Blame view
fs/aio.c
56.4 KB
1da177e4c
|
1 2 3 4 5 6 7 |
/* * 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. |
bfe4037e7
|
8 |
* Copyright 2018 Christoph Hellwig. |
1da177e4c
|
9 10 11 |
* * See ../COPYING for licensing terms. */ |
caf4167aa
|
12 |
#define pr_fmt(fmt) "%s: " fmt, __func__ |
1da177e4c
|
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> |
630d9c472
|
18 |
#include <linux/export.h> |
1da177e4c
|
19 |
#include <linux/syscalls.h> |
b9d128f10
|
20 |
#include <linux/backing-dev.h> |
9018ccc45
|
21 |
#include <linux/refcount.h> |
027445c37
|
22 |
#include <linux/uio.h> |
1da177e4c
|
23 |
|
174cd4b1e
|
24 |
#include <linux/sched/signal.h> |
1da177e4c
|
25 26 27 28 |
#include <linux/fs.h> #include <linux/file.h> #include <linux/mm.h> #include <linux/mman.h> |
3d2d827f5
|
29 |
#include <linux/mmu_context.h> |
e1bdd5f27
|
30 |
#include <linux/percpu.h> |
1da177e4c
|
31 32 33 34 35 36 |
#include <linux/slab.h> #include <linux/timer.h> #include <linux/aio.h> #include <linux/highmem.h> #include <linux/workqueue.h> #include <linux/security.h> |
9c3060bed
|
37 |
#include <linux/eventfd.h> |
cfb1e33ee
|
38 |
#include <linux/blkdev.h> |
9d85cba71
|
39 |
#include <linux/compat.h> |
36bc08cc0
|
40 41 |
#include <linux/migrate.h> #include <linux/ramfs.h> |
723be6e39
|
42 |
#include <linux/percpu-refcount.h> |
71ad7490c
|
43 |
#include <linux/mount.h> |
52db59df1
|
44 |
#include <linux/pseudo_fs.h> |
1da177e4c
|
45 46 |
#include <asm/kmap_types.h> |
7c0f6ba68
|
47 |
#include <linux/uaccess.h> |
a538e3ff9
|
48 |
#include <linux/nospec.h> |
1da177e4c
|
49 |
|
68d70d03f
|
50 |
#include "internal.h" |
f3a2752a4
|
51 |
#define KIOCB_KEY 0 |
4e179bca6
|
52 53 54 55 56 57 |
#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 */ |
fa8a53c39
|
58 59 |
unsigned head; /* Written to by userland or under ring_lock * mutex by aio_read_events_ring(). */ |
4e179bca6
|
60 61 62 63 64 65 66 67 68 69 |
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 */ |
a79d40e9b
|
70 71 72 73 74 |
/* * Plugging is meant to work with larger batches of IOs. If we don't * have more than the below, then don't bother setting up a plug. */ #define AIO_PLUG_THRESHOLD 2 |
4e179bca6
|
75 |
#define AIO_RING_PAGES 8 |
4e179bca6
|
76 |
|
db446a08c
|
77 |
struct kioctx_table { |
d0264c01e
|
78 79 80 |
struct rcu_head rcu; unsigned nr; struct kioctx __rcu *table[]; |
db446a08c
|
81 |
}; |
e1bdd5f27
|
82 83 84 |
struct kioctx_cpu { unsigned reqs_available; }; |
dc48e56d7
|
85 86 87 88 |
struct ctx_rq_wait { struct completion comp; atomic_t count; }; |
4e179bca6
|
89 |
struct kioctx { |
723be6e39
|
90 |
struct percpu_ref users; |
36f558890
|
91 |
atomic_t dead; |
4e179bca6
|
92 |
|
e34ecee2a
|
93 |
struct percpu_ref reqs; |
4e179bca6
|
94 |
unsigned long user_id; |
4e179bca6
|
95 |
|
e1bdd5f27
|
96 97 98 99 100 101 102 |
struct __percpu kioctx_cpu *cpu; /* * For percpu reqs_available, number of slots we move to/from global * counter at a time: */ unsigned req_batch; |
3e845ce01
|
103 104 105 106 |
/* * This is what userspace passed to io_setup(), it's not used for * anything but counting against the global max_reqs quota. * |
58c85dc20
|
107 |
* The real limit is nr_events - 1, which will be larger (see |
3e845ce01
|
108 109 |
* aio_setup_ring()) */ |
4e179bca6
|
110 |
unsigned max_reqs; |
58c85dc20
|
111 112 |
/* Size of ringbuffer, in units of struct io_event */ unsigned nr_events; |
4e179bca6
|
113 |
|
58c85dc20
|
114 115 116 117 118 |
unsigned long mmap_base; unsigned long mmap_size; struct page **ring_pages; long nr_pages; |
f729863a8
|
119 |
struct rcu_work free_rwork; /* see free_ioctx() */ |
4e23bcaeb
|
120 |
|
e02ba72aa
|
121 122 123 |
/* * signals when all in-flight requests are done */ |
dc48e56d7
|
124 |
struct ctx_rq_wait *rq_wait; |
e02ba72aa
|
125 |
|
4e23bcaeb
|
126 |
struct { |
34e83fc61
|
127 128 129 130 131 |
/* * This counts the number of available slots in the ringbuffer, * so we avoid overflowing it: it's decremented (if positive) * when allocating a kiocb and incremented when the resulting * io_event is pulled off the ringbuffer. |
e1bdd5f27
|
132 133 |
* * We batch accesses to it with a percpu version. |
34e83fc61
|
134 135 |
*/ atomic_t reqs_available; |
4e23bcaeb
|
136 137 138 139 140 141 |
} ____cacheline_aligned_in_smp; struct { spinlock_t ctx_lock; struct list_head active_reqs; /* used for cancellation */ } ____cacheline_aligned_in_smp; |
58c85dc20
|
142 143 |
struct { struct mutex ring_lock; |
4e23bcaeb
|
144 145 |
wait_queue_head_t wait; } ____cacheline_aligned_in_smp; |
58c85dc20
|
146 147 148 |
struct { unsigned tail; |
d856f32a8
|
149 |
unsigned completed_events; |
58c85dc20
|
150 |
spinlock_t completion_lock; |
4e23bcaeb
|
151 |
} ____cacheline_aligned_in_smp; |
58c85dc20
|
152 153 |
struct page *internal_pages[AIO_RING_PAGES]; |
36bc08cc0
|
154 |
struct file *aio_ring_file; |
db446a08c
|
155 156 |
unsigned id; |
4e179bca6
|
157 |
}; |
84c4e1f89
|
158 159 160 161 |
/* * First field must be the file pointer in all the * iocb unions! See also 'struct kiocb' in <linux/fs.h> */ |
a3c0d439e
|
162 |
struct fsync_iocb { |
a3c0d439e
|
163 |
struct file *file; |
84c4e1f89
|
164 |
struct work_struct work; |
a3c0d439e
|
165 166 |
bool datasync; }; |
bfe4037e7
|
167 168 169 170 |
struct poll_iocb { struct file *file; struct wait_queue_head *head; __poll_t events; |
af5c72b1f
|
171 |
bool done; |
bfe4037e7
|
172 173 174 175 |
bool cancelled; struct wait_queue_entry wait; struct work_struct work; }; |
84c4e1f89
|
176 177 178 179 180 181 |
/* * NOTE! Each of the iocb union members has the file pointer * as the first entry in their struct definition. So you can * access the file pointer through any of the sub-structs, * or directly as just 'ki_filp' in this struct. */ |
04b2fa9f8
|
182 |
struct aio_kiocb { |
54843f875
|
183 |
union { |
84c4e1f89
|
184 |
struct file *ki_filp; |
54843f875
|
185 |
struct kiocb rw; |
a3c0d439e
|
186 |
struct fsync_iocb fsync; |
bfe4037e7
|
187 |
struct poll_iocb poll; |
54843f875
|
188 |
}; |
04b2fa9f8
|
189 190 191 |
struct kioctx *ki_ctx; kiocb_cancel_fn *ki_cancel; |
a9339b785
|
192 |
struct io_event ki_res; |
04b2fa9f8
|
193 194 195 |
struct list_head ki_list; /* the aio core uses this * for cancellation */ |
9018ccc45
|
196 |
refcount_t ki_refcnt; |
04b2fa9f8
|
197 198 199 200 201 202 203 |
/* * If the aio_resfd field of the userspace iocb is not zero, * this is the underlying eventfd context to deliver events to. */ struct eventfd_ctx *ki_eventfd; }; |
1da177e4c
|
204 |
/*------ sysctl variables----*/ |
d55b5fdaf
|
205 206 207 |
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 */ |
1da177e4c
|
208 |
/*----end sysctl variables---*/ |
e18b890bb
|
209 210 |
static struct kmem_cache *kiocb_cachep; static struct kmem_cache *kioctx_cachep; |
1da177e4c
|
211 |
|
71ad7490c
|
212 213 214 215 216 217 218 |
static struct vfsmount *aio_mnt; static const struct file_operations aio_ring_fops; static const struct address_space_operations aio_ctx_aops; static struct file *aio_private_file(struct kioctx *ctx, loff_t nr_pages) { |
71ad7490c
|
219 |
struct file *file; |
71ad7490c
|
220 |
struct inode *inode = alloc_anon_inode(aio_mnt->mnt_sb); |
7f62656be
|
221 222 |
if (IS_ERR(inode)) return ERR_CAST(inode); |
71ad7490c
|
223 224 225 226 |
inode->i_mapping->a_ops = &aio_ctx_aops; inode->i_mapping->private_data = ctx; inode->i_size = PAGE_SIZE * nr_pages; |
d93aa9d82
|
227 228 |
file = alloc_file_pseudo(inode, aio_mnt, "[aio]", O_RDWR, &aio_ring_fops); |
c9c554f21
|
229 |
if (IS_ERR(file)) |
71ad7490c
|
230 |
iput(inode); |
71ad7490c
|
231 232 |
return file; } |
52db59df1
|
233 |
static int aio_init_fs_context(struct fs_context *fc) |
71ad7490c
|
234 |
{ |
52db59df1
|
235 236 237 238 |
if (!init_pseudo(fc, AIO_RING_MAGIC)) return -ENOMEM; fc->s_iflags |= SB_I_NOEXEC; return 0; |
71ad7490c
|
239 |
} |
1da177e4c
|
240 241 242 243 244 245 |
/* 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) { |
71ad7490c
|
246 247 |
static struct file_system_type aio_fs = { .name = "aio", |
52db59df1
|
248 |
.init_fs_context = aio_init_fs_context, |
71ad7490c
|
249 250 251 252 253 |
.kill_sb = kill_anon_super, }; aio_mnt = kern_mount(&aio_fs); if (IS_ERR(aio_mnt)) panic("Failed to create aio fs mount."); |
04b2fa9f8
|
254 |
kiocb_cachep = KMEM_CACHE(aio_kiocb, SLAB_HWCACHE_ALIGN|SLAB_PANIC); |
0a31bd5f2
|
255 |
kioctx_cachep = KMEM_CACHE(kioctx,SLAB_HWCACHE_ALIGN|SLAB_PANIC); |
1da177e4c
|
256 257 |
return 0; } |
385773e04
|
258 |
__initcall(aio_setup); |
1da177e4c
|
259 |
|
5e9ae2e5d
|
260 261 262 |
static void put_aio_ring_file(struct kioctx *ctx) { struct file *aio_ring_file = ctx->aio_ring_file; |
de04e7693
|
263 |
struct address_space *i_mapping; |
5e9ae2e5d
|
264 |
if (aio_ring_file) { |
450630975
|
265 |
truncate_setsize(file_inode(aio_ring_file), 0); |
5e9ae2e5d
|
266 267 |
/* Prevent further access to the kioctx from migratepages */ |
450630975
|
268 |
i_mapping = aio_ring_file->f_mapping; |
de04e7693
|
269 270 |
spin_lock(&i_mapping->private_lock); i_mapping->private_data = NULL; |
5e9ae2e5d
|
271 |
ctx->aio_ring_file = NULL; |
de04e7693
|
272 |
spin_unlock(&i_mapping->private_lock); |
5e9ae2e5d
|
273 274 275 276 |
fput(aio_ring_file); } } |
1da177e4c
|
277 278 |
static void aio_free_ring(struct kioctx *ctx) { |
36bc08cc0
|
279 |
int i; |
1da177e4c
|
280 |
|
fa8a53c39
|
281 282 283 284 |
/* Disconnect the kiotx from the ring file. This prevents future * accesses to the kioctx from page migration. */ put_aio_ring_file(ctx); |
36bc08cc0
|
285 |
for (i = 0; i < ctx->nr_pages; i++) { |
8e321fefb
|
286 |
struct page *page; |
36bc08cc0
|
287 288 289 |
pr_debug("pid(%d) [%d] page->count=%d ", current->pid, i, page_count(ctx->ring_pages[i])); |
8e321fefb
|
290 291 292 293 294 |
page = ctx->ring_pages[i]; if (!page) continue; ctx->ring_pages[i] = NULL; put_page(page); |
36bc08cc0
|
295 |
} |
1da177e4c
|
296 |
|
ddb8c45ba
|
297 |
if (ctx->ring_pages && ctx->ring_pages != ctx->internal_pages) { |
58c85dc20
|
298 |
kfree(ctx->ring_pages); |
ddb8c45ba
|
299 300 |
ctx->ring_pages = NULL; } |
36bc08cc0
|
301 |
} |
5477e70a6
|
302 |
static int aio_ring_mremap(struct vm_area_struct *vma) |
e4a0d3e72
|
303 |
{ |
5477e70a6
|
304 |
struct file *file = vma->vm_file; |
e4a0d3e72
|
305 306 |
struct mm_struct *mm = vma->vm_mm; struct kioctx_table *table; |
b2edffdd9
|
307 |
int i, res = -EINVAL; |
e4a0d3e72
|
308 309 310 311 312 313 |
spin_lock(&mm->ioctx_lock); rcu_read_lock(); table = rcu_dereference(mm->ioctx_table); for (i = 0; i < table->nr; i++) { struct kioctx *ctx; |
d0264c01e
|
314 |
ctx = rcu_dereference(table->table[i]); |
e4a0d3e72
|
315 |
if (ctx && ctx->aio_ring_file == file) { |
b2edffdd9
|
316 317 318 319 |
if (!atomic_read(&ctx->dead)) { ctx->user_id = ctx->mmap_base = vma->vm_start; res = 0; } |
e4a0d3e72
|
320 321 322 323 324 325 |
break; } } rcu_read_unlock(); spin_unlock(&mm->ioctx_lock); |
b2edffdd9
|
326 |
return res; |
e4a0d3e72
|
327 |
} |
5477e70a6
|
328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 |
static const struct vm_operations_struct aio_ring_vm_ops = { .mremap = aio_ring_mremap, #if IS_ENABLED(CONFIG_MMU) .fault = filemap_fault, .map_pages = filemap_map_pages, .page_mkwrite = filemap_page_mkwrite, #endif }; static int aio_ring_mmap(struct file *file, struct vm_area_struct *vma) { vma->vm_flags |= VM_DONTEXPAND; vma->vm_ops = &aio_ring_vm_ops; return 0; } |
36bc08cc0
|
343 344 345 |
static const struct file_operations aio_ring_fops = { .mmap = aio_ring_mmap, }; |
0c45355fc
|
346 |
#if IS_ENABLED(CONFIG_MIGRATION) |
36bc08cc0
|
347 348 349 |
static int aio_migratepage(struct address_space *mapping, struct page *new, struct page *old, enum migrate_mode mode) { |
5e9ae2e5d
|
350 |
struct kioctx *ctx; |
36bc08cc0
|
351 |
unsigned long flags; |
fa8a53c39
|
352 |
pgoff_t idx; |
36bc08cc0
|
353 |
int rc; |
2916ecc0f
|
354 355 356 357 358 359 360 |
/* * We cannot support the _NO_COPY case here, because copy needs to * happen under the ctx->completion_lock. That does not work with the * migration workflow of MIGRATE_SYNC_NO_COPY. */ if (mode == MIGRATE_SYNC_NO_COPY) return -EINVAL; |
8e321fefb
|
361 |
rc = 0; |
fa8a53c39
|
362 |
/* mapping->private_lock here protects against the kioctx teardown. */ |
8e321fefb
|
363 364 |
spin_lock(&mapping->private_lock); ctx = mapping->private_data; |
fa8a53c39
|
365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 |
if (!ctx) { rc = -EINVAL; goto out; } /* The ring_lock mutex. The prevents aio_read_events() from writing * to the ring's head, and prevents page migration from mucking in * a partially initialized kiotx. */ if (!mutex_trylock(&ctx->ring_lock)) { rc = -EAGAIN; goto out; } idx = old->index; if (idx < (pgoff_t)ctx->nr_pages) { /* Make sure the old page hasn't already been changed */ if (ctx->ring_pages[idx] != old) rc = -EAGAIN; |
8e321fefb
|
384 385 |
} else rc = -EINVAL; |
8e321fefb
|
386 387 |
if (rc != 0) |
fa8a53c39
|
388 |
goto out_unlock; |
8e321fefb
|
389 |
|
36bc08cc0
|
390 391 |
/* Writeback must be complete */ BUG_ON(PageWriteback(old)); |
8e321fefb
|
392 |
get_page(new); |
36bc08cc0
|
393 |
|
371096949
|
394 |
rc = migrate_page_move_mapping(mapping, new, old, 1); |
36bc08cc0
|
395 |
if (rc != MIGRATEPAGE_SUCCESS) { |
8e321fefb
|
396 |
put_page(new); |
fa8a53c39
|
397 |
goto out_unlock; |
36bc08cc0
|
398 |
} |
fa8a53c39
|
399 400 401 |
/* Take completion_lock to prevent other writes to the ring buffer * while the old page is copied to the new. This prevents new * events from being lost. |
5e9ae2e5d
|
402 |
*/ |
fa8a53c39
|
403 404 405 406 407 |
spin_lock_irqsave(&ctx->completion_lock, flags); migrate_page_copy(new, old); BUG_ON(ctx->ring_pages[idx] != old); ctx->ring_pages[idx] = new; spin_unlock_irqrestore(&ctx->completion_lock, flags); |
36bc08cc0
|
408 |
|
fa8a53c39
|
409 410 |
/* The old page is no longer accessible. */ put_page(old); |
8e321fefb
|
411 |
|
fa8a53c39
|
412 413 414 415 |
out_unlock: mutex_unlock(&ctx->ring_lock); out: spin_unlock(&mapping->private_lock); |
36bc08cc0
|
416 |
return rc; |
1da177e4c
|
417 |
} |
0c45355fc
|
418 |
#endif |
1da177e4c
|
419 |
|
36bc08cc0
|
420 |
static const struct address_space_operations aio_ctx_aops = { |
835f252c6
|
421 |
.set_page_dirty = __set_page_dirty_no_writeback, |
0c45355fc
|
422 |
#if IS_ENABLED(CONFIG_MIGRATION) |
36bc08cc0
|
423 |
.migratepage = aio_migratepage, |
0c45355fc
|
424 |
#endif |
36bc08cc0
|
425 |
}; |
2a8a98673
|
426 |
static int aio_setup_ring(struct kioctx *ctx, unsigned int nr_events) |
1da177e4c
|
427 428 |
{ struct aio_ring *ring; |
41003a7bc
|
429 |
struct mm_struct *mm = current->mm; |
3dc9acb67
|
430 |
unsigned long size, unused; |
1da177e4c
|
431 |
int nr_pages; |
36bc08cc0
|
432 433 |
int i; struct file *file; |
1da177e4c
|
434 435 436 437 438 439 |
/* 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; |
1da177e4c
|
440 |
|
36bc08cc0
|
441 |
nr_pages = PFN_UP(size); |
1da177e4c
|
442 443 |
if (nr_pages < 0) return -EINVAL; |
71ad7490c
|
444 |
file = aio_private_file(ctx, nr_pages); |
36bc08cc0
|
445 446 |
if (IS_ERR(file)) { ctx->aio_ring_file = NULL; |
fa8a53c39
|
447 |
return -ENOMEM; |
36bc08cc0
|
448 |
} |
3dc9acb67
|
449 450 451 452 453 454 455 456 457 458 459 460 461 |
ctx->aio_ring_file = file; nr_events = (PAGE_SIZE * nr_pages - sizeof(struct aio_ring)) / sizeof(struct io_event); ctx->ring_pages = ctx->internal_pages; if (nr_pages > AIO_RING_PAGES) { ctx->ring_pages = kcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL); if (!ctx->ring_pages) { put_aio_ring_file(ctx); return -ENOMEM; } } |
36bc08cc0
|
462 463 |
for (i = 0; i < nr_pages; i++) { struct page *page; |
450630975
|
464 |
page = find_or_create_page(file->f_mapping, |
36bc08cc0
|
465 466 467 468 469 470 471 |
i, GFP_HIGHUSER | __GFP_ZERO); if (!page) break; pr_debug("pid(%d) page[%d]->count=%d ", current->pid, i, page_count(page)); SetPageUptodate(page); |
36bc08cc0
|
472 |
unlock_page(page); |
3dc9acb67
|
473 474 |
ctx->ring_pages[i] = page; |
36bc08cc0
|
475 |
} |
3dc9acb67
|
476 |
ctx->nr_pages = i; |
1da177e4c
|
477 |
|
3dc9acb67
|
478 479 |
if (unlikely(i != nr_pages)) { aio_free_ring(ctx); |
fa8a53c39
|
480 |
return -ENOMEM; |
1da177e4c
|
481 |
} |
58c85dc20
|
482 483 484 |
ctx->mmap_size = nr_pages * PAGE_SIZE; pr_debug("attempting mmap of %lu bytes ", ctx->mmap_size); |
36bc08cc0
|
485 |
|
013373e8b
|
486 487 488 489 490 |
if (down_write_killable(&mm->mmap_sem)) { ctx->mmap_size = 0; aio_free_ring(ctx); return -EINTR; } |
36bc08cc0
|
491 492 |
ctx->mmap_base = do_mmap_pgoff(ctx->aio_ring_file, 0, ctx->mmap_size, PROT_READ | PROT_WRITE, |
897ab3e0c
|
493 |
MAP_SHARED, 0, &unused, NULL); |
3dc9acb67
|
494 |
up_write(&mm->mmap_sem); |
58c85dc20
|
495 |
if (IS_ERR((void *)ctx->mmap_base)) { |
58c85dc20
|
496 |
ctx->mmap_size = 0; |
1da177e4c
|
497 |
aio_free_ring(ctx); |
fa8a53c39
|
498 |
return -ENOMEM; |
1da177e4c
|
499 |
} |
58c85dc20
|
500 501 |
pr_debug("mmap address: 0x%08lx ", ctx->mmap_base); |
d6c355c7d
|
502 |
|
58c85dc20
|
503 504 |
ctx->user_id = ctx->mmap_base; ctx->nr_events = nr_events; /* trusted copy */ |
1da177e4c
|
505 |
|
58c85dc20
|
506 |
ring = kmap_atomic(ctx->ring_pages[0]); |
1da177e4c
|
507 |
ring->nr = nr_events; /* user copy */ |
db446a08c
|
508 |
ring->id = ~0U; |
1da177e4c
|
509 510 511 512 513 |
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); |
e8e3c3d66
|
514 |
kunmap_atomic(ring); |
58c85dc20
|
515 |
flush_dcache_page(ctx->ring_pages[0]); |
1da177e4c
|
516 517 518 |
return 0; } |
1da177e4c
|
519 520 521 |
#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) |
04b2fa9f8
|
522 |
void kiocb_set_cancel_fn(struct kiocb *iocb, kiocb_cancel_fn *cancel) |
0460fef2a
|
523 |
{ |
54843f875
|
524 |
struct aio_kiocb *req = container_of(iocb, struct aio_kiocb, rw); |
0460fef2a
|
525 526 |
struct kioctx *ctx = req->ki_ctx; unsigned long flags; |
75321b50a
|
527 528 |
if (WARN_ON_ONCE(!list_empty(&req->ki_list))) return; |
0460fef2a
|
529 |
|
75321b50a
|
530 531 |
spin_lock_irqsave(&ctx->ctx_lock, flags); list_add_tail(&req->ki_list, &ctx->active_reqs); |
0460fef2a
|
532 |
req->ki_cancel = cancel; |
0460fef2a
|
533 534 535 |
spin_unlock_irqrestore(&ctx->ctx_lock, flags); } EXPORT_SYMBOL(kiocb_set_cancel_fn); |
a6d7cff47
|
536 537 538 |
/* * free_ioctx() should be RCU delayed to synchronize against the RCU * protected lookup_ioctx() and also needs process context to call |
f729863a8
|
539 |
* aio_free_ring(). Use rcu_work. |
a6d7cff47
|
540 |
*/ |
e34ecee2a
|
541 |
static void free_ioctx(struct work_struct *work) |
36f558890
|
542 |
{ |
f729863a8
|
543 544 |
struct kioctx *ctx = container_of(to_rcu_work(work), struct kioctx, free_rwork); |
e34ecee2a
|
545 546 |
pr_debug("freeing %p ", ctx); |
e1bdd5f27
|
547 |
|
e34ecee2a
|
548 |
aio_free_ring(ctx); |
e1bdd5f27
|
549 |
free_percpu(ctx->cpu); |
9a1049da9
|
550 551 |
percpu_ref_exit(&ctx->reqs); percpu_ref_exit(&ctx->users); |
36f558890
|
552 553 |
kmem_cache_free(kioctx_cachep, ctx); } |
e34ecee2a
|
554 555 556 |
static void free_ioctx_reqs(struct percpu_ref *ref) { struct kioctx *ctx = container_of(ref, struct kioctx, reqs); |
e02ba72aa
|
557 |
/* At this point we know that there are no any in-flight requests */ |
dc48e56d7
|
558 559 |
if (ctx->rq_wait && atomic_dec_and_test(&ctx->rq_wait->count)) complete(&ctx->rq_wait->comp); |
e02ba72aa
|
560 |
|
a6d7cff47
|
561 |
/* Synchronize against RCU protected table->table[] dereferences */ |
f729863a8
|
562 563 |
INIT_RCU_WORK(&ctx->free_rwork, free_ioctx); queue_rcu_work(system_wq, &ctx->free_rwork); |
e34ecee2a
|
564 |
} |
36f558890
|
565 566 567 568 569 |
/* * When this function runs, the kioctx has been removed from the "hash table" * and ctx->users has dropped to 0, so we know no more kiocbs can be submitted - * now it's safe to cancel any that need to be. */ |
e34ecee2a
|
570 |
static void free_ioctx_users(struct percpu_ref *ref) |
36f558890
|
571 |
{ |
e34ecee2a
|
572 |
struct kioctx *ctx = container_of(ref, struct kioctx, users); |
04b2fa9f8
|
573 |
struct aio_kiocb *req; |
36f558890
|
574 575 576 577 578 |
spin_lock_irq(&ctx->ctx_lock); while (!list_empty(&ctx->active_reqs)) { req = list_first_entry(&ctx->active_reqs, |
04b2fa9f8
|
579 |
struct aio_kiocb, ki_list); |
888933f8f
|
580 |
req->ki_cancel(&req->rw); |
4faa99965
|
581 |
list_del_init(&req->ki_list); |
36f558890
|
582 583 584 |
} spin_unlock_irq(&ctx->ctx_lock); |
e34ecee2a
|
585 586 |
percpu_ref_kill(&ctx->reqs); percpu_ref_put(&ctx->reqs); |
36f558890
|
587 |
} |
db446a08c
|
588 589 590 591 592 593 594 |
static int ioctx_add_table(struct kioctx *ctx, struct mm_struct *mm) { unsigned i, new_nr; struct kioctx_table *table, *old; struct aio_ring *ring; spin_lock(&mm->ioctx_lock); |
855ef0dec
|
595 |
table = rcu_dereference_raw(mm->ioctx_table); |
db446a08c
|
596 597 598 599 |
while (1) { if (table) for (i = 0; i < table->nr; i++) |
d0264c01e
|
600 |
if (!rcu_access_pointer(table->table[i])) { |
db446a08c
|
601 |
ctx->id = i; |
d0264c01e
|
602 |
rcu_assign_pointer(table->table[i], ctx); |
db446a08c
|
603 |
spin_unlock(&mm->ioctx_lock); |
fa8a53c39
|
604 605 606 607 |
/* While kioctx setup is in progress, * we are protected from page migration * changes ring_pages by ->ring_lock. */ |
db446a08c
|
608 609 610 611 612 613 614 |
ring = kmap_atomic(ctx->ring_pages[0]); ring->id = ctx->id; kunmap_atomic(ring); return 0; } new_nr = (table ? table->nr : 1) * 4; |
db446a08c
|
615 616 617 618 619 620 621 622 623 624 |
spin_unlock(&mm->ioctx_lock); table = kzalloc(sizeof(*table) + sizeof(struct kioctx *) * new_nr, GFP_KERNEL); if (!table) return -ENOMEM; table->nr = new_nr; spin_lock(&mm->ioctx_lock); |
855ef0dec
|
625 |
old = rcu_dereference_raw(mm->ioctx_table); |
db446a08c
|
626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 |
if (!old) { rcu_assign_pointer(mm->ioctx_table, table); } else if (table->nr > old->nr) { memcpy(table->table, old->table, old->nr * sizeof(struct kioctx *)); rcu_assign_pointer(mm->ioctx_table, table); kfree_rcu(old, rcu); } else { kfree(table); table = old; } } } |
e34ecee2a
|
641 642 643 644 645 646 647 648 649 |
static void aio_nr_sub(unsigned nr) { spin_lock(&aio_nr_lock); if (WARN_ON(aio_nr - nr > aio_nr)) aio_nr = 0; else aio_nr -= nr; spin_unlock(&aio_nr_lock); } |
1da177e4c
|
650 651 652 653 654 |
/* ioctx_alloc * Allocates and initializes an ioctx. Returns an ERR_PTR if it failed. */ static struct kioctx *ioctx_alloc(unsigned nr_events) { |
41003a7bc
|
655 |
struct mm_struct *mm = current->mm; |
1da177e4c
|
656 |
struct kioctx *ctx; |
e23754f88
|
657 |
int err = -ENOMEM; |
1da177e4c
|
658 |
|
e1bdd5f27
|
659 |
/* |
2a8a98673
|
660 661 662 663 664 665 |
* Store the original nr_events -- what userspace passed to io_setup(), * for counting against the global limit -- before it changes. */ unsigned int max_reqs = nr_events; /* |
e1bdd5f27
|
666 667 668 669 670 671 672 673 674 675 |
* We keep track of the number of available ringbuffer slots, to prevent * overflow (reqs_available), and we also use percpu counters for this. * * So since up to half the slots might be on other cpu's percpu counters * and unavailable, double nr_events so userspace sees what they * expected: additionally, we move req_batch slots to/from percpu * counters at a time, so make sure that isn't 0: */ nr_events = max(nr_events, num_possible_cpus() * 4); nr_events *= 2; |
1da177e4c
|
676 |
/* Prevent overflows */ |
08397acdd
|
677 |
if (nr_events > (0x10000000U / sizeof(struct io_event))) { |
1da177e4c
|
678 679 680 681 |
pr_debug("ENOMEM: nr_events too high "); return ERR_PTR(-EINVAL); } |
2a8a98673
|
682 |
if (!nr_events || (unsigned long)max_reqs > aio_max_nr) |
1da177e4c
|
683 |
return ERR_PTR(-EAGAIN); |
c37622296
|
684 |
ctx = kmem_cache_zalloc(kioctx_cachep, GFP_KERNEL); |
1da177e4c
|
685 686 |
if (!ctx) return ERR_PTR(-ENOMEM); |
2a8a98673
|
687 |
ctx->max_reqs = max_reqs; |
1da177e4c
|
688 |
|
1da177e4c
|
689 |
spin_lock_init(&ctx->ctx_lock); |
0460fef2a
|
690 |
spin_lock_init(&ctx->completion_lock); |
58c85dc20
|
691 |
mutex_init(&ctx->ring_lock); |
fa8a53c39
|
692 693 694 |
/* Protect against page migration throughout kiotx setup by keeping * the ring_lock mutex held until setup is complete. */ mutex_lock(&ctx->ring_lock); |
1da177e4c
|
695 696 697 |
init_waitqueue_head(&ctx->wait); INIT_LIST_HEAD(&ctx->active_reqs); |
1da177e4c
|
698 |
|
2aad2a86f
|
699 |
if (percpu_ref_init(&ctx->users, free_ioctx_users, 0, GFP_KERNEL)) |
fa8a53c39
|
700 |
goto err; |
2aad2a86f
|
701 |
if (percpu_ref_init(&ctx->reqs, free_ioctx_reqs, 0, GFP_KERNEL)) |
fa8a53c39
|
702 |
goto err; |
e1bdd5f27
|
703 704 |
ctx->cpu = alloc_percpu(struct kioctx_cpu); if (!ctx->cpu) |
e34ecee2a
|
705 |
goto err; |
1da177e4c
|
706 |
|
2a8a98673
|
707 |
err = aio_setup_ring(ctx, nr_events); |
fa8a53c39
|
708 |
if (err < 0) |
e34ecee2a
|
709 |
goto err; |
e1bdd5f27
|
710 |
|
34e83fc61
|
711 |
atomic_set(&ctx->reqs_available, ctx->nr_events - 1); |
e1bdd5f27
|
712 |
ctx->req_batch = (ctx->nr_events - 1) / (num_possible_cpus() * 4); |
6878ea72a
|
713 714 |
if (ctx->req_batch < 1) ctx->req_batch = 1; |
34e83fc61
|
715 |
|
1da177e4c
|
716 |
/* limit the number of system wide aios */ |
9fa1cb397
|
717 |
spin_lock(&aio_nr_lock); |
2a8a98673
|
718 719 |
if (aio_nr + ctx->max_reqs > aio_max_nr || aio_nr + ctx->max_reqs < aio_nr) { |
9fa1cb397
|
720 |
spin_unlock(&aio_nr_lock); |
e34ecee2a
|
721 |
err = -EAGAIN; |
d1b943271
|
722 |
goto err_ctx; |
2dd542b7a
|
723 724 |
} aio_nr += ctx->max_reqs; |
9fa1cb397
|
725 |
spin_unlock(&aio_nr_lock); |
1da177e4c
|
726 |
|
1881686f8
|
727 728 |
percpu_ref_get(&ctx->users); /* io_setup() will drop this ref */ percpu_ref_get(&ctx->reqs); /* free_ioctx_users() will drop this */ |
723be6e39
|
729 |
|
da90382c2
|
730 731 |
err = ioctx_add_table(ctx, mm); if (err) |
e34ecee2a
|
732 |
goto err_cleanup; |
da90382c2
|
733 |
|
fa8a53c39
|
734 735 |
/* Release the ring_lock mutex now that all setup is complete. */ mutex_unlock(&ctx->ring_lock); |
caf4167aa
|
736 737 |
pr_debug("allocated ioctx %p[%ld]: mm=%p mask=0x%x ", |
58c85dc20
|
738 |
ctx, ctx->user_id, mm, ctx->nr_events); |
1da177e4c
|
739 |
return ctx; |
e34ecee2a
|
740 741 |
err_cleanup: aio_nr_sub(ctx->max_reqs); |
d1b943271
|
742 |
err_ctx: |
deeb8525f
|
743 744 745 |
atomic_set(&ctx->dead, 1); if (ctx->mmap_size) vm_munmap(ctx->mmap_base, ctx->mmap_size); |
d1b943271
|
746 |
aio_free_ring(ctx); |
e34ecee2a
|
747 |
err: |
fa8a53c39
|
748 |
mutex_unlock(&ctx->ring_lock); |
e1bdd5f27
|
749 |
free_percpu(ctx->cpu); |
9a1049da9
|
750 751 |
percpu_ref_exit(&ctx->reqs); percpu_ref_exit(&ctx->users); |
1da177e4c
|
752 |
kmem_cache_free(kioctx_cachep, ctx); |
caf4167aa
|
753 754 |
pr_debug("error allocating ioctx %d ", err); |
e23754f88
|
755 |
return ERR_PTR(err); |
1da177e4c
|
756 |
} |
36f558890
|
757 758 759 760 761 |
/* kill_ioctx * Cancels all outstanding aio requests on an aio context. Used * when the processes owning a context have all exited to encourage * the rapid destruction of the kioctx. */ |
fb2d44838
|
762 |
static int kill_ioctx(struct mm_struct *mm, struct kioctx *ctx, |
dc48e56d7
|
763 |
struct ctx_rq_wait *wait) |
36f558890
|
764 |
{ |
fa88b6f88
|
765 |
struct kioctx_table *table; |
db446a08c
|
766 |
|
b2edffdd9
|
767 768 769 |
spin_lock(&mm->ioctx_lock); if (atomic_xchg(&ctx->dead, 1)) { spin_unlock(&mm->ioctx_lock); |
fa88b6f88
|
770 |
return -EINVAL; |
b2edffdd9
|
771 |
} |
db446a08c
|
772 |
|
855ef0dec
|
773 |
table = rcu_dereference_raw(mm->ioctx_table); |
d0264c01e
|
774 775 |
WARN_ON(ctx != rcu_access_pointer(table->table[ctx->id])); RCU_INIT_POINTER(table->table[ctx->id], NULL); |
fa88b6f88
|
776 |
spin_unlock(&mm->ioctx_lock); |
4fcc712f5
|
777 |
|
a6d7cff47
|
778 |
/* free_ioctx_reqs() will do the necessary RCU synchronization */ |
fa88b6f88
|
779 |
wake_up_all(&ctx->wait); |
4fcc712f5
|
780 |
|
fa88b6f88
|
781 782 783 784 785 786 787 788 |
/* * It'd be more correct to do this in free_ioctx(), after all * the outstanding kiocbs have finished - but by then io_destroy * has already returned, so io_setup() could potentially return * -EAGAIN with no ioctxs actually in use (as far as userspace * could tell). */ aio_nr_sub(ctx->max_reqs); |
4fcc712f5
|
789 |
|
fa88b6f88
|
790 791 |
if (ctx->mmap_size) vm_munmap(ctx->mmap_base, ctx->mmap_size); |
fb2d44838
|
792 |
|
dc48e56d7
|
793 |
ctx->rq_wait = wait; |
fa88b6f88
|
794 795 |
percpu_ref_kill(&ctx->users); return 0; |
1da177e4c
|
796 |
} |
36f558890
|
797 798 799 800 801 802 803 |
/* * exit_aio: called when the last user of mm goes away. At this point, there is * no way for any new requests to be submited or any of the io_* syscalls to be * called on the context. * * There may be outstanding kiocbs, but free_ioctx() will explicitly wait on * them. |
1da177e4c
|
804 |
*/ |
fc9b52cd8
|
805 |
void exit_aio(struct mm_struct *mm) |
1da177e4c
|
806 |
{ |
4b70ac5fd
|
807 |
struct kioctx_table *table = rcu_dereference_raw(mm->ioctx_table); |
dc48e56d7
|
808 809 |
struct ctx_rq_wait wait; int i, skipped; |
db446a08c
|
810 |
|
4b70ac5fd
|
811 812 |
if (!table) return; |
db446a08c
|
813 |
|
dc48e56d7
|
814 815 816 817 |
atomic_set(&wait.count, table->nr); init_completion(&wait.comp); skipped = 0; |
4b70ac5fd
|
818 |
for (i = 0; i < table->nr; ++i) { |
d0264c01e
|
819 820 |
struct kioctx *ctx = rcu_dereference_protected(table->table[i], true); |
abf137dd7
|
821 |
|
dc48e56d7
|
822 823 |
if (!ctx) { skipped++; |
4b70ac5fd
|
824 |
continue; |
dc48e56d7
|
825 |
} |
936af1576
|
826 |
/* |
4b70ac5fd
|
827 828 829 830 831 |
* We don't need to bother with munmap() here - exit_mmap(mm) * is coming and it'll unmap everything. And we simply can't, * this is not necessarily our ->mm. * Since kill_ioctx() uses non-zero ->mmap_size as indicator * that it needs to unmap the area, just set it to 0. |
936af1576
|
832 |
*/ |
58c85dc20
|
833 |
ctx->mmap_size = 0; |
dc48e56d7
|
834 835 |
kill_ioctx(mm, ctx, &wait); } |
36f558890
|
836 |
|
dc48e56d7
|
837 |
if (!atomic_sub_and_test(skipped, &wait.count)) { |
6098b45b3
|
838 |
/* Wait until all IO for the context are done. */ |
dc48e56d7
|
839 |
wait_for_completion(&wait.comp); |
1da177e4c
|
840 |
} |
4b70ac5fd
|
841 842 843 |
RCU_INIT_POINTER(mm->ioctx_table, NULL); kfree(table); |
1da177e4c
|
844 |
} |
e1bdd5f27
|
845 846 847 |
static void put_reqs_available(struct kioctx *ctx, unsigned nr) { struct kioctx_cpu *kcpu; |
263782c1c
|
848 |
unsigned long flags; |
e1bdd5f27
|
849 |
|
263782c1c
|
850 |
local_irq_save(flags); |
be6fb451a
|
851 |
kcpu = this_cpu_ptr(ctx->cpu); |
e1bdd5f27
|
852 |
kcpu->reqs_available += nr; |
263782c1c
|
853 |
|
e1bdd5f27
|
854 855 856 857 |
while (kcpu->reqs_available >= ctx->req_batch * 2) { kcpu->reqs_available -= ctx->req_batch; atomic_add(ctx->req_batch, &ctx->reqs_available); } |
263782c1c
|
858 |
local_irq_restore(flags); |
e1bdd5f27
|
859 |
} |
432c79978
|
860 |
static bool __get_reqs_available(struct kioctx *ctx) |
e1bdd5f27
|
861 862 863 |
{ struct kioctx_cpu *kcpu; bool ret = false; |
263782c1c
|
864 |
unsigned long flags; |
e1bdd5f27
|
865 |
|
263782c1c
|
866 |
local_irq_save(flags); |
be6fb451a
|
867 |
kcpu = this_cpu_ptr(ctx->cpu); |
e1bdd5f27
|
868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 |
if (!kcpu->reqs_available) { int old, avail = atomic_read(&ctx->reqs_available); do { if (avail < ctx->req_batch) goto out; old = avail; avail = atomic_cmpxchg(&ctx->reqs_available, avail, avail - ctx->req_batch); } while (avail != old); kcpu->reqs_available += ctx->req_batch; } ret = true; kcpu->reqs_available--; out: |
263782c1c
|
886 |
local_irq_restore(flags); |
e1bdd5f27
|
887 888 |
return ret; } |
d856f32a8
|
889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 |
/* refill_reqs_available * Updates the reqs_available reference counts used for tracking the * number of free slots in the completion ring. This can be called * from aio_complete() (to optimistically update reqs_available) or * from aio_get_req() (the we're out of events case). It must be * called holding ctx->completion_lock. */ static void refill_reqs_available(struct kioctx *ctx, unsigned head, unsigned tail) { unsigned events_in_ring, completed; /* Clamp head since userland can write to it. */ head %= ctx->nr_events; if (head <= tail) events_in_ring = tail - head; else events_in_ring = ctx->nr_events - (head - tail); completed = ctx->completed_events; if (events_in_ring < completed) completed -= events_in_ring; else completed = 0; if (!completed) return; ctx->completed_events -= completed; put_reqs_available(ctx, completed); } /* user_refill_reqs_available * Called to refill reqs_available when aio_get_req() encounters an * out of space in the completion ring. */ static void user_refill_reqs_available(struct kioctx *ctx) { spin_lock_irq(&ctx->completion_lock); if (ctx->completed_events) { struct aio_ring *ring; unsigned head; /* Access of ring->head may race with aio_read_events_ring() * here, but that's okay since whether we read the old version * or the new version, and either will be valid. The important * part is that head cannot pass tail since we prevent * aio_complete() from updating tail by holding * ctx->completion_lock. Even if head is invalid, the check * against ctx->completed_events below will make sure we do the * safe/right thing. */ ring = kmap_atomic(ctx->ring_pages[0]); head = ring->head; kunmap_atomic(ring); refill_reqs_available(ctx, head, ctx->tail); } spin_unlock_irq(&ctx->completion_lock); } |
432c79978
|
950 951 952 953 954 955 956 |
static bool get_reqs_available(struct kioctx *ctx) { if (__get_reqs_available(ctx)) return true; user_refill_reqs_available(ctx); return __get_reqs_available(ctx); } |
1da177e4c
|
957 |
/* aio_get_req |
57282d8fd
|
958 959 |
* Allocate a slot for an aio request. * Returns NULL if no requests are free. |
b53119f13
|
960 961 962 |
* * The refcount is initialized to 2 - one for the async op completion, * one for the synchronous code that does this. |
1da177e4c
|
963 |
*/ |
04b2fa9f8
|
964 |
static inline struct aio_kiocb *aio_get_req(struct kioctx *ctx) |
1da177e4c
|
965 |
{ |
04b2fa9f8
|
966 |
struct aio_kiocb *req; |
a1c8eae75
|
967 |
|
2bc4ca9bb
|
968 |
req = kmem_cache_alloc(kiocb_cachep, GFP_KERNEL); |
1da177e4c
|
969 |
if (unlikely(!req)) |
432c79978
|
970 |
return NULL; |
1da177e4c
|
971 |
|
fa0ca2aee
|
972 |
if (unlikely(!get_reqs_available(ctx))) { |
6af1c849d
|
973 |
kmem_cache_free(kiocb_cachep, req); |
fa0ca2aee
|
974 975 |
return NULL; } |
e34ecee2a
|
976 |
percpu_ref_get(&ctx->reqs); |
2bc4ca9bb
|
977 |
req->ki_ctx = ctx; |
75321b50a
|
978 |
INIT_LIST_HEAD(&req->ki_list); |
b53119f13
|
979 |
refcount_set(&req->ki_refcnt, 2); |
2bc4ca9bb
|
980 |
req->ki_eventfd = NULL; |
080d676de
|
981 |
return req; |
1da177e4c
|
982 |
} |
d5470b596
|
983 |
static struct kioctx *lookup_ioctx(unsigned long ctx_id) |
1da177e4c
|
984 |
{ |
db446a08c
|
985 |
struct aio_ring __user *ring = (void __user *)ctx_id; |
abf137dd7
|
986 |
struct mm_struct *mm = current->mm; |
65c24491b
|
987 |
struct kioctx *ctx, *ret = NULL; |
db446a08c
|
988 989 990 991 992 |
struct kioctx_table *table; unsigned id; if (get_user(id, &ring->id)) return NULL; |
1da177e4c
|
993 |
|
abf137dd7
|
994 |
rcu_read_lock(); |
db446a08c
|
995 |
table = rcu_dereference(mm->ioctx_table); |
abf137dd7
|
996 |
|
db446a08c
|
997 998 |
if (!table || id >= table->nr) goto out; |
1da177e4c
|
999 |
|
a538e3ff9
|
1000 |
id = array_index_nospec(id, table->nr); |
d0264c01e
|
1001 |
ctx = rcu_dereference(table->table[id]); |
f30d704fe
|
1002 |
if (ctx && ctx->user_id == ctx_id) { |
baf10564f
|
1003 1004 |
if (percpu_ref_tryget_live(&ctx->users)) ret = ctx; |
db446a08c
|
1005 1006 |
} out: |
abf137dd7
|
1007 |
rcu_read_unlock(); |
65c24491b
|
1008 |
return ret; |
1da177e4c
|
1009 |
} |
b53119f13
|
1010 1011 |
static inline void iocb_destroy(struct aio_kiocb *iocb) { |
742597034
|
1012 1013 |
if (iocb->ki_eventfd) eventfd_ctx_put(iocb->ki_eventfd); |
b53119f13
|
1014 1015 1016 1017 1018 |
if (iocb->ki_filp) fput(iocb->ki_filp); percpu_ref_put(&iocb->ki_ctx->reqs); kmem_cache_free(kiocb_cachep, iocb); } |
1da177e4c
|
1019 1020 |
/* aio_complete * Called when the io request on the given iocb is complete. |
1da177e4c
|
1021 |
*/ |
2bb874c0d
|
1022 |
static void aio_complete(struct aio_kiocb *iocb) |
1da177e4c
|
1023 1024 |
{ struct kioctx *ctx = iocb->ki_ctx; |
1da177e4c
|
1025 |
struct aio_ring *ring; |
21b40200c
|
1026 |
struct io_event *ev_page, *event; |
d856f32a8
|
1027 |
unsigned tail, pos, head; |
1da177e4c
|
1028 |
unsigned long flags; |
1da177e4c
|
1029 |
|
1da177e4c
|
1030 |
/* |
0460fef2a
|
1031 |
* Add a completion event to the ring buffer. Must be done holding |
4b30f07e7
|
1032 |
* ctx->completion_lock to prevent other code from messing with the tail |
0460fef2a
|
1033 1034 1035 |
* pointer since we might be called from irq context. */ spin_lock_irqsave(&ctx->completion_lock, flags); |
58c85dc20
|
1036 |
tail = ctx->tail; |
21b40200c
|
1037 |
pos = tail + AIO_EVENTS_OFFSET; |
58c85dc20
|
1038 |
if (++tail >= ctx->nr_events) |
4bf69b2a0
|
1039 |
tail = 0; |
1da177e4c
|
1040 |
|
58c85dc20
|
1041 |
ev_page = kmap_atomic(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); |
21b40200c
|
1042 |
event = ev_page + pos % AIO_EVENTS_PER_PAGE; |
a9339b785
|
1043 |
*event = iocb->ki_res; |
1da177e4c
|
1044 |
|
21b40200c
|
1045 |
kunmap_atomic(ev_page); |
58c85dc20
|
1046 |
flush_dcache_page(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); |
21b40200c
|
1047 |
|
a9339b785
|
1048 1049 1050 1051 |
pr_debug("%p[%u]: %p: %p %Lx %Lx %Lx ", ctx, tail, iocb, (void __user *)(unsigned long)iocb->ki_res.obj, iocb->ki_res.data, iocb->ki_res.res, iocb->ki_res.res2); |
1da177e4c
|
1052 1053 1054 1055 1056 |
/* after flagging the request as done, we * must never even look at it again */ smp_wmb(); /* make event visible before updating tail */ |
58c85dc20
|
1057 |
ctx->tail = tail; |
1da177e4c
|
1058 |
|
58c85dc20
|
1059 |
ring = kmap_atomic(ctx->ring_pages[0]); |
d856f32a8
|
1060 |
head = ring->head; |
21b40200c
|
1061 |
ring->tail = tail; |
e8e3c3d66
|
1062 |
kunmap_atomic(ring); |
58c85dc20
|
1063 |
flush_dcache_page(ctx->ring_pages[0]); |
1da177e4c
|
1064 |
|
d856f32a8
|
1065 1066 1067 |
ctx->completed_events++; if (ctx->completed_events > 1) refill_reqs_available(ctx, head, tail); |
0460fef2a
|
1068 |
spin_unlock_irqrestore(&ctx->completion_lock, flags); |
21b40200c
|
1069 1070 |
pr_debug("added to ring %p at [%u] ", iocb, tail); |
8d1c98b0b
|
1071 1072 1073 1074 1075 1076 |
/* * 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. */ |
742597034
|
1077 |
if (iocb->ki_eventfd) |
8d1c98b0b
|
1078 |
eventfd_signal(iocb->ki_eventfd, 1); |
6cb2a2104
|
1079 1080 1081 1082 1083 1084 1085 |
/* * 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(); |
1da177e4c
|
1086 1087 |
if (waitqueue_active(&ctx->wait)) wake_up(&ctx->wait); |
2bb874c0d
|
1088 1089 1090 1091 1092 1093 1094 1095 |
} static inline void iocb_put(struct aio_kiocb *iocb) { if (refcount_dec_and_test(&iocb->ki_refcnt)) { aio_complete(iocb); iocb_destroy(iocb); } |
1da177e4c
|
1096 |
} |
2be4e7dee
|
1097 |
/* aio_read_events_ring |
a31ad380b
|
1098 1099 |
* Pull an event off of the ioctx's event ring. Returns the number of * events fetched |
1da177e4c
|
1100 |
*/ |
a31ad380b
|
1101 1102 |
static long aio_read_events_ring(struct kioctx *ctx, struct io_event __user *event, long nr) |
1da177e4c
|
1103 |
{ |
1da177e4c
|
1104 |
struct aio_ring *ring; |
5ffac122d
|
1105 |
unsigned head, tail, pos; |
a31ad380b
|
1106 1107 |
long ret = 0; int copy_ret; |
9c9ce763b
|
1108 1109 1110 1111 1112 1113 1114 |
/* * The mutex can block and wake us up and that will cause * wait_event_interruptible_hrtimeout() to schedule without sleeping * and repeat. This should be rare enough that it doesn't cause * peformance issues. See the comment in read_events() for more detail. */ sched_annotate_sleep(); |
58c85dc20
|
1115 |
mutex_lock(&ctx->ring_lock); |
1da177e4c
|
1116 |
|
fa8a53c39
|
1117 |
/* Access to ->ring_pages here is protected by ctx->ring_lock. */ |
58c85dc20
|
1118 |
ring = kmap_atomic(ctx->ring_pages[0]); |
a31ad380b
|
1119 |
head = ring->head; |
5ffac122d
|
1120 |
tail = ring->tail; |
a31ad380b
|
1121 |
kunmap_atomic(ring); |
2ff396be6
|
1122 1123 1124 1125 1126 |
/* * Ensure that once we've read the current tail pointer, that * we also see the events that were stored up to the tail. */ smp_rmb(); |
5ffac122d
|
1127 1128 |
pr_debug("h%u t%u m%u ", head, tail, ctx->nr_events); |
1da177e4c
|
1129 |
|
5ffac122d
|
1130 |
if (head == tail) |
1da177e4c
|
1131 |
goto out; |
edfbbf388
|
1132 1133 |
head %= ctx->nr_events; tail %= ctx->nr_events; |
a31ad380b
|
1134 1135 1136 1137 |
while (ret < nr) { long avail; struct io_event *ev; struct page *page; |
5ffac122d
|
1138 1139 |
avail = (head <= tail ? tail : ctx->nr_events) - head; if (head == tail) |
a31ad380b
|
1140 |
break; |
a31ad380b
|
1141 |
pos = head + AIO_EVENTS_OFFSET; |
58c85dc20
|
1142 |
page = ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]; |
a31ad380b
|
1143 |
pos %= AIO_EVENTS_PER_PAGE; |
d2988bd41
|
1144 1145 |
avail = min(avail, nr - ret); avail = min_t(long, avail, AIO_EVENTS_PER_PAGE - pos); |
a31ad380b
|
1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 |
ev = kmap(page); copy_ret = copy_to_user(event + ret, ev + pos, sizeof(*ev) * avail); kunmap(page); if (unlikely(copy_ret)) { ret = -EFAULT; goto out; } ret += avail; head += avail; |
58c85dc20
|
1158 |
head %= ctx->nr_events; |
1da177e4c
|
1159 |
} |
1da177e4c
|
1160 |
|
58c85dc20
|
1161 |
ring = kmap_atomic(ctx->ring_pages[0]); |
a31ad380b
|
1162 |
ring->head = head; |
91d80a84b
|
1163 |
kunmap_atomic(ring); |
58c85dc20
|
1164 |
flush_dcache_page(ctx->ring_pages[0]); |
a31ad380b
|
1165 |
|
5ffac122d
|
1166 1167 |
pr_debug("%li h%u t%u ", ret, head, tail); |
a31ad380b
|
1168 |
out: |
58c85dc20
|
1169 |
mutex_unlock(&ctx->ring_lock); |
a31ad380b
|
1170 |
|
1da177e4c
|
1171 1172 |
return ret; } |
a31ad380b
|
1173 1174 |
static bool aio_read_events(struct kioctx *ctx, long min_nr, long nr, struct io_event __user *event, long *i) |
1da177e4c
|
1175 |
{ |
a31ad380b
|
1176 |
long ret = aio_read_events_ring(ctx, event + *i, nr - *i); |
1da177e4c
|
1177 |
|
a31ad380b
|
1178 1179 |
if (ret > 0) *i += ret; |
1da177e4c
|
1180 |
|
a31ad380b
|
1181 1182 |
if (unlikely(atomic_read(&ctx->dead))) ret = -EINVAL; |
1da177e4c
|
1183 |
|
a31ad380b
|
1184 1185 |
if (!*i) *i = ret; |
1da177e4c
|
1186 |
|
a31ad380b
|
1187 |
return ret < 0 || *i >= min_nr; |
1da177e4c
|
1188 |
} |
a31ad380b
|
1189 |
static long read_events(struct kioctx *ctx, long min_nr, long nr, |
1da177e4c
|
1190 |
struct io_event __user *event, |
fa2e62a54
|
1191 |
ktime_t until) |
1da177e4c
|
1192 |
{ |
a31ad380b
|
1193 |
long ret = 0; |
1da177e4c
|
1194 |
|
a31ad380b
|
1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 |
/* * Note that aio_read_events() is being called as the conditional - i.e. * we're calling it after prepare_to_wait() has set task state to * TASK_INTERRUPTIBLE. * * But aio_read_events() can block, and if it blocks it's going to flip * the task state back to TASK_RUNNING. * * This should be ok, provided it doesn't flip the state back to * TASK_RUNNING and return 0 too much - that causes us to spin. That * will only happen if the mutex_lock() call blocks, and we then find * the ringbuffer empty. So in practice we should be ok, but it's * something to be aware of when touching this code. */ |
2456e8553
|
1209 |
if (until == 0) |
5f785de58
|
1210 1211 1212 1213 1214 |
aio_read_events(ctx, min_nr, nr, event, &ret); else wait_event_interruptible_hrtimeout(ctx->wait, aio_read_events(ctx, min_nr, nr, event, &ret), until); |
a31ad380b
|
1215 |
return ret; |
1da177e4c
|
1216 |
} |
1da177e4c
|
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 |
/* 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. */ |
002c8976e
|
1230 |
SYSCALL_DEFINE2(io_setup, unsigned, nr_events, aio_context_t __user *, ctxp) |
1da177e4c
|
1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 |
{ struct kioctx *ioctx = NULL; unsigned long ctx; long ret; ret = get_user(ctx, ctxp); if (unlikely(ret)) goto out; ret = -EINVAL; |
d55b5fdaf
|
1241 |
if (unlikely(ctx || nr_events == 0)) { |
acd88d4e1
|
1242 1243 |
pr_debug("EINVAL: ctx %lu nr_events %u ", |
d55b5fdaf
|
1244 |
ctx, nr_events); |
1da177e4c
|
1245 1246 1247 1248 1249 1250 1251 |
goto out; } ioctx = ioctx_alloc(nr_events); ret = PTR_ERR(ioctx); if (!IS_ERR(ioctx)) { ret = put_user(ioctx->user_id, ctxp); |
a2e1859ad
|
1252 |
if (ret) |
e02ba72aa
|
1253 |
kill_ioctx(current->mm, ioctx, NULL); |
723be6e39
|
1254 |
percpu_ref_put(&ioctx->users); |
1da177e4c
|
1255 1256 1257 1258 1259 |
} out: return ret; } |
c00d2c7e8
|
1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 |
#ifdef CONFIG_COMPAT COMPAT_SYSCALL_DEFINE2(io_setup, unsigned, nr_events, u32 __user *, ctx32p) { struct kioctx *ioctx = NULL; unsigned long ctx; long ret; ret = get_user(ctx, ctx32p); if (unlikely(ret)) goto out; ret = -EINVAL; if (unlikely(ctx || nr_events == 0)) { pr_debug("EINVAL: ctx %lu nr_events %u ", ctx, nr_events); goto out; } ioctx = ioctx_alloc(nr_events); ret = PTR_ERR(ioctx); if (!IS_ERR(ioctx)) { /* truncating is ok because it's a user address */ ret = put_user((u32)ioctx->user_id, ctx32p); if (ret) kill_ioctx(current->mm, ioctx, NULL); percpu_ref_put(&ioctx->users); } out: return ret; } #endif |
1da177e4c
|
1293 1294 1295 |
/* sys_io_destroy: * Destroy the aio_context specified. May cancel any outstanding * AIOs and block on completion. Will fail with -ENOSYS if not |
642b5123a
|
1296 |
* implemented. May fail with -EINVAL if the context pointed to |
1da177e4c
|
1297 1298 |
* is invalid. */ |
002c8976e
|
1299 |
SYSCALL_DEFINE1(io_destroy, aio_context_t, ctx) |
1da177e4c
|
1300 1301 1302 |
{ struct kioctx *ioctx = lookup_ioctx(ctx); if (likely(NULL != ioctx)) { |
dc48e56d7
|
1303 |
struct ctx_rq_wait wait; |
fb2d44838
|
1304 |
int ret; |
e02ba72aa
|
1305 |
|
dc48e56d7
|
1306 1307 |
init_completion(&wait.comp); atomic_set(&wait.count, 1); |
e02ba72aa
|
1308 1309 1310 1311 |
/* Pass requests_done to kill_ioctx() where it can be set * in a thread-safe way. If we try to set it here then we have * a race condition if two io_destroy() called simultaneously. */ |
dc48e56d7
|
1312 |
ret = kill_ioctx(current->mm, ioctx, &wait); |
723be6e39
|
1313 |
percpu_ref_put(&ioctx->users); |
e02ba72aa
|
1314 1315 1316 1317 1318 |
/* Wait until all IO for the context are done. Otherwise kernel * keep using user-space buffers even if user thinks the context * is destroyed. */ |
fb2d44838
|
1319 |
if (!ret) |
dc48e56d7
|
1320 |
wait_for_completion(&wait.comp); |
e02ba72aa
|
1321 |
|
fb2d44838
|
1322 |
return ret; |
1da177e4c
|
1323 |
} |
acd88d4e1
|
1324 1325 |
pr_debug("EINVAL: invalid context id "); |
1da177e4c
|
1326 1327 |
return -EINVAL; } |
3c96c7f4c
|
1328 1329 1330 1331 1332 1333 1334 1335 1336 |
static void aio_remove_iocb(struct aio_kiocb *iocb) { struct kioctx *ctx = iocb->ki_ctx; unsigned long flags; spin_lock_irqsave(&ctx->ctx_lock, flags); list_del(&iocb->ki_list); spin_unlock_irqrestore(&ctx->ctx_lock, flags); } |
54843f875
|
1337 1338 1339 |
static void aio_complete_rw(struct kiocb *kiocb, long res, long res2) { struct aio_kiocb *iocb = container_of(kiocb, struct aio_kiocb, rw); |
3c96c7f4c
|
1340 1341 |
if (!list_empty_careful(&iocb->ki_list)) aio_remove_iocb(iocb); |
54843f875
|
1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 |
if (kiocb->ki_flags & IOCB_WRITE) { struct inode *inode = file_inode(kiocb->ki_filp); /* * Tell lockdep we inherited freeze protection from submission * thread. */ if (S_ISREG(inode->i_mode)) __sb_writers_acquired(inode->i_sb, SB_FREEZE_WRITE); file_end_write(kiocb->ki_filp); } |
2bb874c0d
|
1353 1354 1355 |
iocb->ki_res.res = res; iocb->ki_res.res2 = res2; iocb_put(iocb); |
54843f875
|
1356 |
} |
88a6f18b9
|
1357 |
static int aio_prep_rw(struct kiocb *req, const struct iocb *iocb) |
54843f875
|
1358 1359 |
{ int ret; |
54843f875
|
1360 |
req->ki_complete = aio_complete_rw; |
ec51f8ee1
|
1361 |
req->private = NULL; |
54843f875
|
1362 1363 1364 1365 |
req->ki_pos = iocb->aio_offset; req->ki_flags = iocb_flags(req->ki_filp); if (iocb->aio_flags & IOCB_FLAG_RESFD) req->ki_flags |= IOCB_EVENTFD; |
fc28724d6
|
1366 |
req->ki_hint = ki_hint_validate(file_write_hint(req->ki_filp)); |
d9a08a9e6
|
1367 1368 1369 1370 1371 1372 1373 1374 |
if (iocb->aio_flags & IOCB_FLAG_IOPRIO) { /* * If the IOCB_FLAG_IOPRIO flag of aio_flags is set, then * aio_reqprio is interpreted as an I/O scheduling * class and priority. */ ret = ioprio_check_cap(iocb->aio_reqprio); if (ret) { |
9a6d9a62e
|
1375 1376 |
pr_debug("aio ioprio check cap error: %d ", ret); |
84c4e1f89
|
1377 |
return ret; |
d9a08a9e6
|
1378 1379 1380 1381 |
} req->ki_ioprio = iocb->aio_reqprio; } else |
76dc89139
|
1382 |
req->ki_ioprio = get_current_ioprio(); |
d9a08a9e6
|
1383 |
|
54843f875
|
1384 1385 |
ret = kiocb_set_rw_flags(req, iocb->aio_rw_flags); if (unlikely(ret)) |
84c4e1f89
|
1386 |
return ret; |
154989e45
|
1387 1388 1389 |
req->ki_flags &= ~IOCB_HIPRI; /* no one is going to poll for this I/O */ return 0; |
54843f875
|
1390 |
} |
87e5e6dab
|
1391 1392 1393 |
static ssize_t aio_setup_rw(int rw, const struct iocb *iocb, struct iovec **iovec, bool vectored, bool compat, struct iov_iter *iter) |
eed4e51fb
|
1394 |
{ |
89319d31d
|
1395 1396 1397 1398 1399 1400 1401 1402 |
void __user *buf = (void __user *)(uintptr_t)iocb->aio_buf; size_t len = iocb->aio_nbytes; if (!vectored) { ssize_t ret = import_single_range(rw, buf, len, *iovec, iter); *iovec = NULL; return ret; } |
9d85cba71
|
1403 1404 |
#ifdef CONFIG_COMPAT if (compat) |
89319d31d
|
1405 1406 |
return compat_import_iovec(rw, buf, len, UIO_FASTIOV, iovec, iter); |
9d85cba71
|
1407 |
#endif |
89319d31d
|
1408 |
return import_iovec(rw, buf, len, UIO_FASTIOV, iovec, iter); |
eed4e51fb
|
1409 |
} |
9061d14a8
|
1410 |
static inline void aio_rw_done(struct kiocb *req, ssize_t ret) |
89319d31d
|
1411 1412 1413 |
{ switch (ret) { case -EIOCBQUEUED: |
9061d14a8
|
1414 |
break; |
89319d31d
|
1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 |
case -ERESTARTSYS: case -ERESTARTNOINTR: case -ERESTARTNOHAND: case -ERESTART_RESTARTBLOCK: /* * There's no easy way to restart the syscall since other AIO's * may be already running. Just fail this IO with EINTR. */ ret = -EINTR; /*FALLTHRU*/ default: |
bc9bff616
|
1426 |
req->ki_complete(req, ret, 0); |
89319d31d
|
1427 1428 |
} } |
958c13ce1
|
1429 |
static int aio_read(struct kiocb *req, const struct iocb *iocb, |
88a6f18b9
|
1430 |
bool vectored, bool compat) |
1da177e4c
|
1431 |
{ |
00fefb9cf
|
1432 |
struct iovec inline_vecs[UIO_FASTIOV], *iovec = inline_vecs; |
293bc9822
|
1433 |
struct iov_iter iter; |
54843f875
|
1434 |
struct file *file; |
958c13ce1
|
1435 |
int ret; |
1da177e4c
|
1436 |
|
54843f875
|
1437 1438 1439 1440 |
ret = aio_prep_rw(req, iocb); if (ret) return ret; file = req->ki_filp; |
89319d31d
|
1441 |
if (unlikely(!(file->f_mode & FMODE_READ))) |
84c4e1f89
|
1442 |
return -EBADF; |
54843f875
|
1443 |
ret = -EINVAL; |
89319d31d
|
1444 |
if (unlikely(!file->f_op->read_iter)) |
84c4e1f89
|
1445 |
return -EINVAL; |
73a7075e3
|
1446 |
|
89319d31d
|
1447 |
ret = aio_setup_rw(READ, iocb, &iovec, vectored, compat, &iter); |
87e5e6dab
|
1448 |
if (ret < 0) |
84c4e1f89
|
1449 |
return ret; |
89319d31d
|
1450 1451 |
ret = rw_verify_area(READ, file, &req->ki_pos, iov_iter_count(&iter)); if (!ret) |
9061d14a8
|
1452 |
aio_rw_done(req, call_read_iter(file, req, &iter)); |
89319d31d
|
1453 1454 1455 |
kfree(iovec); return ret; } |
73a7075e3
|
1456 |
|
958c13ce1
|
1457 |
static int aio_write(struct kiocb *req, const struct iocb *iocb, |
88a6f18b9
|
1458 |
bool vectored, bool compat) |
89319d31d
|
1459 |
{ |
89319d31d
|
1460 1461 |
struct iovec inline_vecs[UIO_FASTIOV], *iovec = inline_vecs; struct iov_iter iter; |
54843f875
|
1462 |
struct file *file; |
958c13ce1
|
1463 |
int ret; |
41ef4eb8e
|
1464 |
|
54843f875
|
1465 1466 1467 1468 |
ret = aio_prep_rw(req, iocb); if (ret) return ret; file = req->ki_filp; |
89319d31d
|
1469 |
if (unlikely(!(file->f_mode & FMODE_WRITE))) |
84c4e1f89
|
1470 |
return -EBADF; |
89319d31d
|
1471 |
if (unlikely(!file->f_op->write_iter)) |
84c4e1f89
|
1472 |
return -EINVAL; |
1da177e4c
|
1473 |
|
89319d31d
|
1474 |
ret = aio_setup_rw(WRITE, iocb, &iovec, vectored, compat, &iter); |
87e5e6dab
|
1475 |
if (ret < 0) |
84c4e1f89
|
1476 |
return ret; |
89319d31d
|
1477 1478 |
ret = rw_verify_area(WRITE, file, &req->ki_pos, iov_iter_count(&iter)); if (!ret) { |
70fe2f481
|
1479 |
/* |
92ce47285
|
1480 |
* Open-code file_start_write here to grab freeze protection, |
54843f875
|
1481 1482 1483 1484 |
* which will be released by another thread in * aio_complete_rw(). Fool lockdep by telling it the lock got * released so that it doesn't complain about the held lock when * we return to userspace. |
70fe2f481
|
1485 |
*/ |
92ce47285
|
1486 1487 |
if (S_ISREG(file_inode(file)->i_mode)) { __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true); |
a12f1ae61
|
1488 |
__sb_writers_release(file_inode(file)->i_sb, SB_FREEZE_WRITE); |
92ce47285
|
1489 1490 |
} req->ki_flags |= IOCB_WRITE; |
9061d14a8
|
1491 |
aio_rw_done(req, call_write_iter(file, req, &iter)); |
41ef4eb8e
|
1492 |
} |
89319d31d
|
1493 1494 |
kfree(iovec); return ret; |
1da177e4c
|
1495 |
} |
a3c0d439e
|
1496 1497 |
static void aio_fsync_work(struct work_struct *work) { |
2bb874c0d
|
1498 |
struct aio_kiocb *iocb = container_of(work, struct aio_kiocb, fsync.work); |
a3c0d439e
|
1499 |
|
2bb874c0d
|
1500 1501 |
iocb->ki_res.res = vfs_fsync(iocb->fsync.file, iocb->fsync.datasync); iocb_put(iocb); |
a3c0d439e
|
1502 |
} |
88a6f18b9
|
1503 1504 |
static int aio_fsync(struct fsync_iocb *req, const struct iocb *iocb, bool datasync) |
a3c0d439e
|
1505 1506 1507 1508 |
{ if (unlikely(iocb->aio_buf || iocb->aio_offset || iocb->aio_nbytes || iocb->aio_rw_flags)) return -EINVAL; |
a11e1d432
|
1509 |
|
84c4e1f89
|
1510 |
if (unlikely(!req->file->f_op->fsync)) |
a3c0d439e
|
1511 |
return -EINVAL; |
a3c0d439e
|
1512 1513 1514 1515 |
req->datasync = datasync; INIT_WORK(&req->work, aio_fsync_work); schedule_work(&req->work); |
9061d14a8
|
1516 |
return 0; |
a3c0d439e
|
1517 |
} |
01d7a3568
|
1518 1519 1520 1521 1522 1523 1524 |
static void aio_poll_put_work(struct work_struct *work) { struct poll_iocb *req = container_of(work, struct poll_iocb, work); struct aio_kiocb *iocb = container_of(req, struct aio_kiocb, poll); iocb_put(iocb); } |
bfe4037e7
|
1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 |
static void aio_poll_complete_work(struct work_struct *work) { struct poll_iocb *req = container_of(work, struct poll_iocb, work); struct aio_kiocb *iocb = container_of(req, struct aio_kiocb, poll); struct poll_table_struct pt = { ._key = req->events }; struct kioctx *ctx = iocb->ki_ctx; __poll_t mask = 0; if (!READ_ONCE(req->cancelled)) mask = vfs_poll(req->file, &pt) & req->events; /* * Note that ->ki_cancel callers also delete iocb from active_reqs after * calling ->ki_cancel. We need the ctx_lock roundtrip here to * synchronize with them. In the cancellation case the list_del_init * itself is not actually needed, but harmless so we keep it in to * avoid further branches in the fast path. */ spin_lock_irq(&ctx->ctx_lock); if (!mask && !READ_ONCE(req->cancelled)) { add_wait_queue(req->head, &req->wait); spin_unlock_irq(&ctx->ctx_lock); return; } list_del_init(&iocb->ki_list); |
af5c72b1f
|
1550 1551 |
iocb->ki_res.res = mangle_poll(mask); req->done = true; |
bfe4037e7
|
1552 |
spin_unlock_irq(&ctx->ctx_lock); |
af5c72b1f
|
1553 |
iocb_put(iocb); |
bfe4037e7
|
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 |
} /* assumes we are called with irqs disabled */ static int aio_poll_cancel(struct kiocb *iocb) { struct aio_kiocb *aiocb = container_of(iocb, struct aio_kiocb, rw); struct poll_iocb *req = &aiocb->poll; spin_lock(&req->head->lock); WRITE_ONCE(req->cancelled, true); if (!list_empty(&req->wait.entry)) { list_del_init(&req->wait.entry); schedule_work(&aiocb->poll.work); } spin_unlock(&req->head->lock); return 0; } static int aio_poll_wake(struct wait_queue_entry *wait, unsigned mode, int sync, void *key) { struct poll_iocb *req = container_of(wait, struct poll_iocb, wait); |
e8693bcfa
|
1577 |
struct aio_kiocb *iocb = container_of(req, struct aio_kiocb, poll); |
bfe4037e7
|
1578 |
__poll_t mask = key_to_poll(key); |
d3d6a18d7
|
1579 |
unsigned long flags; |
bfe4037e7
|
1580 |
|
bfe4037e7
|
1581 |
/* for instances that support it check for an event match first: */ |
af5c72b1f
|
1582 1583 |
if (mask && !(mask & req->events)) return 0; |
e8693bcfa
|
1584 |
|
af5c72b1f
|
1585 1586 1587 |
list_del_init(&req->wait.entry); if (mask && spin_trylock_irqsave(&iocb->ki_ctx->ctx_lock, flags)) { |
01d7a3568
|
1588 |
struct kioctx *ctx = iocb->ki_ctx; |
d3d6a18d7
|
1589 1590 1591 1592 1593 1594 |
/* * Try to complete the iocb inline if we can. Use * irqsave/irqrestore because not all filesystems (e.g. fuse) * call this function with IRQs disabled and because IRQs * have to be disabled before ctx_lock is obtained. */ |
af5c72b1f
|
1595 1596 1597 |
list_del(&iocb->ki_list); iocb->ki_res.res = mangle_poll(mask); req->done = true; |
01d7a3568
|
1598 1599 1600 1601 1602 1603 1604 1605 |
if (iocb->ki_eventfd && eventfd_signal_count()) { iocb = NULL; INIT_WORK(&req->work, aio_poll_put_work); schedule_work(&req->work); } spin_unlock_irqrestore(&ctx->ctx_lock, flags); if (iocb) iocb_put(iocb); |
af5c72b1f
|
1606 1607 |
} else { schedule_work(&req->work); |
e8693bcfa
|
1608 |
} |
bfe4037e7
|
1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 |
return 1; } struct aio_poll_table { struct poll_table_struct pt; struct aio_kiocb *iocb; int error; }; static void aio_poll_queue_proc(struct file *file, struct wait_queue_head *head, struct poll_table_struct *p) { struct aio_poll_table *pt = container_of(p, struct aio_poll_table, pt); /* multiple wait queues per file are not supported */ if (unlikely(pt->iocb->poll.head)) { pt->error = -EINVAL; return; } pt->error = 0; pt->iocb->poll.head = head; add_wait_queue(head, &pt->iocb->poll.wait); } |
958c13ce1
|
1634 |
static int aio_poll(struct aio_kiocb *aiocb, const struct iocb *iocb) |
bfe4037e7
|
1635 1636 1637 1638 |
{ struct kioctx *ctx = aiocb->ki_ctx; struct poll_iocb *req = &aiocb->poll; struct aio_poll_table apt; |
af5c72b1f
|
1639 |
bool cancel = false; |
bfe4037e7
|
1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 |
__poll_t mask; /* reject any unknown events outside the normal event mask. */ if ((u16)iocb->aio_buf != iocb->aio_buf) return -EINVAL; /* reject fields that are not defined for poll */ if (iocb->aio_offset || iocb->aio_nbytes || iocb->aio_rw_flags) return -EINVAL; INIT_WORK(&req->work, aio_poll_complete_work); req->events = demangle_poll(iocb->aio_buf) | EPOLLERR | EPOLLHUP; |
bfe4037e7
|
1651 |
|
2bc4ca9bb
|
1652 |
req->head = NULL; |
af5c72b1f
|
1653 |
req->done = false; |
2bc4ca9bb
|
1654 |
req->cancelled = false; |
bfe4037e7
|
1655 1656 1657 1658 1659 1660 1661 1662 |
apt.pt._qproc = aio_poll_queue_proc; apt.pt._key = req->events; apt.iocb = aiocb; apt.error = -EINVAL; /* same as no support for IOCB_CMD_POLL */ /* initialized the list so that we can do list_empty checks */ INIT_LIST_HEAD(&req->wait.entry); init_waitqueue_func_entry(&req->wait, aio_poll_wake); |
bfe4037e7
|
1663 |
mask = vfs_poll(req->file, &apt.pt) & req->events; |
bfe4037e7
|
1664 |
spin_lock_irq(&ctx->ctx_lock); |
af5c72b1f
|
1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 |
if (likely(req->head)) { spin_lock(&req->head->lock); if (unlikely(list_empty(&req->wait.entry))) { if (apt.error) cancel = true; apt.error = 0; mask = 0; } if (mask || apt.error) { list_del_init(&req->wait.entry); } else if (cancel) { WRITE_ONCE(req->cancelled, true); } else if (!req->done) { /* actually waiting for an event */ list_add_tail(&aiocb->ki_list, &ctx->active_reqs); aiocb->ki_cancel = aio_poll_cancel; } spin_unlock(&req->head->lock); } if (mask) { /* no async, we'd stolen it */ aiocb->ki_res.res = mangle_poll(mask); |
bfe4037e7
|
1685 |
apt.error = 0; |
bfe4037e7
|
1686 |
} |
bfe4037e7
|
1687 |
spin_unlock_irq(&ctx->ctx_lock); |
bfe4037e7
|
1688 |
if (mask) |
af5c72b1f
|
1689 1690 |
iocb_put(aiocb); return apt.error; |
bfe4037e7
|
1691 |
} |
88a6f18b9
|
1692 |
static int __io_submit_one(struct kioctx *ctx, const struct iocb *iocb, |
7316b49c2
|
1693 1694 |
struct iocb __user *user_iocb, struct aio_kiocb *req, bool compat) |
1da177e4c
|
1695 |
{ |
84c4e1f89
|
1696 |
req->ki_filp = fget(iocb->aio_fildes); |
84c4e1f89
|
1697 |
if (unlikely(!req->ki_filp)) |
7316b49c2
|
1698 |
return -EBADF; |
84c4e1f89
|
1699 |
|
88a6f18b9
|
1700 |
if (iocb->aio_flags & IOCB_FLAG_RESFD) { |
742597034
|
1701 |
struct eventfd_ctx *eventfd; |
9c3060bed
|
1702 1703 1704 1705 1706 1707 |
/* * 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. */ |
742597034
|
1708 |
eventfd = eventfd_ctx_fdget(iocb->aio_resfd); |
7316b49c2
|
1709 |
if (IS_ERR(eventfd)) |
18bfb9c6a
|
1710 |
return PTR_ERR(eventfd); |
7316b49c2
|
1711 |
|
742597034
|
1712 |
req->ki_eventfd = eventfd; |
9830f4be1
|
1713 |
} |
7316b49c2
|
1714 |
if (unlikely(put_user(KIOCB_KEY, &user_iocb->aio_key))) { |
caf4167aa
|
1715 1716 |
pr_debug("EFAULT: aio_key "); |
7316b49c2
|
1717 |
return -EFAULT; |
1da177e4c
|
1718 |
} |
a9339b785
|
1719 1720 1721 1722 |
req->ki_res.obj = (u64)(unsigned long)user_iocb; req->ki_res.data = iocb->aio_data; req->ki_res.res = 0; req->ki_res.res2 = 0; |
1da177e4c
|
1723 |
|
88a6f18b9
|
1724 |
switch (iocb->aio_lio_opcode) { |
89319d31d
|
1725 |
case IOCB_CMD_PREAD: |
7316b49c2
|
1726 |
return aio_read(&req->rw, iocb, false, compat); |
89319d31d
|
1727 |
case IOCB_CMD_PWRITE: |
7316b49c2
|
1728 |
return aio_write(&req->rw, iocb, false, compat); |
89319d31d
|
1729 |
case IOCB_CMD_PREADV: |
7316b49c2
|
1730 |
return aio_read(&req->rw, iocb, true, compat); |
89319d31d
|
1731 |
case IOCB_CMD_PWRITEV: |
7316b49c2
|
1732 |
return aio_write(&req->rw, iocb, true, compat); |
a3c0d439e
|
1733 |
case IOCB_CMD_FSYNC: |
7316b49c2
|
1734 |
return aio_fsync(&req->fsync, iocb, false); |
a3c0d439e
|
1735 |
case IOCB_CMD_FDSYNC: |
7316b49c2
|
1736 |
return aio_fsync(&req->fsync, iocb, true); |
bfe4037e7
|
1737 |
case IOCB_CMD_POLL: |
7316b49c2
|
1738 |
return aio_poll(req, iocb); |
89319d31d
|
1739 |
default: |
88a6f18b9
|
1740 1741 |
pr_debug("invalid aio operation %d ", iocb->aio_lio_opcode); |
7316b49c2
|
1742 |
return -EINVAL; |
89319d31d
|
1743 |
} |
1da177e4c
|
1744 |
} |
88a6f18b9
|
1745 1746 1747 |
static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb, bool compat) { |
7316b49c2
|
1748 |
struct aio_kiocb *req; |
88a6f18b9
|
1749 |
struct iocb iocb; |
7316b49c2
|
1750 |
int err; |
88a6f18b9
|
1751 1752 1753 |
if (unlikely(copy_from_user(&iocb, user_iocb, sizeof(iocb)))) return -EFAULT; |
7316b49c2
|
1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 |
/* enforce forwards compatibility on users */ if (unlikely(iocb.aio_reserved2)) { pr_debug("EINVAL: reserve field set "); 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: overflow check "); return -EINVAL; } req = aio_get_req(ctx); if (unlikely(!req)) return -EAGAIN; err = __io_submit_one(ctx, &iocb, user_iocb, req, compat); /* Done with the synchronous reference */ iocb_put(req); /* * If err is 0, we'd either done aio_complete() ourselves or have * arranged for that to be done asynchronously. Anything non-zero * means that we need to destroy req ourselves. */ if (unlikely(err)) { iocb_destroy(req); put_reqs_available(ctx, 1); } return err; |
88a6f18b9
|
1791 |
} |
67ba049f9
|
1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 |
/* 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) |
1da177e4c
|
1806 1807 1808 |
{ struct kioctx *ctx; long ret = 0; |
080d676de
|
1809 |
int i = 0; |
9f5b94254
|
1810 |
struct blk_plug plug; |
1da177e4c
|
1811 1812 1813 |
if (unlikely(nr < 0)) return -EINVAL; |
1da177e4c
|
1814 1815 |
ctx = lookup_ioctx(ctx_id); if (unlikely(!ctx)) { |
caf4167aa
|
1816 1817 |
pr_debug("EINVAL: invalid context id "); |
1da177e4c
|
1818 1819 |
return -EINVAL; } |
1da92779e
|
1820 1821 |
if (nr > ctx->nr_events) nr = ctx->nr_events; |
a79d40e9b
|
1822 1823 |
if (nr > AIO_PLUG_THRESHOLD) blk_start_plug(&plug); |
67ba049f9
|
1824 |
for (i = 0; i < nr; i++) { |
1da177e4c
|
1825 |
struct iocb __user *user_iocb; |
1da177e4c
|
1826 |
|
67ba049f9
|
1827 |
if (unlikely(get_user(user_iocb, iocbpp + i))) { |
1da177e4c
|
1828 1829 1830 |
ret = -EFAULT; break; } |
67ba049f9
|
1831 |
ret = io_submit_one(ctx, user_iocb, false); |
1da177e4c
|
1832 1833 1834 |
if (ret) break; } |
a79d40e9b
|
1835 1836 |
if (nr > AIO_PLUG_THRESHOLD) blk_finish_plug(&plug); |
1da177e4c
|
1837 |
|
723be6e39
|
1838 |
percpu_ref_put(&ctx->users); |
1da177e4c
|
1839 1840 |
return i ? i : ret; } |
c00d2c7e8
|
1841 |
#ifdef CONFIG_COMPAT |
c00d2c7e8
|
1842 |
COMPAT_SYSCALL_DEFINE3(io_submit, compat_aio_context_t, ctx_id, |
67ba049f9
|
1843 |
int, nr, compat_uptr_t __user *, iocbpp) |
c00d2c7e8
|
1844 |
{ |
67ba049f9
|
1845 1846 1847 1848 |
struct kioctx *ctx; long ret = 0; int i = 0; struct blk_plug plug; |
c00d2c7e8
|
1849 1850 1851 |
if (unlikely(nr < 0)) return -EINVAL; |
67ba049f9
|
1852 1853 1854 1855 1856 1857 |
ctx = lookup_ioctx(ctx_id); if (unlikely(!ctx)) { pr_debug("EINVAL: invalid context id "); return -EINVAL; } |
1da92779e
|
1858 1859 |
if (nr > ctx->nr_events) nr = ctx->nr_events; |
a79d40e9b
|
1860 1861 |
if (nr > AIO_PLUG_THRESHOLD) blk_start_plug(&plug); |
67ba049f9
|
1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 |
for (i = 0; i < nr; i++) { compat_uptr_t user_iocb; if (unlikely(get_user(user_iocb, iocbpp + i))) { ret = -EFAULT; break; } ret = io_submit_one(ctx, compat_ptr(user_iocb), true); if (ret) break; } |
a79d40e9b
|
1874 1875 |
if (nr > AIO_PLUG_THRESHOLD) blk_finish_plug(&plug); |
67ba049f9
|
1876 1877 1878 |
percpu_ref_put(&ctx->users); return i ? i : ret; |
c00d2c7e8
|
1879 1880 |
} #endif |
1da177e4c
|
1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 |
/* 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. */ |
002c8976e
|
1891 1892 |
SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb, struct io_event __user *, result) |
1da177e4c
|
1893 |
{ |
1da177e4c
|
1894 |
struct kioctx *ctx; |
04b2fa9f8
|
1895 |
struct aio_kiocb *kiocb; |
888933f8f
|
1896 |
int ret = -EINVAL; |
1da177e4c
|
1897 |
u32 key; |
a9339b785
|
1898 |
u64 obj = (u64)(unsigned long)iocb; |
1da177e4c
|
1899 |
|
f3a2752a4
|
1900 |
if (unlikely(get_user(key, &iocb->aio_key))) |
1da177e4c
|
1901 |
return -EFAULT; |
f3a2752a4
|
1902 1903 |
if (unlikely(key != KIOCB_KEY)) return -EINVAL; |
1da177e4c
|
1904 1905 1906 1907 1908 1909 |
ctx = lookup_ioctx(ctx_id); if (unlikely(!ctx)) return -EINVAL; spin_lock_irq(&ctx->ctx_lock); |
833f4154e
|
1910 1911 |
/* TODO: use a hash or array, this sucks. */ list_for_each_entry(kiocb, &ctx->active_reqs, ki_list) { |
a9339b785
|
1912 |
if (kiocb->ki_res.obj == obj) { |
833f4154e
|
1913 1914 1915 1916 |
ret = kiocb->ki_cancel(&kiocb->rw); list_del_init(&kiocb->ki_list); break; } |
888933f8f
|
1917 |
} |
1da177e4c
|
1918 |
spin_unlock_irq(&ctx->ctx_lock); |
906b973cf
|
1919 |
if (!ret) { |
bec68faaf
|
1920 1921 1922 1923 |
/* * The result argument is no longer used - the io_event is * always delivered via the ring buffer. -EINPROGRESS indicates * cancellation is progress: |
906b973cf
|
1924 |
*/ |
bec68faaf
|
1925 |
ret = -EINPROGRESS; |
906b973cf
|
1926 |
} |
1da177e4c
|
1927 |
|
723be6e39
|
1928 |
percpu_ref_put(&ctx->users); |
1da177e4c
|
1929 1930 1931 |
return ret; } |
fa2e62a54
|
1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 |
static long do_io_getevents(aio_context_t ctx_id, long min_nr, long nr, struct io_event __user *events, struct timespec64 *ts) { ktime_t until = ts ? timespec64_to_ktime(*ts) : KTIME_MAX; struct kioctx *ioctx = lookup_ioctx(ctx_id); long ret = -EINVAL; if (likely(ioctx)) { if (likely(min_nr <= nr && min_nr >= 0)) ret = read_events(ioctx, min_nr, nr, events, until); percpu_ref_put(&ioctx->users); } return ret; } |
1da177e4c
|
1950 1951 |
/* io_getevents: * Attempts to read at least min_nr events and up to nr events from |
642b5123a
|
1952 1953 1954 1955 1956 1957 1958 1959 |
* the completion queue for the aio_context specified by ctx_id. If * it succeeds, the number of read events is returned. May fail with * -EINVAL if ctx_id is invalid, if min_nr is out of range, if nr is * out of range, if timeout is out of range. May fail with -EFAULT * if any of the memory specified is invalid. May return 0 or * < min_nr if the timeout specified by timeout has elapsed * before sufficient events are available, where timeout == NULL * specifies an infinite timeout. Note that the timeout pointed to by |
6900807c6
|
1960 |
* timeout is relative. Will fail with -ENOSYS if not implemented. |
1da177e4c
|
1961 |
*/ |
3ca47e958
|
1962 |
#ifdef CONFIG_64BIT |
7a35397f8
|
1963 |
|
002c8976e
|
1964 1965 1966 1967 |
SYSCALL_DEFINE5(io_getevents, aio_context_t, ctx_id, long, min_nr, long, nr, struct io_event __user *, events, |
7a35397f8
|
1968 |
struct __kernel_timespec __user *, timeout) |
1da177e4c
|
1969 |
{ |
fa2e62a54
|
1970 |
struct timespec64 ts; |
7a074e96d
|
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 |
int ret; if (timeout && unlikely(get_timespec64(&ts, timeout))) return -EFAULT; ret = do_io_getevents(ctx_id, min_nr, nr, events, timeout ? &ts : NULL); if (!ret && signal_pending(current)) ret = -EINTR; return ret; } |
1da177e4c
|
1981 |
|
7a35397f8
|
1982 |
#endif |
9ba546c01
|
1983 1984 1985 1986 |
struct __aio_sigset { const sigset_t __user *sigmask; size_t sigsetsize; }; |
7a074e96d
|
1987 1988 1989 1990 1991 |
SYSCALL_DEFINE6(io_pgetevents, aio_context_t, ctx_id, long, min_nr, long, nr, struct io_event __user *, events, |
7a35397f8
|
1992 |
struct __kernel_timespec __user *, timeout, |
7a074e96d
|
1993 1994 1995 |
const struct __aio_sigset __user *, usig) { struct __aio_sigset ksig = { NULL, }; |
7a074e96d
|
1996 |
struct timespec64 ts; |
97abc889e
|
1997 |
bool interrupted; |
7a074e96d
|
1998 1999 2000 2001 2002 2003 2004 |
int ret; if (timeout && unlikely(get_timespec64(&ts, timeout))) return -EFAULT; if (usig && copy_from_user(&ksig, usig, sizeof(ksig))) return -EFAULT; |
b772434be
|
2005 |
ret = set_user_sigmask(ksig.sigmask, ksig.sigsetsize); |
7a35397f8
|
2006 2007 |
if (ret) return ret; |
7a074e96d
|
2008 2009 |
ret = do_io_getevents(ctx_id, min_nr, nr, events, timeout ? &ts : NULL); |
97abc889e
|
2010 2011 |
interrupted = signal_pending(current); |
b772434be
|
2012 |
restore_saved_sigmask_unless(interrupted); |
97abc889e
|
2013 |
if (interrupted && !ret) |
7a35397f8
|
2014 |
ret = -ERESTARTNOHAND; |
7a074e96d
|
2015 |
|
7a35397f8
|
2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 |
return ret; } #if defined(CONFIG_COMPAT_32BIT_TIME) && !defined(CONFIG_64BIT) SYSCALL_DEFINE6(io_pgetevents_time32, aio_context_t, ctx_id, long, min_nr, long, nr, struct io_event __user *, events, struct old_timespec32 __user *, timeout, const struct __aio_sigset __user *, usig) { struct __aio_sigset ksig = { NULL, }; |
7a35397f8
|
2030 |
struct timespec64 ts; |
97abc889e
|
2031 |
bool interrupted; |
7a35397f8
|
2032 2033 2034 2035 2036 2037 2038 |
int ret; if (timeout && unlikely(get_old_timespec32(&ts, timeout))) return -EFAULT; if (usig && copy_from_user(&ksig, usig, sizeof(ksig))) return -EFAULT; |
ded653ccb
|
2039 |
|
b772434be
|
2040 |
ret = set_user_sigmask(ksig.sigmask, ksig.sigsetsize); |
ded653ccb
|
2041 2042 |
if (ret) return ret; |
7a074e96d
|
2043 2044 |
ret = do_io_getevents(ctx_id, min_nr, nr, events, timeout ? &ts : NULL); |
97abc889e
|
2045 2046 |
interrupted = signal_pending(current); |
b772434be
|
2047 |
restore_saved_sigmask_unless(interrupted); |
97abc889e
|
2048 |
if (interrupted && !ret) |
854a6ed56
|
2049 |
ret = -ERESTARTNOHAND; |
fa2e62a54
|
2050 |
|
7a074e96d
|
2051 |
return ret; |
1da177e4c
|
2052 |
} |
c00d2c7e8
|
2053 |
|
7a35397f8
|
2054 2055 2056 |
#endif #if defined(CONFIG_COMPAT_32BIT_TIME) |
8dabe7245
|
2057 2058 2059 2060 2061 |
SYSCALL_DEFINE5(io_getevents_time32, __u32, ctx_id, __s32, min_nr, __s32, nr, struct io_event __user *, events, struct old_timespec32 __user *, timeout) |
c00d2c7e8
|
2062 |
{ |
fa2e62a54
|
2063 |
struct timespec64 t; |
7a074e96d
|
2064 |
int ret; |
9afc5eee6
|
2065 |
if (timeout && get_old_timespec32(&t, timeout)) |
7a074e96d
|
2066 2067 2068 2069 2070 2071 2072 |
return -EFAULT; ret = do_io_getevents(ctx_id, min_nr, nr, events, timeout ? &t : NULL); if (!ret && signal_pending(current)) ret = -EINTR; return ret; } |
7a35397f8
|
2073 2074 2075 |
#endif #ifdef CONFIG_COMPAT |
c00d2c7e8
|
2076 |
|
7a074e96d
|
2077 |
struct __compat_aio_sigset { |
97eba80fc
|
2078 |
compat_uptr_t sigmask; |
7a074e96d
|
2079 2080 |
compat_size_t sigsetsize; }; |
7a35397f8
|
2081 |
#if defined(CONFIG_COMPAT_32BIT_TIME) |
7a074e96d
|
2082 2083 2084 2085 2086 |
COMPAT_SYSCALL_DEFINE6(io_pgetevents, compat_aio_context_t, ctx_id, compat_long_t, min_nr, compat_long_t, nr, struct io_event __user *, events, |
9afc5eee6
|
2087 |
struct old_timespec32 __user *, timeout, |
7a074e96d
|
2088 2089 |
const struct __compat_aio_sigset __user *, usig) { |
97eba80fc
|
2090 |
struct __compat_aio_sigset ksig = { 0, }; |
7a074e96d
|
2091 |
struct timespec64 t; |
97abc889e
|
2092 |
bool interrupted; |
7a074e96d
|
2093 |
int ret; |
9afc5eee6
|
2094 |
if (timeout && get_old_timespec32(&t, timeout)) |
7a074e96d
|
2095 2096 2097 2098 |
return -EFAULT; if (usig && copy_from_user(&ksig, usig, sizeof(ksig))) return -EFAULT; |
97eba80fc
|
2099 |
ret = set_compat_user_sigmask(compat_ptr(ksig.sigmask), ksig.sigsetsize); |
ded653ccb
|
2100 2101 |
if (ret) return ret; |
c00d2c7e8
|
2102 |
|
7a074e96d
|
2103 |
ret = do_io_getevents(ctx_id, min_nr, nr, events, timeout ? &t : NULL); |
97abc889e
|
2104 2105 |
interrupted = signal_pending(current); |
b772434be
|
2106 |
restore_saved_sigmask_unless(interrupted); |
97abc889e
|
2107 |
if (interrupted && !ret) |
854a6ed56
|
2108 |
ret = -ERESTARTNOHAND; |
fa2e62a54
|
2109 |
|
7a074e96d
|
2110 |
return ret; |
c00d2c7e8
|
2111 |
} |
7a35397f8
|
2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 |
#endif COMPAT_SYSCALL_DEFINE6(io_pgetevents_time64, compat_aio_context_t, ctx_id, compat_long_t, min_nr, compat_long_t, nr, struct io_event __user *, events, struct __kernel_timespec __user *, timeout, const struct __compat_aio_sigset __user *, usig) { |
97eba80fc
|
2123 |
struct __compat_aio_sigset ksig = { 0, }; |
7a35397f8
|
2124 |
struct timespec64 t; |
97abc889e
|
2125 |
bool interrupted; |
7a35397f8
|
2126 2127 2128 2129 2130 2131 2132 |
int ret; if (timeout && get_timespec64(&t, timeout)) return -EFAULT; if (usig && copy_from_user(&ksig, usig, sizeof(ksig))) return -EFAULT; |
97eba80fc
|
2133 |
ret = set_compat_user_sigmask(compat_ptr(ksig.sigmask), ksig.sigsetsize); |
7a35397f8
|
2134 2135 2136 2137 |
if (ret) return ret; ret = do_io_getevents(ctx_id, min_nr, nr, events, timeout ? &t : NULL); |
97abc889e
|
2138 2139 |
interrupted = signal_pending(current); |
b772434be
|
2140 |
restore_saved_sigmask_unless(interrupted); |
97abc889e
|
2141 |
if (interrupted && !ret) |
7a35397f8
|
2142 |
ret = -ERESTARTNOHAND; |
fa2e62a54
|
2143 |
|
7a074e96d
|
2144 |
return ret; |
c00d2c7e8
|
2145 2146 |
} #endif |