Re: [PATCH v6 1/4] drm/xe/vf: Update pause/unpause() helpers with VF naming

[Matthew Brost <matthew.brost@intel.com>]

On Wed, Oct 22, 2025 at 07:01:31PM +0200, Raag Jadav wrote:
> On Wed, Oct 22, 2025 at 08:23:46AM -0700, Matthew Brost wrote:
> > On Wed, Oct 22, 2025 at 03:12:43PM +0530, Raag Jadav wrote:
> > > Now that pause/unpause() helpers have been updated for VF migration
> > > usecase, update their naming to match the functionality.
> 
> ...
> 
> > >  /**
> > > - * xe_guc_submit_pause - Stop further runs of submission tasks on given GuC.
> > > + * xe_guc_submit_vf_pause - Stop further runs of submission tasks for VF.
> > >   * @guc: the &xe_guc struct instance whose scheduler is to be disabled
> > >   */
> > > -void xe_guc_submit_pause(struct xe_guc *guc)
> > > +void xe_guc_submit_vf_pause(struct xe_guc *guc)
> > >  {
> > >  	struct xe_exec_queue *q;
> > >  	unsigned long index;
> > 
> > Can we now have xe_gt_assert in the renamed functions that IS_SRIOV_VF
> > is true?
> 
> I thought vf_recovery() already does it for us?
> 

Yes, but the general idea behind asserts — or lockdep, for that matter —
is to place them throughout the call stack. They serve as
self-documenting mechanisms for usage, help prevent misuse, and are
compiled out in production builds, making them harmless to include
everywhere.

In this case, the assert documents that xe_guc_submit_vf_* should only
be called by a VF. If someone tries to misuse it from a different call
site (not vf_recovery), the function will raise a complaint if it isn't
a VF.

Matt

> > Feel free add those in next rev and retain:
> > 
> > Reviewed-by: Matthew Brost <matthew.brost@intel.com>
> 
> Thank you.
> 
> Raag