Crashdec doesn't require SCRATCH registers anymore for a6xx and newer
architectures. So skip dumping them during recovery.

Suggested-by: Rob Clark <rob.clark@oss.qualcomm.com>
Signed-off-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
---
 drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
index 6f7ed07670b1..0995aa1f7286 100644
--- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
@@ -1546,7 +1546,7 @@ static void a6xx_recover(struct msm_gpu *gpu)
 	struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu);
 	struct a6xx_gpu *a6xx_gpu = to_a6xx_gpu(adreno_gpu);
 	struct a6xx_gmu *gmu = &a6xx_gpu->gmu;
-	int i, active_submits;
+	int active_submits;
 
 	adreno_dump_info(gpu);
 
@@ -1554,10 +1554,6 @@ static void a6xx_recover(struct msm_gpu *gpu)
 		/* Sometimes crashstate capture is skipped, so SQE should be halted here again */
 		gpu_write(gpu, REG_A6XX_CP_SQE_CNTL, 3);
 
-		for (i = 0; i < 8; i++)
-			DRM_DEV_INFO(&gpu->pdev->dev, "CP_SCRATCH_REG%d: %u\n", i,
-				gpu_read(gpu, REG_A6XX_CP_SCRATCH_REG(i)));
-
 		if (hang_debug)
 			a6xx_dump(gpu);
 

-- 
2.51.0

On 11/14/25 12:29 AM, Akhil P Oommen wrote:
> Crashdec doesn't require SCRATCH registers anymore for a6xx and newer
> architectures. So skip dumping them during recovery.
> 
> Suggested-by: Rob Clark <rob.clark@oss.qualcomm.com>
> Signed-off-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
> ---

Looks like this changed in:

https://gitlab.freedesktop.org/mesa/mesa/-/commit/95104707f189b2e1b06c855b563c1203b33da354

and needs_seqno is never true now?

Konrad

On Mon, Nov 17, 2025 at 6:00 AM Konrad Dybcio
<konrad.dybcio@oss.qualcomm.com> wrote:
>
> On 11/14/25 12:29 AM, Akhil P Oommen wrote:
> > Crashdec doesn't require SCRATCH registers anymore for a6xx and newer
> > architectures. So skip dumping them during recovery.
> >
> > Suggested-by: Rob Clark <rob.clark@oss.qualcomm.com>
> > Signed-off-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
> > ---
>
> Looks like this changed in:
>
> https://gitlab.freedesktop.org/mesa/mesa/-/commit/95104707f189b2e1b06c855b563c1203b33da354

Crashdec finding estimated hang location based on CP registers dates
back prior to importing that code into mesa ~5yrs ago, fwiw.. and was
improved over the years to better handle IB prefetch since then.
Fwiw.

It does mean for gen8 we are going to need a CPU based fallback to
crashdumper to make smmu fault devcoredumps useful (since now
crashdumper is needed to capture CP regs), but we need that anyways.

BR,
-R

> and needs_seqno is never true now?
>
> Konrad