[PATCH 1/8] drm/i915: implement WaDisableL3CacheAging on VLV

[PATCH 1/8] drm/i915: implement WaDisableL3CacheAging on VLV

[Jesse Barnes]

Needs to be set on every context restore as well, so set it as part of
the initial state so we can save/restore it.  Note this removes the IVB
workaround value from VLV and uses the default value, just adding in the
L3 cache aging disable bit, since the IVB value is wrong for VLV.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/i915_reg.h |    1 +
 drivers/gpu/drm/i915/intel_pm.c |    2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index c31ee5b..876ef96 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -3449,6 +3449,7 @@
 
 #define GEN7_L3CNTLREG1				0xB01C
 #define  GEN7_WA_FOR_GEN7_L3_CONTROL			0x3C4FFF8C
+#define  GEN7_L3AGDIS				(1<<19)
 
 #define GEN7_L3_CHICKEN_MODE_REGISTER		0xB030
 #define  GEN7_WA_L3_CHICKEN_MODE				0x20000000
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 2b3cddf..049c7e2 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -3613,7 +3613,7 @@ static void valleyview_init_clock_gating(struct drm_device *dev)
 		   GEN7_CSC1_RHWO_OPT_DISABLE_IN_RCC);
 
 	/* WaApplyL3ControlAndL3ChickenMode requires those two on Ivy Bridge */
-	I915_WRITE(GEN7_L3CNTLREG1, GEN7_WA_FOR_GEN7_L3_CONTROL);
+	I915_WRITE(GEN7_L3CNTLREG1, I915_READ(GEN7_L3CNTLREG1) | GEN7_L3AGDIS);
 	I915_WRITE(GEN7_L3_CHICKEN_MODE_REGISTER, GEN7_WA_L3_CHICKEN_MODE);
 
 	/* WaForceL3Serialization */
-- 
1.7.9.5

[PATCH 2/8] drm/i915: implement WaDisableDopClockGatingisable on VLV and IVB

[Jesse Barnes]

v2: use correct register
v3: remove extra hunks, pull in register definitions & offset check directly
v4: add GT1 vs GT2 distinction for IVB portion (Ben)

References: https://bugs.freedesktop.org/show_bug.cgi?id=50233
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/i915_drv.c |    7 +++++++
 drivers/gpu/drm/i915/i915_drv.h |    3 +++
 drivers/gpu/drm/i915/i915_reg.h |    4 ++++
 drivers/gpu/drm/i915/intel_pm.c |   13 ++++++++++++-
 4 files changed, 26 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 9e7e647..39282a4 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -1129,6 +1129,13 @@ static bool IS_DISPLAYREG(u32 reg)
 	if (reg == GEN6_GDRST)
 		return false;
 
+	switch (reg) {
+	case GEN7_ROW_CHICKEN2:
+		return false;
+	default:
+		break;
+	}
+
 	return true;
 }
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 4728d30..beef67f 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1128,6 +1128,9 @@ struct drm_i915_file_private {
 #define IS_IRONLAKE_D(dev)	((dev)->pci_device == 0x0042)
 #define IS_IRONLAKE_M(dev)	((dev)->pci_device == 0x0046)
 #define IS_IVYBRIDGE(dev)	(INTEL_INFO(dev)->is_ivybridge)
+#define IS_IVB_GT1(dev)		((dev)->pci_device == 0x0156 || \
+				 (dev)->pci_device == 0x0152 ||	\
+				 (dev)->pci_device == 0x015a)
 #define IS_VALLEYVIEW(dev)	(INTEL_INFO(dev)->is_valleyview)
 #define IS_HASWELL(dev)	(INTEL_INFO(dev)->is_haswell)
 #define IS_MOBILE(dev)		(INTEL_INFO(dev)->is_mobile)
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 876ef96..b07d309 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -4266,6 +4266,10 @@
 #define GEN7_L3LOG_BASE			0xB070
 #define GEN7_L3LOG_SIZE			0x80
 
+#define GEN7_ROW_CHICKEN2		0xe4f4
+#define GEN7_ROW_CHICKEN2_GT2		0xf4f4
+#define   DOP_CLOCK_GATING_DISABLE	(1<<0)
+
 #define G4X_AUD_VID_DID			0x62020
 #define INTEL_AUDIO_DEVCL		0x808629FB
 #define INTEL_AUDIO_DEVBLC		0x80862801
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 049c7e2..15328a3 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -3535,7 +3535,14 @@ static void ivybridge_init_clock_gating(struct drm_device *dev)
 	I915_WRITE(GEN7_L3CNTLREG1,
 			GEN7_WA_FOR_GEN7_L3_CONTROL);
 	I915_WRITE(GEN7_L3_CHICKEN_MODE_REGISTER,
-			GEN7_WA_L3_CHICKEN_MODE);
+		   GEN7_WA_L3_CHICKEN_MODE);
+	if (IS_IVB_GT1(dev))
+		I915_WRITE(GEN7_ROW_CHICKEN2,
+			   _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE));
+	else
+		I915_WRITE(GEN7_ROW_CHICKEN2_GT2,
+			   _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE));
+
 
 	/* WaForceL3Serialization */
 	I915_WRITE(GEN7_L3SQCREG4, I915_READ(GEN7_L3SQCREG4) &
@@ -3620,6 +3627,10 @@ static void valleyview_init_clock_gating(struct drm_device *dev)
 	I915_WRITE(GEN7_L3SQCREG4, I915_READ(GEN7_L3SQCREG4) &
 		   ~L3SQ_URB_READ_CAM_MATCH_DISABLE);
 
+	/* WaDisableDopClockGating */
+	I915_WRITE(GEN7_ROW_CHICKEN2,
+		   _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE));
+
 	/* This is required by WaCatErrorRejectionIssue */
 	I915_WRITE(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG,
 		   I915_READ(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG) |
-- 
1.7.9.5

[PATCH 3/8] drm/i915: implement WaForceL3Serialization on VLV and IVB

[Jesse Barnes]

References: https://bugs.freedesktop.org/show_bug.cgi?id=50250
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/intel_pm.c |    4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 15328a3..969687f 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -3631,6 +3631,10 @@ static void valleyview_init_clock_gating(struct drm_device *dev)
 	I915_WRITE(GEN7_ROW_CHICKEN2,
 		   _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE));
 
+	/* WaForceL3Serialization */
+	I915_WRITE(GEN7_L3SQCREG4, I915_READ(GEN7_L3SQCREG4) &
+		   ~L3SQ_URB_READ_CAM_MATCH_DISABLE);
+
 	/* This is required by WaCatErrorRejectionIssue */
 	I915_WRITE(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG,
 		   I915_READ(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG) |
-- 
1.7.9.5

[PATCH 4/8] drm/i915: implement WaDisableVLVClockGating_VBIIssue on VLV

[Jesse Barnes]

This allows us to get the right vblank interrupt frequency.

v2: pull in register definition

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/i915_reg.h |    2 ++
 drivers/gpu/drm/i915/intel_pm.c |    7 +++++++
 2 files changed, 9 insertions(+)

diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index b07d309..ecb28be 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -552,6 +552,8 @@
 #define IIR		0x020a4
 #define IMR		0x020a8
 #define ISR		0x020ac
+#define VLV_GUNIT_CLOCK_GATE	0x182060
+#define   GCFG_DIS		(1<<8)
 #define VLV_IIR_RW	0x182084
 #define VLV_IER		0x1820a0
 #define VLV_IIR		0x1820a4
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 969687f..4c86549 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -3691,6 +3691,13 @@ static void valleyview_init_clock_gating(struct drm_device *dev)
 		   PIPEA_HLINE_INT_EN | PIPEA_VBLANK_INT_EN |
 		   SPRITEB_FLIPDONE_INT_EN | SPRITEA_FLIPDONE_INT_EN |
 		   PLANEA_FLIPDONE_INT_EN);
+
+	/*
+	 * WaDisableVLVClockGating_VBIIssue
+	 * Disable clock gating on th GCFG unit to prevent a delay
+	 * in the reporting of vblank events.
+	 */
+	I915_WRITE(VLV_GUNIT_CLOCK_GATE, GCFG_DIS);
 }
 
 static void g4x_init_clock_gating(struct drm_device *dev)
-- 
1.7.9.5

[PATCH 5/8] drm/i915: implement WaDisablePSDDualDispatchEnable on IVB & VLV

[Jesse Barnes]

Workaround for dual port PS dispatch on GT1.

v2: pull in register definition & offset handling
v3: use IVB GT1 macro to get the right regs (Ben)
v4: add for VLV too (Ben)
v5: don't read the reg, it's masked so we'll only enable the one extra bit (Chris)
v6: use a _GT2 suffix for the second reg (Chris)

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/i915_drv.c |    1 +
 drivers/gpu/drm/i915/i915_reg.h |    5 +++++
 drivers/gpu/drm/i915/intel_pm.c |   11 +++++++++++
 3 files changed, 17 insertions(+)

diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 39282a4..6f03b26 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -1131,6 +1131,7 @@ static bool IS_DISPLAYREG(u32 reg)
 
 	switch (reg) {
 	case GEN7_ROW_CHICKEN2:
+	case GEN7_HALF_SLICE_CHICKEN1:
 		return false;
 	default:
 		break;
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index ecb28be..34067b5 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -4268,6 +4268,11 @@
 #define GEN7_L3LOG_BASE			0xB070
 #define GEN7_L3LOG_SIZE			0x80
 
+#define GEN7_HALF_SLICE_CHICKEN1	0xe100 /* IVB GT1 + VLV */
+#define GEN7_HALF_SLICE_CHICKEN1_GT2	0xf100
+#define   GEN7_MAX_PS_THREAD_DEP		(8<<12)
+#define   GEN7_PSD_SINGLE_PORT_DISPATCH_ENABLE	(1<<3)
+
 #define GEN7_ROW_CHICKEN2		0xe4f4
 #define GEN7_ROW_CHICKEN2_GT2		0xf4f4
 #define   DOP_CLOCK_GATING_DISABLE	(1<<0)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 4c86549..0c1b270 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -3527,6 +3527,14 @@ static void ivybridge_init_clock_gating(struct drm_device *dev)
 		   CHICKEN3_DGMG_REQ_OUT_FIX_DISABLE |
 		   CHICKEN3_DGMG_DONE_FIX_DISABLE);
 
+	/* WaDisablePSDDualDispatchEnable */
+	if (IS_IVB_GT1(dev))
+		I915_WRITE(GEN7_HALF_SLICE_CHICKEN1,
+			   _MASKED_BIT_ENABLE(GEN7_PSD_SINGLE_PORT_DISPATCH_ENABLE));
+	else
+		I915_WRITE(GEN7_HALF_SLICE_CHICKEN1_GT2,
+			   _MASKED_BIT_ENABLE(GEN7_PSD_SINGLE_PORT_DISPATCH_ENABLE));
+
 	/* Apply the WaDisableRHWOOptimizationForRenderHang workaround. */
 	I915_WRITE(GEN7_COMMON_SLICE_CHICKEN1,
 		   GEN7_CSC1_RHWO_OPT_DISABLE_IN_RCC);
@@ -3615,6 +3623,9 @@ static void valleyview_init_clock_gating(struct drm_device *dev)
 		   CHICKEN3_DGMG_REQ_OUT_FIX_DISABLE |
 		   CHICKEN3_DGMG_DONE_FIX_DISABLE);
 
+	I915_WRITE(GEN7_HALF_SLICE_CHICKEN1,
+		   _MASKED_BIT_ENABLE(GEN7_PSD_SINGLE_PORT_DISPATCH_ENABLE));
+
 	/* Apply the WaDisableRHWOOptimizationForRenderHang workaround. */
 	I915_WRITE(GEN7_COMMON_SLICE_CHICKEN1,
 		   GEN7_CSC1_RHWO_OPT_DISABLE_IN_RCC);
-- 
1.7.9.5

[PATCH 6/8] drm/i915: TLB invalidation with MI_FLUSH_SW requires a post-sync op

[Jesse Barnes]

So store into the scratch space of the HWS to make sure the invalidate
occurs.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/i915_reg.h         |    6 ++++--
 drivers/gpu/drm/i915/intel_ringbuffer.c |   22 ++++++++++++++++++----
 drivers/gpu/drm/i915/intel_ringbuffer.h |    1 +
 3 files changed, 23 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 34067b5..c6f63a4 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -241,8 +241,10 @@
  */
 #define MI_LOAD_REGISTER_IMM(x)	MI_INSTR(0x22, 2*x-1)
 #define MI_FLUSH_DW		MI_INSTR(0x26, 1) /* for GEN6 */
-#define   MI_INVALIDATE_TLB	(1<<18)
-#define   MI_INVALIDATE_BSD	(1<<7)
+#define   MI_FLUSH_DW_STORE_INDEX	(1<<21)
+#define   MI_INVALIDATE_TLB		(1<<18)
+#define   MI_FLUSH_DW_OP_STOREDW	(1<<14)
+#define   MI_INVALIDATE_BSD		(1<<7)
 #define MI_BATCH_BUFFER		MI_INSTR(0x30, 1)
 #define   MI_BATCH_NON_SECURE		(1)
 /* for snb/ivb/vlv this also means "batch in ppgtt" when ppgtt is enabled. */
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 6c6f95a..e7daa90 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -1396,10 +1396,17 @@ static int gen6_ring_flush(struct intel_ring_buffer *ring,
 		return ret;
 
 	cmd = MI_FLUSH_DW;
+	/*
+	 * Bspec vol 1c.5 - video engine command streamer:
+	 * "If ENABLED, all TLBs will be invalidated once the flush
+	 * operation is complete. This bit is only valid when the
+	 * Post-Sync Operation field is a value of 1h or 3h."
+	 */
 	if (invalidate & I915_GEM_GPU_DOMAINS)
-		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD;
+		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD |
+			MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW;
 	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring, 0);
+	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_INDEX << 3);
 	intel_ring_emit(ring, 0);
 	intel_ring_emit(ring, MI_NOOP);
 	intel_ring_advance(ring);
@@ -1461,10 +1468,17 @@ static int blt_ring_flush(struct intel_ring_buffer *ring,
 		return ret;
 
 	cmd = MI_FLUSH_DW;
+	/*
+	 * Bspec vol 1c.3 - blitter engine command streamer:
+	 * "If ENABLED, all TLBs will be invalidated once the flush
+	 * operation is complete. This bit is only valid when the
+	 * Post-Sync Operation field is a value of 1h or 3h."
+	 */
 	if (invalidate & I915_GEM_DOMAIN_RENDER)
-		cmd |= MI_INVALIDATE_TLB;
+		cmd |= MI_INVALIDATE_TLB | MI_FLUSH_DW_STORE_INDEX |
+			MI_FLUSH_DW_OP_STOREDW;
 	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring, 0);
+	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_INDEX << 3);
 	intel_ring_emit(ring, 0);
 	intel_ring_emit(ring, MI_NOOP);
 	intel_ring_advance(ring);
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 3745d1d..d089520 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -183,6 +183,7 @@ intel_read_status_page(struct intel_ring_buffer *ring,
  * The area from dword 0x20 to 0x3ff is available for driver usage.
  */
 #define I915_GEM_HWS_INDEX		0x20
+#define I915_GEM_HWS_SCRATCH_INDEX	0x28
 
 void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring);
 
-- 
1.7.9.5

Re: [PATCH 6/8] drm/i915: TLB invalidation with MI_FLUSH_SW requires a post-sync op

[Chris Wilson]

On Thu, 18 Oct 2012 13:07:17 -0500, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> So store into the scratch space of the HWS to make sure the invalidate
> occurs.

Whoops, instant hang. Probably doesn't agree with being called FLUSH_SW
and not FLUSH_DW! ;-)

> +	/*
> +	 * Bspec vol 1c.5 - video engine command streamer:
> +	 * "If ENABLED, all TLBs will be invalidated once the flush
> +	 * operation is complete. This bit is only valid when the
> +	 * Post-Sync Operation field is a value of 1h or 3h."
> +	 */
>  	if (invalidate & I915_GEM_GPU_DOMAINS)
> -		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD;
> +		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD |
> +			MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW;
>  	intel_ring_emit(ring, cmd);
> -	intel_ring_emit(ring, 0);
> +	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_INDEX << 3);
And here is where the error lies. Perhaps this would be clearer if you
do:

#define MI_FLUSH_DW_USE_PPGTT 0
#define MI_FLUSH_DW_USE_GTT (1<<2)

#define I915_GEM_HWS_SCRATCH_INDEX 0x30
#define I915_GEM_HWS_SCRATCH_ADDR (I915_GEM_HWS_SCRATCH_INDEX << MI_STORE_DWORD_INDEX_SHIFT)

Then:
	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT);

Hangs begone!
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre

Re: [PATCH 6/8] drm/i915: TLB invalidation with MI_FLUSH_SW requires a post-sync op

[Jesse Barnes]

On Tue, 23 Oct 2012 12:22:16 +0100
Chris Wilson <chris@chris-wilson.co.uk> wrote:

> On Thu, 18 Oct 2012 13:07:17 -0500, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> > So store into the scratch space of the HWS to make sure the invalidate
> > occurs.
> 
> Whoops, instant hang. Probably doesn't agree with being called FLUSH_SW
> and not FLUSH_DW! ;-)
> 
> > +	/*
> > +	 * Bspec vol 1c.5 - video engine command streamer:
> > +	 * "If ENABLED, all TLBs will be invalidated once the flush
> > +	 * operation is complete. This bit is only valid when the
> > +	 * Post-Sync Operation field is a value of 1h or 3h."
> > +	 */
> >  	if (invalidate & I915_GEM_GPU_DOMAINS)
> > -		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD;
> > +		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD |
> > +			MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW;
> >  	intel_ring_emit(ring, cmd);
> > -	intel_ring_emit(ring, 0);
> > +	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_INDEX << 3);
> And here is where the error lies. Perhaps this would be clearer if you
> do:
> 
> #define MI_FLUSH_DW_USE_PPGTT 0
> #define MI_FLUSH_DW_USE_GTT (1<<2)
> 
> #define I915_GEM_HWS_SCRATCH_INDEX 0x30
> #define I915_GEM_HWS_SCRATCH_ADDR (I915_GEM_HWS_SCRATCH_INDEX << MI_STORE_DWORD_INDEX_SHIFT)
> 
> Then:
> 	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT);
> 
> Hangs begone!

Ah cool, was hoping it was something simple.  Damn PPGTT vs GTT always
gets us.

I'll respin with the change.

-- 
Jesse Barnes, Intel Open Source Technology Center

[PATCH 7/8] drm/i915: PIPE_CONTROL TLB invalidate requires CS stall

[Jesse Barnes]

"If ENABLED, PIPE_CONTROL command will flush the in flight data  written
out by render engine to Global Observation point on flush done. Also
Requires stall bit ([20] of DW1) set."

So set the stall bit to ensure proper invalidation.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/intel_ringbuffer.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index e7daa90..1e09c62 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -246,7 +246,7 @@ gen6_render_ring_flush(struct intel_ring_buffer *ring,
 		/*
 		 * TLB invalidate requires a post-sync write.
 		 */
-		flags |= PIPE_CONTROL_QW_WRITE;
+		flags |= PIPE_CONTROL_QW_WRITE | PIPE_CONTROL_CS_STALL;
 	}
 
 	ret = intel_ring_begin(ring, 4);
-- 
1.7.9.5

Re: [PATCH 7/8] drm/i915: PIPE_CONTROL TLB invalidate requires CS stall

[Chris Wilson]

On Thu, 18 Oct 2012 13:07:18 -0500, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> "If ENABLED, PIPE_CONTROL command will flush the in flight data  written
> out by render engine to Global Observation point on flush done. Also
> Requires stall bit ([20] of DW1) set."

That quotation doesn't make sense in the context of TLB invalidation,
and the programming guide here very carefully avoids the mention of
requiring any stall bit set for the post-sync op of TLB invalidation.

Maybe quote chapter and verse as well?
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre

Re: [PATCH 7/8] drm/i915: PIPE_CONTROL TLB invalidate requires CS stall

[Jesse Barnes]

On Tue, 23 Oct 2012 12:42:07 +0100
Chris Wilson <chris@chris-wilson.co.uk> wrote:

> On Thu, 18 Oct 2012 13:07:18 -0500, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> > "If ENABLED, PIPE_CONTROL command will flush the in flight data  written
> > out by render engine to Global Observation point on flush done. Also
> > Requires stall bit ([20] of DW1) set."
> 
> That quotation doesn't make sense in the context of TLB invalidation,
> and the programming guide here very carefully avoids the mention of
> requiring any stall bit set for the post-sync op of TLB invalidation.
> 
> Maybe quote chapter and verse as well?

I thought the "Also Requires stall bit ([20] of DW1) set." was pretty
clear?

-- 
Jesse Barnes, Intel Open Source Technology Center

[PATCH 8/8] drm/i915: add clock gating regs to VLV offset check function

[Jesse Barnes]

So we can write them properly.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/i915_drv.c |    9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 6f03b26..39c53ad 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -1130,8 +1130,17 @@ static bool IS_DISPLAYREG(u32 reg)
 		return false;
 
 	switch (reg) {
+	case _3D_CHICKEN3:
+	case IVB_CHICKEN3:
+	case GEN7_COMMON_SLICE_CHICKEN1:
+	case GEN7_L3CNTLREG1:
+	case GEN7_L3_CHICKEN_MODE_REGISTER:
 	case GEN7_ROW_CHICKEN2:
+	case GEN7_L3SQCREG4:
+	case GEN7_SQ_CHICKEN_MBCUNIT_CONFIG:
 	case GEN7_HALF_SLICE_CHICKEN1:
+	case GEN6_MBCTL:
+	case GEN6_UCGCTL2:
 		return false;
 	default:
 		break;
-- 
1.7.9.5

[PATCH 1/8] drm/i915: implement WaDisableL3CacheAging on VLV

[Jesse Barnes]

Needs to be set on every context restore as well, so set it as part of
the initial state so we can save/restore it.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/i915_reg.h |    1 +
 drivers/gpu/drm/i915/intel_pm.c |    2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index c31ee5b..876ef96 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -3449,6 +3449,7 @@
 
 #define GEN7_L3CNTLREG1				0xB01C
 #define  GEN7_WA_FOR_GEN7_L3_CONTROL			0x3C4FFF8C
+#define  GEN7_L3AGDIS				(1<<19)
 
 #define GEN7_L3_CHICKEN_MODE_REGISTER		0xB030
 #define  GEN7_WA_L3_CHICKEN_MODE				0x20000000
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 2b3cddf..049c7e2 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -3613,7 +3613,7 @@ static void valleyview_init_clock_gating(struct drm_device *dev)
 		   GEN7_CSC1_RHWO_OPT_DISABLE_IN_RCC);
 
 	/* WaApplyL3ControlAndL3ChickenMode requires those two on Ivy Bridge */
-	I915_WRITE(GEN7_L3CNTLREG1, GEN7_WA_FOR_GEN7_L3_CONTROL);
+	I915_WRITE(GEN7_L3CNTLREG1, I915_READ(GEN7_L3CNTLREG1) | GEN7_L3AGDIS);
 	I915_WRITE(GEN7_L3_CHICKEN_MODE_REGISTER, GEN7_WA_L3_CHICKEN_MODE);
 
 	/* WaForceL3Serialization */
-- 
1.7.9.5

[PATCH 6/8] drm/i915: TLB invalidation with MI_FLUSH_SW requires a post-sync op

[Jesse Barnes]

So store into the scratch space of the HWS to make sure the invalidate
occurs.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
---
 drivers/gpu/drm/i915/i915_reg.h         |    6 ++++--
 drivers/gpu/drm/i915/intel_ringbuffer.c |   22 ++++++++++++++++++----
 drivers/gpu/drm/i915/intel_ringbuffer.h |    1 +
 3 files changed, 23 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 12816bb..9d7e4f5 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -241,8 +241,10 @@
  */
 #define MI_LOAD_REGISTER_IMM(x)	MI_INSTR(0x22, 2*x-1)
 #define MI_FLUSH_DW		MI_INSTR(0x26, 1) /* for GEN6 */
-#define   MI_INVALIDATE_TLB	(1<<18)
-#define   MI_INVALIDATE_BSD	(1<<7)
+#define   MI_FLUSH_DW_STORE_INDEX	(1<<21)
+#define   MI_INVALIDATE_TLB		(1<<18)
+#define   MI_FLUSH_DW_OP_STOREDW	(1<<14)
+#define   MI_INVALIDATE_BSD		(1<<7)
 #define MI_BATCH_BUFFER		MI_INSTR(0x30, 1)
 #define   MI_BATCH_NON_SECURE		(1)
 /* for snb/ivb/vlv this also means "batch in ppgtt" when ppgtt is enabled. */
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 6c6f95a..e7daa90 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -1396,10 +1396,17 @@ static int gen6_ring_flush(struct intel_ring_buffer *ring,
 		return ret;
 
 	cmd = MI_FLUSH_DW;
+	/*
+	 * Bspec vol 1c.5 - video engine command streamer:
+	 * "If ENABLED, all TLBs will be invalidated once the flush
+	 * operation is complete. This bit is only valid when the
+	 * Post-Sync Operation field is a value of 1h or 3h."
+	 */
 	if (invalidate & I915_GEM_GPU_DOMAINS)
-		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD;
+		cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD |
+			MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW;
 	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring, 0);
+	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_INDEX << 3);
 	intel_ring_emit(ring, 0);
 	intel_ring_emit(ring, MI_NOOP);
 	intel_ring_advance(ring);
@@ -1461,10 +1468,17 @@ static int blt_ring_flush(struct intel_ring_buffer *ring,
 		return ret;
 
 	cmd = MI_FLUSH_DW;
+	/*
+	 * Bspec vol 1c.3 - blitter engine command streamer:
+	 * "If ENABLED, all TLBs will be invalidated once the flush
+	 * operation is complete. This bit is only valid when the
+	 * Post-Sync Operation field is a value of 1h or 3h."
+	 */
 	if (invalidate & I915_GEM_DOMAIN_RENDER)
-		cmd |= MI_INVALIDATE_TLB;
+		cmd |= MI_INVALIDATE_TLB | MI_FLUSH_DW_STORE_INDEX |
+			MI_FLUSH_DW_OP_STOREDW;
 	intel_ring_emit(ring, cmd);
-	intel_ring_emit(ring, 0);
+	intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_INDEX << 3);
 	intel_ring_emit(ring, 0);
 	intel_ring_emit(ring, MI_NOOP);
 	intel_ring_advance(ring);
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 3745d1d..d089520 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -183,6 +183,7 @@ intel_read_status_page(struct intel_ring_buffer *ring,
  * The area from dword 0x20 to 0x3ff is available for driver usage.
  */
 #define I915_GEM_HWS_INDEX		0x20
+#define I915_GEM_HWS_SCRATCH_INDEX	0x28
 
 void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring);
 
-- 
1.7.9.5