Re: [bug report][regression] blktests block/008 lead kerne panic at RIP: 0010:amd_iommu_enable_faulting+0x0/0x10

[Dimitri Sivanich <sivanich@hpe.com>]

On Wed, May 29, 2024 at 08:13:42AM +0000, Tian, Kevin wrote:
> > From: Vasant Hegde <vasant.hegde@amd.com>
> > Sent: Wednesday, May 29, 2024 2:26 PM
> > 
> > Hi Yi,
> > 
> > +Dimitri, Lu, Tian.
> > 
> > 
> > On 5/29/2024 11:46 AM, Yi Zhang wrote:
> > > On Wed, May 29, 2024 at 1:40 AM Vasant Hegde <vasant.hegde@amd.com>
> > wrote:
> > >>
> > >> Hi Yi,
> > >>
> > >>
> > >> On 5/28/2024 11:00 PM, Vasant Hegde wrote:
> > >>> Hi Yi,
> > >>>
> > >>>
> > >>> On 5/28/2024 10:30 AM, Joerg Roedel wrote:
> > >>>> Adding Vasant.
> > >>>>
> > >>>> On Tue, May 28, 2024 at 10:23:10AM +0800, Yi Zhang wrote:
> > >>>>> Hello
> > >>>>> I found this regression panic issue on the latest 6.10-rc1 and it
> > >>>>> cannot be reproduced on 6.9, please help check and let me know if
> > you
> > >>>>> need any info/testing for it, thanks.
> > >>>
> > >>> I have tried to reproduce this issue on my system. So far I am not able to
> > >>> reproduce it.
> > >>>
> > >>> Will you be able to bisect the kernel?
> > >>
> > >> I see that below patch touched this code path. Can you revert below patch
> > and
> > >> test it again?
> > >
> > > Yes, the panic cannot be reproduced now after revert this patch.
> > 
> > Thanks for verifying. AMD code path (amd_iommu_enable_faulting()) just
> > return
> > zero. It doesn't do anything else. I am not familiar with cpuhp_setup_state()
> > code path.
> > 
> > @Dimitri, Can you look into this issue?
> > 
> 
> -int __init amd_iommu_enable_faulting(void)
> +int __init amd_iommu_enable_faulting(unsigned int cpu)
> 
> likely it's due to '__init' not being removed...

Yes, agreed.  I will submit a patch with this change soon.