* FAILED: patch "[PATCH] drm/xe: Make WA BB part of LRC BO" failed to apply to 6.15-stable tree
@ 2025-07-12 14:30 gregkh
2025-07-23 15:10 ` [PATCH 6.15.y] drm/xe: Make WA BB part of LRC BO Sasha Levin
0 siblings, 1 reply; 2+ messages in thread
From: gregkh @ 2025-07-12 14:30 UTC (permalink / raw)
To: matthew.brost, lucas.demarchi, shuicheng.lin; +Cc: stable
The patch below does not apply to the 6.15-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable@vger.kernel.org>.
To reproduce the conflict and resubmit, you may use the following commands:
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.15.y
git checkout FETCH_HEAD
git cherry-pick -x afcad92411772a1f361339f22c49f855c6cc7d0f
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable@vger.kernel.org>' --in-reply-to '2025071259-unrelated-extrovert-b5a9@gregkh' --subject-prefix 'PATCH 6.15.y' HEAD^..
Possible dependencies:
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From afcad92411772a1f361339f22c49f855c6cc7d0f Mon Sep 17 00:00:00 2001
From: Matthew Brost <matthew.brost@intel.com>
Date: Wed, 11 Jun 2025 20:19:25 -0700
Subject: [PATCH] drm/xe: Make WA BB part of LRC BO
No idea why, but without this GuC context switches randomly fail when
running IGTs in a loop. Need to follow up why this fixes the
aforementioned issue but can live with a stable driver for now.
Fixes: 617d824c5323 ("drm/xe: Add WA BB to capture active context utilization")
Cc: stable@vger.kernel.org
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Tested-by: Shuicheng Lin <shuicheng.lin@intel.com>
Link: https://lore.kernel.org/r/20250612031925.4009701-1-matthew.brost@intel.com
(cherry picked from commit 3a1edef8f4b58b0ba826bc68bf4bce4bdf59ecf3)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
diff --git a/drivers/gpu/drm/xe/xe_lrc.c b/drivers/gpu/drm/xe/xe_lrc.c
index bf7c3981897d..6e7b70532d11 100644
--- a/drivers/gpu/drm/xe/xe_lrc.c
+++ b/drivers/gpu/drm/xe/xe_lrc.c
@@ -40,6 +40,7 @@
#define LRC_PPHWSP_SIZE SZ_4K
#define LRC_INDIRECT_RING_STATE_SIZE SZ_4K
+#define LRC_WA_BB_SIZE SZ_4K
static struct xe_device *
lrc_to_xe(struct xe_lrc *lrc)
@@ -910,7 +911,11 @@ static void xe_lrc_finish(struct xe_lrc *lrc)
{
xe_hw_fence_ctx_finish(&lrc->fence_ctx);
xe_bo_unpin_map_no_vm(lrc->bo);
- xe_bo_unpin_map_no_vm(lrc->bb_per_ctx_bo);
+}
+
+static size_t wa_bb_offset(struct xe_lrc *lrc)
+{
+ return lrc->bo->size - LRC_WA_BB_SIZE;
}
/*
@@ -943,15 +948,16 @@ static void xe_lrc_finish(struct xe_lrc *lrc)
#define CONTEXT_ACTIVE 1ULL
static int xe_lrc_setup_utilization(struct xe_lrc *lrc)
{
+ const size_t max_size = LRC_WA_BB_SIZE;
u32 *cmd, *buf = NULL;
- if (lrc->bb_per_ctx_bo->vmap.is_iomem) {
- buf = kmalloc(lrc->bb_per_ctx_bo->size, GFP_KERNEL);
+ if (lrc->bo->vmap.is_iomem) {
+ buf = kmalloc(max_size, GFP_KERNEL);
if (!buf)
return -ENOMEM;
cmd = buf;
} else {
- cmd = lrc->bb_per_ctx_bo->vmap.vaddr;
+ cmd = lrc->bo->vmap.vaddr + wa_bb_offset(lrc);
}
*cmd++ = MI_STORE_REGISTER_MEM | MI_SRM_USE_GGTT | MI_SRM_ADD_CS_OFFSET;
@@ -974,13 +980,14 @@ static int xe_lrc_setup_utilization(struct xe_lrc *lrc)
*cmd++ = MI_BATCH_BUFFER_END;
if (buf) {
- xe_map_memcpy_to(gt_to_xe(lrc->gt), &lrc->bb_per_ctx_bo->vmap, 0,
- buf, (cmd - buf) * sizeof(*cmd));
+ xe_map_memcpy_to(gt_to_xe(lrc->gt), &lrc->bo->vmap,
+ wa_bb_offset(lrc), buf,
+ (cmd - buf) * sizeof(*cmd));
kfree(buf);
}
- xe_lrc_write_ctx_reg(lrc, CTX_BB_PER_CTX_PTR,
- xe_bo_ggtt_addr(lrc->bb_per_ctx_bo) | 1);
+ xe_lrc_write_ctx_reg(lrc, CTX_BB_PER_CTX_PTR, xe_bo_ggtt_addr(lrc->bo) +
+ wa_bb_offset(lrc) + 1);
return 0;
}
@@ -1018,20 +1025,13 @@ static int xe_lrc_init(struct xe_lrc *lrc, struct xe_hw_engine *hwe,
* FIXME: Perma-pinning LRC as we don't yet support moving GGTT address
* via VM bind calls.
*/
- lrc->bo = xe_bo_create_pin_map(xe, tile, NULL, lrc_size,
+ lrc->bo = xe_bo_create_pin_map(xe, tile, NULL,
+ lrc_size + LRC_WA_BB_SIZE,
ttm_bo_type_kernel,
bo_flags);
if (IS_ERR(lrc->bo))
return PTR_ERR(lrc->bo);
- lrc->bb_per_ctx_bo = xe_bo_create_pin_map(xe, tile, NULL, SZ_4K,
- ttm_bo_type_kernel,
- bo_flags);
- if (IS_ERR(lrc->bb_per_ctx_bo)) {
- err = PTR_ERR(lrc->bb_per_ctx_bo);
- goto err_lrc_finish;
- }
-
lrc->size = lrc_size;
lrc->ring.size = ring_size;
lrc->ring.tail = 0;
@@ -1819,7 +1819,8 @@ struct xe_lrc_snapshot *xe_lrc_snapshot_capture(struct xe_lrc *lrc)
snapshot->seqno = xe_lrc_seqno(lrc);
snapshot->lrc_bo = xe_bo_get(lrc->bo);
snapshot->lrc_offset = xe_lrc_pphwsp_offset(lrc);
- snapshot->lrc_size = lrc->bo->size - snapshot->lrc_offset;
+ snapshot->lrc_size = lrc->bo->size - snapshot->lrc_offset -
+ LRC_WA_BB_SIZE;
snapshot->lrc_snapshot = NULL;
snapshot->ctx_timestamp = lower_32_bits(xe_lrc_ctx_timestamp(lrc));
snapshot->ctx_job_timestamp = xe_lrc_ctx_job_timestamp(lrc);
diff --git a/drivers/gpu/drm/xe/xe_lrc_types.h b/drivers/gpu/drm/xe/xe_lrc_types.h
index ae24cf6f8dd9..883e550a9423 100644
--- a/drivers/gpu/drm/xe/xe_lrc_types.h
+++ b/drivers/gpu/drm/xe/xe_lrc_types.h
@@ -53,9 +53,6 @@ struct xe_lrc {
/** @ctx_timestamp: readout value of CTX_TIMESTAMP on last update */
u64 ctx_timestamp;
-
- /** @bb_per_ctx_bo: buffer object for per context batch wa buffer */
- struct xe_bo *bb_per_ctx_bo;
};
struct xe_lrc_snapshot;
^ permalink raw reply related [flat|nested] 2+ messages in thread
* [PATCH 6.15.y] drm/xe: Make WA BB part of LRC BO
2025-07-12 14:30 FAILED: patch "[PATCH] drm/xe: Make WA BB part of LRC BO" failed to apply to 6.15-stable tree gregkh
@ 2025-07-23 15:10 ` Sasha Levin
0 siblings, 0 replies; 2+ messages in thread
From: Sasha Levin @ 2025-07-23 15:10 UTC (permalink / raw)
To: stable; +Cc: Matthew Brost, Lucas De Marchi, Shuicheng Lin, Sasha Levin
From: Matthew Brost <matthew.brost@intel.com>
[ Upstream commit afcad92411772a1f361339f22c49f855c6cc7d0f ]
No idea why, but without this GuC context switches randomly fail when
running IGTs in a loop. Need to follow up why this fixes the
aforementioned issue but can live with a stable driver for now.
Fixes: 617d824c5323 ("drm/xe: Add WA BB to capture active context utilization")
Cc: stable@vger.kernel.org
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Tested-by: Shuicheng Lin <shuicheng.lin@intel.com>
Link: https://lore.kernel.org/r/20250612031925.4009701-1-matthew.brost@intel.com
(cherry picked from commit 3a1edef8f4b58b0ba826bc68bf4bce4bdf59ecf3)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
[ adapted xe_bo_create_pin_map() call ]
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
drivers/gpu/drm/xe/xe_lrc.c | 37 ++++++++++++++++---------------
drivers/gpu/drm/xe/xe_lrc_types.h | 3 ---
2 files changed, 19 insertions(+), 21 deletions(-)
diff --git a/drivers/gpu/drm/xe/xe_lrc.c b/drivers/gpu/drm/xe/xe_lrc.c
index 16e20b5ad325f..c3c9df8ba7bbf 100644
--- a/drivers/gpu/drm/xe/xe_lrc.c
+++ b/drivers/gpu/drm/xe/xe_lrc.c
@@ -39,6 +39,7 @@
#define LRC_ENGINE_INSTANCE GENMASK_ULL(53, 48)
#define LRC_INDIRECT_RING_STATE_SIZE SZ_4K
+#define LRC_WA_BB_SIZE SZ_4K
static struct xe_device *
lrc_to_xe(struct xe_lrc *lrc)
@@ -910,7 +911,11 @@ static void xe_lrc_finish(struct xe_lrc *lrc)
xe_bo_unpin(lrc->bo);
xe_bo_unlock(lrc->bo);
xe_bo_put(lrc->bo);
- xe_bo_unpin_map_no_vm(lrc->bb_per_ctx_bo);
+}
+
+static size_t wa_bb_offset(struct xe_lrc *lrc)
+{
+ return lrc->bo->size - LRC_WA_BB_SIZE;
}
/*
@@ -943,15 +948,16 @@ static void xe_lrc_finish(struct xe_lrc *lrc)
#define CONTEXT_ACTIVE 1ULL
static int xe_lrc_setup_utilization(struct xe_lrc *lrc)
{
+ const size_t max_size = LRC_WA_BB_SIZE;
u32 *cmd, *buf = NULL;
- if (lrc->bb_per_ctx_bo->vmap.is_iomem) {
- buf = kmalloc(lrc->bb_per_ctx_bo->size, GFP_KERNEL);
+ if (lrc->bo->vmap.is_iomem) {
+ buf = kmalloc(max_size, GFP_KERNEL);
if (!buf)
return -ENOMEM;
cmd = buf;
} else {
- cmd = lrc->bb_per_ctx_bo->vmap.vaddr;
+ cmd = lrc->bo->vmap.vaddr + wa_bb_offset(lrc);
}
*cmd++ = MI_STORE_REGISTER_MEM | MI_SRM_USE_GGTT | MI_SRM_ADD_CS_OFFSET;
@@ -974,13 +980,14 @@ static int xe_lrc_setup_utilization(struct xe_lrc *lrc)
*cmd++ = MI_BATCH_BUFFER_END;
if (buf) {
- xe_map_memcpy_to(gt_to_xe(lrc->gt), &lrc->bb_per_ctx_bo->vmap, 0,
- buf, (cmd - buf) * sizeof(*cmd));
+ xe_map_memcpy_to(gt_to_xe(lrc->gt), &lrc->bo->vmap,
+ wa_bb_offset(lrc), buf,
+ (cmd - buf) * sizeof(*cmd));
kfree(buf);
}
- xe_lrc_write_ctx_reg(lrc, CTX_BB_PER_CTX_PTR,
- xe_bo_ggtt_addr(lrc->bb_per_ctx_bo) | 1);
+ xe_lrc_write_ctx_reg(lrc, CTX_BB_PER_CTX_PTR, xe_bo_ggtt_addr(lrc->bo) +
+ wa_bb_offset(lrc) + 1);
return 0;
}
@@ -1016,20 +1023,13 @@ static int xe_lrc_init(struct xe_lrc *lrc, struct xe_hw_engine *hwe,
* FIXME: Perma-pinning LRC as we don't yet support moving GGTT address
* via VM bind calls.
*/
- lrc->bo = xe_bo_create_pin_map(xe, tile, vm, lrc_size,
+ lrc->bo = xe_bo_create_pin_map(xe, tile, vm,
+ lrc_size + LRC_WA_BB_SIZE,
ttm_bo_type_kernel,
bo_flags);
if (IS_ERR(lrc->bo))
return PTR_ERR(lrc->bo);
- lrc->bb_per_ctx_bo = xe_bo_create_pin_map(xe, tile, NULL, SZ_4K,
- ttm_bo_type_kernel,
- bo_flags);
- if (IS_ERR(lrc->bb_per_ctx_bo)) {
- err = PTR_ERR(lrc->bb_per_ctx_bo);
- goto err_lrc_finish;
- }
-
lrc->size = lrc_size;
lrc->ring.size = ring_size;
lrc->ring.tail = 0;
@@ -1819,7 +1819,8 @@ struct xe_lrc_snapshot *xe_lrc_snapshot_capture(struct xe_lrc *lrc)
snapshot->seqno = xe_lrc_seqno(lrc);
snapshot->lrc_bo = xe_bo_get(lrc->bo);
snapshot->lrc_offset = xe_lrc_pphwsp_offset(lrc);
- snapshot->lrc_size = lrc->bo->size - snapshot->lrc_offset;
+ snapshot->lrc_size = lrc->bo->size - snapshot->lrc_offset -
+ LRC_WA_BB_SIZE;
snapshot->lrc_snapshot = NULL;
snapshot->ctx_timestamp = lower_32_bits(xe_lrc_ctx_timestamp(lrc));
snapshot->ctx_job_timestamp = xe_lrc_ctx_job_timestamp(lrc);
diff --git a/drivers/gpu/drm/xe/xe_lrc_types.h b/drivers/gpu/drm/xe/xe_lrc_types.h
index ae24cf6f8dd99..883e550a94234 100644
--- a/drivers/gpu/drm/xe/xe_lrc_types.h
+++ b/drivers/gpu/drm/xe/xe_lrc_types.h
@@ -53,9 +53,6 @@ struct xe_lrc {
/** @ctx_timestamp: readout value of CTX_TIMESTAMP on last update */
u64 ctx_timestamp;
-
- /** @bb_per_ctx_bo: buffer object for per context batch wa buffer */
- struct xe_bo *bb_per_ctx_bo;
};
struct xe_lrc_snapshot;
--
2.39.5
^ permalink raw reply related [flat|nested] 2+ messages in thread
end of thread, other threads:[~2025-07-23 15:10 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-07-12 14:30 FAILED: patch "[PATCH] drm/xe: Make WA BB part of LRC BO" failed to apply to 6.15-stable tree gregkh
2025-07-23 15:10 ` [PATCH 6.15.y] drm/xe: Make WA BB part of LRC BO Sasha Levin
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox