From: Ryan Brue <ryanbrue.dev@gmail.com>
To: Alessio Belle <alessio.belle@imgtec.com>,
Luigi Santivetti <luigi.santivetti@imgtec.com>,
Maarten Lankhorst <maarten.lankhorst@linux.intel.com>,
Maxime Ripard <mripard@kernel.org>,
Thomas Zimmermann <tzimmermann@suse.de>,
David Airlie <airlied@gmail.com>,
Simona Vetter <simona@ffwll.ch>,
Donald Robson <donald.robson@imgtec.com>,
Sarah Walker <sarah.walker@imgtec.com>
Cc: imagination@lists.freedesktop.org,
dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org,
stable@vger.kernel.org, Brajesh Gupta <brajesh.gupta@imgtec.com>,
Ryan Brue <ryanbrue.dev@gmail.com>
Subject: [PATCH v2] drm/imagination: Fix the value of ROGUE_COMPUTE_FLAG_PREVENT_ALL_OVERLAP
Date: Fri, 20 Nov 2026 12:10:21 -0600 [thread overview]
Message-ID: <20261120-rbrue-suez-pvr-compute-overlap-flag-v2-1-ca2b0d3adb4e@gmail.com> (raw)
The firmware takes "prevent all overlap" from bit 1 of a compute
command's flags; bit 2 is FORCE_TPU_CLK. The kernel has used bit 2 since
the FWIF headers were added, so the firmware has never seen
DRM_PVR_SUBMIT_JOB_COMPUTE_CMD_PREVENT_ALL_OVERLAP.
On a GX6250 (MT8173), a compute job overlapping fragment work from
another context hangs both data masters. With every compute job
submitted with bit 1 set, a test that hit the hang on every run no longer
does; with bit 2 set instead it still hangs.
Fixes: a26f067feac1 ("drm/imagination: Add FWIF headers")
Cc: stable@vger.kernel.org
Reviewed-by: Brajesh Gupta <brajesh.gupta@imgtec.com>
Signed-off-by: Ryan Brue <ryanbrue.dev@gmail.com>
---
Found while chasing GPU hangs on an MT8173 tablet (GX6250, BVNC
4.40.2.51) under Mesa's PowerVR Vulkan driver: a compute job running
next to fragment work from another context hangs the GPU, and Mesa's
request to serialise compute never reached the firmware.
The value comes from Imagination's firmware interface as carried in
Mesa (src/imagination/vulkan/winsys/pvrsrvkm/fw-api/pvr_rogue_fwif.h:
PREVENT_ALL_OVERLAP 0x2, FORCE_TPU_CLK 0x4), and the board confirms it:
forcing bit 1 into every compute command stops the hangs, forcing bit 2
does not. The other compute, fragment and transfer flag values in
pvr_rogue_fwif_client.h match that header.
This patch was developed with the help of an LLM.
---
Changes in v2:
- Add r-b, and Cc stable (Brajesh)
- Link to v1: https://lore.kernel.org/r/20261118-rbrue-suez-pvr-compute-overlap-flag-v1-1-ae8b48a17dfb@gmail.com
---
drivers/gpu/drm/imagination/pvr_rogue_fwif_client.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/imagination/pvr_rogue_fwif_client.h b/drivers/gpu/drm/imagination/pvr_rogue_fwif_client.h
index 6e224400083a..967e903bdca6 100644
--- a/drivers/gpu/drm/imagination/pvr_rogue_fwif_client.h
+++ b/drivers/gpu/drm/imagination/pvr_rogue_fwif_client.h
@@ -74,7 +74,7 @@
* Flags supported by the compute DM command i.e. &struct rogue_fwif_cmd_compute.
*/
-#define ROGUE_COMPUTE_FLAG_PREVENT_ALL_OVERLAP BIT_MASK(2)
+#define ROGUE_COMPUTE_FLAG_PREVENT_ALL_OVERLAP BIT_MASK(1)
/*!< Use single core in a multi core setup. */
#define ROGUE_COMPUTE_FLAG_SINGLE_CORE BIT_MASK(5)
---
base-commit: aebb7274476403040d582c61478df8f4a7bf9cd3
change-id: 20261118-rbrue-suez-pvr-compute-overlap-flag-3695ceab9ab6
Best regards,
--
Ryan Brue <ryanbrue.dev@gmail.com>
reply other threads:[~2026-11-20 18:10 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
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=20261120-rbrue-suez-pvr-compute-overlap-flag-v2-1-ca2b0d3adb4e@gmail.com \
--to=ryanbrue.dev@gmail.com \
--cc=airlied@gmail.com \
--cc=alessio.belle@imgtec.com \
--cc=brajesh.gupta@imgtec.com \
--cc=donald.robson@imgtec.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=imagination@lists.freedesktop.org \
--cc=linux-kernel@vger.kernel.org \
--cc=luigi.santivetti@imgtec.com \
--cc=maarten.lankhorst@linux.intel.com \
--cc=mripard@kernel.org \
--cc=sarah.walker@imgtec.com \
--cc=simona@ffwll.ch \
--cc=stable@vger.kernel.org \
--cc=tzimmermann@suse.de \
/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
all inboxes | Powered by JetHome®