aio.c 38.7 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
/*
 *	An async IO implementation for Linux
 *	Written by Benjamin LaHaise <bcrl@kvack.org>
 *
 *	Implements an efficient asynchronous io interface.
 *
 *	Copyright 2000, 2001, 2002 Red Hat, Inc.  All Rights Reserved.
 *
 *	See ../COPYING for licensing terms.
 */
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/errno.h>
#include <linux/time.h>
#include <linux/aio_abi.h>
16
#include <linux/export.h>
Linus Torvalds's avatar
Linus Torvalds committed
17
#include <linux/syscalls.h>
18
#include <linux/backing-dev.h>
19
#include <linux/uio.h>
Linus Torvalds's avatar
Linus Torvalds committed
20
21
22
23
24
25
26
27

#define DEBUG 0

#include <linux/sched.h>
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/mm.h>
#include <linux/mman.h>
28
#include <linux/mmu_context.h>
Linus Torvalds's avatar
Linus Torvalds committed
29
30
31
32
33
34
#include <linux/slab.h>
#include <linux/timer.h>
#include <linux/aio.h>
#include <linux/highmem.h>
#include <linux/workqueue.h>
#include <linux/security.h>
35
#include <linux/eventfd.h>
Jeff Moyer's avatar
Jeff Moyer committed
36
#include <linux/blkdev.h>
37
#include <linux/compat.h>
Linus Torvalds's avatar
Linus Torvalds committed
38
39
40
41
42
43
44
45
46
47

#include <asm/kmap_types.h>
#include <asm/uaccess.h>

#if DEBUG > 1
#define dprintk		printk
#else
#define dprintk(x...)	do { ; } while (0)
#endif

48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
#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;
	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 aio_ring_info {
	unsigned long		mmap_base;
	unsigned long		mmap_size;

	struct page		**ring_pages;
	spinlock_t		ring_lock;
	long			nr_pages;

	unsigned		nr, tail;

	struct page		*internal_pages[AIO_RING_PAGES];
};

static inline unsigned aio_ring_avail(struct aio_ring_info *info,
					struct aio_ring *ring)
{
	return (ring->head + info->nr - 1 - ring->tail) % info->nr;
}

struct kioctx {
	atomic_t		users;
	int			dead;

	/* This needs improving */
	unsigned long		user_id;
	struct hlist_node	list;

	wait_queue_head_t	wait;

	spinlock_t		ctx_lock;

	int			reqs_active;
	struct list_head	active_reqs;	/* used for cancellation */

	/* sys_io_setup currently limits this to an unsigned int */
	unsigned		max_reqs;

	struct aio_ring_info	ring_info;

	struct rcu_head		rcu_head;
};

Linus Torvalds's avatar
Linus Torvalds committed
109
/*------ sysctl variables----*/
110
111
112
static DEFINE_SPINLOCK(aio_nr_lock);
unsigned long aio_nr;		/* current system wide number of aio requests */
unsigned long aio_max_nr = 0x10000; /* system wide maximum number of aio requests */
Linus Torvalds's avatar
Linus Torvalds committed
113
114
/*----end sysctl variables---*/

115
116
static struct kmem_cache	*kiocb_cachep;
static struct kmem_cache	*kioctx_cachep;
Linus Torvalds's avatar
Linus Torvalds committed
117
118
119
120
121
122
123

/* aio_setup
 *	Creates the slab caches used by the aio routines, panic on
 *	failure as this is done early during the boot sequence.
 */
static int __init aio_setup(void)
{
124
125
	kiocb_cachep = KMEM_CACHE(kiocb, SLAB_HWCACHE_ALIGN|SLAB_PANIC);
	kioctx_cachep = KMEM_CACHE(kioctx,SLAB_HWCACHE_ALIGN|SLAB_PANIC);
Linus Torvalds's avatar
Linus Torvalds committed
126
127
128
129
130

	pr_debug("aio_setup: sizeof(struct page) = %d\n", (int)sizeof(struct page));

	return 0;
}
131
__initcall(aio_setup);
Linus Torvalds's avatar
Linus Torvalds committed
132
133
134
135
136
137
138
139
140

static void aio_free_ring(struct kioctx *ctx)
{
	struct aio_ring_info *info = &ctx->ring_info;
	long i;

	for (i=0; i<info->nr_pages; i++)
		put_page(info->ring_pages[i]);

141
	if (info->mmap_size) {
Al Viro's avatar
Al Viro committed
142
		vm_munmap(info->mmap_base, info->mmap_size);
143
	}
Linus Torvalds's avatar
Linus Torvalds committed
144
145
146
147
148
149
150
151
152
153
154
155

	if (info->ring_pages && info->ring_pages != info->internal_pages)
		kfree(info->ring_pages);
	info->ring_pages = NULL;
	info->nr = 0;
}

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;
Zach Brown's avatar
Zach Brown committed
156
	struct mm_struct *mm = current->mm;
157
	unsigned long size, populate;
Linus Torvalds's avatar
Linus Torvalds committed
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
	int nr_pages;

	/* 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;

	if (nr_pages < 0)
		return -EINVAL;

	nr_events = (PAGE_SIZE * nr_pages - sizeof(struct aio_ring)) / sizeof(struct io_event);

	info->nr = 0;
	info->ring_pages = info->internal_pages;
	if (nr_pages > AIO_RING_PAGES) {
175
		info->ring_pages = kcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL);
Linus Torvalds's avatar
Linus Torvalds committed
176
177
178
179
180
181
		if (!info->ring_pages)
			return -ENOMEM;
	}

	info->mmap_size = nr_pages * PAGE_SIZE;
	dprintk("attempting mmap of %lu bytes\n", info->mmap_size);
Zach Brown's avatar
Zach Brown committed
182
	down_write(&mm->mmap_sem);
183
184
	info->mmap_base = do_mmap_pgoff(NULL, 0, info->mmap_size, 
					PROT_READ|PROT_WRITE,
185
186
					MAP_ANONYMOUS|MAP_PRIVATE, 0,
					&populate);
Linus Torvalds's avatar
Linus Torvalds committed
187
	if (IS_ERR((void *)info->mmap_base)) {
Zach Brown's avatar
Zach Brown committed
188
		up_write(&mm->mmap_sem);
Linus Torvalds's avatar
Linus Torvalds committed
189
190
191
192
193
194
		info->mmap_size = 0;
		aio_free_ring(ctx);
		return -EAGAIN;
	}

	dprintk("mmap address: 0x%08lx\n", info->mmap_base);
Zach Brown's avatar
Zach Brown committed
195
	info->nr_pages = get_user_pages(current, mm, info->mmap_base, nr_pages,
Linus Torvalds's avatar
Linus Torvalds committed
196
					1, 0, info->ring_pages, NULL);
Zach Brown's avatar
Zach Brown committed
197
	up_write(&mm->mmap_sem);
Linus Torvalds's avatar
Linus Torvalds committed
198
199
200
201
202

	if (unlikely(info->nr_pages != nr_pages)) {
		aio_free_ring(ctx);
		return -EAGAIN;
	}
203
	if (populate)
204
		mm_populate(info->mmap_base, populate);
Linus Torvalds's avatar
Linus Torvalds committed
205
206
207
208
209

	ctx->user_id = info->mmap_base;

	info->nr = nr_events;		/* trusted copy */

210
	ring = kmap_atomic(info->ring_pages[0]);
Linus Torvalds's avatar
Linus Torvalds committed
211
212
213
214
215
216
217
	ring->nr = nr_events;	/* user copy */
	ring->id = ctx->user_id;
	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);
218
	kunmap_atomic(ring);
Linus Torvalds's avatar
Linus Torvalds committed
219
220
221
222
223
224

	return 0;
}


/* aio_ring_event: returns a pointer to the event at the given index from
225
 * kmap_atomic().  Release the pointer with put_aio_ring_event();
Linus Torvalds's avatar
Linus Torvalds committed
226
227
228
229
230
 */
#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)

231
#define aio_ring_event(info, nr) ({					\
Linus Torvalds's avatar
Linus Torvalds committed
232
233
234
	unsigned pos = (nr) + AIO_EVENTS_OFFSET;			\
	struct io_event *__event;					\
	__event = kmap_atomic(						\
235
			(info)->ring_pages[pos / AIO_EVENTS_PER_PAGE]); \
Linus Torvalds's avatar
Linus Torvalds committed
236
237
238
239
	__event += pos % AIO_EVENTS_PER_PAGE;				\
	__event;							\
})

240
#define put_aio_ring_event(event) do {		\
Linus Torvalds's avatar
Linus Torvalds committed
241
242
	struct io_event *__event = (event);	\
	(void)__event;				\
243
	kunmap_atomic((void *)((unsigned long)__event & PAGE_MASK)); \
Linus Torvalds's avatar
Linus Torvalds committed
244
245
} while(0)

246
247
248
249
250
static void ctx_rcu_free(struct rcu_head *head)
{
	struct kioctx *ctx = container_of(head, struct kioctx, rcu_head);
	kmem_cache_free(kioctx_cachep, ctx);
}
251
252
253
254
255
256
257

/* __put_ioctx
 *	Called when the last user of an aio context has gone away,
 *	and the struct needs to be freed.
 */
static void __put_ioctx(struct kioctx *ctx)
{
258
	unsigned nr_events = ctx->max_reqs;
259
260
261
	BUG_ON(ctx->reqs_active);

	aio_free_ring(ctx);
262
263
264
265
266
267
	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);
	}
268
	pr_debug("__put_ioctx: freeing %p\n", ctx);
269
	call_rcu(&ctx->rcu_head, ctx_rcu_free);
270
271
}

Nick Piggin's avatar
Nick Piggin committed
272
273
274
275
276
277
278
279
280
281
282
static inline int try_get_ioctx(struct kioctx *kioctx)
{
	return atomic_inc_not_zero(&kioctx->users);
}

static inline void put_ioctx(struct kioctx *kioctx)
{
	BUG_ON(atomic_read(&kioctx->users) <= 0);
	if (unlikely(atomic_dec_and_test(&kioctx->users)))
		__put_ioctx(kioctx);
}
283

Kent Overstreet's avatar
Kent Overstreet committed
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
static int kiocb_cancel(struct kioctx *ctx, struct kiocb *kiocb,
			struct io_event *res)
{
	int (*cancel)(struct kiocb *, struct io_event *);
	int ret = -EINVAL;

	cancel = kiocb->ki_cancel;
	kiocbSetCancelled(kiocb);
	if (cancel) {
		kiocb->ki_users++;
		spin_unlock_irq(&ctx->ctx_lock);

		memset(res, 0, sizeof(*res));
		res->obj = (u64)(unsigned long)kiocb->ki_obj.user;
		res->data = kiocb->ki_user_data;
		ret = cancel(kiocb, res);

		spin_lock_irq(&ctx->ctx_lock);
	}

	return ret;
}

Linus Torvalds's avatar
Linus Torvalds committed
307
308
309
310
311
/* ioctx_alloc
 *	Allocates and initializes an ioctx.  Returns an ERR_PTR if it failed.
 */
static struct kioctx *ioctx_alloc(unsigned nr_events)
{
Zach Brown's avatar
Zach Brown committed
312
	struct mm_struct *mm = current->mm;
Linus Torvalds's avatar
Linus Torvalds committed
313
	struct kioctx *ctx;
314
	int err = -ENOMEM;
Linus Torvalds's avatar
Linus Torvalds committed
315
316
317
318
319
320
321
322

	/* Prevent overflows */
	if ((nr_events > (0x10000000U / sizeof(struct io_event))) ||
	    (nr_events > (0x10000000U / sizeof(struct kiocb)))) {
		pr_debug("ENOMEM: nr_events too high\n");
		return ERR_PTR(-EINVAL);
	}

323
	if (!nr_events || (unsigned long)nr_events > aio_max_nr)
Linus Torvalds's avatar
Linus Torvalds committed
324
325
		return ERR_PTR(-EAGAIN);

326
	ctx = kmem_cache_zalloc(kioctx_cachep, GFP_KERNEL);
Linus Torvalds's avatar
Linus Torvalds committed
327
328
329
330
331
	if (!ctx)
		return ERR_PTR(-ENOMEM);

	ctx->max_reqs = nr_events;

Al Viro's avatar
Al Viro committed
332
	atomic_set(&ctx->users, 2);
Linus Torvalds's avatar
Linus Torvalds committed
333
334
335
336
337
338
339
340
341
342
	spin_lock_init(&ctx->ctx_lock);
	spin_lock_init(&ctx->ring_info.ring_lock);
	init_waitqueue_head(&ctx->wait);

	INIT_LIST_HEAD(&ctx->active_reqs);

	if (aio_setup_ring(ctx) < 0)
		goto out_freectx;

	/* limit the number of system wide aios */
343
	spin_lock(&aio_nr_lock);
344
345
	if (aio_nr + nr_events > aio_max_nr ||
	    aio_nr + nr_events < aio_nr) {
346
		spin_unlock(&aio_nr_lock);
Linus Torvalds's avatar
Linus Torvalds committed
347
		goto out_cleanup;
348
349
	}
	aio_nr += ctx->max_reqs;
350
	spin_unlock(&aio_nr_lock);
Linus Torvalds's avatar
Linus Torvalds committed
351

Jeff Moyer's avatar
Jeff Moyer committed
352
	/* now link into global list. */
353
354
355
	spin_lock(&mm->ioctx_lock);
	hlist_add_head_rcu(&ctx->list, &mm->ioctx_list);
	spin_unlock(&mm->ioctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
356
357

	dprintk("aio: allocated ioctx %p[%ld]: mm=%p mask=0x%x\n",
Zach Brown's avatar
Zach Brown committed
358
		ctx, ctx->user_id, mm, ctx->ring_info.nr);
Linus Torvalds's avatar
Linus Torvalds committed
359
360
361
	return ctx;

out_cleanup:
362
363
	err = -EAGAIN;
	aio_free_ring(ctx);
Linus Torvalds's avatar
Linus Torvalds committed
364
365
out_freectx:
	kmem_cache_free(kioctx_cachep, ctx);
366
367
	dprintk("aio: error allocating ioctx %d\n", err);
	return ERR_PTR(err);
Linus Torvalds's avatar
Linus Torvalds committed
368
369
}

370
/* kill_ctx
Linus Torvalds's avatar
Linus Torvalds committed
371
372
373
374
 *	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.
 */
375
static void kill_ctx(struct kioctx *ctx)
Linus Torvalds's avatar
Linus Torvalds committed
376
{
377
378
	struct task_struct *tsk = current;
	DECLARE_WAITQUEUE(wait, tsk);
Linus Torvalds's avatar
Linus Torvalds committed
379
	struct io_event res;
Kent Overstreet's avatar
Kent Overstreet committed
380
	struct kiocb *req;
381

Linus Torvalds's avatar
Linus Torvalds committed
382
383
384
	spin_lock_irq(&ctx->ctx_lock);
	ctx->dead = 1;
	while (!list_empty(&ctx->active_reqs)) {
Kent Overstreet's avatar
Kent Overstreet committed
385
386
387
388
389
		req = list_first_entry(&ctx->active_reqs,
					struct kiocb, ki_list);

		list_del_init(&req->ki_list);
		kiocb_cancel(ctx, req, &res);
Linus Torvalds's avatar
Linus Torvalds committed
390
391
392
	}

	if (!ctx->reqs_active)
393
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
394
395
396
397

	add_wait_queue(&ctx->wait, &wait);
	set_task_state(tsk, TASK_UNINTERRUPTIBLE);
	while (ctx->reqs_active) {
398
		spin_unlock_irq(&ctx->ctx_lock);
399
		io_schedule();
Linus Torvalds's avatar
Linus Torvalds committed
400
		set_task_state(tsk, TASK_UNINTERRUPTIBLE);
401
		spin_lock_irq(&ctx->ctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
402
403
404
	}
	__set_task_state(tsk, TASK_RUNNING);
	remove_wait_queue(&ctx->wait, &wait);
405
406
407

out:
	spin_unlock_irq(&ctx->ctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
408
409
410
411
412
}

/* wait_on_sync_kiocb:
 *	Waits on the given sync kiocb to complete.
 */
413
ssize_t wait_on_sync_kiocb(struct kiocb *iocb)
Linus Torvalds's avatar
Linus Torvalds committed
414
415
416
417
418
{
	while (iocb->ki_users) {
		set_current_state(TASK_UNINTERRUPTIBLE);
		if (!iocb->ki_users)
			break;
419
		io_schedule();
Linus Torvalds's avatar
Linus Torvalds committed
420
421
422
423
	}
	__set_current_state(TASK_RUNNING);
	return iocb->ki_user_data;
}
424
EXPORT_SYMBOL(wait_on_sync_kiocb);
Linus Torvalds's avatar
Linus Torvalds committed
425
426
427
428
429
430
431
432

/* 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).
 */
433
void exit_aio(struct mm_struct *mm)
Linus Torvalds's avatar
Linus Torvalds committed
434
{
435
436
437
438
439
440
	struct kioctx *ctx;

	while (!hlist_empty(&mm->ioctx_list)) {
		ctx = hlist_entry(mm->ioctx_list.first, struct kioctx, list);
		hlist_del_rcu(&ctx->list);

441
		kill_ctx(ctx);
Linus Torvalds's avatar
Linus Torvalds committed
442
443
444
445
446
447

		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);
448
449
450
451
452
453
454
455
456
		/*
		 * 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.
		 */
		ctx->ring_info.mmap_size = 0;
Linus Torvalds's avatar
Linus Torvalds committed
457
458
459
460
461
462
463
464
465
466
467
468
469
470
		put_ioctx(ctx);
	}
}

/* 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.
 */
471
static struct kiocb *__aio_get_req(struct kioctx *ctx)
Linus Torvalds's avatar
Linus Torvalds committed
472
473
474
475
476
477
478
{
	struct kiocb *req = NULL;

	req = kmem_cache_alloc(kiocb_cachep, GFP_KERNEL);
	if (unlikely(!req))
		return NULL;

Zach Brown's avatar
Zach Brown committed
479
	req->ki_flags = 0;
Linus Torvalds's avatar
Linus Torvalds committed
480
481
482
483
484
485
486
	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;
487
	req->ki_iovec = NULL;
488
	req->ki_eventfd = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
489

Jeff Moyer's avatar
Jeff Moyer committed
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
	return req;
}

/*
 * struct kiocb's are allocated in batches to reduce the number of
 * times the ctx lock is acquired and released.
 */
#define KIOCB_BATCH_SIZE	32L
struct kiocb_batch {
	struct list_head head;
	long count; /* number of requests left to allocate */
};

static void kiocb_batch_init(struct kiocb_batch *batch, long total)
{
	INIT_LIST_HEAD(&batch->head);
	batch->count = total;
}

509
static void kiocb_batch_free(struct kioctx *ctx, struct kiocb_batch *batch)
Jeff Moyer's avatar
Jeff Moyer committed
510
511
512
{
	struct kiocb *req, *n;

513
514
515
516
	if (list_empty(&batch->head))
		return;

	spin_lock_irq(&ctx->ctx_lock);
Jeff Moyer's avatar
Jeff Moyer committed
517
518
	list_for_each_entry_safe(req, n, &batch->head, ki_batch) {
		list_del(&req->ki_batch);
519
		list_del(&req->ki_list);
Jeff Moyer's avatar
Jeff Moyer committed
520
		kmem_cache_free(kiocb_cachep, req);
521
		ctx->reqs_active--;
Jeff Moyer's avatar
Jeff Moyer committed
522
	}
523
524
	if (unlikely(!ctx->reqs_active && ctx->dead))
		wake_up_all(&ctx->wait);
525
	spin_unlock_irq(&ctx->ctx_lock);
Jeff Moyer's avatar
Jeff Moyer committed
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
}

/*
 * Allocate a batch of kiocbs.  This avoids taking and dropping the
 * context lock a lot during setup.
 */
static int kiocb_batch_refill(struct kioctx *ctx, struct kiocb_batch *batch)
{
	unsigned short allocated, to_alloc;
	long avail;
	struct kiocb *req, *n;
	struct aio_ring *ring;

	to_alloc = min(batch->count, KIOCB_BATCH_SIZE);
	for (allocated = 0; allocated < to_alloc; allocated++) {
		req = __aio_get_req(ctx);
		if (!req)
			/* allocation failed, go with what we've got */
			break;
		list_add(&req->ki_batch, &batch->head);
	}

	if (allocated == 0)
		goto out;

Linus Torvalds's avatar
Linus Torvalds committed
551
	spin_lock_irq(&ctx->ctx_lock);
Jeff Moyer's avatar
Jeff Moyer committed
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
	ring = kmap_atomic(ctx->ring_info.ring_pages[0]);

	avail = aio_ring_avail(&ctx->ring_info, ring) - ctx->reqs_active;
	BUG_ON(avail < 0);
	if (avail < allocated) {
		/* Trim back the number of requests. */
		list_for_each_entry_safe(req, n, &batch->head, ki_batch) {
			list_del(&req->ki_batch);
			kmem_cache_free(kiocb_cachep, req);
			if (--allocated <= avail)
				break;
		}
	}

	batch->count -= allocated;
	list_for_each_entry(req, &batch->head, ki_batch) {
Linus Torvalds's avatar
Linus Torvalds committed
568
569
570
571
		list_add(&req->ki_list, &ctx->active_reqs);
		ctx->reqs_active++;
	}

Jeff Moyer's avatar
Jeff Moyer committed
572
573
	kunmap_atomic(ring);
	spin_unlock_irq(&ctx->ctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
574

Jeff Moyer's avatar
Jeff Moyer committed
575
576
out:
	return allocated;
Linus Torvalds's avatar
Linus Torvalds committed
577
578
}

Jeff Moyer's avatar
Jeff Moyer committed
579
580
static inline struct kiocb *aio_get_req(struct kioctx *ctx,
					struct kiocb_batch *batch)
Linus Torvalds's avatar
Linus Torvalds committed
581
582
{
	struct kiocb *req;
Jeff Moyer's avatar
Jeff Moyer committed
583
584
585
586
587
588

	if (list_empty(&batch->head))
		if (kiocb_batch_refill(ctx, batch) == 0)
			return NULL;
	req = list_first_entry(&batch->head, struct kiocb, ki_batch);
	list_del(&req->ki_batch);
Linus Torvalds's avatar
Linus Torvalds committed
589
590
591
592
593
	return req;
}

static inline void really_put_req(struct kioctx *ctx, struct kiocb *req)
{
594
595
	assert_spin_locked(&ctx->ctx_lock);

596
597
	if (req->ki_eventfd != NULL)
		eventfd_ctx_put(req->ki_eventfd);
Linus Torvalds's avatar
Linus Torvalds committed
598
599
	if (req->ki_dtor)
		req->ki_dtor(req);
600
601
	if (req->ki_iovec != &req->ki_inline_vec)
		kfree(req->ki_iovec);
Linus Torvalds's avatar
Linus Torvalds committed
602
603
604
605
	kmem_cache_free(kiocb_cachep, req);
	ctx->reqs_active--;

	if (unlikely(!ctx->reqs_active && ctx->dead))
606
		wake_up_all(&ctx->wait);
Linus Torvalds's avatar
Linus Torvalds committed
607
608
609
610
611
}

/* __aio_put_req
 *	Returns true if this put was the last user of the request.
 */
612
static void __aio_put_req(struct kioctx *ctx, struct kiocb *req)
Linus Torvalds's avatar
Linus Torvalds committed
613
{
Al Viro's avatar
Al Viro committed
614
615
	dprintk(KERN_DEBUG "aio_put(%p): f_count=%ld\n",
		req, atomic_long_read(&req->ki_filp->f_count));
Linus Torvalds's avatar
Linus Torvalds committed
616

617
618
	assert_spin_locked(&ctx->ctx_lock);

619
	req->ki_users--;
620
	BUG_ON(req->ki_users < 0);
Linus Torvalds's avatar
Linus Torvalds committed
621
	if (likely(req->ki_users))
622
		return;
Linus Torvalds's avatar
Linus Torvalds committed
623
624
625
626
	list_del(&req->ki_list);		/* remove from active_reqs */
	req->ki_cancel = NULL;
	req->ki_retry = NULL;

627
628
629
	fput(req->ki_filp);
	req->ki_filp = NULL;
	really_put_req(ctx, req);
Linus Torvalds's avatar
Linus Torvalds committed
630
631
632
633
634
635
}

/* aio_put_req
 *	Returns true if this put was the last user of the kiocb,
 *	false if the request is still in use.
 */
636
void aio_put_req(struct kiocb *req)
Linus Torvalds's avatar
Linus Torvalds committed
637
638
639
{
	struct kioctx *ctx = req->ki_ctx;
	spin_lock_irq(&ctx->ctx_lock);
640
	__aio_put_req(ctx, req);
Linus Torvalds's avatar
Linus Torvalds committed
641
642
	spin_unlock_irq(&ctx->ctx_lock);
}
643
EXPORT_SYMBOL(aio_put_req);
Linus Torvalds's avatar
Linus Torvalds committed
644

645
static struct kioctx *lookup_ioctx(unsigned long ctx_id)
Linus Torvalds's avatar
Linus Torvalds committed
646
{
647
	struct mm_struct *mm = current->mm;
648
	struct kioctx *ctx, *ret = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
649

650
651
	rcu_read_lock();

652
	hlist_for_each_entry_rcu(ctx, &mm->ioctx_list, list) {
Nick Piggin's avatar
Nick Piggin committed
653
654
655
656
657
658
659
		/*
		 * RCU protects us against accessing freed memory but
		 * we have to be careful not to get a reference when the
		 * reference count already dropped to 0 (ctx->dead test
		 * is unreliable because of races).
		 */
		if (ctx->user_id == ctx_id && !ctx->dead && try_get_ioctx(ctx)){
660
			ret = ctx;
Linus Torvalds's avatar
Linus Torvalds committed
661
662
			break;
		}
663
	}
Linus Torvalds's avatar
Linus Torvalds committed
664

665
	rcu_read_unlock();
666
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
667
668
669
670
671
}

/* aio_complete
 *	Called when the io request on the given iocb is complete.
 */
672
void aio_complete(struct kiocb *iocb, long res, long res2)
Linus Torvalds's avatar
Linus Torvalds committed
673
674
675
676
677
678
679
680
{
	struct kioctx	*ctx = iocb->ki_ctx;
	struct aio_ring_info	*info;
	struct aio_ring	*ring;
	struct io_event	*event;
	unsigned long	flags;
	unsigned long	tail;

681
682
683
684
685
686
	/*
	 * Special case handling for sync iocbs:
	 *  - events go directly into the iocb for fast handling
	 *  - the sync task with the iocb in its stack holds the single iocb
	 *    ref, no other paths have a way to get another ref
	 *  - the sync task helpfully left a reference to itself in the iocb
Linus Torvalds's avatar
Linus Torvalds committed
687
688
	 */
	if (is_sync_kiocb(iocb)) {
689
		BUG_ON(iocb->ki_users != 1);
Linus Torvalds's avatar
Linus Torvalds committed
690
		iocb->ki_user_data = res;
691
		iocb->ki_users = 0;
Linus Torvalds's avatar
Linus Torvalds committed
692
		wake_up_process(iocb->ki_obj.tsk);
693
		return;
Linus Torvalds's avatar
Linus Torvalds committed
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
	}

	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);

	/*
	 * cancelled requests don't get events, userland was given one
	 * when the event got cancelled.
	 */
	if (kiocbIsCancelled(iocb))
		goto put_rq;

713
	ring = kmap_atomic(info->ring_pages[0]);
Linus Torvalds's avatar
Linus Torvalds committed
714
715

	tail = info->tail;
716
	event = aio_ring_event(info, tail);
717
718
	if (++tail >= info->nr)
		tail = 0;
Linus Torvalds's avatar
Linus Torvalds committed
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736

	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);

	/* 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;
	ring->tail = tail;

737
738
	put_aio_ring_event(event);
	kunmap_atomic(ring);
Linus Torvalds's avatar
Linus Torvalds committed
739
740

	pr_debug("added to ring %p at [%lu]\n", iocb, tail);
Davide Libenzi's avatar
Davide Libenzi committed
741
742
743
744
745
746

	/*
	 * Check if the user asked us to deliver the result through an
	 * eventfd. The eventfd_signal() function is safe to be called
	 * from IRQ context.
	 */
747
	if (iocb->ki_eventfd != NULL)
Davide Libenzi's avatar
Davide Libenzi committed
748
749
		eventfd_signal(iocb->ki_eventfd, 1);

Linus Torvalds's avatar
Linus Torvalds committed
750
751
put_rq:
	/* everything turned out well, dispose of the aiocb. */
752
	__aio_put_req(ctx, iocb);
Linus Torvalds's avatar
Linus Torvalds committed
753

754
755
756
757
758
759
760
761
	/*
	 * We have to order our ring_info tail store above and test
	 * of the wait list below outside the wait lock.  This is
	 * like in wake_up_bit() where clearing a bit has to be
	 * ordered with the unlocked test.
	 */
	smp_mb();

Linus Torvalds's avatar
Linus Torvalds committed
762
763
764
	if (waitqueue_active(&ctx->wait))
		wake_up(&ctx->wait);

765
	spin_unlock_irqrestore(&ctx->ctx_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
766
}
767
EXPORT_SYMBOL(aio_complete);
Linus Torvalds's avatar
Linus Torvalds committed
768
769
770
771
772
773
774
775
776
777
778
779
780
781

/* 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).
 */
static int aio_read_evt(struct kioctx *ioctx, struct io_event *ent)
{
	struct aio_ring_info *info = &ioctx->ring_info;
	struct aio_ring *ring;
	unsigned long head;
	int ret = 0;

782
	ring = kmap_atomic(info->ring_pages[0]);
Linus Torvalds's avatar
Linus Torvalds committed
783
784
785
786
787
788
789
790
791
792
793
	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) {
794
		struct io_event *evp = aio_ring_event(info, head);
Linus Torvalds's avatar
Linus Torvalds committed
795
796
797
798
799
		*ent = *evp;
		head = (head + 1) % info->nr;
		smp_mb(); /* finish reading the event before updatng the head */
		ring->head = head;
		ret = 1;
800
		put_aio_ring_event(evp);
Linus Torvalds's avatar
Linus Torvalds committed
801
802
803
804
805
806
	}
	spin_unlock(&info->ring_lock);

out:
	dprintk("leaving aio_read_evt: %d  h%lu t%lu\n", ret,
		 (unsigned long)ring->head, (unsigned long)ring->tail);
807
	kunmap_atomic(ring);
Linus Torvalds's avatar
Linus Torvalds committed
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
	return ret;
}

struct aio_timeout {
	struct timer_list	timer;
	int			timed_out;
	struct task_struct	*p;
};

static void timeout_func(unsigned long data)
{
	struct aio_timeout *to = (struct aio_timeout *)data;

	to->timed_out = 1;
	wake_up_process(to->p);
}

static inline void init_timeout(struct aio_timeout *to)
{
827
	setup_timer_on_stack(&to->timer, timeout_func, (unsigned long) to);
Linus Torvalds's avatar
Linus Torvalds committed
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
	to->timed_out = 0;
	to->p = current;
}

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);
}

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;

	/* needed to zero any padding within an entry (there shouldn't be 
	 * any, but C is fun!
	 */
	memset(&ent, 0, sizeof(ent));
	ret = 0;
	while (likely(i < nr)) {
		ret = aio_read_evt(ctx, &ent);
		if (unlikely(ret <= 0))
			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;

		/* Good, event copied to userland, update counts. */
		event ++;
		i ++;
	}

	if (min_nr <= i)
		return i;
	if (ret)
		return ret;

	/* End fast path */

	init_timeout(&to);
	if (timeout) {
		struct timespec	ts;
		ret = -EFAULT;
		if (unlikely(copy_from_user(&ts, timeout, sizeof(ts))))
			goto out;

		set_timeout(start_jiffies, &to, &ts);
	}

	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;
912
913
914
915
			if (unlikely(ctx->dead)) {
				ret = -EINVAL;
				break;
			}
Linus Torvalds's avatar
Linus Torvalds committed
916
917
			if (to.timed_out)	/* Only check after read evt */
				break;
918
919
920
921
922
923
			/* Try to only show up in io wait if there are ops
			 *  in flight */
			if (ctx->reqs_active)
				io_schedule();
			else
				schedule();
Linus Torvalds's avatar
Linus Torvalds committed
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
			if (signal_pending(tsk)) {
				ret = -EINTR;
				break;
			}
			/*ret = aio_read_evt(ctx, &ent);*/
		} while (1) ;

		set_task_state(tsk, TASK_RUNNING);
		remove_wait_queue(&ctx->wait, &wait);

		if (unlikely(ret <= 0))
			break;

		ret = -EFAULT;
		if (unlikely(copy_to_user(event, &ent, sizeof(ent)))) {
			dprintk("aio: lost an event due to EFAULT.\n");
			break;
		}

		/* Good, event copied to userland, update counts. */
		event ++;
		i ++;
	}

	if (timeout)
		clear_timeout(&to);
out:
951
	destroy_timer_on_stack(&to.timer);
Linus Torvalds's avatar
Linus Torvalds committed
952
953
954
955
956
957
958
959
960
961
962
963
	return i ? i : ret;
}

/* 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)
{
	struct mm_struct *mm = current->mm;
	int was_dead;

	/* delete the entry from the list is someone else hasn't already */
964
	spin_lock(&mm->ioctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
965
966
	was_dead = ioctx->dead;
	ioctx->dead = 1;
967
968
	hlist_del_rcu(&ioctx->list);
	spin_unlock(&mm->ioctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
969
970
971
972
973

	dprintk("aio_release(%p)\n", ioctx);
	if (likely(!was_dead))
		put_ioctx(ioctx);	/* twice for the list */

974
	kill_ctx(ioctx);
975
976
977
978
979
980

	/*
	 * 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.
	 */
981
	wake_up_all(&ioctx->wait);
Linus Torvalds's avatar
Linus Torvalds committed
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
}

/* sys_io_setup:
 *	Create an aio_context capable of receiving at least nr_events.
 *	ctxp must not point to an aio_context that already exists, and
 *	must be initialized to 0 prior to the call.  On successful
 *	creation of the aio_context, *ctxp is filled in with the resulting 
 *	handle.  May fail with -EINVAL if *ctxp is not initialized,
 *	if the specified nr_events exceeds internal limits.  May fail 
 *	with -EAGAIN if the specified nr_events exceeds the user's limit 
 *	of available events.  May fail with -ENOMEM if insufficient kernel
 *	resources are available.  May fail with -EFAULT if an invalid
 *	pointer is passed for ctxp.  Will fail with -ENOSYS if not
 *	implemented.
 */
997
SYSCALL_DEFINE2(io_setup, unsigned, nr_events, aio_context_t __user *, ctxp)
Linus Torvalds's avatar
Linus Torvalds committed
998
999
1000
1001
1002
1003
1004
1005
1006
1007
{
	struct kioctx *ioctx = NULL;
	unsigned long ctx;
	long ret;

	ret = get_user(ctx, ctxp);
	if (unlikely(ret))
		goto out;

	ret = -EINVAL;
1008
1009
1010
	if (unlikely(ctx || nr_events == 0)) {
		pr_debug("EINVAL: io_setup: ctx %lu nr_events %u\n",
		         ctx, nr_events);
Linus Torvalds's avatar
Linus Torvalds committed
1011
1012
1013
1014
1015
1016
1017
		goto out;
	}

	ioctx = ioctx_alloc(nr_events);
	ret = PTR_ERR(ioctx);
	if (!IS_ERR(ioctx)) {
		ret = put_user(ioctx->user_id, ctxp);
1018
1019
1020
		if (ret)
			io_destroy(ioctx);
		put_ioctx(ioctx);
Linus Torvalds's avatar
Linus Torvalds committed
1021
1022
1023
1024
1025
1026
1027
1028
1029
	}

out:
	return ret;
}

/* sys_io_destroy:
 *	Destroy the aio_context specified.  May cancel any outstanding 
 *	AIOs and block on completion.  Will fail with -ENOSYS if not
1030
 *	implemented.  May fail with -EINVAL if the context pointed to
Linus Torvalds's avatar
Linus Torvalds committed
1031
1032
 *	is invalid.
 */
1033
SYSCALL_DEFINE1(io_destroy, aio_context_t, ctx)
Linus Torvalds's avatar
Linus Torvalds committed
1034
1035
1036
1037
{
	struct kioctx *ioctx = lookup_ioctx(ctx);
	if (likely(NULL != ioctx)) {
		io_destroy(ioctx);
1038
		put_ioctx(ioctx);
Linus Torvalds's avatar
Linus Torvalds committed
1039
1040
1041
1042
1043
1044
		return 0;
	}
	pr_debug("EINVAL: io_destroy: invalid context id\n");
	return -EINVAL;
}

1045
static void aio_advance_iovec(struct kiocb *iocb, ssize_t ret)
Linus Torvalds's avatar
Linus Torvalds committed
1046
{
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
	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++;
1060
		}
1061
	}
Linus Torvalds's avatar
Linus Torvalds committed
1062

1063
1064
1065
	/* the caller should not have done more io than what fit in
	 * the remaining iovecs */
	BUG_ON(ret > 0 && iocb->ki_left == 0);
Linus Torvalds's avatar
Linus Torvalds committed
1066
1067
}

1068
static ssize_t aio_rw_vect_retry(struct kiocb *iocb)
Linus Torvalds's avatar
Linus Torvalds committed
1069
1070
{
	struct file *file = iocb->ki_filp;
1071
1072
1073
1074
	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);
Linus Torvalds's avatar
Linus Torvalds committed
1075
	ssize_t ret = 0;
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
	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;
	}
Linus Torvalds's avatar
Linus Torvalds committed
1086

1087
1088
1089
1090
	/* This matches the pread()/pwrite() logic */
	if (iocb->ki_pos < 0)
		return -EINVAL;

1091
1092
	if (opcode == IOCB_CMD_PWRITEV)
		file_start_write(file);
1093
	do {
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
		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))));
1105
1106
	if (opcode == IOCB_CMD_PWRITEV)
		file_end_write(file);
Linus Torvalds's avatar
Linus Torvalds committed
1107

1108
1109
	/* This means we must have transferred all that we could */
	/* No need to retry anymore */
Linus Torvalds's avatar
Linus Torvalds committed
1110
1111
1112
	if ((ret == 0) || (iocb->ki_left == 0))
		ret = iocb->ki_nbytes - iocb->ki_left;

1113
1114
1115
	/* If we managed to write some out we return that, rather than
	 * the eventual error. */
	if (opcode == IOCB_CMD_PWRITEV
Zach Brown's avatar
Zach Brown committed
1116
	    && ret < 0 && ret != -EIOCBQUEUED
1117
1118
1119
	    && iocb->ki_nbytes - iocb->ki_left)
		ret = iocb->ki_nbytes - iocb->ki_left;

Linus Torvalds's avatar
Linus Torvalds committed
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
	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;
}

static ssize_t aio_fsync(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, 0);
	return ret;
}

1143
static ssize_t aio_setup_vectored_rw(int type, struct kiocb *kiocb, bool compat)
1144
1145
1146
{
	ssize_t ret;

1147
1148
1149
1150
1151
#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,
1152
				&kiocb->ki_iovec);
1153
1154
1155
1156
1157
	else
#endif
		ret = rw_copy_check_uvector(type,
				(struct iovec __user *)kiocb->ki_buf,
				kiocb->ki_nbytes, 1, &kiocb->ki_inline_vec,
1158
				&kiocb->ki_iovec);
1159
1160
1161
	if (ret < 0)
		goto out;

1162
1163
1164
1165
	ret = rw_verify_area(type, kiocb->ki_filp, &kiocb->ki_pos, ret);
	if (ret < 0)
		goto out;

1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
	kiocb->ki_nr_segs = kiocb->ki_nbytes;
	kiocb->ki_cur_seg = 0;
	/* ki_nbytes/left now reflect bytes instead of segs */
	kiocb->ki_nbytes = ret;
	kiocb->ki_left = ret;

	ret = 0;
out:
	return ret;
}

1177
static ssize_t aio_setup_single_vector(int type, struct file * file, struct kiocb *kiocb)
1178
{
1179
1180
1181
1182
1183
1184
	int bytes;

	bytes = rw_verify_area(type, file, &kiocb->ki_pos, kiocb->ki_left);
	if (bytes < 0)
		return bytes;

1185
1186
	kiocb->ki_iovec = &kiocb->ki_inline_vec;
	kiocb->ki_iovec->iov_base = kiocb->ki_buf;
1187
	kiocb->ki_iovec->iov_len = bytes;
1188
1189
1190
1191
1192
	kiocb->ki_nr_segs = 1;
	kiocb->ki_cur_seg = 0;
	return 0;
}

Linus Torvalds's avatar
Linus Torvalds committed
1193
1194
1195
1196
1197
/*
 * aio_setup_iocb:
 *	Performs the initial checks and aio retry method
 *	setup for the kiocb at the time of io submission.
 */
1198
static ssize_t aio_setup_iocb(struct kiocb *kiocb, bool compat)
Linus Torvalds's avatar
Linus Torvalds committed
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
{
	struct file *file = kiocb->ki_filp;
	ssize_t ret = 0;

	switch (kiocb->ki_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;
1212
		ret = aio_setup_single_vector(READ, file, kiocb);
1213
1214
		if (ret)
			break;
Linus Torvalds's avatar
Linus Torvalds committed
1215
1216
		ret = -EINVAL;
		if (file->f_op->aio_read)
1217
			kiocb->ki_retry = aio_rw_vect_retry;
Linus Torvalds's avatar
Linus Torvalds committed
1218
1219
1220
1221
1222
1223
1224
1225
1226
		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;
1227
		ret = aio_setup_single_vector(WRITE, file, kiocb);
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
		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;
1238
		ret = aio_setup_vectored_rw(READ, kiocb, compat);
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
		if (ret)
			break;
		ret = -EINVAL;
		if (file->f_op->aio_read)
			kiocb->ki_retry = aio_rw_vect_retry;
		break;
	case IOCB_CMD_PWRITEV:
		ret = -EBADF;
		if (unlikely(!(file->f_mode & FMODE_WRITE)))
			break;
1249
		ret = aio_setup_vectored_rw(WRITE, kiocb, compat);
1250
1251
		if (ret)
			break;
Linus Torvalds's avatar
Linus Torvalds committed
1252
1253
		ret = -EINVAL;
		if (file->f_op->aio_write)
1254
			kiocb->ki_retry = aio_rw_vect_retry;
Linus Torvalds's avatar
Linus Torvalds committed
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
		break;
	case IOCB_CMD_FDSYNC:
		ret = -EINVAL;
		if (file->f_op->aio_fsync)
			kiocb->ki_retry = aio_fdsync;
		break;
	case IOCB_CMD_FSYNC:
		ret = -EINVAL;
		if (file->f_op->aio_fsync)
			kiocb->ki_retry = aio_fsync;
		break;
	default:
		dprintk("EINVAL: io_submit: no operation provided\n");
		ret = -EINVAL;
	}

	if (!kiocb->ki_retry)
		return ret;

	return 0;
}

1277
static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb,
Jeff Moyer's avatar
Jeff Moyer committed
1278
1279
			 struct iocb *iocb, struct kiocb_batch *batch,
			 bool compat)
Linus Torvalds's avatar
Linus Torvalds committed
1280
1281
1282
1283
1284
1285
{
	struct kiocb *req;
	struct file *file;
	ssize_t ret;

	/* enforce forwards compatibility on users */
1286
	if (unlikely(iocb->aio_reserved1 || iocb->aio_reserved2)) {
Linus Torvalds's avatar
Linus Torvalds committed
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
		pr_debug("EINVAL: io_submit: reserve field set\n");
		return -EINVAL;
	}

	/* prevent overflows */
	if (unlikely(
	    (iocb->aio_buf != (unsigned long)iocb->aio_buf) ||
	    (iocb->aio_nbytes != (size_t)iocb->aio_nbytes) ||
	    ((ssize_t)iocb->aio_nbytes < 0)
	   )) {
		pr_debug("EINVAL: io_submit: overflow check\n");
		return -EINVAL;
	}

	file = fget(iocb->aio_fildes);
	if (unlikely(!file))
		return -EBADF;

Jeff Moyer's avatar
Jeff Moyer committed
1305
	req = aio_get_req(ctx, batch);  /* returns with 2 references to req */
Linus Torvalds's avatar
Linus Torvalds committed
1306
1307
1308
1309
	if (unlikely(!req)) {
		fput(file);
		return -EAGAIN;
	}
1310
	req->ki_filp = file;
1311
1312
1313
1314
1315
1316
1317
	if (iocb->aio_flags & IOCB_FLAG_RESFD) {
		/*
		 * If the IOCB_FLAG_RESFD flag of aio_flags is set, get an
		 * instance of the file* now. The file descriptor must be
		 * an eventfd() fd, and will be signaled for each completed
		 * event using the eventfd_signal() function.
		 */
1318
		req->ki_eventfd = eventfd_ctx_fdget((int) iocb->aio_resfd);
1319
		if (IS_ERR(req->ki_eventfd)) {
1320
			ret = PTR_ERR(req->ki_eventfd);
1321
			req->ki_eventfd = NULL;
1322
1323
1324
			goto out_put_req;
		}
	}
Linus Torvalds's avatar
Linus Torvalds committed
1325

1326
	ret = put_user(req->ki_key, &user_iocb->aio_key);
Linus Torvalds's avatar
Linus Torvalds committed
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
	if (unlikely(ret)) {
		dprintk("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_buf = (char __user *)(unsigned long)iocb->aio_buf;
	req->ki_left = req->ki_nbytes = iocb->aio_nbytes;
	req->ki_opcode = iocb->aio_lio_opcode;

1340
	ret = aio_setup_iocb(req, compat);
Linus Torvalds's avatar
Linus Torvalds committed
1341
1342
1343
1344
1345

	if (ret)
		goto out_put_req;

	spin_lock_irq(&ctx->ctx_lock);
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
	/*
	 * We could have raced with io_destroy() and are currently holding a
	 * reference to ctx which should be destroyed. We cannot submit IO
	 * since ctx gets freed as soon as io_submit() puts its reference.  The
	 * check here is reliable: io_destroy() sets ctx->dead before waiting
	 * for outstanding IO and the barrier between these two is realized by
	 * unlock of mm->ioctx_lock and lock of ctx->ctx_lock.  Analogously we
	 * increment ctx->reqs_active before checking for ctx->dead and the
	 * barrier is realized by unlock and lock of ctx->ctx_lock. Thus if we
	 * don't see ctx->dead set here, io_destroy() waits for our IO to
	 * finish.
	 */
Zach Brown's avatar
Zach Brown committed
1358
	if (ctx->dead)
1359
		ret = -EINVAL;
Zach Brown's avatar
Zach Brown committed
1360
1361
	spin_unlock_irq(&ctx->ctx_lock);
	if (ret)
1362
		goto out_put_req;
Zach Brown's avatar
Zach Brown committed
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378

	if (unlikely(kiocbIsCancelled(req)))
		ret = -EINTR;
	else
		ret = req->ki_retry(req);

	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);
1379
	}
Jeff Moyer's avatar
Jeff Moyer committed
1380

Linus Torvalds's avatar
Linus Torvalds committed
1381
1382
1383
1384
1385
1386
1387
1388
1389
	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 */
	return ret;
}

1390
1391
long do_io_submit(aio_context_t ctx_id, long nr,
		  struct iocb __user *__user *iocbpp, bool compat)
Linus Torvalds's avatar
Linus Torvalds committed
1392
1393
1394
{
	struct kioctx *ctx;
	long ret = 0;
Jeff Moyer's avatar
Jeff Moyer committed
1395
	int i = 0;
Shaohua Li's avatar
Shaohua Li committed
1396
	struct blk_plug plug;
Jeff Moyer's avatar
Jeff Moyer committed
1397
	struct kiocb_batch batch;
Linus Torvalds's avatar
Linus Torvalds committed
1398
1399
1400
1401

	if (unlikely(nr < 0))
		return -EINVAL;

1402
1403
1404
	if (unlikely(nr > LONG_MAX/sizeof(*iocbpp)))
		nr = LONG_MAX/sizeof(*iocbpp);

Linus Torvalds's avatar
Linus Torvalds committed
1405
1406
1407
1408
1409
1410
1411
1412
1413
	if (unlikely(!access_ok(VERIFY_READ, iocbpp, (nr*sizeof(*iocbpp)))))
		return -EFAULT;

	ctx = lookup_ioctx(ctx_id);
	if (unlikely(!ctx)) {
		pr_debug("EINVAL: io_submit: invalid context id\n");
		return -EINVAL;
	}

Jeff Moyer's avatar
Jeff Moyer committed
1414
1415
	kiocb_batch_init(&batch, nr);

Shaohua Li's avatar
Shaohua Li committed
1416
1417
	blk_start_plug(&plug);

Linus Torvalds's avatar
Linus Torvalds committed
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
	/*
	 * AKPM: should this return a partial result if some of the IOs were
	 * successfully submitted?
	 */
	for (i=0; i<nr; i++) {
		struct iocb __user *user_iocb;
		struct iocb tmp;

		if (unlikely(__get_user(user_iocb, iocbpp + i))) {
			ret = -EFAULT;
			break;
		}

		if (unlikely(copy_from_user(&tmp, user_iocb, sizeof(tmp)))) {
			ret = -EFAULT;
			break;
		}

Jeff Moyer's avatar
Jeff Moyer committed
1436
		ret = io_submit_one(ctx, user_iocb, &tmp, &batch, compat);
Linus Torvalds's avatar
Linus Torvalds committed
1437
1438
1439
		if (ret)
			break;
	}
Shaohua Li's avatar
Shaohua Li committed
1440
	blk_finish_plug(&plug);
Linus Torvalds's avatar
Linus Torvalds committed
1441

1442
	kiocb_batch_free(ctx, &batch);
Linus Torvalds's avatar
Linus Torvalds committed
1443
1444
1445
1446
	put_ioctx(ctx);
	return i ? i : ret;
}

1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
/* sys_io_submit:
 *	Queue the nr iocbs pointed to by iocbpp for processing.  Returns
 *	the number of iocbs queued.  May return -EINVAL if the aio_context
 *	specified by ctx_id is invalid, if nr is < 0, if the iocb at
 *	*iocbpp[0] is not properly initialized, if the operation specified
 *	is invalid for the file descriptor in the iocb.  May fail with
 *	-EFAULT if any of the data structures point to invalid data.  May
 *	fail with -EBADF if the file descriptor specified in the first
 *	iocb is invalid.  May fail with -EAGAIN if insufficient resources
 *	are available to queue any iocbs.  Will return 0 if nr is 0.  Will
 *	fail with -ENOSYS if not implemented.
 */
SYSCALL_DEFINE3(io_submit, aio_context_t, ctx_id, long, nr,
		struct iocb __user * __user *, iocbpp)
{
	return do_io_submit(ctx_id, nr, iocbpp, 0);
}

Linus Torvalds's avatar
Linus Torvalds committed
1465
1466
1467
/* lookup_kiocb
 *	Finds a given iocb for cancellation.
 */
1468
1469
static struct kiocb *lookup_kiocb(struct kioctx *ctx, struct iocb __user *iocb,
				  u32 key)
Linus Torvalds's avatar
Linus Torvalds committed
1470
1471
{
	struct list_head *pos;
1472
1473
1474

	assert_spin_locked(&ctx->ctx_lock);

Linus Torvalds's avatar
Linus Torvalds committed
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
	/* 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)
			return kiocb;
	}
	return NULL;
}

/* sys_io_cancel:
 *	Attempts to cancel an iocb previously passed to io_submit.  If
 *	the operation is successfully cancelled, the resulting event is
 *	copied into the memory pointed to by result without being placed
 *	into the completion queue and 0 is returned.  May fail with
 *	-EFAULT if any of the data structures pointed to are invalid.
 *	May fail with -EINVAL if aio_context specified by ctx_id is
 *	invalid.  May fail with -EAGAIN if the iocb specified was not
 *	cancelled.  Will fail with -ENOSYS if not implemented.
 */
1494
1495
SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb,
		struct io_event __user *, result)
Linus Torvalds's avatar
Linus Torvalds committed
1496
{
Kent Overstreet's avatar
Kent Overstreet committed
1497
	struct io_event res;
Linus Torvalds's avatar
Linus Torvalds committed
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
	struct kioctx *ctx;
	struct kiocb *kiocb;
	u32 key;
	int ret;

	ret = get_user(key, &iocb->aio_key);
	if (unlikely(ret))
		return -EFAULT;

	ctx = lookup_ioctx(ctx_id);
	if (unlikely(!ctx))
		return -EINVAL;

	spin_lock_irq(&ctx->ctx_lock);
Kent Overstreet's avatar
Kent Overstreet committed
1512

Linus Torvalds's avatar
Linus Torvalds committed
1513
	kiocb = lookup_kiocb(ctx, iocb, key);
Kent Overstreet's avatar