW_GTT_CACHE_EN, 0xFFFFFFFF,
+ can_use_gtt_cache ? GTT_CACHE_EN_ALL : 0);
+
return 0;
}
@@ -862,6 +884,27 @@ static int display_wa_add(struct drm_i915_private *dev_priv,
static int bdw_display_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ enum pipe pipe;
+
+ /* WaPsrDPAMaskVBlankInSRD:bdw */
+ DISPLAY_WA_SET_BIT(CHICKEN_PAR1_1, DPA_MASK_VBLANK_SRD);
+
+ /* WaPsrDPRSUnmaskVBlankInSRD:bdw */
+ for_each_pipe(dev_priv, pipe) {
+ DISPLAY_WA_SET_BIT(CHICKEN_PIPESL_1(pipe),
+ BDW_DPRS_MASK_VBLANK_SRD);
+ }
+
+ /* WaKVMNotificationOnConfigChange:bdw */
+ DISPLAY_WA_SET_BIT(CHICKEN_PAR2_1, KVM_CONFIG_CHANGE_NOTIFICATION_SELECT);
+
+ /* WaDisableDopClockGating:bdw
+ *
+ * Also see the CHICKEN2 write in bdw_gt_workarounds_init() to disable
+ * DOP clock gating.
+ */
+ DISPLAY_WA_SET_BIT(GEN6_UCGCTL1, GEN6_EU_TCUNIT_CLOCK_GATE_DISABLE);
+
return 0;
}
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 19/20] drm/i915: Move WaProgramL3SqcReg1Default to the workarounds file
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (17 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 18/20] drm/i915/bdw: " Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:17 ` [PATCH 20/20] drm/i915: Document the i915_workarounds file Oscar Mateo
2017-10-30 20:42 ` ✗ Fi.CI.BAT: warning for Refactor HW workaround code (rev4) Patchwork
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx; +Cc: Rodrigo Vivi
This means moving WaTempDisableDOPClkGating as well.
Notice that BXT implements a similar WA to WaProgramL3SqcReg1Default
but, according to the BSpec, it does not require WaTempDisableDOPClkGating.
v2: Use pre-/post- hook calls (Chris)
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Imre Deak <imre.deak@intel.com>
---
drivers/gpu/drm/i915/i915_drv.h | 10 ++++++
drivers/gpu/drm/i915/intel_pm.c | 46 ++-----------------------
drivers/gpu/drm/i915/intel_workarounds.c | 59 ++++++++++++++++++++++++++++++++
3 files changed, 72 insertions(+), 43 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index f34e318..620a8f5 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1968,8 +1968,18 @@ struct i915_frontbuffer_tracking {
unsigned flip_bits;
};
+struct i915_wa_reg;
+
+typedef void (* wa_hook_func)(struct drm_i915_private *dev_priv,
+ struct i915_wa_reg *wa);
+
struct i915_wa_reg {
i915_reg_t addr;
+
+ wa_hook_func pre_hook;
+ wa_hook_func post_hook;
+ u32 hook_data;
+
u32 value;
/* bitmask representing WA bits */
u32 mask;
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index f1e3a04..0fc0670 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -8391,39 +8391,10 @@ static void lpt_suspend_hw(struct drm_i915_private *dev_priv)
}
}
-static void gen8_set_l3sqc_credits(struct drm_i915_private *dev_priv,
- int general_prio_credits,
- int high_prio_credits)
-{
- u32 misccpctl;
- u32 val;
-
- /* WaTempDisableDOPClkGating:bdw */
- misccpctl = I915_READ(GEN7_MISCCPCTL);
- I915_WRITE(GEN7_MISCCPCTL, misccpctl & ~GEN7_DOP_CLOCK_GATE_ENABLE);
-
- val = I915_READ(GEN8_L3SQCREG1);
- val &= ~L3_PRIO_CREDITS_MASK;
- val |= L3_GENERAL_PRIO_CREDITS(general_prio_credits);
- val |= L3_HIGH_PRIO_CREDITS(high_prio_credits);
- I915_WRITE(GEN8_L3SQCREG1, val);
-
- /*
- * Wait at least 100 clocks before re-enabling clock gating.
- * See the definition of L3SQCREG1 in BSpec.
- */
- POSTING_READ(GEN8_L3SQCREG1);
- udelay(1);
- I915_WRITE(GEN7_MISCCPCTL, misccpctl);
-}
-
static void bdw_init_clock_gating(struct drm_i915_private *dev_priv)
{
ilk_init_lp_watermarks(dev_priv);
- /* WaProgramL3SqcReg1Default:bdw */
- gen8_set_l3sqc_credits(dev_priv, 30, 2);
-
lpt_init_clock_gating(dev_priv);
}
@@ -8658,16 +8629,6 @@ static void vlv_init_clock_gating(struct drm_i915_private *dev_priv)
I915_WRITE(VLV_GUNIT_CLOCK_GATE, GCFG_DIS);
}
-static void chv_init_clock_gating(struct drm_i915_private *dev_priv)
-{
- /*
- * WaProgramL3SqcReg1Default:chv
- * See gfxspecs/Related Documents/Performance Guide/
- * LSQC Setting Recommendations.
- */
- gen8_set_l3sqc_credits(dev_priv, 38, 2);
-}
-
static void g4x_init_clock_gating(struct drm_i915_private *dev_priv)
{
uint32_t dspclk_gate;
@@ -8795,13 +8756,12 @@ static void nop_init_clock_gating(struct drm_i915_private *dev_priv)
void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
{
if (IS_CANNONLAKE(dev_priv) || IS_COFFEELAKE(dev_priv) ||
- IS_GEMINILAKE(dev_priv) || IS_KABYLAKE(dev_priv) ||
- IS_BROXTON(dev_priv) || IS_SKYLAKE(dev_priv))
+ IS_GEMINILAKE(dev_priv) || IS_KABYLAKE(dev_priv) ||
+ IS_BROXTON(dev_priv) || IS_SKYLAKE(dev_priv) ||
+ IS_CHERRYVIEW(dev_priv))
dev_priv->display.init_clock_gating = nop_init_clock_gating;
else if (IS_BROADWELL(dev_priv))
dev_priv->display.init_clock_gating = bdw_init_clock_gating;
- else if (IS_CHERRYVIEW(dev_priv))
- dev_priv->display.init_clock_gating = chv_init_clock_gating;
else if (IS_HASWELL(dev_priv))
dev_priv->display.init_clock_gating = hsw_init_clock_gating;
else if (IS_IVYBRIDGE(dev_priv))
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 0a88c92..04f6af9 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -521,7 +521,14 @@ static void mmio_workarounds_apply(struct drm_i915_private *dev_priv,
for (i = 0; i < count; i++) {
u32 value = I915_READ(wa[i].addr);
+
+ if (wa[i].pre_hook)
+ wa[i].pre_hook(dev_priv, &wa[i]);
+
I915_WRITE(wa[i].addr, (value & ~wa[i].mask) | wa[i].value);
+
+ if (wa[i].post_hook)
+ wa[i].post_hook(dev_priv, &wa[i]);
}
}
@@ -561,8 +568,35 @@ static int gt_wa_add(struct drm_i915_private *dev_priv,
#define GT_WA_SET_FIELD(addr, mask, value) \
GT_WA_REG(addr, (mask), (value))
+/* WaTempDisableDOPClkGating */
+static void disable_dop_clock_gating(struct drm_i915_private *dev_priv,
+ struct i915_wa_reg *wa)
+{
+ u32 misccpctl = I915_READ(GEN7_MISCCPCTL);
+
+ wa->hook_data = misccpctl;
+ I915_WRITE(GEN7_MISCCPCTL, misccpctl & ~GEN7_DOP_CLOCK_GATE_ENABLE);
+}
+
+/* WaTempDisableDOPClkGating */
+static void enable_dop_clock_gating(struct drm_i915_private *dev_priv,
+ struct i915_wa_reg *wa)
+{
+ u32 misccpctl = wa->hook_data;
+
+ /*
+ * Wait at least 100 clocks before re-enabling clock
+ * gating. See the definition of L3SQCREG1 in BSpec.
+ */
+ POSTING_READ(GEN8_L3SQCREG1);
+ udelay(1);
+ I915_WRITE(GEN7_MISCCPCTL, misccpctl);
+}
+
static int bdw_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ u32 idx;
+
/* The GTT cache must be disabled if the system is using 2M pages. */
bool can_use_gtt_cache = !HAS_PAGE_SIZES(dev_priv, I915_GTT_PAGE_SIZE_2M);
@@ -585,11 +619,24 @@ static int bdw_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
GT_WA_SET_FIELD(HSW_GTT_CACHE_EN, 0xFFFFFFFF,
can_use_gtt_cache ? GTT_CACHE_EN_ALL : 0);
+ /*
+ * WaProgramL3SqcReg1Default:bdw
+ * See "gfxspecs/Related Documents/Performance Guide/LSQC Setting
+ * Recommendations" and also WaTempDisableDOPClkGating.
+ */
+ GT_WA_SET_FIELD(GEN8_L3SQCREG1, L3_PRIO_CREDITS_MASK,
+ L3_GENERAL_PRIO_CREDITS(30) | L3_HIGH_PRIO_CREDITS(2));
+ idx = dev_priv->workarounds.gt_wa_count - 1;
+ dev_priv->workarounds.gt_wa_reg[idx].pre_hook = enable_dop_clock_gating;
+ dev_priv->workarounds.gt_wa_reg[idx].post_hook = disable_dop_clock_gating;
+
return 0;
}
static int chv_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
{
+ u32 idx;
+
/* WaVSRefCountFullforceMissDisable:chv */
/* WaDSRefCountFullforceMissDisable:chv */
GT_WA_CLR_BIT(GEN7_FF_THREAD_MODE, GEN8_FF_DS_REF_CNT_FFME |
@@ -611,6 +658,17 @@ static int chv_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
/* WaDisableSDEUnitClockGating:chv */
GT_WA_SET_BIT(GEN8_UCGCTL6, GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
+ /*
+ * WaProgramL3SqcReg1Default:chv
+ * See "gfxspecs/Related Documents/Performance Guide/LSQC Setting
+ * Recommendations" and also WaTempDisableDOPClkGating.
+ */
+ GT_WA_SET_FIELD(GEN8_L3SQCREG1, L3_PRIO_CREDITS_MASK,
+ L3_GENERAL_PRIO_CREDITS(38) | L3_HIGH_PRIO_CREDITS(2));
+ idx = dev_priv->workarounds.gt_wa_count - 1;
+ dev_priv->workarounds.gt_wa_reg[idx].pre_hook = enable_dop_clock_gating;
+ dev_priv->workarounds.gt_wa_reg[idx].post_hook = disable_dop_clock_gating;
+
return 0;
}
@@ -699,6 +757,7 @@ static int bxt_gt_workarounds_init_early(struct drm_i915_private *dev_priv)
}
/* WaProgramL3SqcReg1DefaultForPerf:bxt */
+ /* Does not require WaTempDisableDOPClkGating anymore */
if (IS_BXT_REVID(dev_priv, BXT_REVID_B0, REVID_FOREVER))
GT_WA_SET_FIELD(GEN8_L3SQCREG1, L3_PRIO_CREDITS_MASK,
L3_GENERAL_PRIO_CREDITS(62) |
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* [PATCH 20/20] drm/i915: Document the i915_workarounds file
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (18 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 19/20] drm/i915: Move WaProgramL3SqcReg1Default to the workarounds file Oscar Mateo
@ 2017-10-30 20:17 ` Oscar Mateo
2017-10-30 20:42 ` ✗ Fi.CI.BAT: warning for Refactor HW workaround code (rev4) Patchwork
20 siblings, 0 replies; 26+ messages in thread
From: Oscar Mateo @ 2017-10-30 20:17 UTC (permalink / raw)
To: intel-gfx
Does what it says on the tin (plus a few fixes in some old comments).
v2: Include display WAs as a separate category.
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
---
drivers/gpu/drm/i915/intel_pm.c | 4 +--
drivers/gpu/drm/i915/intel_workarounds.c | 47 +++++++++++++++++++++++++++-----
2 files changed, 41 insertions(+), 10 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 0fc0670..98c2ac8 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -8749,9 +8749,7 @@ static void nop_init_clock_gating(struct drm_i915_private *dev_priv)
* @dev_priv: device private
*
* Setup the hooks that configure which clocks of a given platform can be
- * gated and also apply various GT and display specific workarounds for these
- * platforms. Note that some GT specific workarounds are applied separately
- * when GPU contexts or batchbuffers start their execution.
+ * gated.
*/
void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
{
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 04f6af9..ef71a1d 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -25,6 +25,38 @@
#include "i915_drv.h"
#include "intel_workarounds.h"
+/**
+ * DOC: Hardware workarounds
+ *
+ * This file is a central place to implement most* of the required workarounds
+ * required for HW 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 generated once and then emitted (via Load
+ * Register Immediate commands) everytime a new context is created.
+ * - GT workarounds. The list of these WAs is generated once and then applied
+ * whenever these registers revert to default values (on GPU reset,
+ * suspend/resume**, etc..).
+ * - Display workarounds. The list is generated once (at the same time as the GT
+ * workarounds) but 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.
+ *
+ * * 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).
+ *
+ * ** 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.
+ *
+ */
+
static int ctx_wa_add(struct drm_i915_private *dev_priv,
i915_reg_t addr,
const u32 mask, const u32 val)
@@ -190,9 +222,9 @@ static int gen9_ctx_workarounds_init(struct drm_i915_private *dev_priv)
CTX_WA_SET_BIT_MASKED(GEN7_COMMON_SLICE_CHICKEN1,
GEN9_RHWO_OPTIMIZATION_DISABLE);
/*
- * WA also requires GEN9_SLICE_COMMON_ECO_CHICKEN0[14:14] to be set
- * but we do that in per ctx batchbuffer as there is an issue
- * with this register not getting restored on ctx restore
+ * WA also requires GEN9_SLICE_COMMON_ECO_CHICKEN0[14:14] to be
+ * set but we do that in per ctx batchbuffer as there is an
+ * issue with this register not getting restored on ctx restore.
*/
}
@@ -1340,10 +1372,11 @@ void intel_whitelist_workarounds_apply(struct intel_engine_cs *engine)
* but there is a slight complication as this is applied in WA batch where the
* values are only initialized once so we cannot take register value at the
* beginning and reuse it further; hence we save its value to memory, upload a
- * constant value with bit21 set and then we restore it back with the saved value.
+ * constant value with bit21 set and then we restore it back with the saved
+ * value.
* To simplify the WA, a constant value is formed by using the default value
* of this register. This shouldn't be a problem because we are only modifying
- * it for a short period and this batch in non-premptible. We can ofcourse
+ * it for a short period and this batch in non-premptible. We can of course
* use additional instructions that read the actual value of the register
* at that time and set our bit of interest but it makes the WA complicated.
*
@@ -1379,8 +1412,8 @@ void intel_whitelist_workarounds_apply(struct intel_engine_cs *engine)
* Typically we only have one indirect_ctx and per_ctx batch buffer which are
* initialized at the beginning and shared across all contexts but this field
* helps us to have multiple batches at different offsets and select them based
- * on a criteria. At the moment this batch always start at the beginning of the page
- * and at this point we don't have multiple wa_ctx batch buffers.
+ * on a criteria. At the moment this batch always start at the beginning of the
+ * page and at this point we don't have multiple wa_ctx batch buffers.
*
* The number of WA applied are not known at the beginning; we use this field
* to return the no of DWORDS written.
--
1.9.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply related [flat|nested] 26+ messages in thread
* ✗ Fi.CI.BAT: warning for Refactor HW workaround code (rev4)
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
` (19 preceding siblings ...)
2017-10-30 20:17 ` [PATCH 20/20] drm/i915: Document the i915_workarounds file Oscar Mateo
@ 2017-10-30 20:42 ` Patchwork
20 siblings, 0 replies; 26+ messages in thread
From: Patchwork @ 2017-10-30 20:42 UTC (permalink / raw)
To: Oscar Mateo; +Cc: intel-gfx
== Series Details ==
Series: Refactor HW workaround code (rev4)
URL : https://patchwork.freedesktop.org/series/31611/
State : warning
== Summary ==
Series 31611v4 Refactor HW workaround code
https://patchwork.freedesktop.org/api/1.0/series/31611/revisions/4/mbox/
Test gem_exec_suspend:
Subgroup basic-s3:
pass -> DMESG-WARN (fi-cfl-s) fdo#103186
Test gem_workarounds:
Subgroup basic-read:
pass -> SKIP (fi-bdw-5557u)
pass -> SKIP (fi-bdw-gvtdvm)
pass -> SKIP (fi-bsw-n3050)
pass -> SKIP (fi-skl-6260u)
pass -> SKIP (fi-skl-6600u)
pass -> SKIP (fi-skl-6700hq)
pass -> SKIP (fi-skl-6700k)
pass -> SKIP (fi-skl-6770hq)
pass -> SKIP (fi-skl-gvtdvm)
pass -> SKIP (fi-bxt-dsi)
pass -> SKIP (fi-bxt-j4205)
pass -> SKIP (fi-kbl-7500u)
pass -> SKIP (fi-kbl-7560u)
pass -> SKIP (fi-kbl-7567u)
pass -> SKIP (fi-kbl-r)
pass -> SKIP (fi-glk-1)
pass -> SKIP (fi-cfl-s)
Test kms_pipe_crc_basic:
Subgroup suspend-read-crc-pipe-c:
incomplete -> PASS (fi-bxt-dsi)
Test pm_rpm:
Subgroup basic-pci-d3-state:
incomplete -> FAIL (fi-bxt-j4205)
Test drv_module_reload:
Subgroup basic-reload:
pass -> DMESG-WARN (fi-bdw-5557u)
pass -> DMESG-WARN (fi-bdw-gvtdvm)
pass -> DMESG-WARN (fi-bsw-n3050) fdo#103479
Subgroup basic-no-display:
pass -> DMESG-WARN (fi-bdw-5557u)
pass -> DMESG-WARN (fi-bdw-gvtdvm)
pass -> DMESG-WARN (fi-bsw-n3050)
Subgroup basic-reload-inject:
pass -> DMESG-WARN (fi-bdw-5557u)
pass -> DMESG-WARN (fi-bdw-gvtdvm)
pass -> DMESG-WARN (fi-bsw-n3050)
fdo#103186 https://bugs.freedesktop.org/show_bug.cgi?id=103186
fdo#103479 https://bugs.freedesktop.org/show_bug.cgi?id=103479
fi-bdw-5557u total:289 pass:262 dwarn:3 dfail:0 fail:2 skip:22 time:456s
fi-bdw-gvtdvm total:289 pass:261 dwarn:3 dfail:0 fail:0 skip:25 time:453s
fi-blb-e6850 total:289 pass:223 dwarn:1 dfail:0 fail:0 skip:65 time:373s
fi-bsw-n3050 total:289 pass:237 dwarn:3 dfail:0 fail:2 skip:47 time:538s
fi-bwr-2160 total:289 pass:183 dwarn:0 dfail:0 fail:0 skip:106 time:267s
fi-bxt-dsi total:289 pass:256 dwarn:0 dfail:0 fail:2 skip:31 time:515s
fi-bxt-j4205 total:289 pass:257 dwarn:0 dfail:0 fail:2 skip:30 time:515s
fi-byt-j1900 total:289 pass:251 dwarn:1 dfail:0 fail:2 skip:35 time:509s
fi-byt-n2820 total:289 pass:247 dwarn:1 dfail:0 fail:2 skip:39 time:497s
fi-cfl-s total:289 pass:250 dwarn:4 dfail:0 fail:2 skip:33 time:564s
fi-elk-e7500 total:289 pass:229 dwarn:0 dfail:0 fail:0 skip:60 time:421s
fi-gdg-551 total:289 pass:178 dwarn:1 dfail:0 fail:1 skip:109 time:249s
fi-glk-1 total:289 pass:258 dwarn:0 dfail:0 fail:2 skip:29 time:598s
fi-hsw-4770 total:289 pass:260 dwarn:0 dfail:0 fail:2 skip:27 time:446s
fi-hsw-4770r total:289 pass:260 dwarn:0 dfail:0 fail:2 skip:27 time:442s
fi-ilk-650 total:289 pass:228 dwarn:0 dfail:0 fail:0 skip:61 time:415s
fi-ivb-3520m total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:499s
fi-ivb-3770 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:456s
fi-kbl-7500u total:289 pass:261 dwarn:1 dfail:0 fail:2 skip:25 time:509s
fi-kbl-7560u total:289 pass:267 dwarn:0 dfail:0 fail:2 skip:20 time:586s
fi-kbl-7567u total:289 pass:266 dwarn:0 dfail:0 fail:2 skip:21 time:490s
fi-kbl-r total:289 pass:259 dwarn:0 dfail:0 fail:2 skip:28 time:591s
fi-pnv-d510 total:289 pass:222 dwarn:1 dfail:0 fail:0 skip:66 time:540s
fi-skl-6260u total:289 pass:266 dwarn:0 dfail:0 fail:2 skip:21 time:470s
fi-skl-6600u total:289 pass:259 dwarn:0 dfail:0 fail:2 skip:28 time:598s
fi-skl-6700hq total:289 pass:260 dwarn:0 dfail:0 fail:2 skip:27 time:659s
fi-skl-6700k total:289 pass:262 dwarn:0 dfail:0 fail:2 skip:25 time:543s
fi-skl-6770hq total:289 pass:266 dwarn:0 dfail:0 fail:2 skip:21 time:515s
fi-skl-gvtdvm total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:452s
fi-snb-2520m total:289 pass:250 dwarn:0 dfail:0 fail:0 skip:39 time:559s
fi-snb-2600 total:289 pass:249 dwarn:0 dfail:0 fail:0 skip:40 time:414s
302b1723bec5e7f43e423c975b9b87705fb9fdd6 drm-tip: 2017y-10m-30d-18h-28m-39s UTC integration manifest
87080ff7b53b drm/i915: Document the i915_workarounds file
67a5b78838c5 drm/i915: Move WaProgramL3SqcReg1Default to the workarounds file
396de69059e9 drm/i915/bdw: Move GT and Display workarounds from init_clock_gating
c1e60e825bf3 drm/i915/chv: Move GT and Display workarounds from init_clock_gating
145bb7fa659a drm/i915/skl: Move GT and Display workarounds from init_clock_gating
0e648f021c9f drm/i915/bxt: Move GT and Display workarounds from init_clock_gating
ccabeab078fe drm/i915/kbl: Move GT and Display workarounds from init_clock_gating
c6e15d041dce drm/i915/glk: Move GT and Display workarounds from init_clock_gating
c70b885f6126 drm/i915/cfl: Move GT and Display workarounds from init_clock_gating
394694e17f42 drm/i915/gen9: Move GT and Display workarounds from init_clock_gating
154721323716 drm/i915/cnl: Move GT and Display workarounds from init_clock_gating
9c801174be23 drm/i915: Move WA BB stuff to the workarounds file as well
8bec1242e0b4 drm/i915: Print all workaround types correctly in debugfs
6d58ffd7fd3a drm/i915: Create a new category of display WAs
4aa32a01de61 drm/i915: Save all Whitelist WAs and apply them at a later time
ab6c5bfff3ae drm/i915: Save all GT WAs and apply them at a later time
2c91eb7f330f drm/i915: Rename saved workarounds to make it explicit that they are context WAs
e4fe9a68f6ee drm/i915: Split out functions for different kinds of workarounds
ed3b7bd0fae4 drm/i915: Move a bunch of workaround-related code to its own file
0822716dee29 drm/i915: Remove Gen9 WAs with no effect
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_6269/
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [PATCH 05/20] drm/i915: Save all GT WAs and apply them at a later time
2017-10-30 20:17 ` [PATCH 05/20] drm/i915: Save all GT WAs and apply them at a later time Oscar Mateo
@ 2017-10-31 14:14 ` Joonas Lahtinen
2017-10-31 16:30 ` Oscar Mateo
2017-11-03 2:31 ` Chris Wilson
0 siblings, 2 replies; 26+ messages in thread
From: Joonas Lahtinen @ 2017-10-31 14:14 UTC (permalink / raw)
To: Oscar Mateo, intel-gfx
On Mon, 2017-10-30 at 13:17 -0700, Oscar Mateo wrote:
> By doing this, we can dump these workarounds in debugfs for validation (which,
> at the moment, we are only able to do for the contexts WAs).
>
> v2:
> - Wrong macro used for MMIO set bit masked
> - Improved naming
> - Rebased
>
> v3:
> - GT instead of MMIO (Chris, Mika)
> - Leave L3_PRIO_CREDITS_MASK for a separate patch
> - Rebased
>
> v4: Carry the init_early nomenclature over (Chris)
>
> Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
This and the following patch are still a no-go and won't be merged. The
required changes for the series to be accepted (to make it more
declarative) were clearly described previously. If there are further
questions, we should discuss those instead wasting time looking at
respins that do not address the input given.
Respins that ignore review are a waste of everyone's time. So, please
react to the review instead of trying to brute force a series in as you
see fit.
Regards, Joonas
--
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [PATCH 05/20] drm/i915: Save all GT WAs and apply them at a later time
2017-10-31 14:14 ` Joonas Lahtinen
@ 2017-10-31 16:30 ` Oscar Mateo
2017-11-01 10:46 ` David Weinehall
2017-11-03 2:31 ` Chris Wilson
1 sibling, 1 reply; 26+ messages in thread
From: Oscar Mateo @ 2017-10-31 16:30 UTC (permalink / raw)
To: Joonas Lahtinen, intel-gfx
On 10/31/2017 07:14 AM, Joonas Lahtinen wrote:
> On Mon, 2017-10-30 at 13:17 -0700, Oscar Mateo wrote:
>> By doing this, we can dump these workarounds in debugfs for validation (which,
>> at the moment, we are only able to do for the contexts WAs).
>>
>> v2:
>> - Wrong macro used for MMIO set bit masked
>> - Improved naming
>> - Rebased
>>
>> v3:
>> - GT instead of MMIO (Chris, Mika)
>> - Leave L3_PRIO_CREDITS_MASK for a separate patch
>> - Rebased
>>
>> v4: Carry the init_early nomenclature over (Chris)
>>
>> Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
>> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
>> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
>> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
> This and the following patch are still a no-go and won't be merged. The
> required changes for the series to be accepted (to make it more
> declarative) were clearly described previously. If there are further
> questions, we should discuss those instead wasting time looking at
> respins that do not address the input given.
>
> Respins that ignore review are a waste of everyone's time. So, please
> react to the review instead of trying to brute force a series in as you
> see fit.
>
> Regards, Joonas
My apologies, I wasn't trying to "brute force" anything. I haven't been
able to work on this for the last two weeks and I didn't want the issue
to go cold, so I sent a version to attend to Ville's and Chris' review
comments (this is useful anyway, because the same comments apply whether
you go with functions or with a static table). I am working on a version
that implements your suggestions and I will send it as soon as it's ready.
-- Oscar
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [PATCH 05/20] drm/i915: Save all GT WAs and apply them at a later time
2017-10-31 16:30 ` Oscar Mateo
@ 2017-11-01 10:46 ` David Weinehall
0 siblings, 0 replies; 26+ messages in thread
From: David Weinehall @ 2017-11-01 10:46 UTC (permalink / raw)
To: Oscar Mateo; +Cc: intel-gfx
On Tue, Oct 31, 2017 at 09:30:27AM -0700, Oscar Mateo wrote:
>
>
> On 10/31/2017 07:14 AM, Joonas Lahtinen wrote:
> > On Mon, 2017-10-30 at 13:17 -0700, Oscar Mateo wrote:
> > > By doing this, we can dump these workarounds in debugfs for validation (which,
> > > at the moment, we are only able to do for the contexts WAs).
> > >
> > > v2:
> > > - Wrong macro used for MMIO set bit masked
> > > - Improved naming
> > > - Rebased
> > >
> > > v3:
> > > - GT instead of MMIO (Chris, Mika)
> > > - Leave L3_PRIO_CREDITS_MASK for a separate patch
> > > - Rebased
> > >
> > > v4: Carry the init_early nomenclature over (Chris)
> > >
> > > Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
> > > Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
> > > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > > Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
> > This and the following patch are still a no-go and won't be merged. The
> > required changes for the series to be accepted (to make it more
> > declarative) were clearly described previously. If there are further
> > questions, we should discuss those instead wasting time looking at
> > respins that do not address the input given.
> >
> > Respins that ignore review are a waste of everyone's time. So, please
> > react to the review instead of trying to brute force a series in as you
> > see fit.
> >
> > Regards, Joonas
>
> My apologies, I wasn't trying to "brute force" anything. I haven't been able
> to work on this for the last two weeks and I didn't want the issue to go
> cold, so I sent a version to attend to Ville's and Chris' review comments
> (this is useful anyway, because the same comments apply whether you go with
> functions or with a static table). I am working on a version that implements
> your suggestions and I will send it as soon as it's ready.
Typically when submitting a changeset that isn't ready to be merged,
you'd tag it RFC or similar.
Kind regards, David Weinehall
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [PATCH 05/20] drm/i915: Save all GT WAs and apply them at a later time
2017-10-31 14:14 ` Joonas Lahtinen
2017-10-31 16:30 ` Oscar Mateo
@ 2017-11-03 2:31 ` Chris Wilson
1 sibling, 0 replies; 26+ messages in thread
From: Chris Wilson @ 2017-11-03 2:31 UTC (permalink / raw)
To: Joonas Lahtinen, Oscar Mateo, intel-gfx
Quoting Joonas Lahtinen (2017-10-31 14:14:52)
> On Mon, 2017-10-30 at 13:17 -0700, Oscar Mateo wrote:
> > By doing this, we can dump these workarounds in debugfs for validation (which,
> > at the moment, we are only able to do for the contexts WAs).
> >
> > v2:
> > - Wrong macro used for MMIO set bit masked
> > - Improved naming
> > - Rebased
> >
> > v3:
> > - GT instead of MMIO (Chris, Mika)
> > - Leave L3_PRIO_CREDITS_MASK for a separate patch
> > - Rebased
> >
> > v4: Carry the init_early nomenclature over (Chris)
> >
> > Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
> > Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
> > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
>
> This and the following patch are still a no-go and won't be merged. The
> required changes for the series to be accepted (to make it more
> declarative) were clearly described previously. If there are further
> questions, we should discuss those instead wasting time looking at
> respins that do not address the input given.
I would like draw everyone's attention to
https://bugs.freedesktop.org/show_bug.cgi?id=103549
As much as I don't like gem_workarounds for its incestrous relationship
with the kernel it purports to be testing, that bug is exactly the type
of regression it prevents. It could not find this regression because it
requires us to be very formal in our w/a handling, i.e. we had not
declared the w/a for it to check; such formality being sought after here.
Whatever the design outcome, a good test plan is essential.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
^ permalink raw reply [flat|nested] 26+ messages in thread
end of thread, other threads:[~2017-11-03 2:31 UTC | newest]
Thread overview: 26+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-10-30 20:17 [PATCH v4 00/20] Refactor HW workaround code Oscar Mateo
2017-10-30 20:17 ` [PATCH 01/20] drm/i915: Remove Gen9 WAs with no effect Oscar Mateo
2017-10-30 20:17 ` [PATCH 02/20] drm/i915: Move a bunch of workaround-related code to its own file Oscar Mateo
2017-10-30 20:17 ` [PATCH 03/20] drm/i915: Split out functions for different kinds of workarounds Oscar Mateo
2017-10-30 20:17 ` [PATCH 04/20] drm/i915: Rename saved workarounds to make it explicit that they are context WAs Oscar Mateo
2017-10-30 20:17 ` [PATCH 05/20] drm/i915: Save all GT WAs and apply them at a later time Oscar Mateo
2017-10-31 14:14 ` Joonas Lahtinen
2017-10-31 16:30 ` Oscar Mateo
2017-11-01 10:46 ` David Weinehall
2017-11-03 2:31 ` Chris Wilson
2017-10-30 20:17 ` [PATCH 06/20] drm/i915: Save all Whitelist " Oscar Mateo
2017-10-30 20:17 ` [PATCH 07/20] drm/i915: Create a new category of display WAs Oscar Mateo
2017-10-30 20:17 ` [PATCH 08/20] drm/i915: Print all workaround types correctly in debugfs Oscar Mateo
2017-10-30 20:17 ` [PATCH 09/20] drm/i915: Move WA BB stuff to the workarounds file as well Oscar Mateo
2017-10-30 20:17 ` [PATCH 10/20] drm/i915/cnl: Move GT and Display workarounds from init_clock_gating Oscar Mateo
2017-10-30 20:17 ` [PATCH 11/20] drm/i915/gen9: " Oscar Mateo
2017-10-30 20:17 ` [PATCH 12/20] drm/i915/cfl: " Oscar Mateo
2017-10-30 20:17 ` [PATCH 13/20] drm/i915/glk: " Oscar Mateo
2017-10-30 20:17 ` [PATCH 14/20] drm/i915/kbl: " Oscar Mateo
2017-10-30 20:17 ` [PATCH 15/20] drm/i915/bxt: " Oscar Mateo
2017-10-30 20:17 ` [PATCH 16/20] drm/i915/skl: " Oscar Mateo
2017-10-30 20:17 ` [PATCH 17/20] drm/i915/chv: " Oscar Mateo
2017-10-30 20:17 ` [PATCH 18/20] drm/i915/bdw: " Oscar Mateo
2017-10-30 20:17 ` [PATCH 19/20] drm/i915: Move WaProgramL3SqcReg1Default to the workarounds file Oscar Mateo
2017-10-30 20:17 ` [PATCH 20/20] drm/i915: Document the i915_workarounds file Oscar Mateo
2017-10-30 20:42 ` ✗ Fi.CI.BAT: warning for Refactor HW workaround code (rev4) Patchwork
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.