Re: [PATCH 3/3] drm/msm/a8xx: Enable CLX feature on Adreno X2-85

[Rob Clark]

On Mon, Oct 19, 2026 at 6:47 AM Thorsten Leemhuis
<regressions@xxxxxxxxxxxxx> wrote:
>
> On 10/2/26 19:47, Oleg Keri wrote:
> >> This patch landed in next-20261001 and while it works on my Glymur CRD
> >> (COB), on the Asus Zenbook A14 I see the following failure during boot:
> >
> > Same on a Lenovo Yoga Slim 7x Gen 11 (Glymur) with next-20261001, GMU
> > firmware v5.2.38:
> >
> > [    1.632277] platform 3d6c000.gmu: [drm:a6xx_hfi_wait_for_msg_interrupt] *ERROR* Message HFI_H2F_MSG_GX_BW_PERF_VOTE id 11 timed out waiting for response
> > [    1.642345] platform 3d6c000.gmu: [drm:a6xx_gmu_set_oob] *ERROR* Timeout waiting for GMU OOB set GPU_SET: 0x4
> > [    1.642360] msm_dpu ae01000.display-controller: [drm:adreno_load_gpu] *ERROR* gpu hw init failed: -110
> > [    8.053367] platform 3d6c000.gmu: [drm:a6xx_gmu_resume] *ERROR* GMU firmware initialization timed out
> > [    8.053584] msm_dpu ae01000.display-controller: [drm:adreno_load_gpu] *ERROR* Couldn't power up the GPU: -110
> >
> > Compositor fails on the first start and works on the second.
> >
> > Reverting this patch fixes it here as well.
> What's the status here? Was this fixed meanwhile? Or is this regression
> how heading towards mainline now that the merge window is open?

AFAIU this is still open.. but seems to be contained to a non-typical
config (modules + fw built-in).  It seems to be caused by trying to
initialize the GPU (and send the CLX tables to the GMU) before
_something_ else is ready.  Tracking down what that _something_ is is
still ongoing.

OTOH not having CLX could lead to problems exceeding hw current
limits, which itself is a problem.  And why I did not revert CLX.

BR,
-R