aio.c 38.4 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
/*
 *	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.
 */
Kent Overstreet's avatar
Kent Overstreet committed
11
12
#define pr_fmt(fmt) "%s: " fmt, __func__

Linus Torvalds's avatar
Linus Torvalds committed
13
14
15
16
17
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/errno.h>
#include <linux/time.h>
#include <linux/aio_abi.h>
18
#include <linux/export.h>
Linus Torvalds's avatar
Linus Torvalds committed
19
#include <linux/syscalls.h>
20
#include <linux/backing-dev.h>
21
#include <linux/uio.h>
Linus Torvalds's avatar
Linus Torvalds committed
22
23
24
25
26
27

#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

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

42
43
44
45
46
47
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
#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
103
/*------ sysctl variables----*/
104
105
106
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
107
108
/*----end sysctl variables---*/

109
110
static struct kmem_cache	*kiocb_cachep;
static struct kmem_cache	*kioctx_cachep;
Linus Torvalds's avatar
Linus Torvalds committed
111
112
113
114
115
116
117

/* 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)
{
118
119
	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
120

Kent Overstreet's avatar
Kent Overstreet committed
121
	pr_debug("sizeof(struct page) = %zu\n", sizeof(struct page));
Linus Torvalds's avatar
Linus Torvalds committed
122
123
124

	return 0;
}
125
__initcall(aio_setup);
Linus Torvalds's avatar
Linus Torvalds committed
126
127
128
129
130
131
132
133
134

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]);

135
	if (info->mmap_size) {
Al Viro's avatar
Al Viro committed
136
		vm_munmap(info->mmap_base, info->mmap_size);
137
	}
Linus Torvalds's avatar
Linus Torvalds committed
138
139
140
141
142
143
144
145
146
147
148
149

	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
150
	struct mm_struct *mm = current->mm;
151
	unsigned long size, populate;
Linus Torvalds's avatar
Linus Torvalds committed
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
	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) {
169
		info->ring_pages = kcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL);
Linus Torvalds's avatar
Linus Torvalds committed
170
171
172
173
174
		if (!info->ring_pages)
			return -ENOMEM;
	}

	info->mmap_size = nr_pages * PAGE_SIZE;
Kent Overstreet's avatar
Kent Overstreet committed
175
	pr_debug("attempting mmap of %lu bytes\n", info->mmap_size);
Zach Brown's avatar
Zach Brown committed
176
	down_write(&mm->mmap_sem);
177
178
	info->mmap_base = do_mmap_pgoff(NULL, 0, info->mmap_size, 
					PROT_READ|PROT_WRITE,
179
180
					MAP_ANONYMOUS|MAP_PRIVATE, 0,
					&populate);
Linus Torvalds's avatar
Linus Torvalds committed
181
	if (IS_ERR((void *)info->mmap_base)) {
Zach Brown's avatar
Zach Brown committed
182
		up_write(&mm->mmap_sem);
Linus Torvalds's avatar
Linus Torvalds committed
183
184
185
186
187
		info->mmap_size = 0;
		aio_free_ring(ctx);
		return -EAGAIN;
	}

Kent Overstreet's avatar
Kent Overstreet committed
188
	pr_debug("mmap address: 0x%08lx\n", info->mmap_base);
Zach Brown's avatar
Zach Brown committed
189
	info->nr_pages = get_user_pages(current, mm, info->mmap_base, nr_pages,
Linus Torvalds's avatar
Linus Torvalds committed
190
					1, 0, info->ring_pages, NULL);
Zach Brown's avatar
Zach Brown committed
191
	up_write(&mm->mmap_sem);
Linus Torvalds's avatar
Linus Torvalds committed
192
193
194
195
196

	if (unlikely(info->nr_pages != nr_pages)) {
		aio_free_ring(ctx);
		return -EAGAIN;
	}
197
	if (populate)
198
		mm_populate(info->mmap_base, populate);
Linus Torvalds's avatar
Linus Torvalds committed
199
200
201
202
203

	ctx->user_id = info->mmap_base;

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

204
	ring = kmap_atomic(info->ring_pages[0]);
Linus Torvalds's avatar
Linus Torvalds committed
205
206
207
208
209
210
211
	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);
212
	kunmap_atomic(ring);
Linus Torvalds's avatar
Linus Torvalds committed
213
214
215
216
217
218

	return 0;
}


/* aio_ring_event: returns a pointer to the event at the given index from
219
 * kmap_atomic().  Release the pointer with put_aio_ring_event();
Linus Torvalds's avatar
Linus Torvalds committed
220
221
222
223
224
 */
#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)

225
#define aio_ring_event(info, nr) ({					\
Linus Torvalds's avatar
Linus Torvalds committed
226
227
228
	unsigned pos = (nr) + AIO_EVENTS_OFFSET;			\
	struct io_event *__event;					\
	__event = kmap_atomic(						\
229
			(info)->ring_pages[pos / AIO_EVENTS_PER_PAGE]); \
Linus Torvalds's avatar
Linus Torvalds committed
230
231
232
233
	__event += pos % AIO_EVENTS_PER_PAGE;				\
	__event;							\
})

234
#define put_aio_ring_event(event) do {		\
Linus Torvalds's avatar
Linus Torvalds committed
235
236
	struct io_event *__event = (event);	\
	(void)__event;				\
237
	kunmap_atomic((void *)((unsigned long)__event & PAGE_MASK)); \
Linus Torvalds's avatar
Linus Torvalds committed
238
239
} while(0)

240
241
242
243
244
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);
}
245
246
247
248
249
250
251

/* __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)
{
252
	unsigned nr_events = ctx->max_reqs;
253
254
255
	BUG_ON(ctx->reqs_active);

	aio_free_ring(ctx);
256
257
258
259
260
261
	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);
	}
Kent Overstreet's avatar
Kent Overstreet committed
262
	pr_debug("freeing %p\n", ctx);
263
	call_rcu(&ctx->rcu_head, ctx_rcu_free);
264
265
}

Nick Piggin's avatar
Nick Piggin committed
266
267
268
269
270
271
272
273
274
275
276
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);
}
277

Kent Overstreet's avatar
Kent Overstreet committed
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
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
301
302
303
304
305
/* 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
306
	struct mm_struct *mm = current->mm;
Linus Torvalds's avatar
Linus Torvalds committed
307
	struct kioctx *ctx;
308
	int err = -ENOMEM;
Linus Torvalds's avatar
Linus Torvalds committed
309
310
311
312
313
314
315
316

	/* 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);
	}

317
	if (!nr_events || (unsigned long)nr_events > aio_max_nr)
Linus Torvalds's avatar
Linus Torvalds committed
318
319
		return ERR_PTR(-EAGAIN);

320
	ctx = kmem_cache_zalloc(kioctx_cachep, GFP_KERNEL);
Linus Torvalds's avatar
Linus Torvalds committed
321
322
323
324
325
	if (!ctx)
		return ERR_PTR(-ENOMEM);

	ctx->max_reqs = nr_events;

Al Viro's avatar
Al Viro committed
326
	atomic_set(&ctx->users, 2);
Linus Torvalds's avatar
Linus Torvalds committed
327
328
329
330
331
332
333
334
335
336
	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 */
337
	spin_lock(&aio_nr_lock);
338
339
	if (aio_nr + nr_events > aio_max_nr ||
	    aio_nr + nr_events < aio_nr) {
340
		spin_unlock(&aio_nr_lock);
Linus Torvalds's avatar
Linus Torvalds committed
341
		goto out_cleanup;
342
343
	}
	aio_nr += ctx->max_reqs;
344
	spin_unlock(&aio_nr_lock);
Linus Torvalds's avatar
Linus Torvalds committed
345

Jeff Moyer's avatar
Jeff Moyer committed
346
	/* now link into global list. */
347
348
349
	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
350

Kent Overstreet's avatar
Kent Overstreet committed
351
	pr_debug("allocated ioctx %p[%ld]: mm=%p mask=0x%x\n",
Zach Brown's avatar
Zach Brown committed
352
		ctx, ctx->user_id, mm, ctx->ring_info.nr);
Linus Torvalds's avatar
Linus Torvalds committed
353
354
355
	return ctx;

out_cleanup:
356
357
	err = -EAGAIN;
	aio_free_ring(ctx);
Linus Torvalds's avatar
Linus Torvalds committed
358
359
out_freectx:
	kmem_cache_free(kioctx_cachep, ctx);
Kent Overstreet's avatar
Kent Overstreet committed
360
	pr_debug("error allocating ioctx %d\n", err);
361
	return ERR_PTR(err);
Linus Torvalds's avatar
Linus Torvalds committed
362
363
}

364
/* kill_ctx
Linus Torvalds's avatar
Linus Torvalds committed
365
366
367
368
 *	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.
 */
369
static void kill_ctx(struct kioctx *ctx)
Linus Torvalds's avatar
Linus Torvalds committed
370
{
371
372
	struct task_struct *tsk = current;
	DECLARE_WAITQUEUE(wait, tsk);
Linus Torvalds's avatar
Linus Torvalds committed
373
	struct io_event res;
Kent Overstreet's avatar
Kent Overstreet committed
374
	struct kiocb *req;
375

Linus Torvalds's avatar
Linus Torvalds committed
376
377
378
	spin_lock_irq(&ctx->ctx_lock);
	ctx->dead = 1;
	while (!list_empty(&ctx->active_reqs)) {
Kent Overstreet's avatar
Kent Overstreet committed
379
380
381
382
383
		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
384
385
386
	}

	if (!ctx->reqs_active)
387
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
388
389
390
391

	add_wait_queue(&ctx->wait, &wait);
	set_task_state(tsk, TASK_UNINTERRUPTIBLE);
	while (ctx->reqs_active) {
392
		spin_unlock_irq(&ctx->ctx_lock);
393
		io_schedule();
Linus Torvalds's avatar
Linus Torvalds committed
394
		set_task_state(tsk, TASK_UNINTERRUPTIBLE);
395
		spin_lock_irq(&ctx->ctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
396
397
398
	}
	__set_task_state(tsk, TASK_RUNNING);
	remove_wait_queue(&ctx->wait, &wait);
399
400
401

out:
	spin_unlock_irq(&ctx->ctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
402
403
404
405
406
}

/* wait_on_sync_kiocb:
 *	Waits on the given sync kiocb to complete.
 */
407
ssize_t wait_on_sync_kiocb(struct kiocb *iocb)
Linus Torvalds's avatar
Linus Torvalds committed
408
409
410
411
412
{
	while (iocb->ki_users) {
		set_current_state(TASK_UNINTERRUPTIBLE);
		if (!iocb->ki_users)
			break;
413
		io_schedule();
Linus Torvalds's avatar
Linus Torvalds committed
414
415
416
417
	}
	__set_current_state(TASK_RUNNING);
	return iocb->ki_user_data;
}
418
EXPORT_SYMBOL(wait_on_sync_kiocb);
Linus Torvalds's avatar
Linus Torvalds committed
419
420
421
422
423
424
425
426

/* 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).
 */
427
void exit_aio(struct mm_struct *mm)
Linus Torvalds's avatar
Linus Torvalds committed
428
{
429
430
431
432
433
434
	struct kioctx *ctx;

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

435
		kill_ctx(ctx);
Linus Torvalds's avatar
Linus Torvalds committed
436
437
438
439
440
441

		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);
442
443
444
445
446
447
448
449
450
		/*
		 * 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
451
452
453
454
455
456
457
458
459
460
461
462
463
464
		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.
 */
465
static struct kiocb *__aio_get_req(struct kioctx *ctx)
Linus Torvalds's avatar
Linus Torvalds committed
466
467
468
469
470
471
472
{
	struct kiocb *req = NULL;

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

Zach Brown's avatar
Zach Brown committed
473
	req->ki_flags = 0;
Linus Torvalds's avatar
Linus Torvalds committed
474
475
476
477
478
479
480
	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;
481
	req->ki_iovec = NULL;
482
	req->ki_eventfd = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
483

Jeff Moyer's avatar
Jeff Moyer committed
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
	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;
}

503
static void kiocb_batch_free(struct kioctx *ctx, struct kiocb_batch *batch)
Jeff Moyer's avatar
Jeff Moyer committed
504
505
506
{
	struct kiocb *req, *n;

507
508
509
510
	if (list_empty(&batch->head))
		return;

	spin_lock_irq(&ctx->ctx_lock);
Jeff Moyer's avatar
Jeff Moyer committed
511
512
	list_for_each_entry_safe(req, n, &batch->head, ki_batch) {
		list_del(&req->ki_batch);
513
		list_del(&req->ki_list);
Jeff Moyer's avatar
Jeff Moyer committed
514
		kmem_cache_free(kiocb_cachep, req);
515
		ctx->reqs_active--;
Jeff Moyer's avatar
Jeff Moyer committed
516
	}
517
518
	if (unlikely(!ctx->reqs_active && ctx->dead))
		wake_up_all(&ctx->wait);
519
	spin_unlock_irq(&ctx->ctx_lock);
Jeff Moyer's avatar
Jeff Moyer committed
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
}

/*
 * 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
545
	spin_lock_irq(&ctx->ctx_lock);
Jeff Moyer's avatar
Jeff Moyer committed
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
	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
562
563
564
565
		list_add(&req->ki_list, &ctx->active_reqs);
		ctx->reqs_active++;
	}

Jeff Moyer's avatar
Jeff Moyer committed
566
567
	kunmap_atomic(ring);
	spin_unlock_irq(&ctx->ctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
568

Jeff Moyer's avatar
Jeff Moyer committed
569
570
out:
	return allocated;
Linus Torvalds's avatar
Linus Torvalds committed
571
572
}

Jeff Moyer's avatar
Jeff Moyer committed
573
574
static inline struct kiocb *aio_get_req(struct kioctx *ctx,
					struct kiocb_batch *batch)
Linus Torvalds's avatar
Linus Torvalds committed
575
576
{
	struct kiocb *req;
Jeff Moyer's avatar
Jeff Moyer committed
577
578
579
580
581
582

	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
583
584
585
586
587
	return req;
}

static inline void really_put_req(struct kioctx *ctx, struct kiocb *req)
{
588
589
	assert_spin_locked(&ctx->ctx_lock);

590
591
	if (req->ki_eventfd != NULL)
		eventfd_ctx_put(req->ki_eventfd);
Linus Torvalds's avatar
Linus Torvalds committed
592
593
	if (req->ki_dtor)
		req->ki_dtor(req);
594
595
	if (req->ki_iovec != &req->ki_inline_vec)
		kfree(req->ki_iovec);
Linus Torvalds's avatar
Linus Torvalds committed
596
597
598
599
	kmem_cache_free(kiocb_cachep, req);
	ctx->reqs_active--;

	if (unlikely(!ctx->reqs_active && ctx->dead))
600
		wake_up_all(&ctx->wait);
Linus Torvalds's avatar
Linus Torvalds committed
601
602
603
604
605
}

/* __aio_put_req
 *	Returns true if this put was the last user of the request.
 */
606
static void __aio_put_req(struct kioctx *ctx, struct kiocb *req)
Linus Torvalds's avatar
Linus Torvalds committed
607
{
Kent Overstreet's avatar
Kent Overstreet committed
608
609
	pr_debug("(%p): f_count=%ld\n",
		 req, atomic_long_read(&req->ki_filp->f_count));
Linus Torvalds's avatar
Linus Torvalds committed
610

611
612
	assert_spin_locked(&ctx->ctx_lock);

613
	req->ki_users--;
614
	BUG_ON(req->ki_users < 0);
Linus Torvalds's avatar
Linus Torvalds committed
615
	if (likely(req->ki_users))
616
		return;
Linus Torvalds's avatar
Linus Torvalds committed
617
618
619
620
	list_del(&req->ki_list);		/* remove from active_reqs */
	req->ki_cancel = NULL;
	req->ki_retry = NULL;

621
622
623
	fput(req->ki_filp);
	req->ki_filp = NULL;
	really_put_req(ctx, req);
Linus Torvalds's avatar
Linus Torvalds committed
624
625
626
627
628
629
}

/* aio_put_req
 *	Returns true if this put was the last user of the kiocb,
 *	false if the request is still in use.
 */
630
void aio_put_req(struct kiocb *req)
Linus Torvalds's avatar
Linus Torvalds committed
631
632
633
{
	struct kioctx *ctx = req->ki_ctx;
	spin_lock_irq(&ctx->ctx_lock);
634
	__aio_put_req(ctx, req);
Linus Torvalds's avatar
Linus Torvalds committed
635
636
	spin_unlock_irq(&ctx->ctx_lock);
}
637
EXPORT_SYMBOL(aio_put_req);
Linus Torvalds's avatar
Linus Torvalds committed
638

639
static struct kioctx *lookup_ioctx(unsigned long ctx_id)
Linus Torvalds's avatar
Linus Torvalds committed
640
{
641
	struct mm_struct *mm = current->mm;
642
	struct kioctx *ctx, *ret = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
643

644
645
	rcu_read_lock();

646
	hlist_for_each_entry_rcu(ctx, &mm->ioctx_list, list) {
Nick Piggin's avatar
Nick Piggin committed
647
648
649
650
651
652
653
		/*
		 * 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)){
654
			ret = ctx;
Linus Torvalds's avatar
Linus Torvalds committed
655
656
			break;
		}
657
	}
Linus Torvalds's avatar
Linus Torvalds committed
658

659
	rcu_read_unlock();
660
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
661
662
663
664
665
}

/* aio_complete
 *	Called when the io request on the given iocb is complete.
 */
666
void aio_complete(struct kiocb *iocb, long res, long res2)
Linus Torvalds's avatar
Linus Torvalds committed
667
668
669
670
671
672
673
674
{
	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;

675
676
677
678
679
680
	/*
	 * 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
681
682
	 */
	if (is_sync_kiocb(iocb)) {
683
		BUG_ON(iocb->ki_users != 1);
Linus Torvalds's avatar
Linus Torvalds committed
684
		iocb->ki_user_data = res;
685
		iocb->ki_users = 0;
Linus Torvalds's avatar
Linus Torvalds committed
686
		wake_up_process(iocb->ki_obj.tsk);
687
		return;
Linus Torvalds's avatar
Linus Torvalds committed
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
	}

	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;

707
	ring = kmap_atomic(info->ring_pages[0]);
Linus Torvalds's avatar
Linus Torvalds committed
708
709

	tail = info->tail;
710
	event = aio_ring_event(info, tail);
711
712
	if (++tail >= info->nr)
		tail = 0;
Linus Torvalds's avatar
Linus Torvalds committed
713
714
715
716
717
718

	event->obj = (u64)(unsigned long)iocb->ki_obj.user;
	event->data = iocb->ki_user_data;
	event->res = res;
	event->res2 = res2;

Kent Overstreet's avatar
Kent Overstreet committed
719
720
721
	pr_debug("%p[%lu]: %p: %p %Lx %lx %lx\n",
		 ctx, tail, iocb, iocb->ki_obj.user, iocb->ki_user_data,
		 res, res2);
Linus Torvalds's avatar
Linus Torvalds committed
722
723
724
725
726
727
728
729
730

	/* 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;

731
732
	put_aio_ring_event(event);
	kunmap_atomic(ring);
Linus Torvalds's avatar
Linus Torvalds committed
733
734

	pr_debug("added to ring %p at [%lu]\n", iocb, tail);
Davide Libenzi's avatar
Davide Libenzi committed
735
736
737
738
739
740

	/*
	 * 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.
	 */
741
	if (iocb->ki_eventfd != NULL)
Davide Libenzi's avatar
Davide Libenzi committed
742
743
		eventfd_signal(iocb->ki_eventfd, 1);

Linus Torvalds's avatar
Linus Torvalds committed
744
745
put_rq:
	/* everything turned out well, dispose of the aiocb. */
746
	__aio_put_req(ctx, iocb);
Linus Torvalds's avatar
Linus Torvalds committed
747

748
749
750
751
752
753
754
755
	/*
	 * 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
756
757
758
	if (waitqueue_active(&ctx->wait))
		wake_up(&ctx->wait);

759
	spin_unlock_irqrestore(&ctx->ctx_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
760
}
761
EXPORT_SYMBOL(aio_complete);
Linus Torvalds's avatar
Linus Torvalds committed
762
763
764
765
766
767
768
769
770
771
772
773
774
775

/* 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;

776
	ring = kmap_atomic(info->ring_pages[0]);
Kent Overstreet's avatar
Kent Overstreet committed
777
	pr_debug("h%u t%u m%u\n", ring->head, ring->tail, ring->nr);
Linus Torvalds's avatar
Linus Torvalds committed
778
779
780
781
782
783
784
785

	if (ring->head == ring->tail)
		goto out;

	spin_lock(&info->ring_lock);

	head = ring->head % info->nr;
	if (head != ring->tail) {
786
		struct io_event *evp = aio_ring_event(info, head);
Linus Torvalds's avatar
Linus Torvalds committed
787
788
789
790
791
		*ent = *evp;
		head = (head + 1) % info->nr;
		smp_mb(); /* finish reading the event before updatng the head */
		ring->head = head;
		ret = 1;
792
		put_aio_ring_event(evp);
Linus Torvalds's avatar
Linus Torvalds committed
793
794
795
796
	}
	spin_unlock(&info->ring_lock);

out:
797
	kunmap_atomic(ring);
Kent Overstreet's avatar
Kent Overstreet committed
798
	pr_debug("%d  h%u t%u\n", ret, ring->head, ring->tail);
Linus Torvalds's avatar
Linus Torvalds committed
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
	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)
{
818
	setup_timer_on_stack(&to->timer, timeout_func, (unsigned long) to);
Linus Torvalds's avatar
Linus Torvalds committed
819
820
821
822
823
824
825
826
827
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
	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;

Kent Overstreet's avatar
Kent Overstreet committed
861
862
		pr_debug("%Lx %Lx %Lx %Lx\n",
			 ent.data, ent.obj, ent.res, ent.res2);
Linus Torvalds's avatar
Linus Torvalds committed
863
864
865
866

		/* Could we split the check in two? */
		ret = -EFAULT;
		if (unlikely(copy_to_user(event, &ent, sizeof(ent)))) {
Kent Overstreet's avatar
Kent Overstreet committed
867
			pr_debug("lost an event due to EFAULT.\n");
Linus Torvalds's avatar
Linus Torvalds committed
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
			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;
903
904
905
906
			if (unlikely(ctx->dead)) {
				ret = -EINVAL;
				break;
			}
Linus Torvalds's avatar
Linus Torvalds committed
907
908
			if (to.timed_out)	/* Only check after read evt */
				break;
909
910
911
912
913
914
			/* 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
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
			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)))) {
Kent Overstreet's avatar
Kent Overstreet committed
930
			pr_debug("lost an event due to EFAULT.\n");
Linus Torvalds's avatar
Linus Torvalds committed
931
932
933
934
935
936
937
938
939
940
941
			break;
		}

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

	if (timeout)
		clear_timeout(&to);
out:
942
	destroy_timer_on_stack(&to.timer);
Linus Torvalds's avatar
Linus Torvalds committed
943
944
945
946
947
948
949
950
951
952
953
954
	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 */
955
	spin_lock(&mm->ioctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
956
957
	was_dead = ioctx->dead;
	ioctx->dead = 1;
958
959
	hlist_del_rcu(&ioctx->list);
	spin_unlock(&mm->ioctx_lock);
Linus Torvalds's avatar
Linus Torvalds committed
960

Kent Overstreet's avatar
Kent Overstreet committed
961
	pr_debug("(%p)\n", ioctx);
Linus Torvalds's avatar
Linus Torvalds committed
962
963
964
	if (likely(!was_dead))
		put_ioctx(ioctx);	/* twice for the list */

965
	kill_ctx(ioctx);
966
967
968
969
970
971

	/*
	 * 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.
	 */
972
	wake_up_all(&ioctx->wait);
Linus Torvalds's avatar
Linus Torvalds committed
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
}

/* 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.
 */
988
SYSCALL_DEFINE2(io_setup, unsigned, nr_events, aio_context_t __user *, ctxp)
Linus Torvalds's avatar
Linus Torvalds committed
989
990
991
992
993
994
995
996
997
998
{
	struct kioctx *ioctx = NULL;
	unsigned long ctx;
	long ret;

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

	ret = -EINVAL;
999
1000
1001
	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
1002
1003
1004
1005
1006
1007
1008
		goto out;
	}

	ioctx = ioctx_alloc(nr_events);
	ret = PTR_ERR(ioctx);
	if (!IS_ERR(ioctx)) {
		ret = put_user(ioctx->user_id, ctxp);
1009
1010
1011
		if (ret)
			io_destroy(ioctx);
		put_ioctx(ioctx);
Linus Torvalds's avatar
Linus Torvalds committed
1012
1013
1014
1015
1016
1017
1018
1019
1020
	}

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
1021
 *	implemented.  May fail with -EINVAL if the context pointed to
Linus Torvalds's avatar
Linus Torvalds committed
1022
1023
 *	is invalid.
 */
1024
SYSCALL_DEFINE1(io_destroy, aio_context_t, ctx)
Linus Torvalds's avatar
Linus Torvalds committed
1025
1026
1027
1028
{
	struct kioctx *ioctx = lookup_ioctx(ctx);
	if (likely(NULL != ioctx)) {
		io_destroy(ioctx);
1029
		put_ioctx(ioctx);
Linus Torvalds's avatar
Linus Torvalds committed
1030
1031
1032
1033
1034
1035
		return 0;
	}
	pr_debug("EINVAL: io_destroy: invalid context id\n");
	return -EINVAL;
}

1036
static void aio_advance_iovec(struct kiocb *iocb, ssize_t ret)
Linus Torvalds's avatar
Linus Torvalds committed
1037
{
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
	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++;
1051
		}
1052
	}
Linus Torvalds's avatar
Linus Torvalds committed
1053

1054
1055
1056
	/* 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
1057
1058
}

1059
static ssize_t aio_rw_vect_retry(struct kiocb *iocb)
Linus Torvalds's avatar
Linus Torvalds committed
1060
1061
{
	struct file *file = iocb->ki_filp;
1062
1063
1064
1065
	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
1066
	ssize_t ret = 0;
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
	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
1077

1078
1079
1080
1081
	/* This matches the pread()/pwrite() logic */
	if (iocb->ki_pos < 0)
		return -EINVAL;

1082
1083
	if (opcode == IOCB_CMD_PWRITEV)
		file_start_write(file);
1084
	do {
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
		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))));
1096
1097
	if (opcode == IOCB_CMD_PWRITEV)
		file_end_write(file);
Linus Torvalds's avatar
Linus Torvalds committed
1098

1099
1100
	/* This means we must have transferred all that we could */
	/* No need to retry anymore */
Linus Torvalds's avatar
Linus Torvalds committed
1101
1102
1103
	if ((ret == 0) || (iocb->ki_left == 0))
		ret = iocb->ki_nbytes - iocb->ki_left;

1104
1105
1106
	/* 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
1107
	    && ret < 0 && ret != -EIOCBQUEUED
1108
1109
1110
	    && iocb->ki_nbytes - iocb->ki_left)
		ret = iocb->ki_nbytes - iocb->ki_left;

Linus Torvalds's avatar
Linus Torvalds committed
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
	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;
}

1134
static ssize_t aio_setup_vectored_rw(int type, struct kiocb *kiocb, bool compat)
1135
1136
1137
{
	ssize_t ret;

1138
1139
1140
1141
1142
#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,
1143
				&kiocb->ki_iovec);
1144
1145
1146
1147
1148
	else
#endif
		ret = rw_copy_check_uvector(type,
				(struct iovec __user *)kiocb->ki_buf,
				kiocb->ki_nbytes, 1, &kiocb->ki_inline_vec,
1149
				&kiocb->ki_iovec);
1150
1151
1152
	if (ret < 0)
		goto out;

1153
1154
1155
1156
	ret = rw_verify_area(type, kiocb->ki_filp, &kiocb->ki_pos, ret);
	if (ret < 0)
		goto out;

1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
	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;
}

1168
static ssize_t aio_setup_single_vector(int type, struct file * file, struct kiocb *kiocb)
1169
{
1170
1171
1172
1173
1174
1175
	int bytes;

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

1176
1177
	kiocb->ki_iovec = &kiocb->ki_inline_vec;
	kiocb->ki_iovec->iov_base = kiocb->ki_buf;
1178
	kiocb->ki_iovec->iov_len = bytes;
1179
1180
1181
1182
1183
	kiocb->ki_nr_segs = 1;
	kiocb->ki_cur_seg = 0;
	return 0;
}

Linus Torvalds's avatar
Linus Torvalds committed
1184
1185
1186
1187
1188
/*
 * aio_setup_iocb:
 *	Performs the initial checks and aio retry method
 *	setup for the kiocb at the time of io submission.
 */
1189
static ssize_t aio_setup_iocb(struct kiocb *kiocb, bool compat)
Linus Torvalds's avatar
Linus Torvalds committed
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
{
	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;
1203
		ret = aio_setup_single_vector(READ, file, kiocb);
1204
1205
		if (ret)
			break;
Linus Torvalds's avatar
Linus Torvalds committed
1206
1207
		ret = -EINVAL;
		if (file->f_op->aio_read)
1208
			kiocb->ki_retry = aio_rw_vect_retry;
Linus Torvalds's avatar
Linus Torvalds committed
1209
1210
1211
1212
1213
1214
1215
1216
1217
		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;
1218
		ret = aio_setup_single_vector(WRITE, file, kiocb);
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
		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;
1229
		ret = aio_setup_vectored_rw(READ, kiocb, compat);
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
		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;
1240
		ret = aio_setup_vectored_rw(WRITE, kiocb, compat);
1241
1242
		if (ret)
			break;
Linus Torvalds's avatar
Linus Torvalds committed
1243
1244
		ret = -EINVAL;
		if (file->f_op->aio_write)
1245
			kiocb->ki_retry = aio_rw_vect_retry;
Linus Torvalds's avatar
Linus Torvalds committed
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
		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:
Kent Overstreet's avatar
Kent Overstreet committed
1258
		pr_debug("EINVAL: no operation provided\n");
Linus Torvalds's avatar
Linus Torvalds committed
1259
1260
1261
1262
1263
1264
1265
1266
1267
		ret = -EINVAL;
	}

	if (!kiocb->ki_retry)
		return ret;

	return 0;
}

1268
static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb,
Jeff Moyer's avatar
Jeff Moyer committed
1269
1270
			 struct iocb *iocb, struct kiocb_batch *batch,
			 bool compat)
Linus Torvalds's avatar
Linus Torvalds committed
1271
1272
1273
1274
1275
1276
{
	struct kiocb *req;
	struct file *file;
	ssize_t ret;

	/* enforce forwards compatibility on users */
1277
	if (unlikely(iocb->aio_reserved1 || iocb->aio_reserved2)) {
Kent Overstreet's avatar
Kent Overstreet committed
1278
		pr_debug("EINVAL: reserve field set\n");
Linus Torvalds's avatar
Linus Torvalds committed
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
		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
1296
	req = aio_get_req(ctx, batch);  /* returns with 2 references to req */
Linus Torvalds's avatar
Linus Torvalds committed
1297
1298
1299
1300
	if (unlikely(!req)) {
		fput(file);
		return -EAGAIN;
	}
1301
	req->ki_filp = file;
1302
1303
1304
1305
1306
1307
1308
	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.
		 */
1309
		req->ki_eventfd = eventfd_ctx_fdget((int) iocb->aio_resfd);
1310
		if (IS_ERR(req->ki_eventfd)) {
1311
			ret = PTR_ERR(req->ki_eventfd);
1312
			req->ki_eventfd = NULL;
1313
1314
1315
			goto out_put_req;
		}
	}
Linus Torvalds's avatar
Linus Torvalds committed
1316

1317
	ret = put_user(req->ki_key, &user_iocb->aio_key);
Linus Torvalds's avatar
Linus Torvalds committed
1318
	if (unlikely(ret)) {
Kent Overstreet's avatar
Kent Overstreet committed
1319
		pr_debug("EFAULT: aio_key\n");
Linus Torvalds's avatar
Linus Torvalds committed
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
		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;

1331
	ret = aio_setup_iocb(req, compat);
Linus Torvalds's avatar
Linus Torvalds committed
1332
1333
1334
1335
1336

	if (ret)
		goto out_put_req;

	spin_lock_irq(&ctx->ctx_lock);
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
	/*
	 * 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
1349
	if (ctx->dead)
1350
		ret = -EINVAL;
Zach Brown's avatar
Zach Brown committed
1351
1352
	spin_unlock_irq(&ctx->ctx_lock);
	if (ret)
1353
		goto out_put_req;
Zach Brown's avatar
Zach Brown committed
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369

	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);
1370
	}
Jeff Moyer's avatar
Jeff Moyer committed
1371

Linus Torvalds's avatar
Linus Torvalds committed
1372
1373
1374
1375
1376
1377
1378
1379
1380
	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;
}

1381
1382
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
1383
1384
1385
{
	struct kioctx *ctx;
	long ret = 0;
Jeff Moyer's avatar
Jeff Moyer committed
1386
	int i = 0;
Shaohua Li's avatar
Shaohua Li committed
1387
	struct blk_plug plug;
Jeff Moyer's avatar
Jeff Moyer committed
1388
	struct kiocb_batch batch;
Linus Torvalds's avatar
Linus Torvalds committed
1389
1390
1391
1392

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

1393
1394
1395
	if (unlikely(nr > LONG_MAX/sizeof(*iocbpp)))
		nr = LONG_MAX/sizeof(*iocbpp);

Linus Torvalds's avatar
Linus Torvalds committed
1396
1397
1398
1399
1400
	if (unlikely(!access_ok(VERIFY_READ, iocbpp, (nr*sizeof(*iocbpp)))))
		return -EFAULT;

	ctx = lookup_ioctx(ctx_id);
	if (unlikely(!ctx)) {
Kent Overstreet's avatar
Kent Overstreet committed
1401
		pr_debug("EINVAL: invalid context id\n");
Linus Torvalds's avatar
Linus Torvalds committed
1402
1403
1404
		return -EINVAL;
	}

Jeff Moyer's avatar
Jeff Moyer committed
1405
1406
	kiocb_batch_init(&batch, nr);

Shaohua Li's avatar
Shaohua Li committed
1407
1408
	blk_start_plug(&plug);

Linus Torvalds's avatar
Linus Torvalds committed
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
	/*
	 * 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
1427
		ret = io_submit_one(ctx, user_iocb, &tmp, &batch, compat);
Linus Torvalds's avatar
Linus Torvalds committed
1428
1429
1430
		if (ret)
			break;
	}
Shaohua Li's avatar
Shaohua Li committed
1431
	blk_finish_plug(&plug);
Linus Torvalds's avatar
Linus Torvalds committed
1432

1433
	kiocb_batch_free(ctx, &batch);
Linus Torvalds's avatar
Linus Torvalds committed
1434
1435
1436
1437
	put_ioctx(ctx);
	return i ? i : ret;
}

1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
/* 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