[Public]
> From: Alex Deucher <alexdeucher@xxxxxxxxx>
> Sent: Monday, March 17, 2025 9:33 PM
> To: Liang, Prike <Prike.Liang@xxxxxxx>
> Cc: Deucher, Alexander <Alexander.Deucher@xxxxxxx>; amd-
> gfx@xxxxxxxxxxxxxxxxxxxxx
> Subject: Re: [PATCH 03/11] drm/amdgpu/gfx: add generic handling for disable_kq
>
> On Mon, Mar 17, 2025 at 5:07 AM Liang, Prike <Prike.Liang@xxxxxxx> wrote:
> >
> > [Public]
> >
> > When disabling the gfx kernel queue, then the related ring function callback
> should be unassigned, and the clean shader callback should also not be further
> invoked. To avoid the clean shader resource allocated, we may need to drop the
> shader clean initialized at *_ sw_init() when disable gfx kernel queue.
>
> This patch prevents the cleaner shader from being run via sysfs already. We still
> need to allocate the cleaner shader resources though because they are used by
> the MES and CP firmware for user queues.
>
> Alex
Yeah, I revisit the enable_cleaner_shader flag usage logic, and the driver requires it to isolate between the GFX and compute processes.
Reviewed-by: Prike Liang <Prike.Liang@xxxxxxx>
> >
> > Regards,
> > Prike
> >
> > > -----Original Message-----
> > > From: amd-gfx <amd-gfx-bounces@xxxxxxxxxxxxxxxxxxxxx> On Behalf Of
> > > Alex Deucher
> > > Sent: Thursday, March 13, 2025 10:41 PM
> > > To: amd-gfx@xxxxxxxxxxxxxxxxxxxxx
> > > Cc: Deucher, Alexander <Alexander.Deucher@xxxxxxx>
> > > Subject: [PATCH 03/11] drm/amdgpu/gfx: add generic handling for
> > > disable_kq
> > >
> > > Add proper checks for disable_kq functionality in gfx helper
> > > functions. Add special logic for families that require the clear state setup.
> > >
> > > v2: use ring count as per Felix suggestion
> > > v3: fix num_gfx_rings handling in
> > > amdgpu_gfx_graphics_queue_acquire()
> > >
> > > Signed-off-by: Alex Deucher <alexander.deucher@xxxxxxx>
> > > ---
> > > drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 8 ++++++--
> > > drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h | 2 ++
> > > 2 files changed, 8 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> > > b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> > > index 984e6ff6e4632..a08243dd0798e 100644
> > > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> > > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> > > @@ -258,8 +258,9 @@ void amdgpu_gfx_graphics_queue_acquire(struct
> > > amdgpu_device *adev)
> > > }
> > >
> > > /* update the number of active graphics rings */
> > > - adev->gfx.num_gfx_rings =
> > > - bitmap_weight(adev->gfx.me.queue_bitmap,
> > > AMDGPU_MAX_GFX_QUEUES);
> > > + if (adev->gfx.num_gfx_rings)
> > > + adev->gfx.num_gfx_rings =
> > > + bitmap_weight(adev->gfx.me.queue_bitmap,
> > > AMDGPU_MAX_GFX_QUEUES);
> > > }
> > >
> > > static int amdgpu_gfx_kiq_acquire(struct amdgpu_device *adev, @@
> > > -1544,6
> > > +1545,9 @@ static ssize_t amdgpu_gfx_set_run_cleaner_shader(struct
> > > +device
> > > *dev,
> > > if (adev->in_suspend && !adev->in_runpm)
> > > return -EPERM;
> > >
> > > + if (adev->gfx.disable_kq)
> > > + return -ENOTSUPP;
> > > +
> > > ret = kstrtol(buf, 0, &value);
> > >
> > > if (ret)
> > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> > > b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> > > index ddf4533614bac..8fa68a4ac34f1 100644
> > > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> > > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> > > @@ -483,6 +483,8 @@ struct amdgpu_gfx {
> > >
> > > atomic_t total_submission_cnt;
> > > struct delayed_work idle_work;
> > > +
> > > + bool disable_kq;
> > > };
> > >
> > > struct amdgpu_gfx_ras_reg_entry {
> > > --
> > > 2.48.1
> >
