Re: [PATCH 2/4] panthor: save panthor_file in panthor_group

[Chia-I Wu]

Hi,

On Mon, Jun 23, 2025 at 2:07 AM Liviu Dudau <liviu.dudau@xxxxxxx> wrote:
>
> On Mon, Jun 23, 2025 at 08:21:22AM +0200, Boris Brezillon wrote:
> > On Fri, 20 Jun 2025 16:50:51 -0700
> > Chia-I Wu <olvaffe@xxxxxxxxx> wrote:
> >
> > > We would like to access panthor_file from panthor_group on gpu errors.
> > > Because panthour_group can outlive drm_file, add refcount to
> > > panthor_file to ensure its lifetime.
> >
> > I'm not a huge fan of refcounting panthor_file because people tend to
> > put resource they expect to be released when the last handle goes away,
> > and if we don't refcount these sub-resources they might live longer
> > than they are meant to. Also not a huge fan of the circular referencing
> > that exists between file and groups after this change.
> >
> > How about we move the process info to a sub-object that's refcounted
> > and let both panthor_file and panthor_group take a ref on this object
> > instead?
>
> I agree with Boris on this. One alternative is to put the pid and comm in
> the panthor_group struct as panthor_file makes no use of the fields.
I took this suggestion in v2 because, when the task that opened the
node differs from the task that created the group, we are more
interested in the latter.