Re: Linux 6.18.19 -- amdgpu bug and a new warning

[Harshit Mogalapalli]

Hi,

> > A commit in 6.18.19 has introduced a bug and a new warning when doing
> > amdgpu driver re-binding. In addition to the bug, the last line of the
> > output below is a new warning re: the thermal alert
> >
> > This bug doesn't seem to cause any show-stopping problems, but it is a 
> > bug
> > and it persists into 6.18.20.
> >
> > I can do a bisect if needed, but I'm hoping one of our AMD guys can more
> > quickly spot what's going on :)
>
> Are you saying it is from 6.18.18 to 6.18.19 it was introduced?  Nothing 
> immediately jumps out to me.  So I would say bisect please.

I think backporting this would help ?

commit: e12603bf2c3d ("drm/amd/pm: fix amdgpu_irq enabled counter 
unbalanced on smu v11.0")

> >    amdgpu 0000:14:00.0: amdgpu: amdgpu: finishing device.
> >    ------------[ cut here ]------------
> >    WARNING: CPU: 1 PID: 2773 at drivers/gpu/drm/amd/amdgpu/ 
> > amdgpu_irq.c:639 amdgpu_irq_put+0xa4/0xc0 [amdgpu]
...
> >    CPU: 1 UID: 0 PID: 2773 Comm: bind-device.sh Not tainted 6.18.20 #1 
> > PREEMPT(lazy)
..
> >     <TASK>
> >     smu_smc_hw_cleanup+0x61/0x490 [amdgpu]
> >     smu_hw_fini+0xef/0x180 [amdgpu]
> >     amdgpu_ip_block_hw_fini+0x37/0x41 [amdgpu]
Thanks,
Harshit