diff mbox

[01/22] drm/i915/icl: Introduce initial Icelake Workarounds

Message ID 1523053518-20827-1-git-send-email-oscar.mateo@intel.com (mailing list archive)
State New, archived
Headers show

Commit Message

oscar.mateo@intel.com April 6, 2018, 10:24 p.m. UTC
Inherit workarounds from previous platforms that are still valid for
Icelake.

v2: GEN7_ROW_CHICKEN2 is masked
v3:
  - Since it has been fixed already in upstream, removed the TODO
    comment about WA_SET_BIT for WaInPlaceDecompressionHang.
  - Squashed with this patch:
      drm/i915/icl: add icelake_init_clock_gating()
    from Paulo Zanoni <paulo.r.zanoni@intel.com>
  - Squashed with this patch:
      drm/i915/icl: WaForceEnableNonCoherent
    from Oscar Mateo <oscar.mateo@intel.com>
  - WaPushConstantDereferenceHoldDisable is now Wa_1604370585 and
    applies to B0 as well.
  - WaPipeControlBefore3DStateSamplePattern WABB was being applied
    to ICL incorrectly.
v4:
  - Wrap the commit message
  - s/dev_priv/p to please checkpatch

Cc: Tomasz Lis <tomasz.lis@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
---
 drivers/gpu/drm/i915/i915_drv.h        |  9 +++++++++
 drivers/gpu/drm/i915/i915_gem_gtt.c    |  4 ++--
 drivers/gpu/drm/i915/i915_reg.h        |  1 +
 drivers/gpu/drm/i915/intel_engine_cs.c | 25 +++++++++++++++++++++++++
 drivers/gpu/drm/i915/intel_lrc.c       |  2 ++
 drivers/gpu/drm/i915/intel_pm.c        | 19 ++++++++++++++++++-
 6 files changed, 57 insertions(+), 3 deletions(-)

Comments

Chris Wilson April 9, 2018, 7:53 p.m. UTC | #1
Quoting Oscar Mateo (2018-04-06 23:24:57)
> Inherit workarounds from previous platforms that are still valid for
> Icelake.

Speaking of the workarounds, where do we stand with at least landing the
split out of init_workarounds_ring()?

Rebuilding the invariant wa_regs[] on every reset is annoying, and how
many of the general mmio need to be reemitted everytime? I would dearly
like to get the flow of the gt/context workarounds improved.
-Chris
oscar.mateo@intel.com April 9, 2018, 9:33 p.m. UTC | #2
On 4/9/2018 12:53 PM, Chris Wilson wrote:
> Quoting Oscar Mateo (2018-04-06 23:24:57)
>> Inherit workarounds from previous platforms that are still valid for
>> Icelake.
> Speaking of the workarounds, where do we stand with at least landing the
> split out of init_workarounds_ring()?
>
> Rebuilding the invariant wa_regs[] on every reset is annoying, and how
> many of the general mmio need to be reemitted everytime? I would dearly
> like to get the flow of the gt/context workarounds improved.
> -Chris

I'm afraid we stand nowhere. The latest patches I sent were sidetracked 
by a conversation about where i915.rst should reside:

https://patchwork.freedesktop.org/patch/206937/
https://patchwork.freedesktop.org/patch/206557/

I can try once more, see if I am more lucky this time around?
diff mbox

Patch

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 5373b17..1a4801e 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -2458,6 +2458,15 @@  static inline unsigned int i915_sg_segment_size(void)
 #define IS_CNL_REVID(p, since, until) \
 	(IS_CANNONLAKE(p) && IS_REVID(p, since, until))
 
+#define ICL_REVID_A0		0x0
+#define ICL_REVID_A2		0x1
+#define ICL_REVID_B0		0x3
+#define ICL_REVID_B2		0x4
+#define ICL_REVID_C0		0x5
+
+#define IS_ICL_REVID(p, since, until) \
+	(IS_ICELAKE(p) && IS_REVID(p, since, until))
+
 /*
  * The genX designation typically refers to the render engine, so render
  * capability related checks should use IS_GEN, while display and other checks
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 21d72f6..221b873 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -2140,12 +2140,12 @@  static void gtt_write_workarounds(struct drm_i915_private *dev_priv)
 	 * called on driver load and after a GPU reset, so you can place
 	 * workarounds here even if they get overwritten by GPU reset.
 	 */
-	/* WaIncreaseDefaultTLBEntries:chv,bdw,skl,bxt,kbl,glk,cfl,cnl */
+	/* WaIncreaseDefaultTLBEntries:chv,bdw,skl,bxt,kbl,glk,cfl,cnl,icl */
 	if (IS_BROADWELL(dev_priv))
 		I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN8_L3_LRA_1_GPGPU_DEFAULT_VALUE_BDW);
 	else if (IS_CHERRYVIEW(dev_priv))
 		I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN8_L3_LRA_1_GPGPU_DEFAULT_VALUE_CHV);
-	else if (IS_GEN9_BC(dev_priv) || IS_GEN10(dev_priv))
+	else if (IS_GEN9_BC(dev_priv) || IS_GEN10(dev_priv) || IS_GEN11(dev_priv))
 		I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN9_L3_LRA_1_GPGPU_DEFAULT_VALUE_SKL);
 	else if (IS_GEN9_LP(dev_priv))
 		I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN9_L3_LRA_1_GPGPU_DEFAULT_VALUE_BXT);
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 176dca6..6cd2f2a 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -7184,6 +7184,7 @@  enum {
 /* GEN8 chicken */
 #define HDC_CHICKEN0				_MMIO(0x7300)
 #define CNL_HDC_CHICKEN0			_MMIO(0xE5F0)
+#define ICL_HDC_CHICKEN0			_MMIO(0xE5F4)
 #define  HDC_FORCE_CSR_NON_COHERENT_OVR_DISABLE	(1<<15)
 #define  HDC_FENCE_DEST_SLM_DISABLE		(1<<14)
 #define  HDC_DONOT_FETCH_MEM_WHEN_MASKED	(1<<11)
diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
index 12486d8..ba8c137 100644
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -1467,6 +1467,29 @@  static int cfl_init_workarounds(struct intel_engine_cs *engine)
 	return 0;
 }
 
+static int icl_init_workarounds(struct intel_engine_cs *engine)
+{
+	struct drm_i915_private *dev_priv = engine->i915;
+
+	/* Wa_1604370585:icl (pre-prod)
+	 * Formerly known as WaPushConstantDereferenceHoldDisable
+	 */
+	if (IS_ICL_REVID(dev_priv, ICL_REVID_A0, ICL_REVID_B0))
+		WA_SET_BIT_MASKED(GEN7_ROW_CHICKEN2,
+				  PUSH_CONSTANT_DEREF_DISABLE);
+
+	/* WaForceEnableNonCoherent:icl
+	 * This is not the same workaround as in early Gen9 platforms, where
+	 * lacking this could cause system hangs, but coherency performance
+	 * overhead is high and only a few compute workloads really need it
+	 * (the register is whitelisted in hardware now, so UMDs can opt in
+	 * for coherency if they have a good reason).
+	 */
+	WA_SET_BIT_MASKED(ICL_HDC_CHICKEN0, HDC_FORCE_NON_COHERENT);
+
+	return 0;
+}
+
 int init_workarounds_ring(struct intel_engine_cs *engine)
 {
 	struct drm_i915_private *dev_priv = engine->i915;
@@ -1494,6 +1517,8 @@  int init_workarounds_ring(struct intel_engine_cs *engine)
 		err = cfl_init_workarounds(engine);
 	else if (IS_CANNONLAKE(dev_priv))
 		err = cnl_init_workarounds(engine);
+	else if (IS_ICELAKE(dev_priv))
+		err = icl_init_workarounds(engine);
 	else
 		err = 0;
 	if (err)
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index f60b61b..6ec4abf 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -1573,6 +1573,8 @@  static int intel_init_workaround_bb(struct intel_engine_cs *engine)
 		return -EINVAL;
 
 	switch (INTEL_GEN(engine->i915)) {
+	case 11:
+		return 0;
 	case 10:
 		wa_bb_fn[0] = gen10_init_indirectctx_bb;
 		wa_bb_fn[1] = NULL;
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 19e82aa..4f4c7ae 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -8499,6 +8499,21 @@  static void cnp_init_clock_gating(struct drm_i915_private *dev_priv)
 		   CNP_PWM_CGE_GATING_DISABLE);
 }
 
+static void icl_init_clock_gating(struct drm_i915_private *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));
+
+	/* WaInPlaceDecompressionHang:icl */
+	I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, (I915_READ(GEN9_GAMT_ECO_REG_RW_IA) |
+					     GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS));
+
+	/* WaPipelineFlushCoherentLines:icl */
+	I915_WRITE(GEN8_L3SQCREG4, (I915_READ(GEN8_L3SQCREG4) |
+				    GEN8_LQSC_FLUSH_COHERENT_LINES));
+}
+
 static void cnl_init_clock_gating(struct drm_i915_private *dev_priv)
 {
 	u32 val;
@@ -9015,7 +9030,9 @@  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_ICELAKE(dev_priv))
+		dev_priv->display.init_clock_gating = icl_init_clock_gating;
+	else if (IS_CANNONLAKE(dev_priv))
 		dev_priv->display.init_clock_gating = cnl_init_clock_gating;
 	else if (IS_COFFEELAKE(dev_priv))
 		dev_priv->display.init_clock_gating = cfl_init_clock_gating;