Re: [PATCH v1 5/6] drm/xe/bo_evict: Introduce xe_bo_restore_map()

[Raag Jadav <raag.jadav@intel.com>]

On Tue, Feb 24, 2026 at 02:11:38PM -0800, Matthew Brost wrote:
> On Tue, Feb 24, 2026 at 03:55:18PM +0530, Raag Jadav wrote:
> > In preparation of usecases which require remapping kernel bos after FLR,
> > introduce xe_bo_restore_map() helper.
> > 
> > Signed-off-by: Raag Jadav <raag.jadav@intel.com>
> > ---
> >  drivers/gpu/drm/xe/xe_bo_evict.c | 34 ++++++++++++++++++++++++--------
> >  drivers/gpu/drm/xe/xe_bo_evict.h |  2 ++
> >  2 files changed, 28 insertions(+), 8 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/xe/xe_bo_evict.c b/drivers/gpu/drm/xe/xe_bo_evict.c
> > index 7661fca7f278..0624c3039cd7 100644
> > --- a/drivers/gpu/drm/xe/xe_bo_evict.c
> > +++ b/drivers/gpu/drm/xe/xe_bo_evict.c
> > @@ -189,14 +189,8 @@ int xe_bo_evict_all(struct xe_device *xe)
> >  				     xe_bo_evict_pinned);
> >  }
> >  
> > -static int xe_bo_restore_and_map_ggtt(struct xe_bo *bo)
> > +static int xe_bo_map_ggtt(struct xe_bo *bo)
> >  {
> > -	int ret;
> > -
> > -	ret = xe_bo_restore_pinned(bo);
> > -	if (ret)
> > -		return ret;
> > -
> >  	if (bo->flags & XE_BO_FLAG_GGTT) {
> >  		struct xe_tile *tile;
> >  		u8 id;
> > @@ -212,6 +206,30 @@ static int xe_bo_restore_and_map_ggtt(struct xe_bo *bo)
> >  	return 0;
> >  }
> >  
> > +int xe_bo_restore_map(struct xe_device *xe)
> 
> Kernel doc. Maybe also explain why this function is needed during PCIe
> FLR in both the commit message and the kernel doc. I assume it’s
> something along the lines of: “During PCIe FLR we repopulate the GGTTs
> for any kernel BOs required for driver restart that lost GGTTs across FLR."
> 
> Just guessing on above but if we had kernel doc I wouldn't have to.
> 
> > +{
> > +	int ret;
> > +
> > +	ret = xe_bo_apply_to_pinned(xe, &xe->pinned.early.kernel_bo_present,
> > +				    &xe->pinned.early.kernel_bo_present, xe_bo_map_ggtt);
> > +	if (!ret)
> > +		ret = xe_bo_apply_to_pinned(xe, &xe->pinned.late.kernel_bo_present,
> > +					    &xe->pinned.late.kernel_bo_present, xe_bo_map_ggtt);
> > +
> > +	return ret;
> > +}
> > +
> > +static int xe_bo_restore_and_map_ggtt(struct xe_bo *bo)
> > +{
> > +	int ret;
> > +
> > +	ret = xe_bo_restore_pinned(bo);
> > +	if (ret)
> > +		return ret;
> > +
> > +	return xe_bo_map_ggtt(bo);
> > +}
> > +
> >  /**
> >   * xe_bo_restore_early - restore early phase kernel BOs to VRAM
> >   *
> > @@ -270,7 +288,7 @@ int xe_bo_restore_late(struct xe_device *xe)
> >  	return ret;
> >  }
> >  
> > -static void xe_bo_pci_dev_remove_pinned(struct xe_device *xe)
> > +void xe_bo_pci_dev_remove_pinned(struct xe_device *xe)
> 
> Same as above for kernel doc, commit message, reasoning.

Will do.

> Patch looks sane enough though.

Thanks.

Raag

> >  {
> >  	struct xe_tile *tile;
> >  	unsigned int id;
> > diff --git a/drivers/gpu/drm/xe/xe_bo_evict.h b/drivers/gpu/drm/xe/xe_bo_evict.h
> > index e8385cb7f5e9..d4f5b87243e7 100644
> > --- a/drivers/gpu/drm/xe/xe_bo_evict.h
> > +++ b/drivers/gpu/drm/xe/xe_bo_evict.h
> > @@ -14,7 +14,9 @@ int xe_bo_notifier_prepare_all_pinned(struct xe_device *xe);
> >  void xe_bo_notifier_unprepare_all_pinned(struct xe_device *xe);
> >  int xe_bo_restore_early(struct xe_device *xe);
> >  int xe_bo_restore_late(struct xe_device *xe);
> > +int xe_bo_restore_map(struct xe_device *xe);
> >  
> > +void xe_bo_pci_dev_remove_pinned(struct xe_device *xe);
> >  void xe_bo_pci_dev_remove_all(struct xe_device *xe);
> >  
> >  int xe_bo_pinned_init(struct xe_device *xe);
> > -- 
> > 2.43.0
> >