On 13.02.2025 6:19 PM, Rob Clark wrote: > On Thu, Feb 13, 2025 at 8:36 AM Konrad Dybcio > <konrad.dybcio@xxxxxxxxxxxxxxxx> wrote: >> >> On 13.02.2025 5:10 PM, Akhil P Oommen wrote: >>> From: Jie Zhang <quic_jiezh@xxxxxxxxxxx> >>> >>> Adreno 621 has a different memory map for GPUCC block. So update >>> a6xx_gpu_state code to dump the correct set of gpucc registers. >>> >>> Signed-off-by: Jie Zhang <quic_jiezh@xxxxxxxxxxx> >>> Signed-off-by: Akhil P Oommen <quic_akhilpo@xxxxxxxxxxx> >>> --- >> >> So GPU_CC is outside what we consider GPU register region upstream.. >> >> And I've heard voices (+Caleb) lately that we should get some clock register >> dumping infrastructure.. >> >> So while I'm not against this patch fixing a bug, perhaps we can get rid of >> dumping GPU_CC here in the near future > > but we'd still want this to end up in the gpu devcoredump... I suppose if the clock dump is implemented as sysfs, we can export that func and call it from gpu dump code Konrad
