That makes sense, thanks for the explanation. Reviewed-by: Vitaly Prosyak 
<[email protected]>

On 2026-11-16 08:25, Liang, Prike wrote:
> AMD General
>
> Hi Vitaly,
>  
> We should only return the cached VBLANK counter in task context for 
> get_vblank_counter(), based on two key considerations:
>  
>
>  1. It ensures the reset rwsem is never acquired or released from hardirq or 
> softirq contexts.
>  2. It allows the VBLANK IRQ handler to continue reading the live hardware 
> counter even while GPU recovery holds the semaphore. Otherwise, falling back 
> to the cached counter on a failed trylock would cause DRM's VBLANK count to 
> stall.
>
>  
> Regards,
>       Prike
>  
> > -----Original Message-----
> > From: Prosyak, Vitaly <[email protected]>
> > Sent: Thursday, November 12, 2026 10:52 AM
> > To: Liang, Prike <[email protected]>; [email protected]; dri-
> > [email protected]
> > Cc: Deucher, Alexander <[email protected]>; Koenig, Christian
> > <[email protected]>; Wentland, Harry <[email protected]>; Li,
> > Sun peng (Leo) <[email protected]>; [email protected]; Prosyak, Vitaly
> > <[email protected]>; Lin, Ping Lei <[email protected]>
> > Subject: Re: [PATCH RESEND 5/5] drm/amdgpu: keep reading the vblank counter 
> > in
> > IRQ context during reset
> >
> > Hi Prike,
> >
> > Thanks for the fix. Would !in_hardirq() be better than in_task() here?
> >
> > The vblank-disable timer also reaches this callback via
> > vblank_disable_fn() -> drm_vblank_disable_and_save() ->
> > drm_update_vblank_count(). In softirq context, in_task() is false, so the 
> > proposed
> > check also bypasses the reset semaphore for timer reads.
> > Using !in_hardirq() would protect that path while preserving the hard-IRQ 
> > bypass.
> >
> > What do you think?
> >
> > Thanks, Vitaly
> >
> > On 2026-11-11 03:55, Prike Liang wrote:
> > > amdgpu_get_vblank_counter_kms() returns the cached vblank->last
> > > instead of reading the hardware counter whenever it can't take
> > > reset_domain->sem for reading. The vblank interrupt reads the counter
> > > through the same callback via drm_crtc_handle_vblank(), so once GPU
> > > recovery holds the semaphore for writing, every vblank interrupt sees
> > > diff == 0 and the DRM vblank count stops advancing. The CRTCs keep
> > > scanning out until dm_suspend() turns vblank off and stops the
> > > streams, though, which only happens after the rest of the recovery 
> > > preparation.
> > >
> > > Vblank waits in that window don't see a vblank anymore. They only end
> > > when dm_suspend() turns vblank off, or after a second with the "vblank
> > > wait timed out" warning if that takes longer. The fbdev damage worker
> > > is likely to be waiting already, because recovery prints "GPU reset
> > > begin!" right before it takes the semaphore.
> > >
> > > Vblank interrupts only arrive while a CRTC is scanning out, so the
> > > interrupt path can read the hardware counter.
> > > amdgpu_device_skip_hw_access() likewise only checks the reset
> > > semaphore in task context. Only back off in task context, so the
> > > interrupt path keeps the count moving until DM turns vblank off.
> > >
> > > Fixes: f2bc71d16e73 ("drm/amdgpu: serialize vblank counter reads
> > > against GPU reset")
> > > Cc: Vitaly Prosyak <[email protected]>
> > > Cc: James Lin <[email protected]>
> > > Cc: Leo Li <[email protected]>
> > > Assisted-by: Copilot:claude-opus-5.5
> > > Signed-off-by: Prike Liang <[email protected]>
> > > ---
> > >  drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 25
> > > ++++++++++++++++++++++---
> > >  1 file changed, 22 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
> > > b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
> > > index 14a72a406d37..06e539c4c3ea 100644
> > > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
> > > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
> > > @@ -1692,6 +1692,7 @@ u32 amdgpu_get_vblank_counter_kms(struct drm_crtc
> > *crtc)
> > >      struct drm_device *dev = crtc->dev;
> > >      unsigned int pipe = crtc->index;
> > >      struct amdgpu_device *adev = drm_to_adev(dev);
> > > +   bool locked = false;
> > >      int vpos, hpos, stat;
> > >      u32 count;
> > >
> > > @@ -1699,9 +1700,26 @@ u32 amdgpu_get_vblank_counter_kms(struct
> > drm_crtc *crtc)
> > >              DRM_ERROR("Invalid crtc %u\n", pipe);
> > >              return -EINVAL;
> > >      }
> > > -   if (!down_read_trylock(&adev->reset_domain->sem))
> > > -           return drm_crtc_vblank_count(crtc);
> > >
> > > +   /*
> > > +    * A GPU reset is in progress and the display hardware cannot be read
> > > +    * safely. Return the hardware counter value DRM already cached rather
> > > +    * than the software vblank count: the core feeds this callback's
> > > +    * result back into vblank->last, and a software count lives in a
> > > +    * different numeric domain, which later trips
> > > +    * drm_WARN_ON_ONCE(cur_vblank != vblank->last) in
> > > +    * drm_update_vblank_count().
> > > +    *
> > > +    * Only task context has to back off. Vblank interrupts only arrive
> > > +    * while a CRTC is scanning out, so the interrupt path can read the
> > > +    * hardware, and it has to keep counting: vblank waits that already
> > > +    * sleep when a reset takes the semaphore never complete otherwise.
> > > +    */
> > > +   if (in_task()) {
> > > +           if (!down_read_trylock(&adev->reset_domain->sem))
> > > +                   return drm_crtc_vblank_crtc(crtc)->last;
> > > +           locked = true;
> > > +   }
> > >
> > >      /* The hw increments its frame counter at start of vsync, not at 
> > >start
> > >       * of vblank, as is required by DRM core vblank counter handling.
> > > @@ -1747,7 +1765,8 @@ u32 amdgpu_get_vblank_counter_kms(struct drm_crtc
> > *crtc)
> > >              DRM_DEBUG_VBL("NULL mode info! Returned count may be
> > wrong.\n");
> > >      }
> > >
> > > -   up_read(&adev->reset_domain->sem);
> > > +   if (locked)
> > > +           up_read(&adev->reset_domain->sem);
> > >      return count;
> > >  }
> > >

Reply via email to