On Thu, Oct 23, 2025 at 10:25:08PM +0200, Michal Wajdeczko wrote:
>
>
> On 10/22/2025 12:41 AM, Michał Winiarski wrote:
> > From: Lukasz Laguna <[email protected]>
> >
> > Instead of accessing VF's lmem_obj directly, introduce a helper function
> > to make the access more convenient.
> >
> > Signed-off-by: Lukasz Laguna <[email protected]>
> > Signed-off-by: Michał Winiarski <[email protected]>
> > ---
> > drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c | 31 ++++++++++++++++++++++
> > drivers/gpu/drm/xe/xe_gt_sriov_pf_config.h | 1 +
> > 2 files changed, 32 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
> > b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
> > index c857879e28fe5..28d648c386487 100644
> > --- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
> > +++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
> > @@ -1643,6 +1643,37 @@ int xe_gt_sriov_pf_config_bulk_set_lmem(struct xe_gt
> > *gt, unsigned int vfid,
> > "LMEM", n, err);
> > }
> >
> > +static struct xe_bo *pf_get_vf_config_lmem_obj(struct xe_gt *gt, unsigned
> > int vfid)
> > +{
> > + struct xe_gt_sriov_config *config = pf_pick_vf_config(gt, vfid);
> > +
> > + return config->lmem_obj;
> > +}
> > +
> > +/**
> > + * xe_gt_sriov_pf_config_get_lmem_obj - Take a reference to the struct
> > &xe_bo backing VF LMEM.
>
> * xe_gt_sriov_pf_config_get_lmem_obj() - Take ...
Ok.
>
> > + * @gt: the &xe_gt
> > + * @vfid: the VF identifier
>
> since you assert vfid below, add "(can't be 0)"
Ok.
>
> > + *
> > + * This function can only be called on PF.
> > + * The caller is responsible for calling xe_bo_put() on the returned
> > object.
> > + *
> > + * Return: pointer to struct &xe_bo backing VF LMEM (if any).
> > + */
> > +struct xe_bo *xe_gt_sriov_pf_config_get_lmem_obj(struct xe_gt *gt,
> > unsigned int vfid)
> > +{
> > + struct xe_bo *lmem_obj;
> > +
> > + xe_gt_assert(gt, vfid);
> > +
> > + mutex_lock(xe_gt_sriov_pf_master_mutex(gt));
> > + lmem_obj = pf_get_vf_config_lmem_obj(gt, vfid);
> > + xe_bo_get(lmem_obj);
> > + mutex_unlock(xe_gt_sriov_pf_master_mutex(gt));
> > +
> > + return lmem_obj;
>
> or just
>
> {
> guard(mutex)(xe_gt_sriov_pf_master_mutex(gt));
>
> return xe_bo_get(pf_get_vf_config_lmem_obj(gt, vfid));
> }
Ok.
>
> > +}
> > +
> > static u64 pf_query_free_lmem(struct xe_gt *gt)
> > {
> > struct xe_tile *tile = gt->tile;
> > diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.h
> > b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.h
> > index 6916b8f58ebf2..03c5dc0cd5fef 100644
> > --- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.h
> > +++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.h
> > @@ -36,6 +36,7 @@ int xe_gt_sriov_pf_config_set_lmem(struct xe_gt *gt,
> > unsigned int vfid, u64 size
> > int xe_gt_sriov_pf_config_set_fair_lmem(struct xe_gt *gt, unsigned int
> > vfid, unsigned int num_vfs);
> > int xe_gt_sriov_pf_config_bulk_set_lmem(struct xe_gt *gt, unsigned int
> > vfid, unsigned int num_vfs,
> > u64 size);
> > +struct xe_bo *xe_gt_sriov_pf_config_get_lmem_obj(struct xe_gt *gt,
> > unsigned int vfid);
> >
> > u32 xe_gt_sriov_pf_config_get_exec_quantum(struct xe_gt *gt, unsigned int
> > vfid);
> > int xe_gt_sriov_pf_config_set_exec_quantum(struct xe_gt *gt, unsigned int
> > vfid, u32 exec_quantum);
>
> probably we should block VF's reprovisioning during the SAVE/RESTORE,
> but that could be done later as follow up
Yeah - I ended up leaving it out of this series.
But the general work-in-progress idea was to block provisioning access
from userspace if we're in any WIP state.
>
> Reviewed-by: Michal Wajdeczko <[email protected]>
>
Thanks,
-Michał
