On Fri, Oct 08, 2021 at 09:40:43AM -0700, John Harrison wrote:
> On 10/7/2021 18:21, Matthew Brost wrote:
> > On Thu, Oct 07, 2021 at 03:03:04PM -0700, John Harrison wrote:
> > > On 10/4/2021 15:06, Matthew Brost wrote:
> > > > Assign contexts in parent-child relationship consecutive guc_ids. This
> > > > is accomplished by partitioning guc_id space between ones that need to
> > > > be consecutive (1/16 available guc_ids) and ones that do not (15/16 of
> > > > available guc_ids). The consecutive search is implemented via the bitmap
> > > > API.
> > > >
> > > > This is a precursor to the full GuC multi-lrc implementation but aligns
> > > > to how GuC mutli-lrc interface is defined - guc_ids must be consecutive
> > > > when using the GuC multi-lrc interface.
> > > >
> > > > v2:
> > > > (Daniel Vetter)
> > > > - Explicitly state why we assign consecutive guc_ids
> > > > v3:
> > > > (John Harrison)
> > > > - Bring back in spin lock
> > > >
> > > > Signed-off-by: Matthew Brost <matthew.brost@xxxxxxxxx>
> > > > ---
> > > > drivers/gpu/drm/i915/gt/uc/intel_guc.h | 6 +-
> > > > .../gpu/drm/i915/gt/uc/intel_guc_submission.c | 104 ++++++++++++++----
> > > > 2 files changed, 86 insertions(+), 24 deletions(-)
> > > >
> > > > diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.h b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> > > > index 25a598e2b6e8..a9f4ec972bfb 100644
> > > > --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> > > > +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> > > > @@ -76,9 +76,13 @@ struct intel_guc {
> > > > */
> > > > spinlock_t lock;
> > > > /**
> > > > - * @guc_ids: used to allocate new guc_ids
> > > > + * @guc_ids: used to allocate new guc_ids, single-lrc
> > > > */
> > > > struct ida guc_ids;
> > > > + /**
> > > > + * @guc_ids_bitmap: used to allocate new guc_ids, multi-lrc
> > > > + */
> > > > + unsigned long *guc_ids_bitmap;
> > > > /**
> > > > * @guc_id_list: list of intel_context with valid guc_ids but no
> > > > * refs
> > > > diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
> > > > index 1f2809187513..79e7732e83b2 100644
> > > > --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
> > > > +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
> > > > @@ -128,6 +128,16 @@ guc_create_virtual(struct intel_engine_cs **siblings, unsigned int count);
> > > > #define GUC_REQUEST_SIZE 64 /* bytes */
> > > > +/*
> > > > + * We reserve 1/16 of the guc_ids for multi-lrc as these need to be contiguous
> > > > + * per the GuC submission interface. A different allocation algorithm is used
> > > > + * (bitmap vs. ida) between multi-lrc and single-lrc hence the reason to
> > > > + * partition the guc_id space. We believe the number of multi-lrc contexts in
> > > > + * use should be low and 1/16 should be sufficient. Minimum of 32 guc_ids for
> > > > + * multi-lrc.
> > > > + */
> > > > +#define NUMBER_MULTI_LRC_GUC_ID (GUC_MAX_LRC_DESCRIPTORS / 16)
> > > > +
> > > > /*
> > > > * Below is a set of functions which control the GuC scheduling state which
> > > > * require a lock.
> > > > @@ -1206,6 +1216,11 @@ int intel_guc_submission_init(struct intel_guc *guc)
> > > > INIT_WORK(&guc->submission_state.destroyed_worker,
> > > > destroyed_worker_func);
> > > > + guc->submission_state.guc_ids_bitmap =
> > > > + bitmap_zalloc(NUMBER_MULTI_LRC_GUC_ID, GFP_KERNEL);
> > > > + if (!guc->submission_state.guc_ids_bitmap)
> > > > + return -ENOMEM;
> > > > +
> > > > return 0;
> > > > }
> > > > @@ -1217,6 +1232,7 @@ void intel_guc_submission_fini(struct intel_guc *guc)
> > > > guc_lrc_desc_pool_destroy(guc);
> > > > guc_flush_destroyed_contexts(guc);
> > > > i915_sched_engine_put(guc->sched_engine);
> > > > + bitmap_free(guc->submission_state.guc_ids_bitmap);
> > > > }
> > > > static inline void queue_request(struct i915_sched_engine *sched_engine,
> > > > @@ -1268,18 +1284,43 @@ static void guc_submit_request(struct i915_request *rq)
> > > > spin_unlock_irqrestore(&sched_engine->lock, flags);
> > > > }
> > > > -static int new_guc_id(struct intel_guc *guc)
> > > > +static int new_guc_id(struct intel_guc *guc, struct intel_context *ce)
> > > > {
> > > > - return ida_simple_get(&guc->submission_state.guc_ids, 0,
> > > > - GUC_MAX_LRC_DESCRIPTORS, GFP_KERNEL |
> > > > - __GFP_RETRY_MAYFAIL | __GFP_NOWARN);
> > > > + int ret;
> > > > +
> > > > + GEM_BUG_ON(intel_context_is_child(ce));
> > > > +
> > > > + if (intel_context_is_parent(ce))
> > > > + ret = bitmap_find_free_region(guc->submission_state.guc_ids_bitmap,
> > > > + NUMBER_MULTI_LRC_GUC_ID,
> > > > + order_base_2(ce->parallel.number_children
> > > > + + 1));
> > > > + else
> > > > + ret = ida_simple_get(&guc->submission_state.guc_ids,
> > > > + NUMBER_MULTI_LRC_GUC_ID,
> > > > + GUC_MAX_LRC_DESCRIPTORS,
> > > > + GFP_KERNEL | __GFP_RETRY_MAYFAIL |
> > > > + __GFP_NOWARN);
> > > > + if (unlikely(ret < 0))
> > > > + return ret;
> > > > +
> > > > + ce->guc_id.id = ret;
> > > > + return 0;
> > > > }
> > > > static void __release_guc_id(struct intel_guc *guc, struct intel_context *ce)
> > > > {
> > > > + GEM_BUG_ON(intel_context_is_child(ce));
> > > > +
> > > > if (!context_guc_id_invalid(ce)) {
> > > > - ida_simple_remove(&guc->submission_state.guc_ids,
> > > > - ce->guc_id.id);
> > > > + if (intel_context_is_parent(ce))
> > > > + bitmap_release_region(guc->submission_state.guc_ids_bitmap,
> > > > + ce->guc_id.id,
> > > > + order_base_2(ce->parallel.number_children
> > > > + + 1));
> > > There was a discussion on the previous revision about adding a BUG_ON to
> > > ensure that number_children cannot change between the bitmap alloc and the
> > > bitmap release. I'm not seeing the new BUG_ON mentioned in this patch.
> > >
> > I thought you meant to add a BUG_ON to ensure before we release a region
> > / id it is occupied? I looked in both the bitmap API and ida API and
> > neither have a function that checks if region / id is occupied so can't
> > really add a BUG_ON for that.
> >
> > How much you add BUG_ON to ensure the number of children canoot change
> > between alloc and release? I don't follow how that would work.
> >
> > Matt
> I was thinking that where number_children is modified, you have a
> BUG_ON(guc_id_is_valid). That would ensure that the release has to match the
> alloc. Hmm, you already have a BUG_ON about the parent/child not being
> pinned in intel_context_bind_parent_child(), which I guess covers it because
> you shouldn't have a guc_id if you aren't pinned, right? And that is the
> only function which can modify number_children, yes? So maybe it's all good?
>
I think we are all good.
Matt
> John.
>
> >
> > > John.
> > >
> > >
> > > > + else
> > > > + ida_simple_remove(&guc->submission_state.guc_ids,
> > > > + ce->guc_id.id);
> > > > reset_lrc_desc(guc, ce->guc_id.id);
> > > > set_context_guc_id_invalid(ce);
> > > > }
> > > > @@ -1296,49 +1337,64 @@ static void release_guc_id(struct intel_guc *guc, struct intel_context *ce)
> > > > spin_unlock_irqrestore(&guc->submission_state.lock, flags);
> > > > }
> > > > -static int steal_guc_id(struct intel_guc *guc)
> > > > +static int steal_guc_id(struct intel_guc *guc, struct intel_context *ce)
> > > > {
> > > > - struct intel_context *ce;
> > > > - int guc_id;
> > > > + struct intel_context *cn;
> > > > lockdep_assert_held(&guc->submission_state.lock);
> > > > + GEM_BUG_ON(intel_context_is_child(ce));
> > > > + GEM_BUG_ON(intel_context_is_parent(ce));
> > > > if (!list_empty(&guc->submission_state.guc_id_list)) {
> > > > - ce = list_first_entry(&guc->submission_state.guc_id_list,
> > > > + cn = list_first_entry(&guc->submission_state.guc_id_list,
> > > > struct intel_context,
> > > > guc_id.link);
> > > > - GEM_BUG_ON(atomic_read(&ce->guc_id.ref));
> > > > - GEM_BUG_ON(context_guc_id_invalid(ce));
> > > > + GEM_BUG_ON(atomic_read(&cn->guc_id.ref));
> > > > + GEM_BUG_ON(context_guc_id_invalid(cn));
> > > > + GEM_BUG_ON(intel_context_is_child(cn));
> > > > + GEM_BUG_ON(intel_context_is_parent(cn));
> > > > - list_del_init(&ce->guc_id.link);
> > > > - guc_id = ce->guc_id.id;
> > > > + list_del_init(&cn->guc_id.link);
> > > > + ce->guc_id = cn->guc_id;
> > > > spin_lock(&ce->guc_state.lock);
> > > > - clr_context_registered(ce);
> > > > + clr_context_registered(cn);
> > > > spin_unlock(&ce->guc_state.lock);
> > > > - set_context_guc_id_invalid(ce);
> > > > - return guc_id;
> > > > + set_context_guc_id_invalid(cn);
> > > > +
> > > > + return 0;
> > > > } else {
> > > > return -EAGAIN;
> > > > }
> > > > }
> > > > -static int assign_guc_id(struct intel_guc *guc, u16 *out)
> > > > +static int assign_guc_id(struct intel_guc *guc, struct intel_context *ce)
> > > > {
> > > > int ret;
> > > > lockdep_assert_held(&guc->submission_state.lock);
> > > > + GEM_BUG_ON(intel_context_is_child(ce));
> > > > - ret = new_guc_id(guc);
> > > > + ret = new_guc_id(guc, ce);
> > > > if (unlikely(ret < 0)) {
> > > > - ret = steal_guc_id(guc);
> > > > + if (intel_context_is_parent(ce))
> > > > + return -ENOSPC;
> > > > +
> > > > + ret = steal_guc_id(guc, ce);
> > > > if (ret < 0)
> > > > return ret;
> > > > }
> > > > - *out = ret;
> > > > + if (intel_context_is_parent(ce)) {
> > > > + struct intel_context *child;
> > > > + int i = 1;
> > > > +
> > > > + for_each_child(ce, child)
> > > > + child->guc_id.id = ce->guc_id.id + i++;
> > > > + }
> > > > +
> > > > return 0;
> > > > }
> > > > @@ -1356,7 +1412,7 @@ static int pin_guc_id(struct intel_guc *guc, struct intel_context *ce)
> > > > might_lock(&ce->guc_state.lock);
> > > > if (context_guc_id_invalid(ce)) {
> > > > - ret = assign_guc_id(guc, &ce->guc_id.id);
> > > > + ret = assign_guc_id(guc, ce);
> > > > if (ret)
> > > > goto out_unlock;
> > > > ret = 1; /* Indidcates newly assigned guc_id */
> > > > @@ -1398,8 +1454,10 @@ static void unpin_guc_id(struct intel_guc *guc, struct intel_context *ce)
> > > > unsigned long flags;
> > > > GEM_BUG_ON(atomic_read(&ce->guc_id.ref) < 0);
> > > > + GEM_BUG_ON(intel_context_is_child(ce));
> > > > - if (unlikely(context_guc_id_invalid(ce)))
> > > > + if (unlikely(context_guc_id_invalid(ce) ||
> > > > + intel_context_is_parent(ce)))
> > > > return;
> > > > spin_lock_irqsave(&guc->submission_state.lock, flags);
>
