brintos

brintos / linux-shallow public Read only

0
0
Text · 13.5 KiB · 762127d Raw
512 lines · c
1/*2 * SPDX-License-Identifier: MIT3 *4 * Copyright © 2018 Intel Corporation5 */6 7#include <linux/mutex.h>8 9#include "i915_drv.h"10#include "i915_request.h"11#include "i915_scheduler.h"12 13static struct kmem_cache *slab_dependencies;14static struct kmem_cache *slab_priorities;15 16static DEFINE_SPINLOCK(schedule_lock);17 18static const struct i915_request *19node_to_request(const struct i915_sched_node *node)20{21	return container_of(node, const struct i915_request, sched);22}23 24static inline bool node_started(const struct i915_sched_node *node)25{26	return i915_request_started(node_to_request(node));27}28 29static inline bool node_signaled(const struct i915_sched_node *node)30{31	return i915_request_completed(node_to_request(node));32}33 34static inline struct i915_priolist *to_priolist(struct rb_node *rb)35{36	return rb_entry(rb, struct i915_priolist, node);37}38 39static void assert_priolists(struct i915_sched_engine * const sched_engine)40{41	struct rb_node *rb;42	long last_prio;43 44	if (!IS_ENABLED(CONFIG_DRM_I915_DEBUG_GEM))45		return;46 47	GEM_BUG_ON(rb_first_cached(&sched_engine->queue) !=48		   rb_first(&sched_engine->queue.rb_root));49 50	last_prio = INT_MAX;51	for (rb = rb_first_cached(&sched_engine->queue); rb; rb = rb_next(rb)) {52		const struct i915_priolist *p = to_priolist(rb);53 54		GEM_BUG_ON(p->priority > last_prio);55		last_prio = p->priority;56	}57}58 59struct list_head *60i915_sched_lookup_priolist(struct i915_sched_engine *sched_engine, int prio)61{62	struct i915_priolist *p;63	struct rb_node **parent, *rb;64	bool first = true;65 66	lockdep_assert_held(&sched_engine->lock);67	assert_priolists(sched_engine);68 69	if (unlikely(sched_engine->no_priolist))70		prio = I915_PRIORITY_NORMAL;71 72find_priolist:73	/* most positive priority is scheduled first, equal priorities fifo */74	rb = NULL;75	parent = &sched_engine->queue.rb_root.rb_node;76	while (*parent) {77		rb = *parent;78		p = to_priolist(rb);79		if (prio > p->priority) {80			parent = &rb->rb_left;81		} else if (prio < p->priority) {82			parent = &rb->rb_right;83			first = false;84		} else {85			return &p->requests;86		}87	}88 89	if (prio == I915_PRIORITY_NORMAL) {90		p = &sched_engine->default_priolist;91	} else {92		p = kmem_cache_alloc(slab_priorities, GFP_ATOMIC);93		/* Convert an allocation failure to a priority bump */94		if (unlikely(!p)) {95			prio = I915_PRIORITY_NORMAL; /* recurses just once */96 97			/* To maintain ordering with all rendering, after an98			 * allocation failure we have to disable all scheduling.99			 * Requests will then be executed in fifo, and schedule100			 * will ensure that dependencies are emitted in fifo.101			 * There will be still some reordering with existing102			 * requests, so if userspace lied about their103			 * dependencies that reordering may be visible.104			 */105			sched_engine->no_priolist = true;106			goto find_priolist;107		}108	}109 110	p->priority = prio;111	INIT_LIST_HEAD(&p->requests);112 113	rb_link_node(&p->node, rb, parent);114	rb_insert_color_cached(&p->node, &sched_engine->queue, first);115 116	return &p->requests;117}118 119void __i915_priolist_free(struct i915_priolist *p)120{121	kmem_cache_free(slab_priorities, p);122}123 124struct sched_cache {125	struct list_head *priolist;126};127 128static struct i915_sched_engine *129lock_sched_engine(struct i915_sched_node *node,130		  struct i915_sched_engine *locked,131		  struct sched_cache *cache)132{133	const struct i915_request *rq = node_to_request(node);134	struct i915_sched_engine *sched_engine;135 136	GEM_BUG_ON(!locked);137 138	/*139	 * Virtual engines complicate acquiring the engine timeline lock,140	 * as their rq->engine pointer is not stable until under that141	 * engine lock. The simple ploy we use is to take the lock then142	 * check that the rq still belongs to the newly locked engine.143	 */144	while (locked != (sched_engine = READ_ONCE(rq->engine)->sched_engine)) {145		spin_unlock(&locked->lock);146		memset(cache, 0, sizeof(*cache));147		spin_lock(&sched_engine->lock);148		locked = sched_engine;149	}150 151	GEM_BUG_ON(locked != sched_engine);152	return locked;153}154 155static void __i915_schedule(struct i915_sched_node *node,156			    const struct i915_sched_attr *attr)157{158	const int prio = max(attr->priority, node->attr.priority);159	struct i915_sched_engine *sched_engine;160	struct i915_dependency *dep, *p;161	struct i915_dependency stack;162	struct sched_cache cache;163	LIST_HEAD(dfs);164 165	/* Needed in order to use the temporary link inside i915_dependency */166	lockdep_assert_held(&schedule_lock);167	GEM_BUG_ON(prio == I915_PRIORITY_INVALID);168 169	if (node_signaled(node))170		return;171 172	stack.signaler = node;173	list_add(&stack.dfs_link, &dfs);174 175	/*176	 * Recursively bump all dependent priorities to match the new request.177	 *178	 * A naive approach would be to use recursion:179	 * static void update_priorities(struct i915_sched_node *node, prio) {180	 *	list_for_each_entry(dep, &node->signalers_list, signal_link)181	 *		update_priorities(dep->signal, prio)182	 *	queue_request(node);183	 * }184	 * but that may have unlimited recursion depth and so runs a very185	 * real risk of overunning the kernel stack. Instead, we build186	 * a flat list of all dependencies starting with the current request.187	 * As we walk the list of dependencies, we add all of its dependencies188	 * to the end of the list (this may include an already visited189	 * request) and continue to walk onwards onto the new dependencies. The190	 * end result is a topological list of requests in reverse order, the191	 * last element in the list is the request we must execute first.192	 */193	list_for_each_entry(dep, &dfs, dfs_link) {194		struct i915_sched_node *node = dep->signaler;195 196		/* If we are already flying, we know we have no signalers */197		if (node_started(node))198			continue;199 200		/*201		 * Within an engine, there can be no cycle, but we may202		 * refer to the same dependency chain multiple times203		 * (redundant dependencies are not eliminated) and across204		 * engines.205		 */206		list_for_each_entry(p, &node->signalers_list, signal_link) {207			GEM_BUG_ON(p == dep); /* no cycles! */208 209			if (node_signaled(p->signaler))210				continue;211 212			if (prio > READ_ONCE(p->signaler->attr.priority))213				list_move_tail(&p->dfs_link, &dfs);214		}215	}216 217	/*218	 * If we didn't need to bump any existing priorities, and we haven't219	 * yet submitted this request (i.e. there is no potential race with220	 * execlists_submit_request()), we can set our own priority and skip221	 * acquiring the engine locks.222	 */223	if (node->attr.priority == I915_PRIORITY_INVALID) {224		GEM_BUG_ON(!list_empty(&node->link));225		node->attr = *attr;226 227		if (stack.dfs_link.next == stack.dfs_link.prev)228			return;229 230		__list_del_entry(&stack.dfs_link);231	}232 233	memset(&cache, 0, sizeof(cache));234	sched_engine = node_to_request(node)->engine->sched_engine;235	spin_lock(&sched_engine->lock);236 237	/* Fifo and depth-first replacement ensure our deps execute before us */238	sched_engine = lock_sched_engine(node, sched_engine, &cache);239	list_for_each_entry_safe_reverse(dep, p, &dfs, dfs_link) {240		struct i915_request *from = container_of(dep->signaler,241							 struct i915_request,242							 sched);243		INIT_LIST_HEAD(&dep->dfs_link);244 245		node = dep->signaler;246		sched_engine = lock_sched_engine(node, sched_engine, &cache);247		lockdep_assert_held(&sched_engine->lock);248 249		/* Recheck after acquiring the engine->timeline.lock */250		if (prio <= node->attr.priority || node_signaled(node))251			continue;252 253		GEM_BUG_ON(node_to_request(node)->engine->sched_engine !=254			   sched_engine);255 256		/* Must be called before changing the nodes priority */257		if (sched_engine->bump_inflight_request_prio)258			sched_engine->bump_inflight_request_prio(from, prio);259 260		WRITE_ONCE(node->attr.priority, prio);261 262		/*263		 * Once the request is ready, it will be placed into the264		 * priority lists and then onto the HW runlist. Before the265		 * request is ready, it does not contribute to our preemption266		 * decisions and we can safely ignore it, as it will, and267		 * any preemption required, be dealt with upon submission.268		 * See engine->submit_request()269		 */270		if (list_empty(&node->link))271			continue;272 273		if (i915_request_in_priority_queue(node_to_request(node))) {274			if (!cache.priolist)275				cache.priolist =276					i915_sched_lookup_priolist(sched_engine,277								   prio);278			list_move_tail(&node->link, cache.priolist);279		}280 281		/* Defer (tasklet) submission until after all of our updates. */282		if (sched_engine->kick_backend)283			sched_engine->kick_backend(node_to_request(node), prio);284	}285 286	spin_unlock(&sched_engine->lock);287}288 289void i915_schedule(struct i915_request *rq, const struct i915_sched_attr *attr)290{291	spin_lock_irq(&schedule_lock);292	__i915_schedule(&rq->sched, attr);293	spin_unlock_irq(&schedule_lock);294}295 296void i915_sched_node_init(struct i915_sched_node *node)297{298	INIT_LIST_HEAD(&node->signalers_list);299	INIT_LIST_HEAD(&node->waiters_list);300	INIT_LIST_HEAD(&node->link);301 302	i915_sched_node_reinit(node);303}304 305void i915_sched_node_reinit(struct i915_sched_node *node)306{307	node->attr.priority = I915_PRIORITY_INVALID;308	node->semaphores = 0;309	node->flags = 0;310 311	GEM_BUG_ON(!list_empty(&node->signalers_list));312	GEM_BUG_ON(!list_empty(&node->waiters_list));313	GEM_BUG_ON(!list_empty(&node->link));314}315 316static struct i915_dependency *317i915_dependency_alloc(void)318{319	return kmem_cache_alloc(slab_dependencies, GFP_KERNEL);320}321 322static void323i915_dependency_free(struct i915_dependency *dep)324{325	kmem_cache_free(slab_dependencies, dep);326}327 328bool __i915_sched_node_add_dependency(struct i915_sched_node *node,329				      struct i915_sched_node *signal,330				      struct i915_dependency *dep,331				      unsigned long flags)332{333	bool ret = false;334 335	spin_lock_irq(&schedule_lock);336 337	if (!node_signaled(signal)) {338		INIT_LIST_HEAD(&dep->dfs_link);339		dep->signaler = signal;340		dep->waiter = node;341		dep->flags = flags;342 343		/* All set, now publish. Beware the lockless walkers. */344		list_add_rcu(&dep->signal_link, &node->signalers_list);345		list_add_rcu(&dep->wait_link, &signal->waiters_list);346 347		/* Propagate the chains */348		node->flags |= signal->flags;349		ret = true;350	}351 352	spin_unlock_irq(&schedule_lock);353 354	return ret;355}356 357int i915_sched_node_add_dependency(struct i915_sched_node *node,358				   struct i915_sched_node *signal,359				   unsigned long flags)360{361	struct i915_dependency *dep;362 363	dep = i915_dependency_alloc();364	if (!dep)365		return -ENOMEM;366 367	if (!__i915_sched_node_add_dependency(node, signal, dep,368					      flags | I915_DEPENDENCY_ALLOC))369		i915_dependency_free(dep);370 371	return 0;372}373 374void i915_sched_node_fini(struct i915_sched_node *node)375{376	struct i915_dependency *dep, *tmp;377 378	spin_lock_irq(&schedule_lock);379 380	/*381	 * Everyone we depended upon (the fences we wait to be signaled)382	 * should retire before us and remove themselves from our list.383	 * However, retirement is run independently on each timeline and384	 * so we may be called out-of-order.385	 */386	list_for_each_entry_safe(dep, tmp, &node->signalers_list, signal_link) {387		GEM_BUG_ON(!list_empty(&dep->dfs_link));388 389		list_del_rcu(&dep->wait_link);390		if (dep->flags & I915_DEPENDENCY_ALLOC)391			i915_dependency_free(dep);392	}393	INIT_LIST_HEAD(&node->signalers_list);394 395	/* Remove ourselves from everyone who depends upon us */396	list_for_each_entry_safe(dep, tmp, &node->waiters_list, wait_link) {397		GEM_BUG_ON(dep->signaler != node);398		GEM_BUG_ON(!list_empty(&dep->dfs_link));399 400		list_del_rcu(&dep->signal_link);401		if (dep->flags & I915_DEPENDENCY_ALLOC)402			i915_dependency_free(dep);403	}404	INIT_LIST_HEAD(&node->waiters_list);405 406	spin_unlock_irq(&schedule_lock);407}408 409void i915_request_show_with_schedule(struct drm_printer *m,410				     const struct i915_request *rq,411				     const char *prefix,412				     int indent)413{414	struct i915_dependency *dep;415 416	i915_request_show(m, rq, prefix, indent);417	if (i915_request_completed(rq))418		return;419 420	rcu_read_lock();421	for_each_signaler(dep, rq) {422		const struct i915_request *signaler =423			node_to_request(dep->signaler);424 425		/* Dependencies along the same timeline are expected. */426		if (signaler->timeline == rq->timeline)427			continue;428 429		if (__i915_request_is_complete(signaler))430			continue;431 432		i915_request_show(m, signaler, prefix, indent + 2);433	}434	rcu_read_unlock();435}436 437static void default_destroy(struct kref *kref)438{439	struct i915_sched_engine *sched_engine =440		container_of(kref, typeof(*sched_engine), ref);441 442	tasklet_kill(&sched_engine->tasklet); /* flush the callback */443	kfree(sched_engine);444}445 446static bool default_disabled(struct i915_sched_engine *sched_engine)447{448	return false;449}450 451struct i915_sched_engine *452i915_sched_engine_create(unsigned int subclass)453{454	struct i915_sched_engine *sched_engine;455 456	sched_engine = kzalloc(sizeof(*sched_engine), GFP_KERNEL);457	if (!sched_engine)458		return NULL;459 460	kref_init(&sched_engine->ref);461 462	sched_engine->queue = RB_ROOT_CACHED;463	sched_engine->queue_priority_hint = INT_MIN;464	sched_engine->destroy = default_destroy;465	sched_engine->disabled = default_disabled;466 467	INIT_LIST_HEAD(&sched_engine->requests);468	INIT_LIST_HEAD(&sched_engine->hold);469 470	spin_lock_init(&sched_engine->lock);471	lockdep_set_subclass(&sched_engine->lock, subclass);472 473	/*474	 * Due to an interesting quirk in lockdep's internal debug tracking,475	 * after setting a subclass we must ensure the lock is used. Otherwise,476	 * nr_unused_locks is incremented once too often.477	 */478#ifdef CONFIG_DEBUG_LOCK_ALLOC479	local_irq_disable();480	lock_map_acquire(&sched_engine->lock.dep_map);481	lock_map_release(&sched_engine->lock.dep_map);482	local_irq_enable();483#endif484 485	return sched_engine;486}487 488void i915_scheduler_module_exit(void)489{490	kmem_cache_destroy(slab_dependencies);491	kmem_cache_destroy(slab_priorities);492}493 494int __init i915_scheduler_module_init(void)495{496	slab_dependencies = KMEM_CACHE(i915_dependency,497					      SLAB_HWCACHE_ALIGN |498					      SLAB_TYPESAFE_BY_RCU);499	if (!slab_dependencies)500		return -ENOMEM;501 502	slab_priorities = KMEM_CACHE(i915_priolist, 0);503	if (!slab_priorities)504		goto err_priorities;505 506	return 0;507 508err_priorities:509	kmem_cache_destroy(slab_priorities);510	return -ENOMEM;511}512