On Mon, 11 May 2026, Imre Deak <[email protected]> wrote:
> On Mon, May 11, 2026 at 09:50:59AM -0300, Gustavo Sousa wrote:
>> Imre Deak <[email protected]> writes:
>>
>> > On Mon, May 11, 2026 at 09:20:56AM -0300, Gustavo Sousa wrote:
>> >> James Xiong <[email protected]> writes:
>> >>
>> >> > During driver probe, DMC firmware is loaded asynchronously via a
>> >> > workqueue. There is a race between parse_dmc_fw() setting the payload
>> >> > pointer (making has_dmc_id_fw() return true) and
>> >> > intel_dmc_load_program()
>> >> > writing the firmware to hardware registers. If the probe thread calls
>> >> > intel_dmc_enable_pipe() -> assert_dmc_loaded() in this window via
>> >> > intel_modeset_setup_hw_state(), it sees parsed payload but stale HW
>> >> > registers, triggering a ~20% intermittent WARNING on ADL-N warm boot.
>> >> >
>> >> > v2: Fix by calling intel_dmc_wait_fw_load() in
>> >> > intel_modeset_setup_hw_state() before iterating the CRTCs (Gustavo
>> >> > Sousa).
>> >> >
>> >> > v3: Move intel_dmc_wait_fw_load() into intel_dmc_enable_pipe() itself
>> >> > so the function is self-contained (Jani Nikula, Gustavo Sousa).
>> >> >
>> >> > Signed-off-by: James Xiong <[email protected]>
>> >>
>> >> +Imre
>> >>
>> >> Perhaps this deserves a fixes tag?
>> >
>> > Since that would result in backporting the fix to -stable, it depends on
>> > whether enabling the pipe DMC functionality before the firmware is
>> > loaded is an actual functionality problem.
>>
>> I think we depend on pipe DMC for the flip queue functionality, so I
>> suspect it could be an issue.
>
> AFAIU stable rules require knowing that a fix fixes something for a
> user.
The rules may say that, but the reality doesn't match the rules. Lots
more stuff gets backported to stable nowadays.
BR,
Jani
> I suppose one way to make this sure would be reproducing a problem
> with PIPE DMC explicitly enabled early, before the firmware is loaded.
>
>> That said, I'm not very involved in that, so I can say for sure.
>>
>> --
>> Gustavo Sousa
>>
>> >
>> >> If so, maybe it should be this one:
>> >>
>> >> Fixes: 3af2ff0840be ("drm/i915: Enable a PIPEDMC whenever its
>> >> corresponding pipe is enabled")
>> >>
>> >> The change itself looks good to me, so
>> >>
>> >> Reviewed-by: Gustavo Sousa <[email protected]>
>> >>
>> >> > ---
>> >> > drivers/gpu/drm/i915/display/intel_dmc.c | 7 ++++++-
>> >> > 1 file changed, 6 insertions(+), 1 deletion(-)
>> >> >
>> >> > diff --git a/drivers/gpu/drm/i915/display/intel_dmc.c
>> >> > b/drivers/gpu/drm/i915/display/intel_dmc.c
>> >> > index 0df4f42ba3e3..4151eae92744 100644
>> >> > --- a/drivers/gpu/drm/i915/display/intel_dmc.c
>> >> > +++ b/drivers/gpu/drm/i915/display/intel_dmc.c
>> >> > @@ -786,7 +786,12 @@ void intel_dmc_enable_pipe(const struct
>> >> > intel_crtc_state *crtc_state)
>> >> > enum pipe pipe = crtc->pipe;
>> >> > enum intel_dmc_id dmc_id = PIPE_TO_DMC_ID(pipe);
>> >> >
>> >> > - if (!is_valid_dmc_id(dmc_id) || !has_dmc_id_fw(display, dmc_id))
>> >> > + if (!is_valid_dmc_id(dmc_id))
>> >> > + return;
>> >> > +
>> >> > + intel_dmc_wait_fw_load(display);
>> >> > +
>> >> > + if (!has_dmc_id_fw(display, dmc_id))
>> >> > return;
>> >> >
>> >> > if (!can_enable_pipedmc(crtc_state)) {
>> >> > --
>> >> > 2.34.1
--
Jani Nikula, Intel
