Quoting Tvrtko Ursulin (2019-01-15 09:47:45)
>
> On 14/01/2019 21:17, Chris Wilson wrote:
> > Since commit 93065ac753e4 ("mm, oom: distinguish blockable mode for mmu
> > notifiers") we have been able to report failure from
> > mmu_invalidate_range_start which allows us to use a trylock on the
> > struct_mutex to avoid potential recursion and report -EBUSY instead.
> > Furthermore, this allows us to pull the work into the main callback and
> > avoid the sleight-of-hand in using a workqueue to avoid lockdep.
> >
> > However, not all paths to mmu_invalidate_range_start are prepared to
> > handle failure, so instead of reporting the recursion, deal with it by
> > propagating the failure upwards, who can decide themselves to handle it
> > or report it.
> >
> > v2: Mark up the recursive lock behaviour and comment on the various weak
> > points.
> >
> > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108375
> > References: 93065ac753e4 ("mm, oom: distinguish blockable mode for mmu notifiers")
> > Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
> > Cc: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx>
> > ---
> > drivers/gpu/drm/i915/i915_drv.h | 4 +-
> > drivers/gpu/drm/i915/i915_gem.c | 30 +++-
> > drivers/gpu/drm/i915/i915_gem_object.h | 7 +
> > drivers/gpu/drm/i915/i915_gem_userptr.c | 220 +++++++++++-------------
> > 4 files changed, 135 insertions(+), 126 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> > index fa99824f63b3..4b67037c0318 100644
> > --- a/drivers/gpu/drm/i915/i915_drv.h
> > +++ b/drivers/gpu/drm/i915/i915_drv.h
> > @@ -2931,8 +2931,8 @@ enum i915_mm_subclass { /* lockdep subclass for obj->mm.lock/struct_mutex */
> > I915_MM_SHRINKER /* called "recursively" from direct-reclaim-esque */
> > };
> >
> > -void __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
> > - enum i915_mm_subclass subclass);
> > +int __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
> > + enum i915_mm_subclass subclass);
> > void __i915_gem_object_invalidate(struct drm_i915_gem_object *obj);
> >
> > enum i915_map_type {
> > diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> > index 0bfed33178e1..90c167f71345 100644
> > --- a/drivers/gpu/drm/i915/i915_gem.c
> > +++ b/drivers/gpu/drm/i915/i915_gem.c
> > @@ -2303,8 +2303,8 @@ __i915_gem_object_unset_pages(struct drm_i915_gem_object *obj)
> > struct sg_table *pages;
> >
> > pages = fetch_and_zero(&obj->mm.pages);
> > - if (!pages)
> > - return NULL;
> > + if (IS_ERR_OR_NULL(pages))
> > + return pages;
> >
> > spin_lock(&i915->mm.obj_lock);
> > list_del(&obj->mm.link);
> > @@ -2328,22 +2328,23 @@ __i915_gem_object_unset_pages(struct drm_i915_gem_object *obj)
> > return pages;
> > }
> >
> > -void __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
> > - enum i915_mm_subclass subclass)
> > +int __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
> > + enum i915_mm_subclass subclass)
> > {
> > struct sg_table *pages;
> > + int ret;
> >
> > if (i915_gem_object_has_pinned_pages(obj))
> > - return;
> > + return -EBUSY;
> >
> > GEM_BUG_ON(obj->bind_count);
> > - if (!i915_gem_object_has_pages(obj))
> > - return;
> >
> > /* May be called by shrinker from within get_pages() (on another bo) */
> > mutex_lock_nested(&obj->mm.lock, subclass);
> > - if (unlikely(atomic_read(&obj->mm.pages_pin_count)))
> > + if (unlikely(atomic_read(&obj->mm.pages_pin_count))) {
> > + ret = -EBUSY;
> > goto unlock;
> > + }
> >
> > /*
> > * ->put_pages might need to allocate memory for the bit17 swizzle
> > @@ -2351,11 +2352,24 @@ void __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
> > * lists early.
> > */
> > pages = __i915_gem_object_unset_pages(obj);
> > +
> > + /*
> > + * XXX Temporary hijinx to avoid updating all backends to handle
> > + * NULL pages. In the future, when we have more asynchronous
> > + * get_pages backends we should be better able to handle the
> > + * cancellation of the async task in a more uniform manner.
> > + */
> > + if (!pages && !i915_gem_object_needs_async_cancel(obj))
> > + pages = ERR_PTR(-EINVAL);
> > +
> > if (!IS_ERR(pages))
> > obj->ops->put_pages(obj, pages);
> >
> > + ret = 0;
> > unlock:
> > mutex_unlock(&obj->mm.lock);
> > +
> > + return ret;
> > }
> >
> > bool i915_sg_trim(struct sg_table *orig_st)
> > diff --git a/drivers/gpu/drm/i915/i915_gem_object.h b/drivers/gpu/drm/i915/i915_gem_object.h
> > index ff3da64470dd..cb1b0144d274 100644
> > --- a/drivers/gpu/drm/i915/i915_gem_object.h
> > +++ b/drivers/gpu/drm/i915/i915_gem_object.h
> > @@ -57,6 +57,7 @@ struct drm_i915_gem_object_ops {
> > #define I915_GEM_OBJECT_HAS_STRUCT_PAGE BIT(0)
> > #define I915_GEM_OBJECT_IS_SHRINKABLE BIT(1)
> > #define I915_GEM_OBJECT_IS_PROXY BIT(2)
> > +#define I915_GEM_OBJECT_ASYNC_CANCEL BIT(3)
> >
> > /* Interface between the GEM object and its backing storage.
> > * get_pages() is called once prior to the use of the associated set
> > @@ -387,6 +388,12 @@ i915_gem_object_is_proxy(const struct drm_i915_gem_object *obj)
> > return obj->ops->flags & I915_GEM_OBJECT_IS_PROXY;
> > }
> >
> > +static inline bool
> > +i915_gem_object_needs_async_cancel(const struct drm_i915_gem_object *obj)
> > +{
> > + return obj->ops->flags & I915_GEM_OBJECT_ASYNC_CANCEL;
> > +}
> > +
> > static inline bool
> > i915_gem_object_is_active(const struct drm_i915_gem_object *obj)
> > {
> > diff --git a/drivers/gpu/drm/i915/i915_gem_userptr.c b/drivers/gpu/drm/i915/i915_gem_userptr.c
> > index 1fb6a7bb5054..9c2008a480e2 100644
> > --- a/drivers/gpu/drm/i915/i915_gem_userptr.c
> > +++ b/drivers/gpu/drm/i915/i915_gem_userptr.c
> > @@ -49,77 +49,81 @@ struct i915_mmu_notifier {
> > struct hlist_node node;
> > struct mmu_notifier mn;
> > struct rb_root_cached objects;
> > - struct workqueue_struct *wq;
> > + struct i915_mm_struct *mm;
> > };
> >
> > struct i915_mmu_object {
> > struct i915_mmu_notifier *mn;
> > struct drm_i915_gem_object *obj;
> > struct interval_tree_node it;
> > - struct list_head link;
> > - struct work_struct work;
> > - bool attached;
> > };
> >
> > -static void cancel_userptr(struct work_struct *work)
> > +static void add_object(struct i915_mmu_object *mo)
> > {
> > - struct i915_mmu_object *mo = container_of(work, typeof(*mo), work);
> > - struct drm_i915_gem_object *obj = mo->obj;
> > - struct work_struct *active;
> > -
> > - /* Cancel any active worker and force us to re-evaluate gup */
> > - mutex_lock(&obj->mm.lock);
> > - active = fetch_and_zero(&obj->userptr.work);
> > - mutex_unlock(&obj->mm.lock);
> > - if (active)
> > - goto out;
> > -
> > - i915_gem_object_wait(obj, I915_WAIT_ALL, MAX_SCHEDULE_TIMEOUT, NULL);
> > -
> > - mutex_lock(&obj->base.dev->struct_mutex);
> > -
> > - /* We are inside a kthread context and can't be interrupted */
> > - if (i915_gem_object_unbind(obj) == 0)
> > - __i915_gem_object_put_pages(obj, I915_MM_NORMAL);
> > - WARN_ONCE(i915_gem_object_has_pages(obj),
> > - "Failed to release pages: bind_count=%d, pages_pin_count=%d, pin_global=%d\n",
> > - obj->bind_count,
> > - atomic_read(&obj->mm.pages_pin_count),
> > - obj->pin_global);
> > -
> > - mutex_unlock(&obj->base.dev->struct_mutex);
> > -
> > -out:
> > - i915_gem_object_put(obj);
> > + GEM_BUG_ON(!RB_EMPTY_NODE(&mo->it.rb));
> > + interval_tree_insert(&mo->it, &mo->mn->objects);
> > }
> >
> > -static void add_object(struct i915_mmu_object *mo)
> > +static void del_object(struct i915_mmu_object *mo)
> > {
> > - if (mo->attached)
> > + if (RB_EMPTY_NODE(&mo->it.rb))
> > return;
> >
> > - interval_tree_insert(&mo->it, &mo->mn->objects);
> > - mo->attached = true;
> > + interval_tree_remove(&mo->it, &mo->mn->objects);
> > + RB_CLEAR_NODE(&mo->it.rb);
> > }
> >
> > -static void del_object(struct i915_mmu_object *mo)
> > +static void
> > +__i915_gem_userptr_set_active(struct drm_i915_gem_object *obj, bool value)
> > {
> > - if (!mo->attached)
> > + struct i915_mmu_object *mo = obj->userptr.mmu_object;
> > +
> > + /*
> > + * During mm_invalidate_range we need to cancel any userptr that
> > + * overlaps the range being invalidated. Doing so requires the
> > + * struct_mutex, and that risks recursion. In order to cause
> > + * recursion, the user must alias the userptr address space with
> > + * a GTT mmapping (possible with a MAP_FIXED) - then when we have
> > + * to invalidate that mmaping, mm_invalidate_range is called with
> > + * the userptr address *and* the struct_mutex held. To prevent that
> > + * we set a flag under the i915_mmu_notifier spinlock to indicate
> > + * whether this object is valid.
> > + */
> > + if (!mo)
> > return;
> >
> > - interval_tree_remove(&mo->it, &mo->mn->objects);
> > - mo->attached = false;
> > + spin_lock(&mo->mn->lock);
> > + if (value)
> > + add_object(mo);
> > + else
> > + del_object(mo);
> > + spin_unlock(&mo->mn->lock);
> > }
> >
> > -static int i915_gem_userptr_mn_invalidate_range_start(struct mmu_notifier *_mn,
> > - const struct mmu_notifier_range *range)
> > +static struct mutex *__i915_mutex_lock_recursive(struct mutex *m)
> > +{
> > + switch (mutex_trylock_recursive(m)) {
> > + default:
> > + case MUTEX_TRYLOCK_FAILED:
> > + mutex_lock_nested(m, I915_MM_SHRINKER);
>
> Have you found in the meantime that this is actually needed? Although it
> is since we use this subclass to put the pages..
Yes, if we do not attempt to try a full mutex_lock() here, we fail
the invalidation subtests of gem_userptr_blits(). But you mean the
subclass? Given enough time I'll find the path to break it, at the
present time I feel more comfortable cargo-culting the rules devised for
the recursive locking from inside direct-reclaim to another
direct-reclaim path.
The end-goal here is that upon breaking up struct_mutex we replace it
various locks that we are not allowed to allocate under. No more direct
reclaim recursion.
> Should i915_gem_shrinker_tainst_mutex taint obj->mm.lock with the
> shrinker subclass?
The shrinker uses the obj->mm.lock I915_MM_SHRINKER subclass, so it's a
little more faff to taint.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
