From: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>
To: intel-xe@lists.freedesktop.org
Subject: [Intel-xe] [PATCH 2/2] drm/xe: Emit a render cache flush after each rcs/ccs batch
Date: Wed, 7 Jun 2023 19:47:29 +0200 [thread overview]
Message-ID: <20230607174729.54899-3-thomas.hellstrom@linux.intel.com> (raw)
In-Reply-To: <20230607174729.54899-1-thomas.hellstrom@linux.intel.com>
We need to flush render caches before fence signalling, where we might
release the memory for reuse. We can't rely on userspace doing this,
so flush render caches after the batch, but before user fence- and
dma_fence signalling.
Copy the cache flush from i915, but omit PIPE_CONTROL_FLUSH_L3, since it
should be implied by the other flushes. Also omit
PIPE_CONTROL_TLB_INVALIDATE since there should be no apparent need to
invalidate TLB after batch completion.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
---
drivers/gpu/drm/xe/regs/xe_gpu_commands.h | 3 ++
drivers/gpu/drm/xe/xe_ring_ops.c | 35 +++++++++++++++++++++++
drivers/gpu/drm/xe/xe_wa_oob.rules | 1 +
3 files changed, 39 insertions(+)
diff --git a/drivers/gpu/drm/xe/regs/xe_gpu_commands.h b/drivers/gpu/drm/xe/regs/xe_gpu_commands.h
index d2d41f717525..dd3408fd3d33 100644
--- a/drivers/gpu/drm/xe/regs/xe_gpu_commands.h
+++ b/drivers/gpu/drm/xe/regs/xe_gpu_commands.h
@@ -66,6 +66,9 @@
#define PVC_MS_MOCS_INDEX_MASK GENMASK(6, 1)
#define GFX_OP_PIPE_CONTROL(len) ((0x3<<29)|(0x3<<27)|(0x2<<24)|((len)-2))
+
+#define PIPE_CONTROL0_HDC_PIPELINE_FLUSH REG_BIT(9) /* gen12 */
+
#define PIPE_CONTROL_COMMAND_CACHE_INVALIDATE (1<<29)
#define PIPE_CONTROL_TILE_CACHE_FLUSH (1<<28)
#define PIPE_CONTROL_AMFS_FLUSH (1<<25)
diff --git a/drivers/gpu/drm/xe/xe_ring_ops.c b/drivers/gpu/drm/xe/xe_ring_ops.c
index c20fe41c0729..91511f72d971 100644
--- a/drivers/gpu/drm/xe/xe_ring_ops.c
+++ b/drivers/gpu/drm/xe/xe_ring_ops.c
@@ -5,6 +5,7 @@
#include "xe_ring_ops.h"
+#include "generated/xe_wa_oob.h"
#include "regs/xe_gpu_commands.h"
#include "regs/xe_gt_regs.h"
#include "regs/xe_lrc_layout.h"
@@ -16,6 +17,7 @@
#include "xe_sched_job.h"
#include "xe_vm_types.h"
#include "xe_vm.h"
+#include "xe_wa.h"
/*
* 3D-related flags that can't be set on _engines_ that lack access to the 3D
@@ -147,6 +149,37 @@ static int emit_store_imm_ppgtt_posted(u64 addr, u64 value,
return i;
}
+static int emit_render_cache_flush(struct xe_sched_job *job, u32 *dw, int i)
+{
+ struct xe_gt *gt = job->engine->gt;
+ bool lacks_render = !(gt->info.engine_mask & XE_HW_ENGINE_RCS_MASK);
+ u32 flags;
+
+ flags = (PIPE_CONTROL_CS_STALL |
+ PIPE_CONTROL_TILE_CACHE_FLUSH |
+ PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH |
+ PIPE_CONTROL_DEPTH_CACHE_FLUSH |
+ PIPE_CONTROL_DC_FLUSH_ENABLE |
+ PIPE_CONTROL_FLUSH_ENABLE);
+
+ if (XE_WA(gt, 1409600907))
+ flags |= PIPE_CONTROL_DEPTH_STALL;
+
+ if (lacks_render)
+ flags &= ~PIPE_CONTROL_3D_ARCH_FLAGS;
+ else if (job->engine->class == XE_ENGINE_CLASS_COMPUTE)
+ flags &= ~PIPE_CONTROL_3D_ENGINE_FLAGS;
+
+ dw[i++] = GFX_OP_PIPE_CONTROL(6) | PIPE_CONTROL0_HDC_PIPELINE_FLUSH;
+ dw[i++] = flags;
+ dw[i++] = 0;
+ dw[i++] = 0;
+ dw[i++] = 0;
+ dw[i++] = 0;
+
+ return i;
+}
+
static int emit_pipe_imm_ggtt(u32 addr, u32 value, bool stall_only, u32 *dw,
int i)
{
@@ -279,6 +312,8 @@ static void __emit_job_gen12_render_compute(struct xe_sched_job *job,
i = emit_bb_start(batch_addr, ppgtt_flag, dw, i);
+ i = emit_render_cache_flush(job, dw, i);
+
if (job->user_fence.used)
i = emit_store_imm_ppgtt_posted(job->user_fence.addr,
job->user_fence.value,
diff --git a/drivers/gpu/drm/xe/xe_wa_oob.rules b/drivers/gpu/drm/xe/xe_wa_oob.rules
index 1ecb10390b28..15c23813398a 100644
--- a/drivers/gpu/drm/xe/xe_wa_oob.rules
+++ b/drivers/gpu/drm/xe/xe_wa_oob.rules
@@ -14,3 +14,4 @@
SUBPLATFORM(DG2, G12)
18020744125 PLATFORM(PVC)
1509372804 PLATFORM(PVC), GRAPHICS_STEP(A0, C0)
+1409600907 GRAPHICS_VERSION_RANGE(1200, 1250)
--
2.39.2
next prev parent reply other threads:[~2023-06-07 17:47 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-07 17:47 [Intel-xe] [PATCH 0/2] Implement rcs/ccs missing invalidations and flushes Thomas Hellström
2023-06-07 17:47 ` [Intel-xe] [PATCH 1/2] drm/xe: Invalidate TLB also on bind if in scratch page mode Thomas Hellström
2023-06-07 18:01 ` Souza, Jose
2023-06-09 15:51 ` Matthew Brost
2023-06-09 15:54 ` Souza, Jose
2023-06-07 17:47 ` Thomas Hellström [this message]
2023-06-07 18:44 ` [Intel-xe] [PATCH 2/2] drm/xe: Emit a render cache flush after each rcs/ccs batch Souza, Jose
2023-06-07 17:49 ` [Intel-xe] ✓ CI.Patch_applied: success for Implement rcs/ccs missing invalidations and flushes Patchwork
2023-06-07 17:49 ` [Intel-xe] ✗ CI.checkpatch: warning " Patchwork
2023-06-07 17:50 ` [Intel-xe] ✗ CI.KUnit: failure " Patchwork
2023-06-07 18:03 ` [Intel-xe] [PATCH 0/2] " Souza, Jose
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=20230607174729.54899-3-thomas.hellstrom@linux.intel.com \
--to=thomas.hellstrom@linux.intel.com \
--cc=intel-xe@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 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.