All of lore.kernel.org
 help / color / mirror / Atom feed
* [Intel-gfx] s/obj->mm.lock//
@ 2020-07-06  6:19 Chris Wilson
  2020-07-06  6:19 ` [Intel-gfx] [PATCH 01/20] drm/i915: Preallocate stashes for vma page-directories Chris Wilson
                   ` (24 more replies)
  0 siblings, 25 replies; 65+ messages in thread
From: Chris Wilson @ 2020-07-06  6:19 UTC (permalink / raw)
  To: intel-gfx

This is the easy part; pulling reservation of multiple objects under an
ww acquire context. With one simple rule that eviction is handled by the
ww acquire context, we can carefully transition the driver over to using
eviction. Instead of feeding the acquire context everywhere, we make the
caller gather up all the objects they need to acquire into the context,
then acquire their backing store. The major boon here is that by
providing a clean set of objects (that we have not yet started to
acquire any auxiliary attachments for) to the acquire context, it can
handle all the EDEADLK processing for us [since it is a pure locking
operation and does not need to release attachments upon revoking the
locks].

As a sketch of what that would look like, to illustrate the major work
remaining:

static int evict(struct drm_i915_gem_object *obj, struct i915_acquire_ctx *ctx)
{
	struct intel_memory_region *mem = obj->mm->region;
	struct drm_i915_gem_object *swap; // struct i915_mm_bo *swap
	struct i915_request *rq;
	int err;

	/* swap = mem->create_eviction_target(obj); */
	swap = i915_gem_object_create_shmem(mem->i915, obj->base.size);
	if (IS_ERR(swap))
		return PTR_ERR(swap);

	err = dma_resv_lock_interruptible(swap, &ctx->ctx);
	GEM_BUG_ON(err == -EALREADY);
	if (err == -EDEADLK)
		goto out;

	/* Obviously swap has to be carefully chosen so that this may succeed */
	err = __i915_gem_object_get_pages_locked(swap);
	if (err)
		goto out_unlock;

	rq = pinned_evict_copy(ctx, obj, swap);
	if (IS_ERR(rq)) {
		err = PTR_ERR(rq);
		goto out_unlock;
	}

	err = i915_gem_revoke_mm(obj);
	if (err)
		goto out_request;

	/* Alternatively you could wait synchronously! */
	mem->release_blocks(&obj->mm->blocks, rq);
	i915_mm_bo_put(xchg(&obj->mm, i915_mm_bo_get(swap)));

	dma_resv_add_exclusive_fence(obj->base.resv, &rq->fence);
out_request:
	i915_request_put(rq);
out_unlock:
	dma_resv_unlock(swap);
out:
	i915_gem_object_put(swap);
	return err;
}

static int relock_all(struct i915_acquire_ctx *ctx)
{
	struct i915_acquire_link *lnk, *lock;
	int err;

	for (lnk = ctx->locked; lnk; lnk = lnk->next)
		dma_resv_unlock(lnk->obj->base.resv);

	lock = fetch_and_zero(&ctx->locked);
	while ((lnk = lock)) {
		struct drm_i915_gem_object *obj;

		obj = lnk->obj;
		lock = lnk->next;

		if (ctx->locked)
			err = dma_resv_lock_interruptible(obj->base.resv,
							  &ctx->ctx);
		else
			err = dma_resv_lock_slow_interruptible(obj->base.resv,
							       &ctx->ctx);
		GEM_BUG_ON(err == -EALREADY);
		if (err == -EDEADLK) {
			struct i915_acquire *old;

			while ((old = ctx->locked)) {
				dma_resv_unlock(old->obj->base.resv);
				ctx->locked = old->next;
				old->next = lock;
				lock = old;
			}

			lnk->next = lock;
			lock = lnk;
			continue;
		}
		if (err) {
			lock = lnk;
			break;
		}

		lnk->next = ctx->locked;
		ctx->locked = lnk;
	}

	while ((lnk = lock)) {
		lock = lnk->next;
		i915_gem_object_put(lnk->obj);
		i915_acquire_free(lnk);
	}

	return err;
}

int i915_acquire_mm(struct i915_acquire_ctx *ctx)
{
	struct i915_acquire_link *lnk;
	int n, err;

restart:
	for (lnk = ctx->locked; lnk; lnk = lnk->next) {
		for (n = 0;
		     !i915_gem_object_has_pages(lnk->obj);
		     n++) {
			struct drm_i915_gem_object *evictee = NULL;

			mem = get_preferred_memregion_for_object(lnk->obj, n);
			if (!mem)
				return -ENXIO;

			while (!i915_gem_object_get_pages(lnk->obj)) {
				struct i915_acquire_link *this;

				evictee = mem->get_eviction_candidate(mem,
								      evictee);
				if (!evictee)
					break;

				err = dma_resv_lock_interruptible(evictee,
								  &ctx->ctx);
				if (err == -EALREADY)
					continue; /* XXX fragmentation? */
				if (err == 0)
					err = evict(evictee);
				dma_resv_unlock(evictee);

				if (err == -EDEADLK) {
					err = relock_all(ctx);
					if (!err)
						goto restart;
				}
				if (err)
					return err;

				evictee = NULL;
			}
		}
	}

	return 0;
}

With the biggest task being seperating the GEM object from its backing
store, so that we can transparently exchange that backing store underneath
the user handles.

[There's also the challenge of *needing* to take ww_mutex underneath
userptr/mmu-notifier, but at the worst that's a bit of lockdep annotation
and a crafty if (page_maybe_dma_pinned)) in the LRU shrinker.]
-Chris


_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

^ permalink raw reply	[flat|nested] 65+ messages in thread

end of thread, other threads:[~2020-07-27 18:53 UTC | newest]

Thread overview: 65+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-07-06  6:19 [Intel-gfx] s/obj->mm.lock// Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 01/20] drm/i915: Preallocate stashes for vma page-directories Chris Wilson
2020-07-06 18:15   ` Matthew Auld
2020-07-06 18:20     ` Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 02/20] drm/i915: Switch to object allocations for page directories Chris Wilson
2020-07-06 19:06   ` Matthew Auld
2020-07-06 19:31     ` Chris Wilson
2020-07-06 20:01     ` Chris Wilson
2020-07-06 21:08       ` Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 03/20] drm/i915/gem: Don't drop the timeline lock during execbuf Chris Wilson
2020-07-08 16:54   ` Tvrtko Ursulin
2020-07-08 18:08     ` Chris Wilson
2020-07-09 10:52       ` Tvrtko Ursulin
2020-07-09 10:57         ` Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 04/20] drm/i915/gem: Rename execbuf.bind_link to unbound_link Chris Wilson
2020-07-10 11:26   ` Tvrtko Ursulin
2020-07-06  6:19 ` [Intel-gfx] [PATCH 05/20] drm/i915/gem: Break apart the early i915_vma_pin from execbuf object lookup Chris Wilson
2020-07-10 11:27   ` Tvrtko Ursulin
2020-07-06  6:19 ` [Intel-gfx] [PATCH 06/20] drm/i915/gem: Remove the call for no-evict i915_vma_pin Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 07/20] drm/i915: Add list_for_each_entry_safe_continue_reverse Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 08/20] drm/i915: Always defer fenced work to the worker Chris Wilson
2020-07-08 12:18   ` Tvrtko Ursulin
2020-07-08 12:25     ` Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 09/20] drm/i915/gem: Assign context id for async work Chris Wilson
2020-07-08 12:26   ` Tvrtko Ursulin
2020-07-08 12:42     ` Chris Wilson
2020-07-08 14:24       ` Tvrtko Ursulin
2020-07-08 15:36         ` Chris Wilson
2020-07-09 11:01           ` Tvrtko Ursulin
2020-07-09 11:07             ` Chris Wilson
2020-07-09 11:59               ` Tvrtko Ursulin
2020-07-09 12:07                 ` Chris Wilson
2020-07-13 12:22                   ` Tvrtko Ursulin
2020-07-14 14:01                     ` Chris Wilson
2020-07-08 12:45     ` Tvrtko Ursulin
2020-07-06  6:19 ` [Intel-gfx] [PATCH 10/20] drm/i915: Export a preallocate variant of i915_active_acquire() Chris Wilson
2020-07-09 14:36   ` Maarten Lankhorst
2020-07-10 12:24     ` Tvrtko Ursulin
2020-07-10 12:32       ` Maarten Lankhorst
2020-07-13 14:29   ` Tvrtko Ursulin
2020-07-06  6:19 ` [Intel-gfx] [PATCH 11/20] drm/i915/gem: Separate the ww_mutex walker into its own list Chris Wilson
2020-07-13 14:53   ` Tvrtko Ursulin
2020-07-14 14:10     ` Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 12/20] drm/i915/gem: Asynchronous GTT unbinding Chris Wilson
2020-07-14  9:02   ` Tvrtko Ursulin
2020-07-14 15:05     ` Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 13/20] drm/i915/gem: Bind the fence async for execbuf Chris Wilson
2020-07-14 12:19   ` Tvrtko Ursulin
2020-07-14 15:21     ` Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 14/20] drm/i915/gem: Include cmdparser in common execbuf pinning Chris Wilson
2020-07-14 12:48   ` Tvrtko Ursulin
2020-07-06  6:19 ` [Intel-gfx] [PATCH 15/20] drm/i915/gem: Include secure batch " Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 16/20] drm/i915/gem: Reintroduce multiple passes for reloc processing Chris Wilson
2020-07-09 15:39   ` Tvrtko Ursulin
2020-07-06  6:19 ` [Intel-gfx] [PATCH 17/20] drm/i915: Add an implementation for i915_gem_ww_ctx locking, v2 Chris Wilson
2020-07-06 17:21   ` kernel test robot
2020-07-06  6:19 ` [Intel-gfx] [PATCH 18/20] drm/i915/gem: Pull execbuf dma resv under a single critical section Chris Wilson
2020-07-06  6:19 ` [Intel-gfx] [PATCH 19/20] drm/i915/gem: Replace i915_gem_object.mm.mutex with reservation_ww_class Chris Wilson
2020-07-09 14:06   ` Maarten Lankhorst
2020-07-06  6:19 ` [Intel-gfx] [PATCH 20/20] drm/i915: Track i915_vma with its own reference counter Chris Wilson
2020-07-06  6:28 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for series starting with [01/20] drm/i915: Preallocate stashes for vma page-directories Patchwork
2020-07-06  6:29 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2020-07-06  6:51 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2020-07-06  7:55 ` [Intel-gfx] ✗ Fi.CI.IGT: failure " Patchwork
2020-07-27 18:53 ` [Intel-gfx] s/obj->mm.lock// Thomas Hellström (Intel)

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.