diff options
Diffstat (limited to 'fs/aio.c')
| -rw-r--r-- | fs/aio.c | 2235 |
1 files changed, 1021 insertions, 1214 deletions
@@ -8,24 +8,25 @@ * * See ../COPYING for licensing terms. */ +#define pr_fmt(fmt) "%s: " fmt, __func__ + #include <linux/kernel.h> #include <linux/init.h> #include <linux/errno.h> #include <linux/time.h> #include <linux/aio_abi.h> -#include <linux/module.h> +#include <linux/export.h> #include <linux/syscalls.h> #include <linux/backing-dev.h> #include <linux/uio.h> -#define DEBUG 0 - #include <linux/sched.h> #include <linux/fs.h> #include <linux/file.h> #include <linux/mm.h> #include <linux/mman.h> #include <linux/mmu_context.h> +#include <linux/percpu.h> #include <linux/slab.h> #include <linux/timer.h> #include <linux/aio.h> @@ -34,18 +35,120 @@ #include <linux/security.h> #include <linux/eventfd.h> #include <linux/blkdev.h> -#include <linux/mempool.h> -#include <linux/hash.h> #include <linux/compat.h> +#include <linux/migrate.h> +#include <linux/ramfs.h> +#include <linux/percpu-refcount.h> +#include <linux/mount.h> #include <asm/kmap_types.h> #include <asm/uaccess.h> -#if DEBUG > 1 -#define dprintk printk -#else -#define dprintk(x...) do { ; } while (0) -#endif +#include "internal.h" + +#define AIO_RING_MAGIC 0xa10a10a1 +#define AIO_RING_COMPAT_FEATURES 1 +#define AIO_RING_INCOMPAT_FEATURES 0 +struct aio_ring { + unsigned id; /* kernel internal index number */ + unsigned nr; /* number of io_events */ + unsigned head; /* Written to by userland or under ring_lock + * mutex by aio_read_events_ring(). */ + unsigned tail; + + unsigned magic; + unsigned compat_features; + unsigned incompat_features; + unsigned header_length; /* size of aio_ring */ + + + struct io_event io_events[0]; +}; /* 128 bytes + ring size */ + +#define AIO_RING_PAGES 8 + +struct kioctx_table { + struct rcu_head rcu; + unsigned nr; + struct kioctx *table[]; +}; + +struct kioctx_cpu { + unsigned reqs_available; +}; + +struct kioctx { + struct percpu_ref users; + atomic_t dead; + + struct percpu_ref reqs; + + unsigned long user_id; + + struct __percpu kioctx_cpu *cpu; + + /* + * For percpu reqs_available, number of slots we move to/from global + * counter at a time: + */ + unsigned req_batch; + /* + * This is what userspace passed to io_setup(), it's not used for + * anything but counting against the global max_reqs quota. + * + * The real limit is nr_events - 1, which will be larger (see + * aio_setup_ring()) + */ + unsigned max_reqs; + + /* Size of ringbuffer, in units of struct io_event */ + unsigned nr_events; + + unsigned long mmap_base; + unsigned long mmap_size; + + struct page **ring_pages; + long nr_pages; + + struct work_struct free_work; + + /* + * signals when all in-flight requests are done + */ + struct completion *requests_done; + + struct { + /* + * This counts the number of available slots in the ringbuffer, + * so we avoid overflowing it: it's decremented (if positive) + * when allocating a kiocb and incremented when the resulting + * io_event is pulled off the ringbuffer. + * + * We batch accesses to it with a percpu version. + */ + atomic_t reqs_available; + } ____cacheline_aligned_in_smp; + + struct { + spinlock_t ctx_lock; + struct list_head active_reqs; /* used for cancellation */ + } ____cacheline_aligned_in_smp; + + struct { + struct mutex ring_lock; + wait_queue_head_t wait; + } ____cacheline_aligned_in_smp; + + struct { + unsigned tail; + spinlock_t completion_lock; + } ____cacheline_aligned_in_smp; + + struct page *internal_pages[AIO_RING_PAGES]; + struct file *aio_ring_file; + + unsigned id; +}; /*------ sysctl variables----*/ static DEFINE_SPINLOCK(aio_nr_lock); @@ -56,25 +159,51 @@ unsigned long aio_max_nr = 0x10000; /* system wide maximum number of aio request static struct kmem_cache *kiocb_cachep; static struct kmem_cache *kioctx_cachep; -static struct workqueue_struct *aio_wq; +static struct vfsmount *aio_mnt; -/* Used for rare fput completion. */ -static void aio_fput_routine(struct work_struct *); -static DECLARE_WORK(fput_work, aio_fput_routine); +static const struct file_operations aio_ring_fops; +static const struct address_space_operations aio_ctx_aops; -static DEFINE_SPINLOCK(fput_lock); -static LIST_HEAD(fput_head); +static struct file *aio_private_file(struct kioctx *ctx, loff_t nr_pages) +{ + struct qstr this = QSTR_INIT("[aio]", 5); + struct file *file; + struct path path; + struct inode *inode = alloc_anon_inode(aio_mnt->mnt_sb); + if (IS_ERR(inode)) + return ERR_CAST(inode); + + inode->i_mapping->a_ops = &aio_ctx_aops; + inode->i_mapping->private_data = ctx; + inode->i_size = PAGE_SIZE * nr_pages; + + path.dentry = d_alloc_pseudo(aio_mnt->mnt_sb, &this); + if (!path.dentry) { + iput(inode); + return ERR_PTR(-ENOMEM); + } + path.mnt = mntget(aio_mnt); -#define AIO_BATCH_HASH_BITS 3 /* allocated on-stack, so don't go crazy */ -#define AIO_BATCH_HASH_SIZE (1 << AIO_BATCH_HASH_BITS) -struct aio_batch_entry { - struct hlist_node list; - struct address_space *mapping; -}; -mempool_t *abe_pool; + d_instantiate(path.dentry, inode); + file = alloc_file(&path, FMODE_READ | FMODE_WRITE, &aio_ring_fops); + if (IS_ERR(file)) { + path_put(&path); + return file; + } + + file->f_flags = O_RDWR; + file->private_data = ctx; + return file; +} -static void aio_kick_handler(struct work_struct *); -static void aio_queue_work(struct kioctx *); +static struct dentry *aio_mount(struct file_system_type *fs_type, + int flags, const char *dev_name, void *data) +{ + static const struct dentry_operations ops = { + .d_dname = simple_dname, + }; + return mount_pseudo(fs_type, "aio:", NULL, &ops, 0xa10a10a1); +} /* aio_setup * Creates the slab caches used by the aio routines, panic on @@ -82,181 +211,432 @@ static void aio_queue_work(struct kioctx *); */ static int __init aio_setup(void) { + static struct file_system_type aio_fs = { + .name = "aio", + .mount = aio_mount, + .kill_sb = kill_anon_super, + }; + aio_mnt = kern_mount(&aio_fs); + if (IS_ERR(aio_mnt)) + panic("Failed to create aio fs mount."); + kiocb_cachep = KMEM_CACHE(kiocb, SLAB_HWCACHE_ALIGN|SLAB_PANIC); kioctx_cachep = KMEM_CACHE(kioctx,SLAB_HWCACHE_ALIGN|SLAB_PANIC); - aio_wq = create_workqueue("aio"); - abe_pool = mempool_create_kmalloc_pool(1, sizeof(struct aio_batch_entry)); - BUG_ON(!abe_pool); - - pr_debug("aio_setup: sizeof(struct page) = %d\n", (int)sizeof(struct page)); + pr_debug("sizeof(struct page) = %zu\n", sizeof(struct page)); return 0; } __initcall(aio_setup); +static void put_aio_ring_file(struct kioctx *ctx) +{ + struct file *aio_ring_file = ctx->aio_ring_file; + if (aio_ring_file) { + truncate_setsize(aio_ring_file->f_inode, 0); + + /* Prevent further access to the kioctx from migratepages */ + spin_lock(&aio_ring_file->f_inode->i_mapping->private_lock); + aio_ring_file->f_inode->i_mapping->private_data = NULL; + ctx->aio_ring_file = NULL; + spin_unlock(&aio_ring_file->f_inode->i_mapping->private_lock); + + fput(aio_ring_file); + } +} + static void aio_free_ring(struct kioctx *ctx) { - struct aio_ring_info *info = &ctx->ring_info; - long i; + int i; - for (i=0; i<info->nr_pages; i++) - put_page(info->ring_pages[i]); + /* Disconnect the kiotx from the ring file. This prevents future + * accesses to the kioctx from page migration. + */ + put_aio_ring_file(ctx); + + for (i = 0; i < ctx->nr_pages; i++) { + struct page *page; + pr_debug("pid(%d) [%d] page->count=%d\n", current->pid, i, + page_count(ctx->ring_pages[i])); + page = ctx->ring_pages[i]; + if (!page) + continue; + ctx->ring_pages[i] = NULL; + put_page(page); + } - if (info->mmap_size) { - down_write(&ctx->mm->mmap_sem); - do_munmap(ctx->mm, info->mmap_base, info->mmap_size); - up_write(&ctx->mm->mmap_sem); + if (ctx->ring_pages && ctx->ring_pages != ctx->internal_pages) { + kfree(ctx->ring_pages); + ctx->ring_pages = NULL; } +} - if (info->ring_pages && info->ring_pages != info->internal_pages) - kfree(info->ring_pages); - info->ring_pages = NULL; - info->nr = 0; +static int aio_ring_mmap(struct file *file, struct vm_area_struct *vma) +{ + vma->vm_ops = &generic_file_vm_ops; + return 0; +} + +static const struct file_operations aio_ring_fops = { + .mmap = aio_ring_mmap, +}; + +static int aio_set_page_dirty(struct page *page) +{ + return 0; } +#if IS_ENABLED(CONFIG_MIGRATION) +static int aio_migratepage(struct address_space *mapping, struct page *new, + struct page *old, enum migrate_mode mode) +{ + struct kioctx *ctx; + unsigned long flags; + pgoff_t idx; + int rc; + + rc = 0; + + /* mapping->private_lock here protects against the kioctx teardown. */ + spin_lock(&mapping->private_lock); + ctx = mapping->private_data; + if (!ctx) { + rc = -EINVAL; + goto out; + } + + /* The ring_lock mutex. The prevents aio_read_events() from writing + * to the ring's head, and prevents page migration from mucking in + * a partially initialized kiotx. + */ + if (!mutex_trylock(&ctx->ring_lock)) { + rc = -EAGAIN; + goto out; + } + + idx = old->index; + if (idx < (pgoff_t)ctx->nr_pages) { + /* Make sure the old page hasn't already been changed */ + if (ctx->ring_pages[idx] != old) + rc = -EAGAIN; + } else + rc = -EINVAL; + + if (rc != 0) + goto out_unlock; + + /* Writeback must be complete */ + BUG_ON(PageWriteback(old)); + get_page(new); + + rc = migrate_page_move_mapping(mapping, new, old, NULL, mode, 1); + if (rc != MIGRATEPAGE_SUCCESS) { + put_page(new); + goto out_unlock; + } + + /* Take completion_lock to prevent other writes to the ring buffer + * while the old page is copied to the new. This prevents new + * events from being lost. + */ + spin_lock_irqsave(&ctx->completion_lock, flags); + migrate_page_copy(new, old); + BUG_ON(ctx->ring_pages[idx] != old); + ctx->ring_pages[idx] = new; + spin_unlock_irqrestore(&ctx->completion_lock, flags); + + /* The old page is no longer accessible. */ + put_page(old); + +out_unlock: + mutex_unlock(&ctx->ring_lock); +out: + spin_unlock(&mapping->private_lock); + return rc; +} +#endif + +static const struct address_space_operations aio_ctx_aops = { + .set_page_dirty = aio_set_page_dirty, +#if IS_ENABLED(CONFIG_MIGRATION) + .migratepage = aio_migratepage, +#endif +}; + static int aio_setup_ring(struct kioctx *ctx) { struct aio_ring *ring; - struct aio_ring_info *info = &ctx->ring_info; unsigned nr_events = ctx->max_reqs; - unsigned long size; + struct mm_struct *mm = current->mm; + unsigned long size, unused; int nr_pages; + int i; + struct file *file; /* Compensate for the ring buffer's head/tail overlap entry */ nr_events += 2; /* 1 is required, 2 for good luck */ size = sizeof(struct aio_ring); size += sizeof(struct io_event) * nr_events; - nr_pages = (size + PAGE_SIZE-1) >> PAGE_SHIFT; + nr_pages = PFN_UP(size); if (nr_pages < 0) return -EINVAL; - nr_events = (PAGE_SIZE * nr_pages - sizeof(struct aio_ring)) / sizeof(struct io_event); + file = aio_private_file(ctx, nr_pages); + if (IS_ERR(file)) { + ctx->aio_ring_file = NULL; + return -ENOMEM; + } + + ctx->aio_ring_file = file; + nr_events = (PAGE_SIZE * nr_pages - sizeof(struct aio_ring)) + / sizeof(struct io_event); - info->nr = 0; - info->ring_pages = info->internal_pages; + ctx->ring_pages = ctx->internal_pages; if (nr_pages > AIO_RING_PAGES) { - info->ring_pages = kcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL); - if (!info->ring_pages) + ctx->ring_pages = kcalloc(nr_pages, sizeof(struct page *), + GFP_KERNEL); + if (!ctx->ring_pages) { + put_aio_ring_file(ctx); return -ENOMEM; + } + } + + for (i = 0; i < nr_pages; i++) { + struct page *page; + page = find_or_create_page(file->f_inode->i_mapping, + i, GFP_HIGHUSER | __GFP_ZERO); + if (!page) + break; + pr_debug("pid(%d) page[%d]->count=%d\n", + current->pid, i, page_count(page)); + SetPageUptodate(page); + SetPageDirty(page); + unlock_page(page); + + ctx->ring_pages[i] = page; } + ctx->nr_pages = i; - info->mmap_size = nr_pages * PAGE_SIZE; - dprintk("attempting mmap of %lu bytes\n", info->mmap_size); - down_write(&ctx->mm->mmap_sem); - info->mmap_base = do_mmap(NULL, 0, info->mmap_size, - PROT_READ|PROT_WRITE, MAP_ANONYMOUS|MAP_PRIVATE, - 0); - if (IS_ERR((void *)info->mmap_base)) { - up_write(&ctx->mm->mmap_sem); - info->mmap_size = 0; + if (unlikely(i != nr_pages)) { aio_free_ring(ctx); - return -EAGAIN; + return -ENOMEM; } - dprintk("mmap address: 0x%08lx\n", info->mmap_base); - info->nr_pages = get_user_pages(current, ctx->mm, - info->mmap_base, nr_pages, - 1, 0, info->ring_pages, NULL); - up_write(&ctx->mm->mmap_sem); + ctx->mmap_size = nr_pages * PAGE_SIZE; + pr_debug("attempting mmap of %lu bytes\n", ctx->mmap_size); - if (unlikely(info->nr_pages != nr_pages)) { + down_write(&mm->mmap_sem); + ctx->mmap_base = do_mmap_pgoff(ctx->aio_ring_file, 0, ctx->mmap_size, + PROT_READ | PROT_WRITE, + MAP_SHARED, 0, &unused); + up_write(&mm->mmap_sem); + if (IS_ERR((void *)ctx->mmap_base)) { + ctx->mmap_size = 0; aio_free_ring(ctx); - return -EAGAIN; + return -ENOMEM; } - ctx->user_id = info->mmap_base; + pr_debug("mmap address: 0x%08lx\n", ctx->mmap_base); - info->nr = nr_events; /* trusted copy */ + ctx->user_id = ctx->mmap_base; + ctx->nr_events = nr_events; /* trusted copy */ - ring = kmap_atomic(info->ring_pages[0], KM_USER0); + ring = kmap_atomic(ctx->ring_pages[0]); ring->nr = nr_events; /* user copy */ - ring->id = ctx->user_id; + ring->id = ~0U; ring->head = ring->tail = 0; ring->magic = AIO_RING_MAGIC; ring->compat_features = AIO_RING_COMPAT_FEATURES; ring->incompat_features = AIO_RING_INCOMPAT_FEATURES; ring->header_length = sizeof(struct aio_ring); - kunmap_atomic(ring, KM_USER0); + kunmap_atomic(ring); + flush_dcache_page(ctx->ring_pages[0]); return 0; } - -/* aio_ring_event: returns a pointer to the event at the given index from - * kmap_atomic(, km). Release the pointer with put_aio_ring_event(); - */ #define AIO_EVENTS_PER_PAGE (PAGE_SIZE / sizeof(struct io_event)) #define AIO_EVENTS_FIRST_PAGE ((PAGE_SIZE - sizeof(struct aio_ring)) / sizeof(struct io_event)) #define AIO_EVENTS_OFFSET (AIO_EVENTS_PER_PAGE - AIO_EVENTS_FIRST_PAGE) -#define aio_ring_event(info, nr, km) ({ \ - unsigned pos = (nr) + AIO_EVENTS_OFFSET; \ - struct io_event *__event; \ - __event = kmap_atomic( \ - (info)->ring_pages[pos / AIO_EVENTS_PER_PAGE], km); \ - __event += pos % AIO_EVENTS_PER_PAGE; \ - __event; \ -}) - -#define put_aio_ring_event(event, km) do { \ - struct io_event *__event = (event); \ - (void)__event; \ - kunmap_atomic((void *)((unsigned long)__event & PAGE_MASK), km); \ -} while(0) - -static void ctx_rcu_free(struct rcu_head *head) +void kiocb_set_cancel_fn(struct kiocb *req, kiocb_cancel_fn *cancel) { - struct kioctx *ctx = container_of(head, struct kioctx, rcu_head); - unsigned nr_events = ctx->max_reqs; + struct kioctx *ctx = req->ki_ctx; + unsigned long flags; + + spin_lock_irqsave(&ctx->ctx_lock, flags); + + if (!req->ki_list.next) + list_add(&req->ki_list, &ctx->active_reqs); + + req->ki_cancel = cancel; + spin_unlock_irqrestore(&ctx->ctx_lock, flags); +} +EXPORT_SYMBOL(kiocb_set_cancel_fn); + +static int kiocb_cancel(struct kiocb *kiocb) +{ + kiocb_cancel_fn *old, *cancel; + + /* + * Don't want to set kiocb->ki_cancel = KIOCB_CANCELLED unless it + * actually has a cancel function, hence the cmpxchg() + */ + + cancel = ACCESS_ONCE(kiocb->ki_cancel); + do { + if (!cancel || cancel == KIOCB_CANCELLED) + return -EINVAL; + + old = cancel; + cancel = cmpxchg(&kiocb->ki_cancel, old, KIOCB_CANCELLED); + } while (cancel != old); + + return cancel(kiocb); +} + +static void free_ioctx(struct work_struct *work) +{ + struct kioctx *ctx = container_of(work, struct kioctx, free_work); + + pr_debug("freeing %p\n", ctx); + + aio_free_ring(ctx); + free_percpu(ctx->cpu); kmem_cache_free(kioctx_cachep, ctx); +} - if (nr_events) { - spin_lock(&aio_nr_lock); - BUG_ON(aio_nr - nr_events > aio_nr); - aio_nr -= nr_events; - spin_unlock(&aio_nr_lock); - } +static void free_ioctx_reqs(struct percpu_ref *ref) +{ + struct kioctx *ctx = container_of(ref, struct kioctx, reqs); + + /* At this point we know that there are no any in-flight requests */ + if (ctx->requests_done) + complete(ctx->requests_done); + + INIT_WORK(&ctx->free_work, free_ioctx); + schedule_work(&ctx->free_work); } -/* __put_ioctx - * Called when the last user of an aio context has gone away, - * and the struct needs to be freed. +/* + * When this function runs, the kioctx has been removed from the "hash table" + * and ctx->users has dropped to 0, so we know no more kiocbs can be submitted - + * now it's safe to cancel any that need to be. */ -static void __put_ioctx(struct kioctx *ctx) +static void free_ioctx_users(struct percpu_ref *ref) { - BUG_ON(ctx->reqs_active); + struct kioctx *ctx = container_of(ref, struct kioctx, users); + struct kiocb *req; - cancel_delayed_work(&ctx->wq); - cancel_work_sync(&ctx->wq.work); - aio_free_ring(ctx); - mmdrop(ctx->mm); - ctx->mm = NULL; - pr_debug("__put_ioctx: freeing %p\n", ctx); - call_rcu(&ctx->rcu_head, ctx_rcu_free); + spin_lock_irq(&ctx->ctx_lock); + + while (!list_empty(&ctx->active_reqs)) { + req = list_first_entry(&ctx->active_reqs, + struct kiocb, ki_list); + + list_del_init(&req->ki_list); + kiocb_cancel(req); + } + + spin_unlock_irq(&ctx->ctx_lock); + + percpu_ref_kill(&ctx->reqs); + percpu_ref_put(&ctx->reqs); } -#define get_ioctx(kioctx) do { \ - BUG_ON(atomic_read(&(kioctx)->users) <= 0); \ - atomic_inc(&(kioctx)->users); \ -} while (0) -#define put_ioctx(kioctx) do { \ - BUG_ON(atomic_read(&(kioctx)->users) <= 0); \ - if (unlikely(atomic_dec_and_test(&(kioctx)->users))) \ - __put_ioctx(kioctx); \ -} while (0) +static int ioctx_add_table(struct kioctx *ctx, struct mm_struct *mm) +{ + unsigned i, new_nr; + struct kioctx_table *table, *old; + struct aio_ring *ring; + + spin_lock(&mm->ioctx_lock); + rcu_read_lock(); + table = rcu_dereference(mm->ioctx_table); + + while (1) { + if (table) + for (i = 0; i < table->nr; i++) + if (!table->table[i]) { + ctx->id = i; + table->table[i] = ctx; + rcu_read_unlock(); + spin_unlock(&mm->ioctx_lock); + + /* While kioctx setup is in progress, + * we are protected from page migration + * changes ring_pages by ->ring_lock. + */ + ring = kmap_atomic(ctx->ring_pages[0]); + ring->id = ctx->id; + kunmap_atomic(ring); + return 0; + } + + new_nr = (table ? table->nr : 1) * 4; + + rcu_read_unlock(); + spin_unlock(&mm->ioctx_lock); + + table = kzalloc(sizeof(*table) + sizeof(struct kioctx *) * + new_nr, GFP_KERNEL); + if (!table) + return -ENOMEM; + + table->nr = new_nr; + + spin_lock(&mm->ioctx_lock); + rcu_read_lock(); + old = rcu_dereference(mm->ioctx_table); + + if (!old) { + rcu_assign_pointer(mm->ioctx_table, table); + } else if (table->nr > old->nr) { + memcpy(table->table, old->table, + old->nr * sizeof(struct kioctx *)); + + rcu_assign_pointer(mm->ioctx_table, table); + kfree_rcu(old, rcu); + } else { + kfree(table); + table = old; + } + } +} + +static void aio_nr_sub(unsigned nr) +{ + spin_lock(&aio_nr_lock); + if (WARN_ON(aio_nr - nr > aio_nr)) + aio_nr = 0; + else + aio_nr -= nr; + spin_unlock(&aio_nr_lock); +} /* ioctx_alloc * Allocates and initializes an ioctx. Returns an ERR_PTR if it failed. */ static struct kioctx *ioctx_alloc(unsigned nr_events) { - struct mm_struct *mm; + struct mm_struct *mm = current->mm; struct kioctx *ctx; - int did_sync = 0; + int err = -ENOMEM; + + /* + * We keep track of the number of available ringbuffer slots, to prevent + * overflow (reqs_available), and we also use percpu counters for this. + * + * So since up to half the slots might be on other cpu's percpu counters + * and unavailable, double nr_events so userspace sees what they + * expected: additionally, we move req_batch slots to/from percpu + * counters at a time, so make sure that isn't 0: + */ + nr_events = max(nr_events, num_possible_cpus() * 4); + nr_events *= 2; /* Prevent overflows */ if ((nr_events > (0x10000000U / sizeof(struct io_event))) || @@ -265,7 +645,7 @@ static struct kioctx *ioctx_alloc(unsigned nr_events) return ERR_PTR(-EINVAL); } - if ((unsigned long)nr_events > aio_max_nr) + if (!nr_events || (unsigned long)nr_events > (aio_max_nr * 2UL)) return ERR_PTR(-EAGAIN); ctx = kmem_cache_zalloc(kioctx_cachep, GFP_KERNEL); @@ -273,651 +653,303 @@ static struct kioctx *ioctx_alloc(unsigned nr_events) return ERR_PTR(-ENOMEM); ctx->max_reqs = nr_events; - mm = ctx->mm = current->mm; - atomic_inc(&mm->mm_count); - atomic_set(&ctx->users, 1); spin_lock_init(&ctx->ctx_lock); - spin_lock_init(&ctx->ring_info.ring_lock); + spin_lock_init(&ctx->completion_lock); + mutex_init(&ctx->ring_lock); + /* Protect against page migration throughout kiotx setup by keeping + * the ring_lock mutex held until setup is complete. */ + mutex_lock(&ctx->ring_lock); init_waitqueue_head(&ctx->wait); INIT_LIST_HEAD(&ctx->active_reqs); - INIT_LIST_HEAD(&ctx->run_list); - INIT_DELAYED_WORK(&ctx->wq, aio_kick_handler); - if (aio_setup_ring(ctx) < 0) - goto out_freectx; + if (percpu_ref_init(&ctx->users, free_ioctx_users)) + goto err; - /* limit the number of system wide aios */ - do { - spin_lock_bh(&aio_nr_lock); - if (aio_nr + nr_events > aio_max_nr || - aio_nr + nr_events < aio_nr) - ctx->max_reqs = 0; - else - aio_nr += ctx->max_reqs; - spin_unlock_bh(&aio_nr_lock); - if (ctx->max_reqs || did_sync) - break; + if (percpu_ref_init(&ctx->reqs, free_ioctx_reqs)) + goto err; - /* wait for rcu callbacks to have completed before giving up */ - synchronize_rcu(); - did_sync = 1; - ctx->max_reqs = nr_events; - } while (1); + ctx->cpu = alloc_percpu(struct kioctx_cpu); + if (!ctx->cpu) + goto err; - if (ctx->max_reqs == 0) - goto out_cleanup; + err = aio_setup_ring(ctx); + if (err < 0) + goto err; - /* now link into global list. */ - spin_lock(&mm->ioctx_lock); - hlist_add_head_rcu(&ctx->list, &mm->ioctx_list); - spin_unlock(&mm->ioctx_lock); + atomic_set(&ctx->reqs_available, ctx->nr_events - 1); + ctx->req_batch = (ctx->nr_events - 1) / (num_possible_cpus() * 4); + if (ctx->req_batch < 1) + ctx->req_batch = 1; - dprintk("aio: allocated ioctx %p[%ld]: mm=%p mask=0x%x\n", - ctx, ctx->user_id, current->mm, ctx->ring_info.nr); - return ctx; + /* limit the number of system wide aios */ + spin_lock(&aio_nr_lock); + if (aio_nr + nr_events > (aio_max_nr * 2UL) || + aio_nr + nr_events < aio_nr) { + spin_unlock(&aio_nr_lock); + err = -EAGAIN; + goto err_ctx; + } + aio_nr += ctx->max_reqs; + spin_unlock(&aio_nr_lock); -out_cleanup: - __put_ioctx(ctx); - return ERR_PTR(-EAGAIN); + percpu_ref_get(&ctx->users); /* io_setup() will drop this ref */ + percpu_ref_get(&ctx->reqs); /* free_ioctx_users() will drop this */ -out_freectx: - mmdrop(mm); - kmem_cache_free(kioctx_cachep, ctx); - ctx = ERR_PTR(-ENOMEM); + err = ioctx_add_table(ctx, mm); + if (err) + goto err_cleanup; + + /* Release the ring_lock mutex now that all setup is complete. */ + mutex_unlock(&ctx->ring_lock); - dprintk("aio: error allocating ioctx %p\n", ctx); + pr_debug("allocated ioctx %p[%ld]: mm=%p mask=0x%x\n", + ctx, ctx->user_id, mm, ctx->nr_events); return ctx; + +err_cleanup: + aio_nr_sub(ctx->max_reqs); +err_ctx: + aio_free_ring(ctx); +err: + mutex_unlock(&ctx->ring_lock); + free_percpu(ctx->cpu); + free_percpu(ctx->reqs.pcpu_count); + free_percpu(ctx->users.pcpu_count); + kmem_cache_free(kioctx_cachep, ctx); + pr_debug("error allocating ioctx %d\n", err); + return ERR_PTR(err); } -/* aio_cancel_all - * Cancels all outstanding aio requests on an aio context. Used - * when the processes owning a context have all exited to encourage +/* kill_ioctx + * Cancels all outstanding aio requests on an aio context. Used + * when the processes owning a context have all exited to encourage * the rapid destruction of the kioctx. */ -static void aio_cancel_all(struct kioctx *ctx) +static int kill_ioctx(struct mm_struct *mm, struct kioctx *ctx, + struct completion *requests_done) { - int (*cancel)(struct kiocb *, struct io_event *); - struct io_event res; - spin_lock_irq(&ctx->ctx_lock); - ctx->dead = 1; - while (!list_empty(&ctx->active_reqs)) { - struct list_head *pos = ctx->active_reqs.next; - struct kiocb *iocb = list_kiocb(pos); - list_del_init(&iocb->ki_list); - cancel = iocb->ki_cancel; - kiocbSetCancelled(iocb); - if (cancel) { - iocb->ki_users++; - spin_unlock_irq(&ctx->ctx_lock); - cancel(iocb, &res); - spin_lock_irq(&ctx->ctx_lock); - } - } - spin_unlock_irq(&ctx->ctx_lock); -} + struct kioctx_table *table; -static void wait_for_all_aios(struct kioctx *ctx) -{ - struct task_struct *tsk = current; - DECLARE_WAITQUEUE(wait, tsk); + if (atomic_xchg(&ctx->dead, 1)) + return -EINVAL; - spin_lock_irq(&ctx->ctx_lock); - if (!ctx->reqs_active) - goto out; - add_wait_queue(&ctx->wait, &wait); - set_task_state(tsk, TASK_UNINTERRUPTIBLE); - while (ctx->reqs_active) { - spin_unlock_irq(&ctx->ctx_lock); - io_schedule(); - set_task_state(tsk, TASK_UNINTERRUPTIBLE); - spin_lock_irq(&ctx->ctx_lock); - } - __set_task_state(tsk, TASK_RUNNING); - remove_wait_queue(&ctx->wait, &wait); + spin_lock(&mm->ioctx_lock); + rcu_read_lock(); + table = rcu_dereference(mm->ioctx_table); -out: - spin_unlock_irq(&ctx->ctx_lock); + WARN_ON(ctx != table->table[ctx->id]); + table->table[ctx->id] = NULL; + rcu_read_unlock(); + spin_unlock(&mm->ioctx_lock); + + /* percpu_ref_kill() will do the necessary call_rcu() */ + wake_up_all(&ctx->wait); + + /* + * It'd be more correct to do this in free_ioctx(), after all + * the outstanding kiocbs have finished - but by then io_destroy + * has already returned, so io_setup() could potentially return + * -EAGAIN with no ioctxs actually in use (as far as userspace + * could tell). + */ + aio_nr_sub(ctx->max_reqs); + + if (ctx->mmap_size) + vm_munmap(ctx->mmap_base, ctx->mmap_size); + + ctx->requests_done = requests_done; + percpu_ref_kill(&ctx->users); + return 0; } /* wait_on_sync_kiocb: * Waits on the given sync kiocb to complete. */ -ssize_t wait_on_sync_kiocb(struct kiocb *iocb) +ssize_t wait_on_sync_kiocb(struct kiocb *req) { - while (iocb->ki_users) { + while (!req->ki_ctx) { set_current_state(TASK_UNINTERRUPTIBLE); - if (!iocb->ki_users) + if (req->ki_ctx) break; io_schedule(); } __set_current_state(TASK_RUNNING); - return iocb->ki_user_data; + return req->ki_user_data; } EXPORT_SYMBOL(wait_on_sync_kiocb); -/* exit_aio: called when the last user of mm goes away. At this point, - * there is no way for any new requests to be submited or any of the - * io_* syscalls to be called on the context. However, there may be - * outstanding requests which hold references to the context; as they - * go away, they will call put_ioctx and release any pinned memory - * associated with the request (held via struct page * references). +/* + * exit_aio: called when the last user of mm goes away. At this point, there is + * no way for any new requests to be submited or any of the io_* syscalls to be + * called on the context. + * + * There may be outstanding kiocbs, but free_ioctx() will explicitly wait on + * them. */ void exit_aio(struct mm_struct *mm) { + struct kioctx_table *table; struct kioctx *ctx; + unsigned i = 0; - while (!hlist_empty(&mm->ioctx_list)) { - ctx = hlist_entry(mm->ioctx_list.first, struct kioctx, list); - hlist_del_rcu(&ctx->list); + while (1) { + rcu_read_lock(); + table = rcu_dereference(mm->ioctx_table); - aio_cancel_all(ctx); + do { + if (!table || i >= table->nr) { + rcu_read_unlock(); + rcu_assign_pointer(mm->ioctx_table, NULL); + if (table) + kfree(table); + return; + } + + ctx = table->table[i++]; + } while (!ctx); + + rcu_read_unlock(); - wait_for_all_aios(ctx); /* - * Ensure we don't leave the ctx on the aio_wq + * We don't need to bother with munmap() here - + * exit_mmap(mm) is coming and it'll unmap everything. + * Since aio_free_ring() uses non-zero ->mmap_size + * as indicator that it needs to unmap the area, + * just set it to 0; aio_free_ring() is the only + * place that uses ->mmap_size, so it's safe. */ - cancel_work_sync(&ctx->wq.work); - - if (1 != atomic_read(&ctx->users)) - printk(KERN_DEBUG - "exit_aio:ioctx still alive: %d %d %d\n", - atomic_read(&ctx->users), ctx->dead, - ctx->reqs_active); - put_ioctx(ctx); + ctx->mmap_size = 0; + + kill_ioctx(mm, ctx, NULL); } } -/* aio_get_req - * Allocate a slot for an aio request. Increments the users count - * of the kioctx so that the kioctx stays around until all requests are - * complete. Returns NULL if no requests are free. - * - * Returns with kiocb->users set to 2. The io submit code path holds - * an extra reference while submitting the i/o. - * This prevents races between the aio code path referencing the - * req (after submitting it) and aio_complete() freeing the req. - */ -static struct kiocb *__aio_get_req(struct kioctx *ctx) +static void put_reqs_available(struct kioctx *ctx, unsigned nr) { - struct kiocb *req = NULL; - struct aio_ring *ring; - int okay = 0; + struct kioctx_cpu *kcpu; + unsigned long flags; - req = kmem_cache_alloc(kiocb_cachep, GFP_KERNEL); - if (unlikely(!req)) - return NULL; + preempt_disable(); + kcpu = this_cpu_ptr(ctx->cpu); - req->ki_flags = 0; - req->ki_users = 2; - req->ki_key = 0; - req->ki_ctx = ctx; - req->ki_cancel = NULL; - req->ki_retry = NULL; - req->ki_dtor = NULL; - req->private = NULL; - req->ki_iovec = NULL; - INIT_LIST_HEAD(&req->ki_run_list); - req->ki_eventfd = NULL; - - /* Check if the completion queue has enough free space to - * accept an event from this io. - */ - spin_lock_irq(&ctx->ctx_lock); - ring = kmap_atomic(ctx->ring_info.ring_pages[0], KM_USER0); - if (ctx->reqs_active < aio_ring_avail(&ctx->ring_info, ring)) { - list_add(&req->ki_list, &ctx->active_reqs); - ctx->reqs_active++; - okay = 1; - } - kunmap_atomic(ring, KM_USER0); - spin_unlock_irq(&ctx->ctx_lock); + local_irq_save(flags); + kcpu->reqs_available += nr; - if (!okay) { - kmem_cache_free(kiocb_cachep, req); - req = NULL; + while (kcpu->reqs_available >= ctx->req_batch * 2) { + kcpu->reqs_available -= ctx->req_batch; + atomic_add(ctx->req_batch, &ctx->reqs_available); } - return req; -} - -static inline struct kiocb *aio_get_req(struct kioctx *ctx) -{ - struct kiocb *req; - /* Handle a potential starvation case -- should be exceedingly rare as - * requests will be stuck on fput_head only if the aio_fput_routine is - * delayed and the requests were the last user of the struct file. - */ - req = __aio_get_req(ctx); - if (unlikely(NULL == req)) { - aio_fput_routine(NULL); - req = __aio_get_req(ctx); - } - return req; + local_irq_restore(flags); + preempt_enable(); } -static inline void really_put_req(struct kioctx *ctx, struct kiocb *req) +static bool get_reqs_available(struct kioctx *ctx) { - assert_spin_locked(&ctx->ctx_lock); - - if (req->ki_eventfd != NULL) - eventfd_ctx_put(req->ki_eventfd); - if (req->ki_dtor) - req->ki_dtor(req); - if (req->ki_iovec != &req->ki_inline_vec) - kfree(req->ki_iovec); - kmem_cache_free(kiocb_cachep, req); - ctx->reqs_active--; - - if (unlikely(!ctx->reqs_active && ctx->dead)) - wake_up(&ctx->wait); -} + struct kioctx_cpu *kcpu; + bool ret = false; + unsigned long flags; -static void aio_fput_routine(struct work_struct *data) -{ - spin_lock_irq(&fput_lock); - while (likely(!list_empty(&fput_head))) { - struct kiocb *req = list_kiocb(fput_head.next); - struct kioctx *ctx = req->ki_ctx; + preempt_disable(); + kcpu = this_cpu_ptr(ctx->cpu); - list_del(&req->ki_list); - spin_unlock_irq(&fput_lock); + local_irq_save(flags); + if (!kcpu->reqs_available) { + int old, avail = atomic_read(&ctx->reqs_available); - /* Complete the fput(s) */ - if (req->ki_filp != NULL) - fput(req->ki_filp); + do { + if (avail < ctx->req_batch) + goto out; - /* Link the iocb into the context's free list */ - spin_lock_irq(&ctx->ctx_lock); - really_put_req(ctx, req); - spin_unlock_irq(&ctx->ctx_lock); + old = avail; + avail = atomic_cmpxchg(&ctx->reqs_available, + avail, avail - ctx->req_batch); + } while (avail != old); - put_ioctx(ctx); - spin_lock_irq(&fput_lock); + kcpu->reqs_available += ctx->req_batch; } - spin_unlock_irq(&fput_lock); + + ret = true; + kcpu->reqs_available--; +out: + local_irq_restore(flags); + preempt_enable(); + return ret; } -/* __aio_put_req - * Returns true if this put was the last user of the request. +/* aio_get_req + * Allocate a slot for an aio request. + * Returns NULL if no requests are free. */ -static int __aio_put_req(struct kioctx *ctx, struct kiocb *req) +static inline struct kiocb *aio_get_req(struct kioctx *ctx) { - dprintk(KERN_DEBUG "aio_put(%p): f_count=%ld\n", - req, atomic_long_read(&req->ki_filp->f_count)); + struct kiocb *req; - assert_spin_locked(&ctx->ctx_lock); + if (!get_reqs_available(ctx)) + return NULL; - req->ki_users--; - BUG_ON(req->ki_users < 0); - if (likely(req->ki_users)) - return 0; - list_del(&req->ki_list); /* remove from active_reqs */ - req->ki_cancel = NULL; - req->ki_retry = NULL; + req = kmem_cache_alloc(kiocb_cachep, GFP_KERNEL|__GFP_ZERO); + if (unlikely(!req)) + goto out_put; - /* - * Try to optimize the aio and eventfd file* puts, by avoiding to - * schedule work in case it is not final fput() time. In normal cases, - * we would not be holding the last reference to the file*, so - * this function will be executed w/out any aio kthread wakeup. - */ - if (unlikely(!fput_atomic(req->ki_filp))) { - get_ioctx(ctx); - spin_lock(&fput_lock); - list_add(&req->ki_list, &fput_head); - spin_unlock(&fput_lock); - queue_work(aio_wq, &fput_work); - } else { - req->ki_filp = NULL; - really_put_req(ctx, req); - } - return 1; + percpu_ref_get(&ctx->reqs); + + req->ki_ctx = ctx; + return req; +out_put: + put_reqs_available(ctx, 1); + return NULL; } -/* aio_put_req - * Returns true if this put was the last user of the kiocb, - * false if the request is still in use. - */ -int aio_put_req(struct kiocb *req) +static void kiocb_free(struct kiocb *req) { - struct kioctx *ctx = req->ki_ctx; - int ret; - spin_lock_irq(&ctx->ctx_lock); - ret = __aio_put_req(ctx, req); - spin_unlock_irq(&ctx->ctx_lock); - return ret; + if (req->ki_filp) + fput(req->ki_filp); + if (req->ki_eventfd != NULL) + eventfd_ctx_put(req->ki_eventfd); + kmem_cache_free(kiocb_cachep, req); } -EXPORT_SYMBOL(aio_put_req); static struct kioctx *lookup_ioctx(unsigned long ctx_id) { + struct aio_ring __user *ring = (void __user *)ctx_id; struct mm_struct *mm = current->mm; struct kioctx *ctx, *ret = NULL; - struct hlist_node *n; - - rcu_read_lock(); - - hlist_for_each_entry_rcu(ctx, n, &mm->ioctx_list, list) { - if (ctx->user_id == ctx_id && !ctx->dead) { - get_ioctx(ctx); - ret = ctx; - break; - } - } - - rcu_read_unlock(); - return ret; -} - -/* - * Queue up a kiocb to be retried. Assumes that the kiocb - * has already been marked as kicked, and places it on - * the retry run list for the corresponding ioctx, if it - * isn't already queued. Returns 1 if it actually queued - * the kiocb (to tell the caller to activate the work - * queue to process it), or 0, if it found that it was - * already queued. - */ -static inline int __queue_kicked_iocb(struct kiocb *iocb) -{ - struct kioctx *ctx = iocb->ki_ctx; - - assert_spin_locked(&ctx->ctx_lock); + struct kioctx_table *table; + unsigned id; - if (list_empty(&iocb->ki_run_list)) { - list_add_tail(&iocb->ki_run_list, - &ctx->run_list); - return 1; - } - return 0; -} - -/* aio_run_iocb - * This is the core aio execution routine. It is - * invoked both for initial i/o submission and - * subsequent retries via the aio_kick_handler. - * Expects to be invoked with iocb->ki_ctx->lock - * already held. The lock is released and reacquired - * as needed during processing. - * - * Calls the iocb retry method (already setup for the - * iocb on initial submission) for operation specific - * handling, but takes care of most of common retry - * execution details for a given iocb. The retry method - * needs to be non-blocking as far as possible, to avoid - * holding up other iocbs waiting to be serviced by the - * retry kernel thread. - * - * The trickier parts in this code have to do with - * ensuring that only one retry instance is in progress - * for a given iocb at any time. Providing that guarantee - * simplifies the coding of individual aio operations as - * it avoids various potential races. - */ -static ssize_t aio_run_iocb(struct kiocb *iocb) -{ - struct kioctx *ctx = iocb->ki_ctx; - ssize_t (*retry)(struct kiocb *); - ssize_t ret; - - if (!(retry = iocb->ki_retry)) { - printk("aio_run_iocb: iocb->ki_retry = NULL\n"); - return 0; - } - - /* - * We don't want the next retry iteration for this - * operation to start until this one has returned and - * updated the iocb state. However, wait_queue functions - * can trigger a kick_iocb from interrupt context in the - * meantime, indicating that data is available for the next - * iteration. We want to remember that and enable the - * next retry iteration _after_ we are through with - * this one. - * - * So, in order to be able to register a "kick", but - * prevent it from being queued now, we clear the kick - * flag, but make the kick code *think* that the iocb is - * still on the run list until we are actually done. - * When we are done with this iteration, we check if - * the iocb was kicked in the meantime and if so, queue - * it up afresh. - */ - - kiocbClearKicked(iocb); + if (get_user(id, &ring->id)) + return NULL; - /* - * This is so that aio_complete knows it doesn't need to - * pull the iocb off the run list (We can't just call - * INIT_LIST_HEAD because we don't want a kick_iocb to - * queue this on the run list yet) - */ - iocb->ki_run_list.next = iocb->ki_run_list.prev = NULL; - spin_unlock_irq(&ctx->ctx_lock); + rcu_read_lock(); + table = rcu_dereference(mm->ioctx_table); - /* Quit retrying if the i/o has been cancelled */ - if (kiocbIsCancelled(iocb)) { - ret = -EINTR; - aio_complete(iocb, ret, 0); - /* must not access the iocb after this */ + if (!table || id >= table->nr) goto out; - } - /* - * Now we are all set to call the retry method in async - * context. - */ - ret = retry(iocb); - - if (ret != -EIOCBRETRY && ret != -EIOCBQUEUED) { - /* - * There's no easy way to restart the syscall since other AIO's - * may be already running. Just fail this IO with EINTR. - */ - if (unlikely(ret == -ERESTARTSYS || ret == -ERESTARTNOINTR || - ret == -ERESTARTNOHAND || ret == -ERESTART_RESTARTBLOCK)) - ret = -EINTR; - aio_complete(iocb, ret, 0); + ctx = table->table[id]; + if (ctx && ctx->user_id == ctx_id) { + percpu_ref_get(&ctx->users); + ret = ctx; } out: - spin_lock_irq(&ctx->ctx_lock); - - if (-EIOCBRETRY == ret) { - /* - * OK, now that we are done with this iteration - * and know that there is more left to go, - * this is where we let go so that a subsequent - * "kick" can start the next iteration - */ - - /* will make __queue_kicked_iocb succeed from here on */ - INIT_LIST_HEAD(&iocb->ki_run_list); - /* we must queue the next iteration ourselves, if it - * has already been kicked */ - if (kiocbIsKicked(iocb)) { - __queue_kicked_iocb(iocb); - - /* - * __queue_kicked_iocb will always return 1 here, because - * iocb->ki_run_list is empty at this point so it should - * be safe to unconditionally queue the context into the - * work queue. - */ - aio_queue_work(ctx); - } - } + rcu_read_unlock(); return ret; } -/* - * __aio_run_iocbs: - * Process all pending retries queued on the ioctx - * run list. - * Assumes it is operating within the aio issuer's mm - * context. - */ -static int __aio_run_iocbs(struct kioctx *ctx) -{ - struct kiocb *iocb; - struct list_head run_list; - - assert_spin_locked(&ctx->ctx_lock); - - list_replace_init(&ctx->run_list, &run_list); - while (!list_empty(&run_list)) { - iocb = list_entry(run_list.next, struct kiocb, - ki_run_list); - list_del(&iocb->ki_run_list); - /* - * Hold an extra reference while retrying i/o. - */ - iocb->ki_users++; /* grab extra reference */ - aio_run_iocb(iocb); - __aio_put_req(ctx, iocb); - } - if (!list_empty(&ctx->run_list)) - return 1; - return 0; -} - -static void aio_queue_work(struct kioctx * ctx) -{ - unsigned long timeout; - /* - * if someone is waiting, get the work started right - * away, otherwise, use a longer delay - */ - smp_mb(); - if (waitqueue_active(&ctx->wait)) - timeout = 1; - else - timeout = HZ/10; - queue_delayed_work(aio_wq, &ctx->wq, timeout); -} - - -/* - * aio_run_iocbs: - * Process all pending retries queued on the ioctx - * run list. - * Assumes it is operating within the aio issuer's mm - * context. - */ -static inline void aio_run_iocbs(struct kioctx *ctx) -{ - int requeue; - - spin_lock_irq(&ctx->ctx_lock); - - requeue = __aio_run_iocbs(ctx); - spin_unlock_irq(&ctx->ctx_lock); - if (requeue) - aio_queue_work(ctx); -} - -/* - * just like aio_run_iocbs, but keeps running them until - * the list stays empty - */ -static inline void aio_run_all_iocbs(struct kioctx *ctx) -{ - spin_lock_irq(&ctx->ctx_lock); - while (__aio_run_iocbs(ctx)) - ; - spin_unlock_irq(&ctx->ctx_lock); -} - -/* - * aio_kick_handler: - * Work queue handler triggered to process pending - * retries on an ioctx. Takes on the aio issuer's - * mm context before running the iocbs, so that - * copy_xxx_user operates on the issuer's address - * space. - * Run on aiod's context. - */ -static void aio_kick_handler(struct work_struct *work) -{ - struct kioctx *ctx = container_of(work, struct kioctx, wq.work); - mm_segment_t oldfs = get_fs(); - struct mm_struct *mm; - int requeue; - - set_fs(USER_DS); - use_mm(ctx->mm); - spin_lock_irq(&ctx->ctx_lock); - requeue =__aio_run_iocbs(ctx); - mm = ctx->mm; - spin_unlock_irq(&ctx->ctx_lock); - unuse_mm(mm); - set_fs(oldfs); - /* - * we're in a worker thread already, don't use queue_delayed_work, - */ - if (requeue) - queue_delayed_work(aio_wq, &ctx->wq, 0); -} - - -/* - * Called by kick_iocb to queue the kiocb for retry - * and if required activate the aio work queue to process - * it - */ -static void try_queue_kicked_iocb(struct kiocb *iocb) -{ - struct kioctx *ctx = iocb->ki_ctx; - unsigned long flags; - int run = 0; - - spin_lock_irqsave(&ctx->ctx_lock, flags); - /* set this inside the lock so that we can't race with aio_run_iocb() - * testing it and putting the iocb on the run list under the lock */ - if (!kiocbTryKick(iocb)) - run = __queue_kicked_iocb(iocb); - spin_unlock_irqrestore(&ctx->ctx_lock, flags); - if (run) - aio_queue_work(ctx); -} - -/* - * kick_iocb: - * Called typically from a wait queue callback context - * to trigger a retry of the iocb. - * The retry is usually executed by aio workqueue - * threads (See aio_kick_handler). - */ -void kick_iocb(struct kiocb *iocb) -{ - /* sync iocbs are easy: they can only ever be executing from a - * single context. */ - if (is_sync_kiocb(iocb)) { - kiocbSetKicked(iocb); - wake_up_process(iocb->ki_obj.tsk); - return; - } - - try_queue_kicked_iocb(iocb); -} -EXPORT_SYMBOL(kick_iocb); - /* aio_complete * Called when the io request on the given iocb is complete. - * Returns true if this is the last user of the request. The - * only other user of the request can be the cancellation code. */ -int aio_complete(struct kiocb *iocb, long res, long res2) +void aio_complete(struct kiocb *iocb, long res, long res2) { struct kioctx *ctx = iocb->ki_ctx; - struct aio_ring_info *info; struct aio_ring *ring; - struct io_event *event; + struct io_event *ev_page, *event; unsigned long flags; - unsigned long tail; - int ret; + unsigned tail, pos; /* * Special case handling for sync iocbs: @@ -927,61 +959,64 @@ int aio_complete(struct kiocb *iocb, long res, long res2) * - the sync task helpfully left a reference to itself in the iocb */ if (is_sync_kiocb(iocb)) { - BUG_ON(iocb->ki_users != 1); iocb->ki_user_data = res; - iocb->ki_users = 0; + smp_wmb(); + iocb->ki_ctx = ERR_PTR(-EXDEV); wake_up_process(iocb->ki_obj.tsk); - return 1; + return; } - info = &ctx->ring_info; - - /* add a completion event to the ring buffer. - * must be done holding ctx->ctx_lock to prevent - * other code from messing with the tail - * pointer since we might be called from irq - * context. - */ - spin_lock_irqsave(&ctx->ctx_lock, flags); + if (iocb->ki_list.next) { + unsigned long flags; - if (iocb->ki_run_list.prev && !list_empty(&iocb->ki_run_list)) - list_del_init(&iocb->ki_run_list); + spin_lock_irqsave(&ctx->ctx_lock, flags); + list_del(&iocb->ki_list); + spin_unlock_irqrestore(&ctx->ctx_lock, flags); + } /* - * cancelled requests don't get events, userland was given one - * when the event got cancelled. + * Add a completion event to the ring buffer. Must be done holding + * ctx->completion_lock to prevent other code from messing with the tail + * pointer since we might be called from irq context. */ - if (kiocbIsCancelled(iocb)) - goto put_rq; + spin_lock_irqsave(&ctx->completion_lock, flags); - ring = kmap_atomic(info->ring_pages[0], KM_IRQ1); + tail = ctx->tail; + pos = tail + AIO_EVENTS_OFFSET; - tail = info->tail; - event = aio_ring_event(info, tail, KM_IRQ0); - if (++tail >= info->nr) + if (++tail >= ctx->nr_events) tail = 0; + ev_page = kmap_atomic(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); + event = ev_page + pos % AIO_EVENTS_PER_PAGE; + event->obj = (u64)(unsigned long)iocb->ki_obj.user; event->data = iocb->ki_user_data; event->res = res; event->res2 = res2; - dprintk("aio_complete: %p[%lu]: %p: %p %Lx %lx %lx\n", - ctx, tail, iocb, iocb->ki_obj.user, iocb->ki_user_data, - res, res2); + kunmap_atomic(ev_page); + flush_dcache_page(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); + + pr_debug("%p[%u]: %p: %p %Lx %lx %lx\n", + ctx, tail, iocb, iocb->ki_obj.user, iocb->ki_user_data, + res, res2); /* after flagging the request as done, we * must never even look at it again */ smp_wmb(); /* make event visible before updating tail */ - info->tail = tail; + ctx->tail = tail; + + ring = kmap_atomic(ctx->ring_pages[0]); ring->tail = tail; + kunmap_atomic(ring); + flush_dcache_page(ctx->ring_pages[0]); - put_aio_ring_event(event, KM_IRQ0); - kunmap_atomic(ring, KM_IRQ1); + spin_unlock_irqrestore(&ctx->completion_lock, flags); - pr_debug("added to ring %p at [%lu]\n", iocb, tail); + pr_debug("added to ring %p at [%u]\n", iocb, tail); /* * Check if the user asked us to deliver the result through an @@ -991,9 +1026,9 @@ int aio_complete(struct kiocb *iocb, long res, long res2) if (iocb->ki_eventfd != NULL) eventfd_signal(iocb->ki_eventfd, 1); -put_rq: /* everything turned out well, dispose of the aiocb. */ - ret = __aio_put_req(ctx, iocb); + kiocb_free(iocb); + put_reqs_available(ctx, 1); /* * We have to order our ring_info tail store above and test @@ -1006,235 +1041,136 @@ put_rq: if (waitqueue_active(&ctx->wait)) wake_up(&ctx->wait); - spin_unlock_irqrestore(&ctx->ctx_lock, flags); - return ret; + percpu_ref_put(&ctx->reqs); } EXPORT_SYMBOL(aio_complete); -/* aio_read_evt - * Pull an event off of the ioctx's event ring. Returns the number of - * events fetched (0 or 1 ;-) - * FIXME: make this use cmpxchg. - * TODO: make the ringbuffer user mmap()able (requires FIXME). +/* aio_read_events + * Pull an event off of the ioctx's event ring. Returns the number of + * events fetched */ -static int aio_read_evt(struct kioctx *ioctx, struct io_event *ent) +static long aio_read_events_ring(struct kioctx *ctx, + struct io_event __user *event, long nr) { - struct aio_ring_info *info = &ioctx->ring_info; struct aio_ring *ring; - unsigned long head; - int ret = 0; - - ring = kmap_atomic(info->ring_pages[0], KM_USER0); - dprintk("in aio_read_evt h%lu t%lu m%lu\n", - (unsigned long)ring->head, (unsigned long)ring->tail, - (unsigned long)ring->nr); - - if (ring->head == ring->tail) - goto out; - - spin_lock(&info->ring_lock); - - head = ring->head % info->nr; - if (head != ring->tail) { - struct io_event *evp = aio_ring_event(info, head, KM_USER1); - *ent = *evp; - head = (head + 1) % info->nr; - smp_mb(); /* finish reading the event before updatng the head */ - ring->head = head; - ret = 1; - put_aio_ring_event(evp, KM_USER1); - } - spin_unlock(&info->ring_lock); + unsigned head, tail, pos; + long ret = 0; + int copy_ret; -out: - kunmap_atomic(ring, KM_USER0); - dprintk("leaving aio_read_evt: %d h%lu t%lu\n", ret, - (unsigned long)ring->head, (unsigned long)ring->tail); - return ret; -} + mutex_lock(&ctx->ring_lock); -struct aio_timeout { - struct timer_list timer; - int timed_out; - struct task_struct *p; -}; + /* Access to ->ring_pages here is protected by ctx->ring_lock. */ + ring = kmap_atomic(ctx->ring_pages[0]); + head = ring->head; + tail = ring->tail; + kunmap_atomic(ring); -static void timeout_func(unsigned long data) -{ - struct aio_timeout *to = (struct aio_timeout *)data; + pr_debug("h%u t%u m%u\n", head, tail, ctx->nr_events); - to->timed_out = 1; - wake_up_process(to->p); -} + if (head == tail) + goto out; -static inline void init_timeout(struct aio_timeout *to) -{ - setup_timer_on_stack(&to->timer, timeout_func, (unsigned long) to); - to->timed_out = 0; - to->p = current; -} + head %= ctx->nr_events; + tail %= ctx->nr_events; -static inline void set_timeout(long start_jiffies, struct aio_timeout *to, - const struct timespec *ts) -{ - to->timer.expires = start_jiffies + timespec_to_jiffies(ts); - if (time_after(to->timer.expires, jiffies)) - add_timer(&to->timer); - else - to->timed_out = 1; -} - -static inline void clear_timeout(struct aio_timeout *to) -{ - del_singleshot_timer_sync(&to->timer); -} + while (ret < nr) { + long avail; + struct io_event *ev; + struct page *page; -static int read_events(struct kioctx *ctx, - long min_nr, long nr, - struct io_event __user *event, - struct timespec __user *timeout) -{ - long start_jiffies = jiffies; - struct task_struct *tsk = current; - DECLARE_WAITQUEUE(wait, tsk); - int ret; - int i = 0; - struct io_event ent; - struct aio_timeout to; - int retry = 0; - - /* needed to zero any padding within an entry (there shouldn't be - * any, but C is fun! - */ - memset(&ent, 0, sizeof(ent)); -retry: - ret = 0; - while (likely(i < nr)) { - ret = aio_read_evt(ctx, &ent); - if (unlikely(ret <= 0)) + avail = (head <= tail ? tail : ctx->nr_events) - head; + if (head == tail) break; - dprintk("read event: %Lx %Lx %Lx %Lx\n", - ent.data, ent.obj, ent.res, ent.res2); - - /* Could we split the check in two? */ - ret = -EFAULT; - if (unlikely(copy_to_user(event, &ent, sizeof(ent)))) { - dprintk("aio: lost an event due to EFAULT.\n"); - break; - } - ret = 0; + avail = min(avail, nr - ret); + avail = min_t(long, avail, AIO_EVENTS_PER_PAGE - + ((head + AIO_EVENTS_OFFSET) % AIO_EVENTS_PER_PAGE)); - /* Good, event copied to userland, update counts. */ - event ++; - i ++; - } + pos = head + AIO_EVENTS_OFFSET; + page = ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]; + pos %= AIO_EVENTS_PER_PAGE; - if (min_nr <= i) - return i; - if (ret) - return ret; + ev = kmap(page); + copy_ret = copy_to_user(event + ret, ev + pos, + sizeof(*ev) * avail); + kunmap(page); - /* End fast path */ + if (unlikely(copy_ret)) { + ret = -EFAULT; + goto out; + } - /* racey check, but it gets redone */ - if (!retry && unlikely(!list_empty(&ctx->run_list))) { - retry = 1; - aio_run_all_iocbs(ctx); - goto retry; + ret += avail; + head += avail; + head %= ctx->nr_events; } - init_timeout(&to); - if (timeout) { - struct timespec ts; - ret = -EFAULT; - if (unlikely(copy_from_user(&ts, timeout, sizeof(ts)))) - goto out; + ring = kmap_atomic(ctx->ring_pages[0]); + ring->head = head; + kunmap_atomic(ring); + flush_dcache_page(ctx->ring_pages[0]); - set_timeout(start_jiffies, &to, &ts); - } + pr_debug("%li h%u t%u\n", ret, head, tail); +out: + mutex_unlock(&ctx->ring_lock); - while (likely(i < nr)) { - add_wait_queue_exclusive(&ctx->wait, &wait); - do { - set_task_state(tsk, TASK_INTERRUPTIBLE); - ret = aio_read_evt(ctx, &ent); - if (ret) - break; - if (min_nr <= i) - break; - if (unlikely(ctx->dead)) { - ret = -EINVAL; - break; - } - if (to.timed_out) /* Only check after read evt */ - break; - /* Try to only show up in io wait if there are ops - * in flight */ - if (ctx->reqs_active) - io_schedule(); - else - schedule(); - if (signal_pending(tsk)) { - ret = -EINTR; - break; - } - /*ret = aio_read_evt(ctx, &ent);*/ - } while (1) ; + return ret; +} - set_task_state(tsk, TASK_RUNNING); - remove_wait_queue(&ctx->wait, &wait); +static bool aio_read_events(struct kioctx *ctx, long min_nr, long nr, + struct io_event __user *event, long *i) +{ + long ret = aio_read_events_ring(ctx, event + *i, nr - *i); - if (unlikely(ret <= 0)) - break; + if (ret > 0) + *i += ret; - ret = -EFAULT; - if (unlikely(copy_to_user(event, &ent, sizeof(ent)))) { - dprintk("aio: lost an event due to EFAULT.\n"); - break; - } + if (unlikely(atomic_read(&ctx->dead))) + ret = -EINVAL; - /* Good, event copied to userland, update counts. */ - event ++; - i ++; - } + if (!*i) + *i = ret; - if (timeout) - clear_timeout(&to); -out: - destroy_timer_on_stack(&to.timer); - return i ? i : ret; + return ret < 0 || *i >= min_nr; } -/* Take an ioctx and remove it from the list of ioctx's. Protects - * against races with itself via ->dead. - */ -static void io_destroy(struct kioctx *ioctx) +static long read_events(struct kioctx *ctx, long min_nr, long nr, + struct io_event __user *event, + struct timespec __user *timeout) { - struct mm_struct *mm = current->mm; - int was_dead; + ktime_t until = { .tv64 = KTIME_MAX }; + long ret = 0; - /* delete the entry from the list is someone else hasn't already */ - spin_lock(&mm->ioctx_lock); - was_dead = ioctx->dead; - ioctx->dead = 1; - hlist_del_rcu(&ioctx->list); - spin_unlock(&mm->ioctx_lock); + if (timeout) { + struct timespec ts; - dprintk("aio_release(%p)\n", ioctx); - if (likely(!was_dead)) - put_ioctx(ioctx); /* twice for the list */ + if (unlikely(copy_from_user(&ts, timeout, sizeof(ts)))) + return -EFAULT; - aio_cancel_all(ioctx); - wait_for_all_aios(ioctx); + until = timespec_to_ktime(ts); + } /* - * Wake up any waiters. The setting of ctx->dead must be seen - * by other CPUs at this point. Right now, we rely on the - * locking done by the above calls to ensure this consistency. + * Note that aio_read_events() is being called as the conditional - i.e. + * we're calling it after prepare_to_wait() has set task state to + * TASK_INTERRUPTIBLE. + * + * But aio_read_events() can block, and if it blocks it's going to flip + * the task state back to TASK_RUNNING. + * + * This should be ok, provided it doesn't flip the state back to + * TASK_RUNNING and return 0 too much - that causes us to spin. That + * will only happen if the mutex_lock() call blocks, and we then find + * the ringbuffer empty. So in practice we should be ok, but it's + * something to be aware of when touching this code. */ - wake_up(&ioctx->wait); - put_ioctx(ioctx); /* once for the lookup */ + wait_event_interruptible_hrtimeout(ctx->wait, + aio_read_events(ctx, min_nr, nr, event, &ret), until); + + if (!ret && signal_pending(current)) + ret = -EINTR; + + return ret; } /* sys_io_setup: @@ -1271,11 +1207,9 @@ SYSCALL_DEFINE2(io_setup, unsigned, nr_events, aio_context_t __user *, ctxp) ret = PTR_ERR(ioctx); if (!IS_ERR(ioctx)) { ret = put_user(ioctx->user_id, ctxp); - if (!ret) - return 0; - - get_ioctx(ioctx); /* io_destroy() expects us to hold a ref */ - io_destroy(ioctx); + if (ret) + kill_ioctx(current->mm, ioctx, NULL); + percpu_ref_put(&ioctx->users); } out: @@ -1292,144 +1226,73 @@ SYSCALL_DEFINE1(io_destroy, aio_context_t, ctx) { struct kioctx *ioctx = lookup_ioctx(ctx); if (likely(NULL != ioctx)) { - io_destroy(ioctx); - return 0; - } - pr_debug("EINVAL: io_destroy: invalid context id\n"); - return -EINVAL; -} + struct completion requests_done = + COMPLETION_INITIALIZER_ONSTACK(requests_done); + int ret; -static void aio_advance_iovec(struct kiocb *iocb, ssize_t ret) -{ - struct iovec *iov = &iocb->ki_iovec[iocb->ki_cur_seg]; - - BUG_ON(ret <= 0); - - while (iocb->ki_cur_seg < iocb->ki_nr_segs && ret > 0) { - ssize_t this = min((ssize_t)iov->iov_len, ret); - iov->iov_base += this; - iov->iov_len -= this; - iocb->ki_left -= this; - ret -= this; - if (iov->iov_len == 0) { - iocb->ki_cur_seg++; - iov++; - } - } + /* Pass requests_done to kill_ioctx() where it can be set + * in a thread-safe way. If we try to set it here then we have + * a race condition if two io_destroy() called simultaneously. + */ + ret = kill_ioctx(current->mm, ioctx, &requests_done); + percpu_ref_put(&ioctx->users); - /* the caller should not have done more io than what fit in - * the remaining iovecs */ - BUG_ON(ret > 0 && iocb->ki_left == 0); -} + /* Wait until all IO for the context are done. Otherwise kernel + * keep using user-space buffers even if user thinks the context + * is destroyed. + */ + if (!ret) + wait_for_completion(&requests_done); -static ssize_t aio_rw_vect_retry(struct kiocb *iocb) -{ - struct file *file = iocb->ki_filp; - struct address_space *mapping = file->f_mapping; - struct inode *inode = mapping->host; - ssize_t (*rw_op)(struct kiocb *, const struct iovec *, - unsigned long, loff_t); - ssize_t ret = 0; - unsigned short opcode; - - if ((iocb->ki_opcode == IOCB_CMD_PREADV) || - (iocb->ki_opcode == IOCB_CMD_PREAD)) { - rw_op = file->f_op->aio_read; - opcode = IOCB_CMD_PREADV; - } else { - rw_op = file->f_op->aio_write; - opcode = IOCB_CMD_PWRITEV; + return ret; } - - /* This matches the pread()/pwrite() logic */ - if (iocb->ki_pos < 0) - return -EINVAL; - - do { - ret = rw_op(iocb, &iocb->ki_iovec[iocb->ki_cur_seg], - iocb->ki_nr_segs - iocb->ki_cur_seg, - iocb->ki_pos); - if (ret > 0) - aio_advance_iovec(iocb, ret); - - /* retry all partial writes. retry partial reads as long as its a - * regular file. */ - } while (ret > 0 && iocb->ki_left > 0 && - (opcode == IOCB_CMD_PWRITEV || - (!S_ISFIFO(inode->i_mode) && !S_ISSOCK(inode->i_mode)))); - - /* This means we must have transferred all that we could */ - /* No need to retry anymore */ - if ((ret == 0) || (iocb->ki_left == 0)) - ret = iocb->ki_nbytes - iocb->ki_left; - - /* If we managed to write some out we return that, rather than - * the eventual error. */ - if (opcode == IOCB_CMD_PWRITEV - && ret < 0 && ret != -EIOCBQUEUED && ret != -EIOCBRETRY - && iocb->ki_nbytes - iocb->ki_left) - ret = iocb->ki_nbytes - iocb->ki_left; - - return ret; -} - -static ssize_t aio_fdsync(struct kiocb *iocb) -{ - struct file *file = iocb->ki_filp; - ssize_t ret = -EINVAL; - - if (file->f_op->aio_fsync) - ret = file->f_op->aio_fsync(iocb, 1); - return ret; + pr_debug("EINVAL: io_destroy: invalid context id\n"); + return -EINVAL; } -static ssize_t aio_fsync(struct kiocb *iocb) -{ - struct file *file = iocb->ki_filp; - ssize_t ret = -EINVAL; +typedef ssize_t (aio_rw_op)(struct kiocb *, const struct iovec *, + unsigned long, loff_t); +typedef ssize_t (rw_iter_op)(struct kiocb *, struct iov_iter *); - if (file->f_op->aio_fsync) - ret = file->f_op->aio_fsync(iocb, 0); - return ret; -} - -static ssize_t aio_setup_vectored_rw(int type, struct kiocb *kiocb, bool compat) +static ssize_t aio_setup_vectored_rw(struct kiocb *kiocb, + int rw, char __user *buf, + unsigned long *nr_segs, + struct iovec **iovec, + bool compat) { ssize_t ret; + *nr_segs = kiocb->ki_nbytes; + #ifdef CONFIG_COMPAT if (compat) - ret = compat_rw_copy_check_uvector(type, - (struct compat_iovec __user *)kiocb->ki_buf, - kiocb->ki_nbytes, 1, &kiocb->ki_inline_vec, - &kiocb->ki_iovec); + ret = compat_rw_copy_check_uvector(rw, + (struct compat_iovec __user *)buf, + *nr_segs, 1, *iovec, iovec); else #endif - ret = rw_copy_check_uvector(type, - (struct iovec __user *)kiocb->ki_buf, - kiocb->ki_nbytes, 1, &kiocb->ki_inline_vec, - &kiocb->ki_iovec); + ret = rw_copy_check_uvector(rw, + (struct iovec __user *)buf, + *nr_segs, 1, *iovec, iovec); if (ret < 0) - goto out; + return ret; - kiocb->ki_nr_segs = kiocb->ki_nbytes; - kiocb->ki_cur_seg = 0; - /* ki_nbytes/left now reflect bytes instead of segs */ + /* ki_nbytes now reflect bytes instead of segs */ kiocb->ki_nbytes = ret; - kiocb->ki_left = ret; - - ret = 0; -out: - return ret; + return 0; } -static ssize_t aio_setup_single_vector(struct kiocb *kiocb) +static ssize_t aio_setup_single_vector(struct kiocb *kiocb, + int rw, char __user *buf, + unsigned long *nr_segs, + struct iovec *iovec) { - kiocb->ki_iovec = &kiocb->ki_inline_vec; - kiocb->ki_iovec->iov_base = kiocb->ki_buf; - kiocb->ki_iovec->iov_len = kiocb->ki_left; - kiocb->ki_nr_segs = 1; - kiocb->ki_cur_seg = 0; + if (unlikely(!access_ok(!rw, buf, kiocb->ki_nbytes))) + return -EFAULT; + + iovec->iov_base = buf; + iovec->iov_len = kiocb->ki_nbytes; + *nr_segs = 1; return 0; } @@ -1438,156 +1301,125 @@ static ssize_t aio_setup_single_vector(struct kiocb *kiocb) * Performs the initial checks and aio retry method * setup for the kiocb at the time of io submission. */ -static ssize_t aio_setup_iocb(struct kiocb *kiocb, bool compat) +static ssize_t aio_run_iocb(struct kiocb *req, unsigned opcode, + char __user *buf, bool compat) { - struct file *file = kiocb->ki_filp; - ssize_t ret = 0; - - switch (kiocb->ki_opcode) { + struct file *file = req->ki_filp; + ssize_t ret; + unsigned long nr_segs; + int rw; + fmode_t mode; + aio_rw_op *rw_op; + rw_iter_op *iter_op; + struct iovec inline_vec, *iovec = &inline_vec; + struct iov_iter iter; + + switch (opcode) { case IOCB_CMD_PREAD: - ret = -EBADF; - if (unlikely(!(file->f_mode & FMODE_READ))) - break; - ret = -EFAULT; - if (unlikely(!access_ok(VERIFY_WRITE, kiocb->ki_buf, - kiocb->ki_left))) - break; - ret = security_file_permission(file, MAY_READ); - if (unlikely(ret)) - break; - ret = aio_setup_single_vector(kiocb); - if (ret) - break; - ret = -EINVAL; - if (file->f_op->aio_read) - kiocb->ki_retry = aio_rw_vect_retry; - break; - case IOCB_CMD_PWRITE: - ret = -EBADF; - if (unlikely(!(file->f_mode & FMODE_WRITE))) - break; - ret = -EFAULT; - if (unlikely(!access_ok(VERIFY_READ, kiocb->ki_buf, - kiocb->ki_left))) - break; - ret = security_file_permission(file, MAY_WRITE); - if (unlikely(ret)) - break; - ret = aio_setup_single_vector(kiocb); - if (ret) - break; - ret = -EINVAL; - if (file->f_op->aio_write) - kiocb->ki_retry = aio_rw_vect_retry; - break; case IOCB_CMD_PREADV: - ret = -EBADF; - if (unlikely(!(file->f_mode & FMODE_READ))) - break; - ret = security_file_permission(file, MAY_READ); - if (unlikely(ret)) - break; - ret = aio_setup_vectored_rw(READ, kiocb, compat); - if (ret) - break; - ret = -EINVAL; - if (file->f_op->aio_read) - kiocb->ki_retry = aio_rw_vect_retry; - break; + mode = FMODE_READ; + rw = READ; + rw_op = file->f_op->aio_read; + iter_op = file->f_op->read_iter; + goto rw_common; + + case IOCB_CMD_PWRITE: case IOCB_CMD_PWRITEV: - ret = -EBADF; - if (unlikely(!(file->f_mode & FMODE_WRITE))) - break; - ret = security_file_permission(file, MAY_WRITE); - if (unlikely(ret)) - break; - ret = aio_setup_vectored_rw(WRITE, kiocb, compat); - if (ret) + mode = FMODE_WRITE; + rw = WRITE; + rw_op = file->f_op->aio_write; + iter_op = file->f_op->write_iter; + goto rw_common; +rw_common: + if (unlikely(!(file->f_mode & mode))) + return -EBADF; + + if (!rw_op && !iter_op) + return -EINVAL; + + ret = (opcode == IOCB_CMD_PREADV || + opcode == IOCB_CMD_PWRITEV) + ? aio_setup_vectored_rw(req, rw, buf, &nr_segs, + &iovec, compat) + : aio_setup_single_vector(req, rw, buf, &nr_segs, + iovec); + if (!ret) + ret = rw_verify_area(rw, file, &req->ki_pos, req->ki_nbytes); + if (ret < 0) { + if (iovec != &inline_vec) + kfree(iovec); + return ret; + } + + req->ki_nbytes = ret; + + /* XXX: move/kill - rw_verify_area()? */ + /* This matches the pread()/pwrite() logic */ + if (req->ki_pos < 0) { + ret = -EINVAL; break; - ret = -EINVAL; - if (file->f_op->aio_write) - kiocb->ki_retry = aio_rw_vect_retry; + } + + if (rw == WRITE) + file_start_write(file); + + if (iter_op) { + iov_iter_init(&iter, rw, iovec, nr_segs, req->ki_nbytes); + ret = iter_op(req, &iter); + } else { + ret = rw_op(req, iovec, nr_segs, req->ki_pos); + } + + if (rw == WRITE) + file_end_write(file); break; + case IOCB_CMD_FDSYNC: - ret = -EINVAL; - if (file->f_op->aio_fsync) - kiocb->ki_retry = aio_fdsync; + if (!file->f_op->aio_fsync) + return -EINVAL; + + ret = file->f_op->aio_fsync(req, 1); break; + case IOCB_CMD_FSYNC: - ret = -EINVAL; - if (file->f_op->aio_fsync) - kiocb->ki_retry = aio_fsync; + if (!file->f_op->aio_fsync) + return -EINVAL; + + ret = file->f_op->aio_fsync(req, 0); break; + default: - dprintk("EINVAL: io_submit: no operation provided\n"); - ret = -EINVAL; + pr_debug("EINVAL: no operation provided\n"); + return -EINVAL; } - if (!kiocb->ki_retry) - return ret; + if (iovec != &inline_vec) + kfree(iovec); - return 0; -} - -static void aio_batch_add(struct address_space *mapping, - struct hlist_head *batch_hash) -{ - struct aio_batch_entry *abe; - struct hlist_node *pos; - unsigned bucket; - - bucket = hash_ptr(mapping, AIO_BATCH_HASH_BITS); - hlist_for_each_entry(abe, pos, &batch_hash[bucket], list) { - if (abe->mapping == mapping) - return; + if (ret != -EIOCBQUEUED) { + /* + * There's no easy way to restart the syscall since other AIO's + * may be already running. Just fail this IO with EINTR. + */ + if (unlikely(ret == -ERESTARTSYS || ret == -ERESTARTNOINTR || + ret == -ERESTARTNOHAND || + ret == -ERESTART_RESTARTBLOCK)) + ret = -EINTR; + aio_complete(req, ret, 0); } - abe = mempool_alloc(abe_pool, GFP_KERNEL); - - /* - * we should be using igrab here, but - * we don't want to hammer on the global - * inode spinlock just to take an extra - * reference on a file that we must already - * have a reference to. - * - * When we're called, we always have a reference - * on the file, so we must always have a reference - * on the inode, so ihold() is safe here. - */ - ihold(mapping->host); - abe->mapping = mapping; - hlist_add_head(&abe->list, &batch_hash[bucket]); - return; -} - -static void aio_batch_free(struct hlist_head *batch_hash) -{ - struct aio_batch_entry *abe; - struct hlist_node *pos, *n; - int i; - - for (i = 0; i < AIO_BATCH_HASH_SIZE; i++) { - hlist_for_each_entry_safe(abe, pos, n, &batch_hash[i], list) { - blk_run_address_space(abe->mapping); - iput(abe->mapping->host); - hlist_del(&abe->list); - mempool_free(abe, abe_pool); - } - } + return 0; } static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb, - struct iocb *iocb, struct hlist_head *batch_hash, - bool compat) + struct iocb *iocb, bool compat) { struct kiocb *req; - struct file *file; ssize_t ret; /* enforce forwards compatibility on users */ if (unlikely(iocb->aio_reserved1 || iocb->aio_reserved2)) { - pr_debug("EINVAL: io_submit: reserve field set\n"); + pr_debug("EINVAL: reserve field set\n"); return -EINVAL; } @@ -1601,16 +1433,16 @@ static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb, return -EINVAL; } - file = fget(iocb->aio_fildes); - if (unlikely(!file)) - return -EBADF; - - req = aio_get_req(ctx); /* returns with 2 references to req */ - if (unlikely(!req)) { - fput(file); + req = aio_get_req(ctx); + if (unlikely(!req)) return -EAGAIN; + + req->ki_filp = fget(iocb->aio_fildes); + if (unlikely(!req->ki_filp)) { + ret = -EBADF; + goto out_put_req; } - req->ki_filp = file; + if (iocb->aio_flags & IOCB_FLAG_RESFD) { /* * If the IOCB_FLAG_RESFD flag of aio_flags is set, get an @@ -1626,45 +1458,28 @@ static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb, } } - ret = put_user(req->ki_key, &user_iocb->aio_key); + ret = put_user(KIOCB_KEY, &user_iocb->aio_key); if (unlikely(ret)) { - dprintk("EFAULT: aio_key\n"); + pr_debug("EFAULT: aio_key\n"); goto out_put_req; } req->ki_obj.user = user_iocb; req->ki_user_data = iocb->aio_data; req->ki_pos = iocb->aio_offset; + req->ki_nbytes = iocb->aio_nbytes; - req->ki_buf = (char __user *)(unsigned long)iocb->aio_buf; - req->ki_left = req->ki_nbytes = iocb->aio_nbytes; - req->ki_opcode = iocb->aio_lio_opcode; - - ret = aio_setup_iocb(req, compat); - + ret = aio_run_iocb(req, iocb->aio_lio_opcode, + (char __user *)(unsigned long)iocb->aio_buf, + compat); if (ret) goto out_put_req; - spin_lock_irq(&ctx->ctx_lock); - aio_run_iocb(req); - if (!list_empty(&ctx->run_list)) { - /* drain the run list */ - while (__aio_run_iocbs(ctx)) - ; - } - spin_unlock_irq(&ctx->ctx_lock); - if (req->ki_opcode == IOCB_CMD_PREAD || - req->ki_opcode == IOCB_CMD_PREADV || - req->ki_opcode == IOCB_CMD_PWRITE || - req->ki_opcode == IOCB_CMD_PWRITEV) - aio_batch_add(file->f_mapping, batch_hash); - - aio_put_req(req); /* drop extra ref to req */ return 0; - out_put_req: - aio_put_req(req); /* drop extra ref to req */ - aio_put_req(req); /* drop i/o ref to req */ + put_reqs_available(ctx, 1); + percpu_ref_put(&ctx->reqs); + kiocb_free(req); return ret; } @@ -1673,8 +1488,8 @@ long do_io_submit(aio_context_t ctx_id, long nr, { struct kioctx *ctx; long ret = 0; - int i; - struct hlist_head batch_hash[AIO_BATCH_HASH_SIZE] = { { 0, }, }; + int i = 0; + struct blk_plug plug; if (unlikely(nr < 0)) return -EINVAL; @@ -1687,10 +1502,12 @@ long do_io_submit(aio_context_t ctx_id, long nr, ctx = lookup_ioctx(ctx_id); if (unlikely(!ctx)) { - pr_debug("EINVAL: io_submit: invalid context id\n"); + pr_debug("EINVAL: invalid context id\n"); return -EINVAL; } + blk_start_plug(&plug); + /* * AKPM: should this return a partial result if some of the IOs were * successfully submitted? @@ -1709,13 +1526,13 @@ long do_io_submit(aio_context_t ctx_id, long nr, break; } - ret = io_submit_one(ctx, user_iocb, &tmp, batch_hash, compat); + ret = io_submit_one(ctx, user_iocb, &tmp, compat); if (ret) break; } - aio_batch_free(batch_hash); + blk_finish_plug(&plug); - put_ioctx(ctx); + percpu_ref_put(&ctx->users); return i ? i : ret; } @@ -1747,10 +1564,13 @@ static struct kiocb *lookup_kiocb(struct kioctx *ctx, struct iocb __user *iocb, assert_spin_locked(&ctx->ctx_lock); + if (key != KIOCB_KEY) + return NULL; + /* TODO: use a hash or array, this sucks. */ list_for_each(pos, &ctx->active_reqs) { struct kiocb *kiocb = list_kiocb(pos); - if (kiocb->ki_obj.user == iocb && kiocb->ki_key == key) + if (kiocb->ki_obj.user == iocb) return kiocb; } return NULL; @@ -1769,7 +1589,6 @@ static struct kiocb *lookup_kiocb(struct kioctx *ctx, struct iocb __user *iocb, SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb, struct io_event __user *, result) { - int (*cancel)(struct kiocb *iocb, struct io_event *res); struct kioctx *ctx; struct kiocb *kiocb; u32 key; @@ -1784,34 +1603,25 @@ SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb, return -EINVAL; spin_lock_irq(&ctx->ctx_lock); - ret = -EAGAIN; + kiocb = lookup_kiocb(ctx, iocb, key); - if (kiocb && kiocb->ki_cancel) { - cancel = kiocb->ki_cancel; - kiocb->ki_users ++; - kiocbSetCancelled(kiocb); - } else - cancel = NULL; + if (kiocb) + ret = kiocb_cancel(kiocb); + else + ret = -EINVAL; + spin_unlock_irq(&ctx->ctx_lock); - if (NULL != cancel) { - struct io_event tmp; - pr_debug("calling cancel\n"); - memset(&tmp, 0, sizeof(tmp)); - tmp.obj = (u64)(unsigned long)kiocb->ki_obj.user; - tmp.data = kiocb->ki_user_data; - ret = cancel(kiocb, &tmp); - if (!ret) { - /* Cancellation succeeded -- copy the result - * into the user's buffer. - */ - if (copy_to_user(result, &tmp, sizeof(tmp))) - ret = -EFAULT; - } - } else - ret = -EINVAL; + if (!ret) { + /* + * The result argument is no longer used - the io_event is + * always delivered via the ring buffer. -EINPROGRESS indicates + * cancellation is progress: + */ + ret = -EINPROGRESS; + } - put_ioctx(ctx); + percpu_ref_put(&ctx->users); return ret; } @@ -1826,8 +1636,7 @@ SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb, * < min_nr if the timeout specified by timeout has elapsed * before sufficient events are available, where timeout == NULL * specifies an infinite timeout. Note that the timeout pointed to by - * timeout is relative and will be updated if not NULL and the - * operation blocks. Will fail with -ENOSYS if not implemented. + * timeout is relative. Will fail with -ENOSYS if not implemented. */ SYSCALL_DEFINE5(io_getevents, aio_context_t, ctx_id, long, min_nr, @@ -1839,11 +1648,9 @@ SYSCALL_DEFINE5(io_getevents, aio_context_t, ctx_id, long ret = -EINVAL; if (likely(ioctx)) { - if (likely(min_nr <= nr && min_nr >= 0 && nr >= 0)) + if (likely(min_nr <= nr && min_nr >= 0)) ret = read_events(ioctx, min_nr, nr, events, timeout); - put_ioctx(ioctx); + percpu_ref_put(&ioctx->users); } - - asmlinkage_protect(5, ret, ctx_id, min_nr, nr, events, timeout); return ret; } |