From: Jesse Barnes <jbarnes@virtuousgeek.org>
To: Chris Wilson <chris@chris-wilson.co.uk>
Cc: intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 6/8] drm/i915: TLB invalidation with MI_FLUSH_DW requires a post-sync op
Date: Fri, 26 Oct 2012 09:42:42 -0700 [thread overview]
Message-ID: <20121026094242.0c74d329@jbarnes-desktop> (raw)
In-Reply-To: <eeac1e$4peaf2@AZSMGA002.ch.intel.com>
On Fri, 26 Oct 2012 12:13:39 +0100
Chris Wilson <chris@chris-wilson.co.uk> wrote:
> On Thu, 25 Oct 2012 12:15:46 -0700, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> > So store into the scratch space of the HWS to make sure the invalidate
> > occurs.
> >
> > v2: use GTT address space for store, clean up #defines (Chris)
> >
> > Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
> > ---
> > @@ -1460,10 +1467,17 @@ 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 | MI_FLUSH_DW_OP_STOREDW;
> > intel_ring_emit(ring, cmd);
> > - intel_ring_emit(ring, 0);
> > + intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_INDEX | MI_FLUSH_DW_USE_GTT);
>
> s/SCRATCH_INDEX/SCRATCH_ADDR/
> -Chris
>
--
Jesse Barnes, Intel Open Source Technology Center
commit b99c792eddf804150b3341a85c256df50d7ab5c2
Author: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Wed Sep 19 13:02:39 2012 -0700
drm/i915: TLB invalidation with MI_FLUSH_DW requires a post-sync op v3
So store into the scratch space of the HWS to make sure the invalidate
occurs.
v2: use GTT address space for store, clean up #defines (Chris)
v3: use correct #define in blt ring flush (Chris)
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index f29401b..ea97430 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -242,8 +242,12 @@
*/
#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_FLUSH_DW_USE_GTT (1<<2)
+#define MI_FLUSH_DW_USE_PPGTT (0<<2)
#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 785df4f..55abda5 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -1395,10 +1395,17 @@ 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_ADDR | MI_FLUSH_DW_USE_GTT);
intel_ring_emit(ring, 0);
intel_ring_emit(ring, MI_NOOP);
intel_ring_advance(ring);
@@ -1460,10 +1467,17 @@ 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 | MI_FLUSH_DW_OP_STOREDW;
intel_ring_emit(ring, cmd);
- intel_ring_emit(ring, 0);
+ intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT);
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..5af65b8 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -183,6 +183,8 @@ 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 0x30
+#define I915_GEM_HWS_SCRATCH_ADDR (I915_GEM_HWS_SCRATCH_INDEX << MI_STORE_DWORD_INDEX_SHIFT)
void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring);
next prev parent reply other threads:[~2012-10-26 16:42 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-10-25 19:15 [PATCH 1/8] drm/i915: implement WaDisableL3CacheAging on VLV Jesse Barnes
2012-10-25 19:15 ` [PATCH 2/8] drm/i915: implement WaDisableDopClockGatingisable on VLV and IVB Jesse Barnes
2012-10-25 19:15 ` [PATCH 3/8] drm/i915: implement WaForceL3Serialization " Jesse Barnes
2012-10-25 19:15 ` [PATCH 4/8] drm/i915: implement WaDisableVLVClockGating_VBIIssue on VLV Jesse Barnes
2012-11-01 14:48 ` Antti Koskipää
2012-11-01 14:50 ` Jesse Barnes
2012-11-01 14:52 ` Antti Koskipää
2012-10-25 19:15 ` [PATCH 5/8] drm/i915: implement WaDisablePSDDualDispatchEnable on IVB & VLV Jesse Barnes
2012-10-25 19:15 ` [PATCH 6/8] drm/i915: TLB invalidation with MI_FLUSH_DW requires a post-sync op Jesse Barnes
2012-10-26 11:13 ` Chris Wilson
2012-10-26 16:42 ` Jesse Barnes [this message]
2012-11-02 12:41 ` Chris Wilson
2012-10-25 19:15 ` [PATCH 7/8] drm/i915: PIPE_CONTROL TLB invalidate requires CS stall Jesse Barnes
2012-10-25 19:15 ` [PATCH 8/8] drm/i915: add clock gating regs to VLV offset check function Jesse Barnes
2012-11-02 15:34 ` Daniel Vetter
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20121026094242.0c74d329@jbarnes-desktop \
--to=jbarnes@virtuousgeek.org \
--cc=chris@chris-wilson.co.uk \
--cc=intel-gfx@lists.freedesktop.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox