[PATCH AUTOSEL 6.11 71/76] drm/xe/oa: Fix overflow in oa batch buffer

From: Sasha Levin
Date: Fri Oct 04 2024 - 14:37:23 EST


From: José Roberto de Souza <jose.souza@xxxxxxxxx>

[ Upstream commit 6c10ba06bb1b48acce6d4d9c1e33beb9954f1788 ]

By default xe_bb_create_job() appends a MI_BATCH_BUFFER_END to batch
buffer, this is not a problem if batch buffer is only used once but
oa reuses the batch buffer for the same metric and at each call
it appends a MI_BATCH_BUFFER_END, printing the warning below and then
overflowing.

[ 381.072016] ------------[ cut here ]------------
[ 381.072019] xe 0000:00:02.0: [drm] Assertion `bb->len * 4 + bb_prefetch(q->gt) <= size` failed!
platform: LUNARLAKE subplatform: 1
graphics: Xe2_LPG / Xe2_HPG 20.04 step B0
media: Xe2_LPM / Xe2_HPM 20.00 step B0
tile: 0 VRAM 0 B
GT: 0 type 1

So here checking if batch buffer already have MI_BATCH_BUFFER_END if
not append it.

v2:
- simply fix, suggestion from Ashutosh

Cc: Ashutosh Dixit <ashutosh.dixit@xxxxxxxxx>
Signed-off-by: José Roberto de Souza <jose.souza@xxxxxxxxx>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@xxxxxxxxx>
Link: https://patchwork.freedesktop.org/patch/msgid/20240912153842.35813-1-jose.souza@xxxxxxxxx
(cherry picked from commit 9ba0e0f30ca42a98af3689460063edfb6315718a)
Signed-off-by: Lucas De Marchi <lucas.demarchi@xxxxxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
drivers/gpu/drm/xe/xe_bb.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/xe/xe_bb.c b/drivers/gpu/drm/xe/xe_bb.c
index a13e0b3a169ed..ef777dbdf4ecc 100644
--- a/drivers/gpu/drm/xe/xe_bb.c
+++ b/drivers/gpu/drm/xe/xe_bb.c
@@ -65,7 +65,8 @@ __xe_bb_create_job(struct xe_exec_queue *q, struct xe_bb *bb, u64 *addr)
{
u32 size = drm_suballoc_size(bb->bo);

- bb->cs[bb->len++] = MI_BATCH_BUFFER_END;
+ if (bb->len == 0 || bb->cs[bb->len - 1] != MI_BATCH_BUFFER_END)
+ bb->cs[bb->len++] = MI_BATCH_BUFFER_END;

xe_gt_assert(q->gt, bb->len * 4 + bb_prefetch(q->gt) <= size);

--
2.43.0