Blame view
fs/aio.c
43.9 KB
1da177e4c Linux-2.6.12-rc2 |
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. */ |
caf4167aa aio: dprintk() ->... |
11 |
#define pr_fmt(fmt) "%s: " fmt, __func__ |
1da177e4c Linux-2.6.12-rc2 |
12 13 14 15 16 |
#include <linux/kernel.h> #include <linux/init.h> #include <linux/errno.h> #include <linux/time.h> #include <linux/aio_abi.h> |
630d9c472 fs: reduce the us... |
17 |
#include <linux/export.h> |
1da177e4c Linux-2.6.12-rc2 |
18 |
#include <linux/syscalls.h> |
b9d128f10 block: move bdi/a... |
19 |
#include <linux/backing-dev.h> |
027445c37 [PATCH] Vectorize... |
20 |
#include <linux/uio.h> |
1da177e4c Linux-2.6.12-rc2 |
21 |
|
1da177e4c Linux-2.6.12-rc2 |
22 23 24 25 26 |
#include <linux/sched.h> #include <linux/fs.h> #include <linux/file.h> #include <linux/mm.h> #include <linux/mman.h> |
3d2d827f5 mm: move use_mm/u... |
27 |
#include <linux/mmu_context.h> |
e1bdd5f27 aio: percpu reqs_... |
28 |
#include <linux/percpu.h> |
1da177e4c Linux-2.6.12-rc2 |
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> |
9c3060bed signal/timer/even... |
35 |
#include <linux/eventfd.h> |
cfb1e33ee aio: implement re... |
36 |
#include <linux/blkdev.h> |
9d85cba71 aio: fix the comp... |
37 |
#include <linux/compat.h> |
36bc08cc0 fs/aio: Add suppo... |
38 39 |
#include <linux/migrate.h> #include <linux/ramfs.h> |
723be6e39 aio: percpu ioctx... |
40 |
#include <linux/percpu-refcount.h> |
71ad7490c rework aio migrat... |
41 |
#include <linux/mount.h> |
1da177e4c Linux-2.6.12-rc2 |
42 43 44 |
#include <asm/kmap_types.h> #include <asm/uaccess.h> |
1da177e4c Linux-2.6.12-rc2 |
45 |
|
68d70d03f constify rw_verif... |
46 |
#include "internal.h" |
4e179bca6 aio: move private... |
47 48 49 50 51 52 |
#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 aio: v4 ensure ac... |
53 54 |
unsigned head; /* Written to by userland or under ring_lock * mutex by aio_read_events_ring(). */ |
4e179bca6 aio: move private... |
55 56 57 58 59 60 61 62 63 64 65 66 |
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 |
4e179bca6 aio: move private... |
67 |
|
db446a08c aio: convert the ... |
68 69 70 71 72 |
struct kioctx_table { struct rcu_head rcu; unsigned nr; struct kioctx *table[]; }; |
e1bdd5f27 aio: percpu reqs_... |
73 74 75 |
struct kioctx_cpu { unsigned reqs_available; }; |
dc48e56d7 aio: fix serial d... |
76 77 78 79 |
struct ctx_rq_wait { struct completion comp; atomic_t count; }; |
4e179bca6 aio: move private... |
80 |
struct kioctx { |
723be6e39 aio: percpu ioctx... |
81 |
struct percpu_ref users; |
36f558890 aio: refcounting ... |
82 |
atomic_t dead; |
4e179bca6 aio: move private... |
83 |
|
e34ecee2a aio: Fix a trinit... |
84 |
struct percpu_ref reqs; |
4e179bca6 aio: move private... |
85 |
unsigned long user_id; |
4e179bca6 aio: move private... |
86 |
|
e1bdd5f27 aio: percpu reqs_... |
87 88 89 90 91 92 93 |
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 aio: change reqs_... |
94 95 96 97 |
/* * This is what userspace passed to io_setup(), it's not used for * anything but counting against the global max_reqs quota. * |
58c85dc20 aio: kill struct ... |
98 |
* The real limit is nr_events - 1, which will be larger (see |
3e845ce01 aio: change reqs_... |
99 100 |
* aio_setup_ring()) */ |
4e179bca6 aio: move private... |
101 |
unsigned max_reqs; |
58c85dc20 aio: kill struct ... |
102 103 |
/* Size of ringbuffer, in units of struct io_event */ unsigned nr_events; |
4e179bca6 aio: move private... |
104 |
|
58c85dc20 aio: kill struct ... |
105 106 107 108 109 |
unsigned long mmap_base; unsigned long mmap_size; struct page **ring_pages; long nr_pages; |
723be6e39 aio: percpu ioctx... |
110 |
struct work_struct free_work; |
4e23bcaeb aio: give shared ... |
111 |
|
e02ba72aa aio: block io_des... |
112 113 114 |
/* * signals when all in-flight requests are done */ |
dc48e56d7 aio: fix serial d... |
115 |
struct ctx_rq_wait *rq_wait; |
e02ba72aa aio: block io_des... |
116 |
|
4e23bcaeb aio: give shared ... |
117 |
struct { |
34e83fc61 aio: reqs_active ... |
118 119 120 121 122 |
/* * 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 aio: percpu reqs_... |
123 124 |
* * We batch accesses to it with a percpu version. |
34e83fc61 aio: reqs_active ... |
125 126 |
*/ atomic_t reqs_available; |
4e23bcaeb aio: give shared ... |
127 128 129 130 131 132 |
} ____cacheline_aligned_in_smp; struct { spinlock_t ctx_lock; struct list_head active_reqs; /* used for cancellation */ } ____cacheline_aligned_in_smp; |
58c85dc20 aio: kill struct ... |
133 134 |
struct { struct mutex ring_lock; |
4e23bcaeb aio: give shared ... |
135 136 |
wait_queue_head_t wait; } ____cacheline_aligned_in_smp; |
58c85dc20 aio: kill struct ... |
137 138 139 |
struct { unsigned tail; |
d856f32a8 aio: fix reqs_ava... |
140 |
unsigned completed_events; |
58c85dc20 aio: kill struct ... |
141 |
spinlock_t completion_lock; |
4e23bcaeb aio: give shared ... |
142 |
} ____cacheline_aligned_in_smp; |
58c85dc20 aio: kill struct ... |
143 144 |
struct page *internal_pages[AIO_RING_PAGES]; |
36bc08cc0 fs/aio: Add suppo... |
145 |
struct file *aio_ring_file; |
db446a08c aio: convert the ... |
146 147 |
unsigned id; |
4e179bca6 aio: move private... |
148 |
}; |
04b2fa9f8 fs: split generic... |
149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 |
/* * We use ki_cancel == KIOCB_CANCELLED to indicate that a kiocb has been either * cancelled or completed (this makes a certain amount of sense because * successful cancellation - io_cancel() - does deliver the completion to * userspace). * * And since most things don't implement kiocb cancellation and we'd really like * kiocb completion to be lockless when possible, we use ki_cancel to * synchronize cancellation and completion - we only set it to KIOCB_CANCELLED * with xchg() or cmpxchg(), see batch_complete_aio() and kiocb_cancel(). */ #define KIOCB_CANCELLED ((void *) (~0ULL)) struct aio_kiocb { struct kiocb common; struct kioctx *ki_ctx; kiocb_cancel_fn *ki_cancel; struct iocb __user *ki_user_iocb; /* user's aiocb */ __u64 ki_user_data; /* user's data for completion */ struct list_head ki_list; /* the aio core uses this * for cancellation */ /* * 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 Linux-2.6.12-rc2 |
180 |
/*------ sysctl variables----*/ |
d55b5fdaf [PATCH] aio: remo... |
181 182 183 |
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 Linux-2.6.12-rc2 |
184 |
/*----end sysctl variables---*/ |
e18b890bb [PATCH] slab: rem... |
185 186 |
static struct kmem_cache *kiocb_cachep; static struct kmem_cache *kioctx_cachep; |
1da177e4c Linux-2.6.12-rc2 |
187 |
|
71ad7490c rework aio migrat... |
188 189 190 191 192 193 194 195 196 197 198 |
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) { struct qstr this = QSTR_INIT("[aio]", 5); struct file *file; struct path path; struct inode *inode = alloc_anon_inode(aio_mnt->mnt_sb); |
7f62656be aio: checking for... |
199 200 |
if (IS_ERR(inode)) return ERR_CAST(inode); |
71ad7490c rework aio migrat... |
201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 |
inode->i_mapping->a_ops = &aio_ctx_aops; inode->i_mapping->private_data = ctx; inode->i_size = PAGE_SIZE * nr_pages; path.dentry = d_alloc_pseudo(aio_mnt->mnt_sb, &this); if (!path.dentry) { iput(inode); return ERR_PTR(-ENOMEM); } path.mnt = mntget(aio_mnt); d_instantiate(path.dentry, inode); file = alloc_file(&path, FMODE_READ | FMODE_WRITE, &aio_ring_fops); if (IS_ERR(file)) { path_put(&path); return file; } file->f_flags = O_RDWR; |
71ad7490c rework aio migrat... |
221 222 223 224 225 226 227 228 229 |
return file; } static struct dentry *aio_mount(struct file_system_type *fs_type, int flags, const char *dev_name, void *data) { static const struct dentry_operations ops = { .d_dname = simple_dname, }; |
22f6b4d34 aio: mark AIO pse... |
230 231 232 233 234 235 |
struct dentry *root = mount_pseudo(fs_type, "aio:", NULL, &ops, AIO_RING_MAGIC); if (!IS_ERR(root)) root->d_sb->s_iflags |= SB_I_NOEXEC; return root; |
71ad7490c rework aio migrat... |
236 |
} |
1da177e4c Linux-2.6.12-rc2 |
237 238 239 240 241 242 |
/* 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 rework aio migrat... |
243 244 245 246 247 248 249 250 |
static struct file_system_type aio_fs = { .name = "aio", .mount = aio_mount, .kill_sb = kill_anon_super, }; aio_mnt = kern_mount(&aio_fs); if (IS_ERR(aio_mnt)) panic("Failed to create aio fs mount."); |
04b2fa9f8 fs: split generic... |
251 |
kiocb_cachep = KMEM_CACHE(aio_kiocb, SLAB_HWCACHE_ALIGN|SLAB_PANIC); |
0a31bd5f2 KMEM_CACHE(): sim... |
252 |
kioctx_cachep = KMEM_CACHE(kioctx,SLAB_HWCACHE_ALIGN|SLAB_PANIC); |
1da177e4c Linux-2.6.12-rc2 |
253 |
|
caf4167aa aio: dprintk() ->... |
254 255 |
pr_debug("sizeof(struct page) = %zu ", sizeof(struct page)); |
1da177e4c Linux-2.6.12-rc2 |
256 257 258 |
return 0; } |
385773e04 aio.c: move EXPOR... |
259 |
__initcall(aio_setup); |
1da177e4c Linux-2.6.12-rc2 |
260 |
|
5e9ae2e5d aio: fix use-afte... |
261 262 263 |
static void put_aio_ring_file(struct kioctx *ctx) { struct file *aio_ring_file = ctx->aio_ring_file; |
de04e7693 fs/aio.c: elimina... |
264 |
struct address_space *i_mapping; |
5e9ae2e5d aio: fix use-afte... |
265 266 267 268 |
if (aio_ring_file) { truncate_setsize(aio_ring_file->f_inode, 0); /* Prevent further access to the kioctx from migratepages */ |
de04e7693 fs/aio.c: elimina... |
269 270 271 |
i_mapping = aio_ring_file->f_inode->i_mapping; spin_lock(&i_mapping->private_lock); i_mapping->private_data = NULL; |
5e9ae2e5d aio: fix use-afte... |
272 |
ctx->aio_ring_file = NULL; |
de04e7693 fs/aio.c: elimina... |
273 |
spin_unlock(&i_mapping->private_lock); |
5e9ae2e5d aio: fix use-afte... |
274 275 276 277 |
fput(aio_ring_file); } } |
1da177e4c Linux-2.6.12-rc2 |
278 279 |
static void aio_free_ring(struct kioctx *ctx) { |
36bc08cc0 fs/aio: Add suppo... |
280 |
int i; |
1da177e4c Linux-2.6.12-rc2 |
281 |
|
fa8a53c39 aio: v4 ensure ac... |
282 283 284 285 |
/* Disconnect the kiotx from the ring file. This prevents future * accesses to the kioctx from page migration. */ put_aio_ring_file(ctx); |
36bc08cc0 fs/aio: Add suppo... |
286 |
for (i = 0; i < ctx->nr_pages; i++) { |
8e321fefb aio/migratepages:... |
287 |
struct page *page; |
36bc08cc0 fs/aio: Add suppo... |
288 289 290 |
pr_debug("pid(%d) [%d] page->count=%d ", current->pid, i, page_count(ctx->ring_pages[i])); |
8e321fefb aio/migratepages:... |
291 292 293 294 295 |
page = ctx->ring_pages[i]; if (!page) continue; ctx->ring_pages[i] = NULL; put_page(page); |
36bc08cc0 fs/aio: Add suppo... |
296 |
} |
1da177e4c Linux-2.6.12-rc2 |
297 |
|
ddb8c45ba aio: nullify aio-... |
298 |
if (ctx->ring_pages && ctx->ring_pages != ctx->internal_pages) { |
58c85dc20 aio: kill struct ... |
299 |
kfree(ctx->ring_pages); |
ddb8c45ba aio: nullify aio-... |
300 301 |
ctx->ring_pages = NULL; } |
36bc08cc0 fs/aio: Add suppo... |
302 |
} |
5477e70a6 mm: move ->mremap... |
303 |
static int aio_ring_mremap(struct vm_area_struct *vma) |
e4a0d3e72 aio: Make it poss... |
304 |
{ |
5477e70a6 mm: move ->mremap... |
305 |
struct file *file = vma->vm_file; |
e4a0d3e72 aio: Make it poss... |
306 307 |
struct mm_struct *mm = vma->vm_mm; struct kioctx_table *table; |
b2edffdd9 fix mremap() vs. ... |
308 |
int i, res = -EINVAL; |
e4a0d3e72 aio: Make it poss... |
309 310 311 312 313 314 315 316 317 |
spin_lock(&mm->ioctx_lock); rcu_read_lock(); table = rcu_dereference(mm->ioctx_table); for (i = 0; i < table->nr; i++) { struct kioctx *ctx; ctx = table->table[i]; if (ctx && ctx->aio_ring_file == file) { |
b2edffdd9 fix mremap() vs. ... |
318 319 320 321 |
if (!atomic_read(&ctx->dead)) { ctx->user_id = ctx->mmap_base = vma->vm_start; res = 0; } |
e4a0d3e72 aio: Make it poss... |
322 323 324 325 326 327 |
break; } } rcu_read_unlock(); spin_unlock(&mm->ioctx_lock); |
b2edffdd9 fix mremap() vs. ... |
328 |
return res; |
e4a0d3e72 aio: Make it poss... |
329 |
} |
5477e70a6 mm: move ->mremap... |
330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 |
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 fs/aio: Add suppo... |
345 346 347 |
static const struct file_operations aio_ring_fops = { .mmap = aio_ring_mmap, }; |
0c45355fc aio: fix build wh... |
348 |
#if IS_ENABLED(CONFIG_MIGRATION) |
36bc08cc0 fs/aio: Add suppo... |
349 350 351 |
static int aio_migratepage(struct address_space *mapping, struct page *new, struct page *old, enum migrate_mode mode) { |
5e9ae2e5d aio: fix use-afte... |
352 |
struct kioctx *ctx; |
36bc08cc0 fs/aio: Add suppo... |
353 |
unsigned long flags; |
fa8a53c39 aio: v4 ensure ac... |
354 |
pgoff_t idx; |
36bc08cc0 fs/aio: Add suppo... |
355 |
int rc; |
8e321fefb aio/migratepages:... |
356 |
rc = 0; |
fa8a53c39 aio: v4 ensure ac... |
357 |
/* mapping->private_lock here protects against the kioctx teardown. */ |
8e321fefb aio/migratepages:... |
358 359 |
spin_lock(&mapping->private_lock); ctx = mapping->private_data; |
fa8a53c39 aio: v4 ensure ac... |
360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 |
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 aio/migratepages:... |
379 380 |
} else rc = -EINVAL; |
8e321fefb aio/migratepages:... |
381 382 |
if (rc != 0) |
fa8a53c39 aio: v4 ensure ac... |
383 |
goto out_unlock; |
8e321fefb aio/migratepages:... |
384 |
|
36bc08cc0 fs/aio: Add suppo... |
385 386 |
/* Writeback must be complete */ BUG_ON(PageWriteback(old)); |
8e321fefb aio/migratepages:... |
387 |
get_page(new); |
36bc08cc0 fs/aio: Add suppo... |
388 |
|
8e321fefb aio/migratepages:... |
389 |
rc = migrate_page_move_mapping(mapping, new, old, NULL, mode, 1); |
36bc08cc0 fs/aio: Add suppo... |
390 |
if (rc != MIGRATEPAGE_SUCCESS) { |
8e321fefb aio/migratepages:... |
391 |
put_page(new); |
fa8a53c39 aio: v4 ensure ac... |
392 |
goto out_unlock; |
36bc08cc0 fs/aio: Add suppo... |
393 |
} |
fa8a53c39 aio: v4 ensure ac... |
394 395 396 |
/* 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 aio: fix use-afte... |
397 |
*/ |
fa8a53c39 aio: v4 ensure ac... |
398 399 400 401 402 |
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 fs/aio: Add suppo... |
403 |
|
fa8a53c39 aio: v4 ensure ac... |
404 405 |
/* The old page is no longer accessible. */ put_page(old); |
8e321fefb aio/migratepages:... |
406 |
|
fa8a53c39 aio: v4 ensure ac... |
407 408 409 410 |
out_unlock: mutex_unlock(&ctx->ring_lock); out: spin_unlock(&mapping->private_lock); |
36bc08cc0 fs/aio: Add suppo... |
411 |
return rc; |
1da177e4c Linux-2.6.12-rc2 |
412 |
} |
0c45355fc aio: fix build wh... |
413 |
#endif |
1da177e4c Linux-2.6.12-rc2 |
414 |
|
36bc08cc0 fs/aio: Add suppo... |
415 |
static const struct address_space_operations aio_ctx_aops = { |
835f252c6 aio: fix uncorren... |
416 |
.set_page_dirty = __set_page_dirty_no_writeback, |
0c45355fc aio: fix build wh... |
417 |
#if IS_ENABLED(CONFIG_MIGRATION) |
36bc08cc0 fs/aio: Add suppo... |
418 |
.migratepage = aio_migratepage, |
0c45355fc aio: fix build wh... |
419 |
#endif |
36bc08cc0 fs/aio: Add suppo... |
420 |
}; |
1da177e4c Linux-2.6.12-rc2 |
421 422 423 |
static int aio_setup_ring(struct kioctx *ctx) { struct aio_ring *ring; |
1da177e4c Linux-2.6.12-rc2 |
424 |
unsigned nr_events = ctx->max_reqs; |
41003a7bc aio: remove retry... |
425 |
struct mm_struct *mm = current->mm; |
3dc9acb67 aio: clean up and... |
426 |
unsigned long size, unused; |
1da177e4c Linux-2.6.12-rc2 |
427 |
int nr_pages; |
36bc08cc0 fs/aio: Add suppo... |
428 429 |
int i; struct file *file; |
1da177e4c Linux-2.6.12-rc2 |
430 431 432 433 434 435 |
/* 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 Linux-2.6.12-rc2 |
436 |
|
36bc08cc0 fs/aio: Add suppo... |
437 |
nr_pages = PFN_UP(size); |
1da177e4c Linux-2.6.12-rc2 |
438 439 |
if (nr_pages < 0) return -EINVAL; |
71ad7490c rework aio migrat... |
440 |
file = aio_private_file(ctx, nr_pages); |
36bc08cc0 fs/aio: Add suppo... |
441 442 |
if (IS_ERR(file)) { ctx->aio_ring_file = NULL; |
fa8a53c39 aio: v4 ensure ac... |
443 |
return -ENOMEM; |
36bc08cc0 fs/aio: Add suppo... |
444 |
} |
3dc9acb67 aio: clean up and... |
445 446 447 448 449 450 451 452 453 454 455 456 457 |
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 fs/aio: Add suppo... |
458 459 460 461 462 463 464 465 466 467 |
for (i = 0; i < nr_pages; i++) { struct page *page; page = find_or_create_page(file->f_inode->i_mapping, i, GFP_HIGHUSER | __GFP_ZERO); if (!page) break; pr_debug("pid(%d) page[%d]->count=%d ", current->pid, i, page_count(page)); SetPageUptodate(page); |
36bc08cc0 fs/aio: Add suppo... |
468 |
unlock_page(page); |
3dc9acb67 aio: clean up and... |
469 470 |
ctx->ring_pages[i] = page; |
36bc08cc0 fs/aio: Add suppo... |
471 |
} |
3dc9acb67 aio: clean up and... |
472 |
ctx->nr_pages = i; |
1da177e4c Linux-2.6.12-rc2 |
473 |
|
3dc9acb67 aio: clean up and... |
474 475 |
if (unlikely(i != nr_pages)) { aio_free_ring(ctx); |
fa8a53c39 aio: v4 ensure ac... |
476 |
return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
477 |
} |
58c85dc20 aio: kill struct ... |
478 479 480 |
ctx->mmap_size = nr_pages * PAGE_SIZE; pr_debug("attempting mmap of %lu bytes ", ctx->mmap_size); |
36bc08cc0 fs/aio: Add suppo... |
481 |
|
013373e8b aio: make aio_set... |
482 483 484 485 486 |
if (down_write_killable(&mm->mmap_sem)) { ctx->mmap_size = 0; aio_free_ring(ctx); return -EINTR; } |
36bc08cc0 fs/aio: Add suppo... |
487 488 |
ctx->mmap_base = do_mmap_pgoff(ctx->aio_ring_file, 0, ctx->mmap_size, PROT_READ | PROT_WRITE, |
3dc9acb67 aio: clean up and... |
489 490 |
MAP_SHARED, 0, &unused); up_write(&mm->mmap_sem); |
58c85dc20 aio: kill struct ... |
491 |
if (IS_ERR((void *)ctx->mmap_base)) { |
58c85dc20 aio: kill struct ... |
492 |
ctx->mmap_size = 0; |
1da177e4c Linux-2.6.12-rc2 |
493 |
aio_free_ring(ctx); |
fa8a53c39 aio: v4 ensure ac... |
494 |
return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
495 |
} |
58c85dc20 aio: kill struct ... |
496 497 |
pr_debug("mmap address: 0x%08lx ", ctx->mmap_base); |
d6c355c7d aio: fix race in ... |
498 |
|
58c85dc20 aio: kill struct ... |
499 500 |
ctx->user_id = ctx->mmap_base; ctx->nr_events = nr_events; /* trusted copy */ |
1da177e4c Linux-2.6.12-rc2 |
501 |
|
58c85dc20 aio: kill struct ... |
502 |
ring = kmap_atomic(ctx->ring_pages[0]); |
1da177e4c Linux-2.6.12-rc2 |
503 |
ring->nr = nr_events; /* user copy */ |
db446a08c aio: convert the ... |
504 |
ring->id = ~0U; |
1da177e4c Linux-2.6.12-rc2 |
505 506 507 508 509 |
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 fs: remove the se... |
510 |
kunmap_atomic(ring); |
58c85dc20 aio: kill struct ... |
511 |
flush_dcache_page(ctx->ring_pages[0]); |
1da177e4c Linux-2.6.12-rc2 |
512 513 514 |
return 0; } |
1da177e4c Linux-2.6.12-rc2 |
515 516 517 |
#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 fs: split generic... |
518 |
void kiocb_set_cancel_fn(struct kiocb *iocb, kiocb_cancel_fn *cancel) |
0460fef2a aio: use cancella... |
519 |
{ |
04b2fa9f8 fs: split generic... |
520 |
struct aio_kiocb *req = container_of(iocb, struct aio_kiocb, common); |
0460fef2a aio: use cancella... |
521 522 523 524 525 526 527 528 529 530 531 532 533 |
struct kioctx *ctx = req->ki_ctx; unsigned long flags; spin_lock_irqsave(&ctx->ctx_lock, flags); if (!req->ki_list.next) list_add(&req->ki_list, &ctx->active_reqs); req->ki_cancel = cancel; spin_unlock_irqrestore(&ctx->ctx_lock, flags); } EXPORT_SYMBOL(kiocb_set_cancel_fn); |
04b2fa9f8 fs: split generic... |
534 |
static int kiocb_cancel(struct aio_kiocb *kiocb) |
906b973cf aio: add kiocb_ca... |
535 |
{ |
0460fef2a aio: use cancella... |
536 |
kiocb_cancel_fn *old, *cancel; |
906b973cf aio: add kiocb_ca... |
537 |
|
0460fef2a aio: use cancella... |
538 539 540 541 542 543 544 545 |
/* * Don't want to set kiocb->ki_cancel = KIOCB_CANCELLED unless it * actually has a cancel function, hence the cmpxchg() */ cancel = ACCESS_ONCE(kiocb->ki_cancel); do { if (!cancel || cancel == KIOCB_CANCELLED) |
57282d8fd aio: Kill ki_users |
546 |
return -EINVAL; |
906b973cf aio: add kiocb_ca... |
547 |
|
0460fef2a aio: use cancella... |
548 549 550 |
old = cancel; cancel = cmpxchg(&kiocb->ki_cancel, old, KIOCB_CANCELLED); } while (cancel != old); |
906b973cf aio: add kiocb_ca... |
551 |
|
04b2fa9f8 fs: split generic... |
552 |
return cancel(&kiocb->common); |
906b973cf aio: add kiocb_ca... |
553 |
} |
e34ecee2a aio: Fix a trinit... |
554 |
static void free_ioctx(struct work_struct *work) |
36f558890 aio: refcounting ... |
555 |
{ |
e34ecee2a aio: Fix a trinit... |
556 |
struct kioctx *ctx = container_of(work, struct kioctx, free_work); |
e1bdd5f27 aio: percpu reqs_... |
557 |
|
e34ecee2a aio: Fix a trinit... |
558 559 |
pr_debug("freeing %p ", ctx); |
e1bdd5f27 aio: percpu reqs_... |
560 |
|
e34ecee2a aio: Fix a trinit... |
561 |
aio_free_ring(ctx); |
e1bdd5f27 aio: percpu reqs_... |
562 |
free_percpu(ctx->cpu); |
9a1049da9 percpu-refcount: ... |
563 564 |
percpu_ref_exit(&ctx->reqs); percpu_ref_exit(&ctx->users); |
36f558890 aio: refcounting ... |
565 566 |
kmem_cache_free(kioctx_cachep, ctx); } |
e34ecee2a aio: Fix a trinit... |
567 568 569 |
static void free_ioctx_reqs(struct percpu_ref *ref) { struct kioctx *ctx = container_of(ref, struct kioctx, reqs); |
e02ba72aa aio: block io_des... |
570 |
/* At this point we know that there are no any in-flight requests */ |
dc48e56d7 aio: fix serial d... |
571 572 |
if (ctx->rq_wait && atomic_dec_and_test(&ctx->rq_wait->count)) complete(&ctx->rq_wait->comp); |
e02ba72aa aio: block io_des... |
573 |
|
e34ecee2a aio: Fix a trinit... |
574 575 576 |
INIT_WORK(&ctx->free_work, free_ioctx); schedule_work(&ctx->free_work); } |
36f558890 aio: refcounting ... |
577 578 579 580 581 |
/* * 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 aio: Fix a trinit... |
582 |
static void free_ioctx_users(struct percpu_ref *ref) |
36f558890 aio: refcounting ... |
583 |
{ |
e34ecee2a aio: Fix a trinit... |
584 |
struct kioctx *ctx = container_of(ref, struct kioctx, users); |
04b2fa9f8 fs: split generic... |
585 |
struct aio_kiocb *req; |
36f558890 aio: refcounting ... |
586 587 588 589 590 |
spin_lock_irq(&ctx->ctx_lock); while (!list_empty(&ctx->active_reqs)) { req = list_first_entry(&ctx->active_reqs, |
04b2fa9f8 fs: split generic... |
591 |
struct aio_kiocb, ki_list); |
36f558890 aio: refcounting ... |
592 593 |
list_del_init(&req->ki_list); |
d52a8f9ea fs/aio.c: Remove ... |
594 |
kiocb_cancel(req); |
36f558890 aio: refcounting ... |
595 596 597 |
} spin_unlock_irq(&ctx->ctx_lock); |
e34ecee2a aio: Fix a trinit... |
598 599 |
percpu_ref_kill(&ctx->reqs); percpu_ref_put(&ctx->reqs); |
36f558890 aio: refcounting ... |
600 |
} |
db446a08c aio: convert the ... |
601 602 603 604 605 606 607 |
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 aio: kill the mis... |
608 |
table = rcu_dereference_raw(mm->ioctx_table); |
db446a08c aio: convert the ... |
609 610 611 612 613 614 615 616 |
while (1) { if (table) for (i = 0; i < table->nr; i++) if (!table->table[i]) { ctx->id = i; table->table[i] = ctx; spin_unlock(&mm->ioctx_lock); |
fa8a53c39 aio: v4 ensure ac... |
617 618 619 620 |
/* While kioctx setup is in progress, * we are protected from page migration * changes ring_pages by ->ring_lock. */ |
db446a08c aio: convert the ... |
621 622 623 624 625 626 627 |
ring = kmap_atomic(ctx->ring_pages[0]); ring->id = ctx->id; kunmap_atomic(ring); return 0; } new_nr = (table ? table->nr : 1) * 4; |
db446a08c aio: convert the ... |
628 629 630 631 632 633 634 635 636 637 |
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 aio: kill the mis... |
638 |
old = rcu_dereference_raw(mm->ioctx_table); |
db446a08c aio: convert the ... |
639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 |
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 aio: Fix a trinit... |
654 655 656 657 658 659 660 661 662 |
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 Linux-2.6.12-rc2 |
663 664 665 666 667 |
/* ioctx_alloc * Allocates and initializes an ioctx. Returns an ERR_PTR if it failed. */ static struct kioctx *ioctx_alloc(unsigned nr_events) { |
41003a7bc aio: remove retry... |
668 |
struct mm_struct *mm = current->mm; |
1da177e4c Linux-2.6.12-rc2 |
669 |
struct kioctx *ctx; |
e23754f88 aio: don't bother... |
670 |
int err = -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
671 |
|
e1bdd5f27 aio: percpu reqs_... |
672 673 674 675 676 677 678 679 680 681 682 |
/* * 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 Linux-2.6.12-rc2 |
683 |
/* Prevent overflows */ |
08397acdd ioctx_alloc(): re... |
684 |
if (nr_events > (0x10000000U / sizeof(struct io_event))) { |
1da177e4c Linux-2.6.12-rc2 |
685 686 687 688 |
pr_debug("ENOMEM: nr_events too high "); return ERR_PTR(-EINVAL); } |
4cd81c3df aio: double aio_m... |
689 |
if (!nr_events || (unsigned long)nr_events > (aio_max_nr * 2UL)) |
1da177e4c Linux-2.6.12-rc2 |
690 |
return ERR_PTR(-EAGAIN); |
c37622296 [PATCH] Transform... |
691 |
ctx = kmem_cache_zalloc(kioctx_cachep, GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
692 693 |
if (!ctx) return ERR_PTR(-ENOMEM); |
1da177e4c Linux-2.6.12-rc2 |
694 |
ctx->max_reqs = nr_events; |
1da177e4c Linux-2.6.12-rc2 |
695 |
|
1da177e4c Linux-2.6.12-rc2 |
696 |
spin_lock_init(&ctx->ctx_lock); |
0460fef2a aio: use cancella... |
697 |
spin_lock_init(&ctx->completion_lock); |
58c85dc20 aio: kill struct ... |
698 |
mutex_init(&ctx->ring_lock); |
fa8a53c39 aio: v4 ensure ac... |
699 700 701 |
/* Protect against page migration throughout kiotx setup by keeping * the ring_lock mutex held until setup is complete. */ mutex_lock(&ctx->ring_lock); |
1da177e4c Linux-2.6.12-rc2 |
702 703 704 |
init_waitqueue_head(&ctx->wait); INIT_LIST_HEAD(&ctx->active_reqs); |
1da177e4c Linux-2.6.12-rc2 |
705 |
|
2aad2a86f percpu_ref: add P... |
706 |
if (percpu_ref_init(&ctx->users, free_ioctx_users, 0, GFP_KERNEL)) |
fa8a53c39 aio: v4 ensure ac... |
707 |
goto err; |
2aad2a86f percpu_ref: add P... |
708 |
if (percpu_ref_init(&ctx->reqs, free_ioctx_reqs, 0, GFP_KERNEL)) |
fa8a53c39 aio: v4 ensure ac... |
709 |
goto err; |
e1bdd5f27 aio: percpu reqs_... |
710 711 |
ctx->cpu = alloc_percpu(struct kioctx_cpu); if (!ctx->cpu) |
e34ecee2a aio: Fix a trinit... |
712 |
goto err; |
1da177e4c Linux-2.6.12-rc2 |
713 |
|
fa8a53c39 aio: v4 ensure ac... |
714 715 |
err = aio_setup_ring(ctx); if (err < 0) |
e34ecee2a aio: Fix a trinit... |
716 |
goto err; |
e1bdd5f27 aio: percpu reqs_... |
717 |
|
34e83fc61 aio: reqs_active ... |
718 |
atomic_set(&ctx->reqs_available, ctx->nr_events - 1); |
e1bdd5f27 aio: percpu reqs_... |
719 |
ctx->req_batch = (ctx->nr_events - 1) / (num_possible_cpus() * 4); |
6878ea72a aio: be defensive... |
720 721 |
if (ctx->req_batch < 1) ctx->req_batch = 1; |
34e83fc61 aio: reqs_active ... |
722 |
|
1da177e4c Linux-2.6.12-rc2 |
723 |
/* limit the number of system wide aios */ |
9fa1cb397 aio: aio_nr_lock ... |
724 |
spin_lock(&aio_nr_lock); |
4cd81c3df aio: double aio_m... |
725 |
if (aio_nr + nr_events > (aio_max_nr * 2UL) || |
2dd542b7a aio: aio_nr decre... |
726 |
aio_nr + nr_events < aio_nr) { |
9fa1cb397 aio: aio_nr_lock ... |
727 |
spin_unlock(&aio_nr_lock); |
e34ecee2a aio: Fix a trinit... |
728 |
err = -EAGAIN; |
d1b943271 aio: clean up aio... |
729 |
goto err_ctx; |
2dd542b7a aio: aio_nr decre... |
730 731 |
} aio_nr += ctx->max_reqs; |
9fa1cb397 aio: aio_nr_lock ... |
732 |
spin_unlock(&aio_nr_lock); |
1da177e4c Linux-2.6.12-rc2 |
733 |
|
1881686f8 aio: fix kioctx l... |
734 735 |
percpu_ref_get(&ctx->users); /* io_setup() will drop this ref */ percpu_ref_get(&ctx->reqs); /* free_ioctx_users() will drop this */ |
723be6e39 aio: percpu ioctx... |
736 |
|
da90382c2 aio: fix error ha... |
737 738 |
err = ioctx_add_table(ctx, mm); if (err) |
e34ecee2a aio: Fix a trinit... |
739 |
goto err_cleanup; |
da90382c2 aio: fix error ha... |
740 |
|
fa8a53c39 aio: v4 ensure ac... |
741 742 |
/* Release the ring_lock mutex now that all setup is complete. */ mutex_unlock(&ctx->ring_lock); |
caf4167aa aio: dprintk() ->... |
743 744 |
pr_debug("allocated ioctx %p[%ld]: mm=%p mask=0x%x ", |
58c85dc20 aio: kill struct ... |
745 |
ctx, ctx->user_id, mm, ctx->nr_events); |
1da177e4c Linux-2.6.12-rc2 |
746 |
return ctx; |
e34ecee2a aio: Fix a trinit... |
747 748 |
err_cleanup: aio_nr_sub(ctx->max_reqs); |
d1b943271 aio: clean up aio... |
749 |
err_ctx: |
deeb8525f ioctx_alloc(): fi... |
750 751 752 |
atomic_set(&ctx->dead, 1); if (ctx->mmap_size) vm_munmap(ctx->mmap_base, ctx->mmap_size); |
d1b943271 aio: clean up aio... |
753 |
aio_free_ring(ctx); |
e34ecee2a aio: Fix a trinit... |
754 |
err: |
fa8a53c39 aio: v4 ensure ac... |
755 |
mutex_unlock(&ctx->ring_lock); |
e1bdd5f27 aio: percpu reqs_... |
756 |
free_percpu(ctx->cpu); |
9a1049da9 percpu-refcount: ... |
757 758 |
percpu_ref_exit(&ctx->reqs); percpu_ref_exit(&ctx->users); |
1da177e4c Linux-2.6.12-rc2 |
759 |
kmem_cache_free(kioctx_cachep, ctx); |
caf4167aa aio: dprintk() ->... |
760 761 |
pr_debug("error allocating ioctx %d ", err); |
e23754f88 aio: don't bother... |
762 |
return ERR_PTR(err); |
1da177e4c Linux-2.6.12-rc2 |
763 |
} |
36f558890 aio: refcounting ... |
764 765 766 767 768 |
/* 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 aio: report error... |
769 |
static int kill_ioctx(struct mm_struct *mm, struct kioctx *ctx, |
dc48e56d7 aio: fix serial d... |
770 |
struct ctx_rq_wait *wait) |
36f558890 aio: refcounting ... |
771 |
{ |
fa88b6f88 aio: cleanup: fla... |
772 |
struct kioctx_table *table; |
db446a08c aio: convert the ... |
773 |
|
b2edffdd9 fix mremap() vs. ... |
774 775 776 |
spin_lock(&mm->ioctx_lock); if (atomic_xchg(&ctx->dead, 1)) { spin_unlock(&mm->ioctx_lock); |
fa88b6f88 aio: cleanup: fla... |
777 |
return -EINVAL; |
b2edffdd9 fix mremap() vs. ... |
778 |
} |
db446a08c aio: convert the ... |
779 |
|
855ef0dec aio: kill the mis... |
780 |
table = rcu_dereference_raw(mm->ioctx_table); |
fa88b6f88 aio: cleanup: fla... |
781 782 |
WARN_ON(ctx != table->table[ctx->id]); table->table[ctx->id] = NULL; |
fa88b6f88 aio: cleanup: fla... |
783 |
spin_unlock(&mm->ioctx_lock); |
4fcc712f5 aio: fix io_destr... |
784 |
|
fa88b6f88 aio: cleanup: fla... |
785 786 |
/* percpu_ref_kill() will do the necessary call_rcu() */ wake_up_all(&ctx->wait); |
4fcc712f5 aio: fix io_destr... |
787 |
|
fa88b6f88 aio: cleanup: fla... |
788 789 790 791 792 793 794 795 |
/* * 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 aio: fix io_destr... |
796 |
|
fa88b6f88 aio: cleanup: fla... |
797 798 |
if (ctx->mmap_size) vm_munmap(ctx->mmap_base, ctx->mmap_size); |
fb2d44838 aio: report error... |
799 |
|
dc48e56d7 aio: fix serial d... |
800 |
ctx->rq_wait = wait; |
fa88b6f88 aio: cleanup: fla... |
801 802 |
percpu_ref_kill(&ctx->users); return 0; |
1da177e4c Linux-2.6.12-rc2 |
803 |
} |
36f558890 aio: refcounting ... |
804 805 806 807 808 809 810 |
/* * 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 Linux-2.6.12-rc2 |
811 |
*/ |
fc9b52cd8 fs: remove fastca... |
812 |
void exit_aio(struct mm_struct *mm) |
1da177e4c Linux-2.6.12-rc2 |
813 |
{ |
4b70ac5fd aio: change exit_... |
814 |
struct kioctx_table *table = rcu_dereference_raw(mm->ioctx_table); |
dc48e56d7 aio: fix serial d... |
815 816 |
struct ctx_rq_wait wait; int i, skipped; |
db446a08c aio: convert the ... |
817 |
|
4b70ac5fd aio: change exit_... |
818 819 |
if (!table) return; |
db446a08c aio: convert the ... |
820 |
|
dc48e56d7 aio: fix serial d... |
821 822 823 824 |
atomic_set(&wait.count, table->nr); init_completion(&wait.comp); skipped = 0; |
4b70ac5fd aio: change exit_... |
825 826 |
for (i = 0; i < table->nr; ++i) { struct kioctx *ctx = table->table[i]; |
abf137dd7 aio: make the loo... |
827 |
|
dc48e56d7 aio: fix serial d... |
828 829 |
if (!ctx) { skipped++; |
4b70ac5fd aio: change exit_... |
830 |
continue; |
dc48e56d7 aio: fix serial d... |
831 |
} |
936af1576 aio: don't bother... |
832 |
/* |
4b70ac5fd aio: change exit_... |
833 834 835 836 837 |
* 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 aio: don't bother... |
838 |
*/ |
58c85dc20 aio: kill struct ... |
839 |
ctx->mmap_size = 0; |
dc48e56d7 aio: fix serial d... |
840 841 |
kill_ioctx(mm, ctx, &wait); } |
36f558890 aio: refcounting ... |
842 |
|
dc48e56d7 aio: fix serial d... |
843 |
if (!atomic_sub_and_test(skipped, &wait.count)) { |
6098b45b3 aio: block exit_a... |
844 |
/* Wait until all IO for the context are done. */ |
dc48e56d7 aio: fix serial d... |
845 |
wait_for_completion(&wait.comp); |
1da177e4c Linux-2.6.12-rc2 |
846 |
} |
4b70ac5fd aio: change exit_... |
847 848 849 |
RCU_INIT_POINTER(mm->ioctx_table, NULL); kfree(table); |
1da177e4c Linux-2.6.12-rc2 |
850 |
} |
e1bdd5f27 aio: percpu reqs_... |
851 852 853 |
static void put_reqs_available(struct kioctx *ctx, unsigned nr) { struct kioctx_cpu *kcpu; |
263782c1c aio: protect reqs... |
854 |
unsigned long flags; |
e1bdd5f27 aio: percpu reqs_... |
855 |
|
263782c1c aio: protect reqs... |
856 |
local_irq_save(flags); |
be6fb451a aio: remove no lo... |
857 |
kcpu = this_cpu_ptr(ctx->cpu); |
e1bdd5f27 aio: percpu reqs_... |
858 |
kcpu->reqs_available += nr; |
263782c1c aio: protect reqs... |
859 |
|
e1bdd5f27 aio: percpu reqs_... |
860 861 862 863 |
while (kcpu->reqs_available >= ctx->req_batch * 2) { kcpu->reqs_available -= ctx->req_batch; atomic_add(ctx->req_batch, &ctx->reqs_available); } |
263782c1c aio: protect reqs... |
864 |
local_irq_restore(flags); |
e1bdd5f27 aio: percpu reqs_... |
865 866 867 868 869 870 |
} static bool get_reqs_available(struct kioctx *ctx) { struct kioctx_cpu *kcpu; bool ret = false; |
263782c1c aio: protect reqs... |
871 |
unsigned long flags; |
e1bdd5f27 aio: percpu reqs_... |
872 |
|
263782c1c aio: protect reqs... |
873 |
local_irq_save(flags); |
be6fb451a aio: remove no lo... |
874 |
kcpu = this_cpu_ptr(ctx->cpu); |
e1bdd5f27 aio: percpu reqs_... |
875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 |
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 aio: protect reqs... |
893 |
local_irq_restore(flags); |
e1bdd5f27 aio: percpu reqs_... |
894 895 |
return ret; } |
d856f32a8 aio: fix reqs_ava... |
896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 |
/* 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); } |
1da177e4c Linux-2.6.12-rc2 |
957 |
/* aio_get_req |
57282d8fd aio: Kill ki_users |
958 959 |
* Allocate a slot for an aio request. * Returns NULL if no requests are free. |
1da177e4c Linux-2.6.12-rc2 |
960 |
*/ |
04b2fa9f8 fs: split generic... |
961 |
static inline struct aio_kiocb *aio_get_req(struct kioctx *ctx) |
1da177e4c Linux-2.6.12-rc2 |
962 |
{ |
04b2fa9f8 fs: split generic... |
963 |
struct aio_kiocb *req; |
a1c8eae75 aio: kill batch a... |
964 |
|
d856f32a8 aio: fix reqs_ava... |
965 966 967 968 969 |
if (!get_reqs_available(ctx)) { user_refill_reqs_available(ctx); if (!get_reqs_available(ctx)) return NULL; } |
a1c8eae75 aio: kill batch a... |
970 |
|
0460fef2a aio: use cancella... |
971 |
req = kmem_cache_alloc(kiocb_cachep, GFP_KERNEL|__GFP_ZERO); |
1da177e4c Linux-2.6.12-rc2 |
972 |
if (unlikely(!req)) |
a1c8eae75 aio: kill batch a... |
973 |
goto out_put; |
1da177e4c Linux-2.6.12-rc2 |
974 |
|
e34ecee2a aio: Fix a trinit... |
975 |
percpu_ref_get(&ctx->reqs); |
1da177e4c Linux-2.6.12-rc2 |
976 |
req->ki_ctx = ctx; |
080d676de aio: allocate kio... |
977 |
return req; |
a1c8eae75 aio: kill batch a... |
978 |
out_put: |
e1bdd5f27 aio: percpu reqs_... |
979 |
put_reqs_available(ctx, 1); |
a1c8eae75 aio: kill batch a... |
980 |
return NULL; |
1da177e4c Linux-2.6.12-rc2 |
981 |
} |
04b2fa9f8 fs: split generic... |
982 |
static void kiocb_free(struct aio_kiocb *req) |
1da177e4c Linux-2.6.12-rc2 |
983 |
{ |
04b2fa9f8 fs: split generic... |
984 985 |
if (req->common.ki_filp) fput(req->common.ki_filp); |
133890103 eventfd: revised ... |
986 987 |
if (req->ki_eventfd != NULL) eventfd_ctx_put(req->ki_eventfd); |
1da177e4c Linux-2.6.12-rc2 |
988 |
kmem_cache_free(kiocb_cachep, req); |
1da177e4c Linux-2.6.12-rc2 |
989 |
} |
d5470b596 fs/aio.c: make 3 ... |
990 |
static struct kioctx *lookup_ioctx(unsigned long ctx_id) |
1da177e4c Linux-2.6.12-rc2 |
991 |
{ |
db446a08c aio: convert the ... |
992 |
struct aio_ring __user *ring = (void __user *)ctx_id; |
abf137dd7 aio: make the loo... |
993 |
struct mm_struct *mm = current->mm; |
65c24491b aio: lookup_ioctx... |
994 |
struct kioctx *ctx, *ret = NULL; |
db446a08c aio: convert the ... |
995 996 997 998 999 |
struct kioctx_table *table; unsigned id; if (get_user(id, &ring->id)) return NULL; |
1da177e4c Linux-2.6.12-rc2 |
1000 |
|
abf137dd7 aio: make the loo... |
1001 |
rcu_read_lock(); |
db446a08c aio: convert the ... |
1002 |
table = rcu_dereference(mm->ioctx_table); |
abf137dd7 aio: make the loo... |
1003 |
|
db446a08c aio: convert the ... |
1004 1005 |
if (!table || id >= table->nr) goto out; |
1da177e4c Linux-2.6.12-rc2 |
1006 |
|
db446a08c aio: convert the ... |
1007 |
ctx = table->table[id]; |
f30d704fe aio: table lookup... |
1008 |
if (ctx && ctx->user_id == ctx_id) { |
db446a08c aio: convert the ... |
1009 1010 1011 1012 |
percpu_ref_get(&ctx->users); ret = ctx; } out: |
abf137dd7 aio: make the loo... |
1013 |
rcu_read_unlock(); |
65c24491b aio: lookup_ioctx... |
1014 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1015 |
} |
1da177e4c Linux-2.6.12-rc2 |
1016 1017 |
/* aio_complete * Called when the io request on the given iocb is complete. |
1da177e4c Linux-2.6.12-rc2 |
1018 |
*/ |
04b2fa9f8 fs: split generic... |
1019 |
static void aio_complete(struct kiocb *kiocb, long res, long res2) |
1da177e4c Linux-2.6.12-rc2 |
1020 |
{ |
04b2fa9f8 fs: split generic... |
1021 |
struct aio_kiocb *iocb = container_of(kiocb, struct aio_kiocb, common); |
1da177e4c Linux-2.6.12-rc2 |
1022 |
struct kioctx *ctx = iocb->ki_ctx; |
1da177e4c Linux-2.6.12-rc2 |
1023 |
struct aio_ring *ring; |
21b40200c aio: use flush_dc... |
1024 |
struct io_event *ev_page, *event; |
d856f32a8 aio: fix reqs_ava... |
1025 |
unsigned tail, pos, head; |
1da177e4c Linux-2.6.12-rc2 |
1026 |
unsigned long flags; |
1da177e4c Linux-2.6.12-rc2 |
1027 |
|
70fe2f481 aio: fix freeze p... |
1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 |
if (kiocb->ki_flags & IOCB_WRITE) { struct file *file = kiocb->ki_filp; /* * Tell lockdep we inherited freeze protection from submission * thread. */ __sb_writers_acquired(file_inode(file)->i_sb, SB_FREEZE_WRITE); file_end_write(file); } |
20dcae324 [PATCH] aio: remo... |
1038 1039 1040 1041 1042 1043 |
/* * 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 |
1da177e4c Linux-2.6.12-rc2 |
1044 |
*/ |
04b2fa9f8 fs: split generic... |
1045 |
BUG_ON(is_sync_kiocb(kiocb)); |
1da177e4c Linux-2.6.12-rc2 |
1046 |
|
0460fef2a aio: use cancella... |
1047 1048 1049 1050 1051 1052 1053 |
if (iocb->ki_list.next) { unsigned long flags; spin_lock_irqsave(&ctx->ctx_lock, flags); list_del(&iocb->ki_list); spin_unlock_irqrestore(&ctx->ctx_lock, flags); } |
11599ebac aio: make aio_put... |
1054 |
|
1da177e4c Linux-2.6.12-rc2 |
1055 |
/* |
0460fef2a aio: use cancella... |
1056 |
* Add a completion event to the ring buffer. Must be done holding |
4b30f07e7 aio: fix wrong co... |
1057 |
* ctx->completion_lock to prevent other code from messing with the tail |
0460fef2a aio: use cancella... |
1058 1059 1060 |
* pointer since we might be called from irq context. */ spin_lock_irqsave(&ctx->completion_lock, flags); |
58c85dc20 aio: kill struct ... |
1061 |
tail = ctx->tail; |
21b40200c aio: use flush_dc... |
1062 |
pos = tail + AIO_EVENTS_OFFSET; |
58c85dc20 aio: kill struct ... |
1063 |
if (++tail >= ctx->nr_events) |
4bf69b2a0 [PATCH] aio: ring... |
1064 |
tail = 0; |
1da177e4c Linux-2.6.12-rc2 |
1065 |
|
58c85dc20 aio: kill struct ... |
1066 |
ev_page = kmap_atomic(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); |
21b40200c aio: use flush_dc... |
1067 |
event = ev_page + pos % AIO_EVENTS_PER_PAGE; |
04b2fa9f8 fs: split generic... |
1068 |
event->obj = (u64)(unsigned long)iocb->ki_user_iocb; |
1da177e4c Linux-2.6.12-rc2 |
1069 1070 1071 |
event->data = iocb->ki_user_data; event->res = res; event->res2 = res2; |
21b40200c aio: use flush_dc... |
1072 |
kunmap_atomic(ev_page); |
58c85dc20 aio: kill struct ... |
1073 |
flush_dcache_page(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); |
21b40200c aio: use flush_dc... |
1074 1075 1076 |
pr_debug("%p[%u]: %p: %p %Lx %lx %lx ", |
04b2fa9f8 fs: split generic... |
1077 |
ctx, tail, iocb, iocb->ki_user_iocb, iocb->ki_user_data, |
caf4167aa aio: dprintk() ->... |
1078 |
res, res2); |
1da177e4c Linux-2.6.12-rc2 |
1079 1080 1081 1082 1083 |
/* after flagging the request as done, we * must never even look at it again */ smp_wmb(); /* make event visible before updating tail */ |
58c85dc20 aio: kill struct ... |
1084 |
ctx->tail = tail; |
1da177e4c Linux-2.6.12-rc2 |
1085 |
|
58c85dc20 aio: kill struct ... |
1086 |
ring = kmap_atomic(ctx->ring_pages[0]); |
d856f32a8 aio: fix reqs_ava... |
1087 |
head = ring->head; |
21b40200c aio: use flush_dc... |
1088 |
ring->tail = tail; |
e8e3c3d66 fs: remove the se... |
1089 |
kunmap_atomic(ring); |
58c85dc20 aio: kill struct ... |
1090 |
flush_dcache_page(ctx->ring_pages[0]); |
1da177e4c Linux-2.6.12-rc2 |
1091 |
|
d856f32a8 aio: fix reqs_ava... |
1092 1093 1094 |
ctx->completed_events++; if (ctx->completed_events > 1) refill_reqs_available(ctx, head, tail); |
0460fef2a aio: use cancella... |
1095 |
spin_unlock_irqrestore(&ctx->completion_lock, flags); |
21b40200c aio: use flush_dc... |
1096 1097 |
pr_debug("added to ring %p at [%u] ", iocb, tail); |
8d1c98b0b eventfd/kaio inte... |
1098 1099 1100 1101 1102 1103 |
/* * 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. */ |
87c3a86e1 eventfd: remove f... |
1104 |
if (iocb->ki_eventfd != NULL) |
8d1c98b0b eventfd/kaio inte... |
1105 |
eventfd_signal(iocb->ki_eventfd, 1); |
1da177e4c Linux-2.6.12-rc2 |
1106 |
/* everything turned out well, dispose of the aiocb. */ |
57282d8fd aio: Kill ki_users |
1107 |
kiocb_free(iocb); |
1da177e4c Linux-2.6.12-rc2 |
1108 |
|
6cb2a2104 aio: bad AIO race... |
1109 1110 1111 1112 1113 1114 1115 |
/* * 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 Linux-2.6.12-rc2 |
1116 1117 |
if (waitqueue_active(&ctx->wait)) wake_up(&ctx->wait); |
e34ecee2a aio: Fix a trinit... |
1118 |
percpu_ref_put(&ctx->reqs); |
1da177e4c Linux-2.6.12-rc2 |
1119 |
} |
2be4e7dee aio: fix some com... |
1120 |
/* aio_read_events_ring |
a31ad380b aio: make aio_rea... |
1121 1122 |
* Pull an event off of the ioctx's event ring. Returns the number of * events fetched |
1da177e4c Linux-2.6.12-rc2 |
1123 |
*/ |
a31ad380b aio: make aio_rea... |
1124 1125 |
static long aio_read_events_ring(struct kioctx *ctx, struct io_event __user *event, long nr) |
1da177e4c Linux-2.6.12-rc2 |
1126 |
{ |
1da177e4c Linux-2.6.12-rc2 |
1127 |
struct aio_ring *ring; |
5ffac122d aio: Don't use ct... |
1128 |
unsigned head, tail, pos; |
a31ad380b aio: make aio_rea... |
1129 1130 |
long ret = 0; int copy_ret; |
9c9ce763b aio: annotate aio... |
1131 1132 1133 1134 1135 1136 1137 |
/* * 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 aio: kill struct ... |
1138 |
mutex_lock(&ctx->ring_lock); |
1da177e4c Linux-2.6.12-rc2 |
1139 |
|
fa8a53c39 aio: v4 ensure ac... |
1140 |
/* Access to ->ring_pages here is protected by ctx->ring_lock. */ |
58c85dc20 aio: kill struct ... |
1141 |
ring = kmap_atomic(ctx->ring_pages[0]); |
a31ad380b aio: make aio_rea... |
1142 |
head = ring->head; |
5ffac122d aio: Don't use ct... |
1143 |
tail = ring->tail; |
a31ad380b aio: make aio_rea... |
1144 |
kunmap_atomic(ring); |
2ff396be6 aio: add missing ... |
1145 1146 1147 1148 1149 |
/* * 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 aio: Don't use ct... |
1150 1151 |
pr_debug("h%u t%u m%u ", head, tail, ctx->nr_events); |
1da177e4c Linux-2.6.12-rc2 |
1152 |
|
5ffac122d aio: Don't use ct... |
1153 |
if (head == tail) |
1da177e4c Linux-2.6.12-rc2 |
1154 |
goto out; |
edfbbf388 aio: fix kernel m... |
1155 1156 |
head %= ctx->nr_events; tail %= ctx->nr_events; |
a31ad380b aio: make aio_rea... |
1157 1158 1159 1160 |
while (ret < nr) { long avail; struct io_event *ev; struct page *page; |
5ffac122d aio: Don't use ct... |
1161 1162 |
avail = (head <= tail ? tail : ctx->nr_events) - head; if (head == tail) |
a31ad380b aio: make aio_rea... |
1163 1164 1165 1166 1167 1168 1169 |
break; avail = min(avail, nr - ret); avail = min_t(long, avail, AIO_EVENTS_PER_PAGE - ((head + AIO_EVENTS_OFFSET) % AIO_EVENTS_PER_PAGE)); pos = head + AIO_EVENTS_OFFSET; |
58c85dc20 aio: kill struct ... |
1170 |
page = ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]; |
a31ad380b aio: make aio_rea... |
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 |
pos %= AIO_EVENTS_PER_PAGE; 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 aio: kill struct ... |
1185 |
head %= ctx->nr_events; |
1da177e4c Linux-2.6.12-rc2 |
1186 |
} |
1da177e4c Linux-2.6.12-rc2 |
1187 |
|
58c85dc20 aio: kill struct ... |
1188 |
ring = kmap_atomic(ctx->ring_pages[0]); |
a31ad380b aio: make aio_rea... |
1189 |
ring->head = head; |
91d80a84b aio: fix possible... |
1190 |
kunmap_atomic(ring); |
58c85dc20 aio: kill struct ... |
1191 |
flush_dcache_page(ctx->ring_pages[0]); |
a31ad380b aio: make aio_rea... |
1192 |
|
5ffac122d aio: Don't use ct... |
1193 1194 |
pr_debug("%li h%u t%u ", ret, head, tail); |
a31ad380b aio: make aio_rea... |
1195 |
out: |
58c85dc20 aio: kill struct ... |
1196 |
mutex_unlock(&ctx->ring_lock); |
a31ad380b aio: make aio_rea... |
1197 |
|
1da177e4c Linux-2.6.12-rc2 |
1198 1199 |
return ret; } |
a31ad380b aio: make aio_rea... |
1200 1201 |
static bool aio_read_events(struct kioctx *ctx, long min_nr, long nr, struct io_event __user *event, long *i) |
1da177e4c Linux-2.6.12-rc2 |
1202 |
{ |
a31ad380b aio: make aio_rea... |
1203 |
long ret = aio_read_events_ring(ctx, event + *i, nr - *i); |
1da177e4c Linux-2.6.12-rc2 |
1204 |
|
a31ad380b aio: make aio_rea... |
1205 1206 |
if (ret > 0) *i += ret; |
1da177e4c Linux-2.6.12-rc2 |
1207 |
|
a31ad380b aio: make aio_rea... |
1208 1209 |
if (unlikely(atomic_read(&ctx->dead))) ret = -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1210 |
|
a31ad380b aio: make aio_rea... |
1211 1212 |
if (!*i) *i = ret; |
1da177e4c Linux-2.6.12-rc2 |
1213 |
|
a31ad380b aio: make aio_rea... |
1214 |
return ret < 0 || *i >= min_nr; |
1da177e4c Linux-2.6.12-rc2 |
1215 |
} |
a31ad380b aio: make aio_rea... |
1216 |
static long read_events(struct kioctx *ctx, long min_nr, long nr, |
1da177e4c Linux-2.6.12-rc2 |
1217 1218 1219 |
struct io_event __user *event, struct timespec __user *timeout) { |
a31ad380b aio: make aio_rea... |
1220 1221 |
ktime_t until = { .tv64 = KTIME_MAX }; long ret = 0; |
1da177e4c Linux-2.6.12-rc2 |
1222 |
|
1da177e4c Linux-2.6.12-rc2 |
1223 1224 |
if (timeout) { struct timespec ts; |
a31ad380b aio: make aio_rea... |
1225 |
|
1da177e4c Linux-2.6.12-rc2 |
1226 |
if (unlikely(copy_from_user(&ts, timeout, sizeof(ts)))) |
a31ad380b aio: make aio_rea... |
1227 |
return -EFAULT; |
1da177e4c Linux-2.6.12-rc2 |
1228 |
|
a31ad380b aio: make aio_rea... |
1229 |
until = timespec_to_ktime(ts); |
1da177e4c Linux-2.6.12-rc2 |
1230 |
} |
a31ad380b aio: make aio_rea... |
1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 |
/* * 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. */ |
5f785de58 aio: Skip timer f... |
1245 1246 1247 1248 1249 1250 |
if (until.tv64 == 0) 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); |
1da177e4c Linux-2.6.12-rc2 |
1251 |
|
a31ad380b aio: make aio_rea... |
1252 1253 |
if (!ret && signal_pending(current)) ret = -EINTR; |
1da177e4c Linux-2.6.12-rc2 |
1254 |
|
a31ad380b aio: make aio_rea... |
1255 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1256 |
} |
1da177e4c Linux-2.6.12-rc2 |
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 |
/* 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 [CVE-2009-0029] S... |
1270 |
SYSCALL_DEFINE2(io_setup, unsigned, nr_events, aio_context_t __user *, ctxp) |
1da177e4c Linux-2.6.12-rc2 |
1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 |
{ struct kioctx *ioctx = NULL; unsigned long ctx; long ret; ret = get_user(ctx, ctxp); if (unlikely(ret)) goto out; ret = -EINVAL; |
d55b5fdaf [PATCH] aio: remo... |
1281 |
if (unlikely(ctx || nr_events == 0)) { |
acd88d4e1 fs/aio.c: Remove ... |
1282 1283 |
pr_debug("EINVAL: ctx %lu nr_events %u ", |
d55b5fdaf [PATCH] aio: remo... |
1284 |
ctx, nr_events); |
1da177e4c Linux-2.6.12-rc2 |
1285 1286 1287 1288 1289 1290 1291 |
goto out; } ioctx = ioctx_alloc(nr_events); ret = PTR_ERR(ioctx); if (!IS_ERR(ioctx)) { ret = put_user(ioctx->user_id, ctxp); |
a2e1859ad aio: take final p... |
1292 |
if (ret) |
e02ba72aa aio: block io_des... |
1293 |
kill_ioctx(current->mm, ioctx, NULL); |
723be6e39 aio: percpu ioctx... |
1294 |
percpu_ref_put(&ioctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1295 1296 1297 1298 1299 1300 1301 1302 1303 |
} 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 |
642b5123a aio: fix wrong su... |
1304 |
* implemented. May fail with -EINVAL if the context pointed to |
1da177e4c Linux-2.6.12-rc2 |
1305 1306 |
* is invalid. */ |
002c8976e [CVE-2009-0029] S... |
1307 |
SYSCALL_DEFINE1(io_destroy, aio_context_t, ctx) |
1da177e4c Linux-2.6.12-rc2 |
1308 1309 1310 |
{ struct kioctx *ioctx = lookup_ioctx(ctx); if (likely(NULL != ioctx)) { |
dc48e56d7 aio: fix serial d... |
1311 |
struct ctx_rq_wait wait; |
fb2d44838 aio: report error... |
1312 |
int ret; |
e02ba72aa aio: block io_des... |
1313 |
|
dc48e56d7 aio: fix serial d... |
1314 1315 |
init_completion(&wait.comp); atomic_set(&wait.count, 1); |
e02ba72aa aio: block io_des... |
1316 1317 1318 1319 |
/* 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 aio: fix serial d... |
1320 |
ret = kill_ioctx(current->mm, ioctx, &wait); |
723be6e39 aio: percpu ioctx... |
1321 |
percpu_ref_put(&ioctx->users); |
e02ba72aa aio: block io_des... |
1322 1323 1324 1325 1326 |
/* 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 aio: report error... |
1327 |
if (!ret) |
dc48e56d7 aio: fix serial d... |
1328 |
wait_for_completion(&wait.comp); |
e02ba72aa aio: block io_des... |
1329 |
|
fb2d44838 aio: report error... |
1330 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1331 |
} |
acd88d4e1 fs/aio.c: Remove ... |
1332 1333 |
pr_debug("EINVAL: invalid context id "); |
1da177e4c Linux-2.6.12-rc2 |
1334 1335 |
return -EINVAL; } |
89319d31d fs: remove aio_ru... |
1336 1337 |
static int aio_setup_rw(int rw, struct iocb *iocb, struct iovec **iovec, bool vectored, bool compat, struct iov_iter *iter) |
eed4e51fb [PATCH] Add vecto... |
1338 |
{ |
89319d31d fs: remove aio_ru... |
1339 1340 1341 1342 1343 1344 1345 1346 |
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 aio: fix the comp... |
1347 1348 |
#ifdef CONFIG_COMPAT if (compat) |
89319d31d fs: remove aio_ru... |
1349 1350 |
return compat_import_iovec(rw, buf, len, UIO_FASTIOV, iovec, iter); |
9d85cba71 aio: fix the comp... |
1351 |
#endif |
89319d31d fs: remove aio_ru... |
1352 |
return import_iovec(rw, buf, len, UIO_FASTIOV, iovec, iter); |
eed4e51fb [PATCH] Add vecto... |
1353 |
} |
89319d31d fs: remove aio_ru... |
1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 |
static inline ssize_t aio_ret(struct kiocb *req, ssize_t ret) { switch (ret) { case -EIOCBQUEUED: return ret; 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: aio_complete(req, ret, 0); return 0; } } static ssize_t aio_read(struct kiocb *req, struct iocb *iocb, bool vectored, bool compat) |
1da177e4c Linux-2.6.12-rc2 |
1377 |
{ |
41ef4eb8e aio: kill ki_retry |
1378 |
struct file *file = req->ki_filp; |
00fefb9cf aio: use iovec ar... |
1379 |
struct iovec inline_vecs[UIO_FASTIOV], *iovec = inline_vecs; |
293bc9822 new methods: ->re... |
1380 |
struct iov_iter iter; |
89319d31d fs: remove aio_ru... |
1381 |
ssize_t ret; |
1da177e4c Linux-2.6.12-rc2 |
1382 |
|
89319d31d fs: remove aio_ru... |
1383 1384 1385 1386 |
if (unlikely(!(file->f_mode & FMODE_READ))) return -EBADF; if (unlikely(!file->f_op->read_iter)) return -EINVAL; |
73a7075e3 aio: Kill aio_rw_... |
1387 |
|
89319d31d fs: remove aio_ru... |
1388 1389 1390 1391 1392 1393 1394 1395 1396 |
ret = aio_setup_rw(READ, iocb, &iovec, vectored, compat, &iter); if (ret) return ret; ret = rw_verify_area(READ, file, &req->ki_pos, iov_iter_count(&iter)); if (!ret) ret = aio_ret(req, file->f_op->read_iter(req, &iter)); kfree(iovec); return ret; } |
73a7075e3 aio: Kill aio_rw_... |
1397 |
|
89319d31d fs: remove aio_ru... |
1398 1399 1400 1401 1402 1403 1404 |
static ssize_t aio_write(struct kiocb *req, struct iocb *iocb, bool vectored, bool compat) { struct file *file = req->ki_filp; struct iovec inline_vecs[UIO_FASTIOV], *iovec = inline_vecs; struct iov_iter iter; ssize_t ret; |
41ef4eb8e aio: kill ki_retry |
1405 |
|
89319d31d fs: remove aio_ru... |
1406 1407 1408 |
if (unlikely(!(file->f_mode & FMODE_WRITE))) return -EBADF; if (unlikely(!file->f_op->write_iter)) |
41ef4eb8e aio: kill ki_retry |
1409 |
return -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1410 |
|
89319d31d fs: remove aio_ru... |
1411 1412 1413 1414 1415 |
ret = aio_setup_rw(WRITE, iocb, &iovec, vectored, compat, &iter); if (ret) return ret; ret = rw_verify_area(WRITE, file, &req->ki_pos, iov_iter_count(&iter)); if (!ret) { |
70fe2f481 aio: fix freeze p... |
1416 |
req->ki_flags |= IOCB_WRITE; |
89319d31d fs: remove aio_ru... |
1417 1418 |
file_start_write(file); ret = aio_ret(req, file->f_op->write_iter(req, &iter)); |
70fe2f481 aio: fix freeze p... |
1419 1420 1421 1422 1423 1424 |
/* * We release freeze protection in aio_complete(). Fool lockdep * by telling it the lock got released so that it doesn't * complain about held lock when we return to userspace. */ __sb_writers_release(file_inode(file)->i_sb, SB_FREEZE_WRITE); |
41ef4eb8e aio: kill ki_retry |
1425 |
} |
89319d31d fs: remove aio_ru... |
1426 1427 |
kfree(iovec); return ret; |
1da177e4c Linux-2.6.12-rc2 |
1428 |
} |
d5470b596 fs/aio.c: make 3 ... |
1429 |
static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb, |
a1c8eae75 aio: kill batch a... |
1430 |
struct iocb *iocb, bool compat) |
1da177e4c Linux-2.6.12-rc2 |
1431 |
{ |
04b2fa9f8 fs: split generic... |
1432 |
struct aio_kiocb *req; |
89319d31d fs: remove aio_ru... |
1433 |
struct file *file; |
1da177e4c Linux-2.6.12-rc2 |
1434 1435 1436 |
ssize_t ret; /* enforce forwards compatibility on users */ |
9c3060bed signal/timer/even... |
1437 |
if (unlikely(iocb->aio_reserved1 || iocb->aio_reserved2)) { |
caf4167aa aio: dprintk() ->... |
1438 1439 |
pr_debug("EINVAL: reserve field set "); |
1da177e4c Linux-2.6.12-rc2 |
1440 1441 1442 1443 1444 1445 1446 1447 1448 |
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) )) { |
acd88d4e1 fs/aio.c: Remove ... |
1449 1450 |
pr_debug("EINVAL: overflow check "); |
1da177e4c Linux-2.6.12-rc2 |
1451 1452 |
return -EINVAL; } |
41ef4eb8e aio: kill ki_retry |
1453 |
req = aio_get_req(ctx); |
1d98ebfcc aio: do fget() af... |
1454 |
if (unlikely(!req)) |
1da177e4c Linux-2.6.12-rc2 |
1455 |
return -EAGAIN; |
1d98ebfcc aio: do fget() af... |
1456 |
|
89319d31d fs: remove aio_ru... |
1457 |
req->common.ki_filp = file = fget(iocb->aio_fildes); |
04b2fa9f8 fs: split generic... |
1458 |
if (unlikely(!req->common.ki_filp)) { |
1d98ebfcc aio: do fget() af... |
1459 1460 |
ret = -EBADF; goto out_put_req; |
1da177e4c Linux-2.6.12-rc2 |
1461 |
} |
04b2fa9f8 fs: split generic... |
1462 1463 |
req->common.ki_pos = iocb->aio_offset; req->common.ki_complete = aio_complete; |
2ba48ce51 mirror O_APPEND a... |
1464 |
req->common.ki_flags = iocb_flags(req->common.ki_filp); |
1d98ebfcc aio: do fget() af... |
1465 |
|
9c3060bed signal/timer/even... |
1466 1467 1468 1469 1470 1471 1472 |
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. */ |
133890103 eventfd: revised ... |
1473 |
req->ki_eventfd = eventfd_ctx_fdget((int) iocb->aio_resfd); |
801678c5a Remove duplicated... |
1474 |
if (IS_ERR(req->ki_eventfd)) { |
9c3060bed signal/timer/even... |
1475 |
ret = PTR_ERR(req->ki_eventfd); |
87c3a86e1 eventfd: remove f... |
1476 |
req->ki_eventfd = NULL; |
9c3060bed signal/timer/even... |
1477 1478 |
goto out_put_req; } |
04b2fa9f8 fs: split generic... |
1479 1480 |
req->common.ki_flags |= IOCB_EVENTFD; |
9c3060bed signal/timer/even... |
1481 |
} |
1da177e4c Linux-2.6.12-rc2 |
1482 |
|
8a6608907 aio: kill ki_key |
1483 |
ret = put_user(KIOCB_KEY, &user_iocb->aio_key); |
1da177e4c Linux-2.6.12-rc2 |
1484 |
if (unlikely(ret)) { |
caf4167aa aio: dprintk() ->... |
1485 1486 |
pr_debug("EFAULT: aio_key "); |
1da177e4c Linux-2.6.12-rc2 |
1487 1488 |
goto out_put_req; } |
04b2fa9f8 fs: split generic... |
1489 |
req->ki_user_iocb = user_iocb; |
1da177e4c Linux-2.6.12-rc2 |
1490 |
req->ki_user_data = iocb->aio_data; |
1da177e4c Linux-2.6.12-rc2 |
1491 |
|
89319d31d fs: remove aio_ru... |
1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 |
get_file(file); switch (iocb->aio_lio_opcode) { case IOCB_CMD_PREAD: ret = aio_read(&req->common, iocb, false, compat); break; case IOCB_CMD_PWRITE: ret = aio_write(&req->common, iocb, false, compat); break; case IOCB_CMD_PREADV: ret = aio_read(&req->common, iocb, true, compat); break; case IOCB_CMD_PWRITEV: ret = aio_write(&req->common, iocb, true, compat); break; default: pr_debug("invalid aio operation %d ", iocb->aio_lio_opcode); ret = -EINVAL; break; } fput(file); |
1da177e4c Linux-2.6.12-rc2 |
1513 |
|
89319d31d fs: remove aio_ru... |
1514 1515 |
if (ret && ret != -EIOCBQUEUED) goto out_put_req; |
1da177e4c Linux-2.6.12-rc2 |
1516 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
1517 |
out_put_req: |
e1bdd5f27 aio: percpu reqs_... |
1518 |
put_reqs_available(ctx, 1); |
e34ecee2a aio: Fix a trinit... |
1519 |
percpu_ref_put(&ctx->reqs); |
57282d8fd aio: Kill ki_users |
1520 |
kiocb_free(req); |
1da177e4c Linux-2.6.12-rc2 |
1521 1522 |
return ret; } |
9d85cba71 aio: fix the comp... |
1523 1524 |
long do_io_submit(aio_context_t ctx_id, long nr, struct iocb __user *__user *iocbpp, bool compat) |
1da177e4c Linux-2.6.12-rc2 |
1525 1526 1527 |
{ struct kioctx *ctx; long ret = 0; |
080d676de aio: allocate kio... |
1528 |
int i = 0; |
9f5b94254 fs: make aio plug |
1529 |
struct blk_plug plug; |
1da177e4c Linux-2.6.12-rc2 |
1530 1531 1532 |
if (unlikely(nr < 0)) return -EINVAL; |
75e1c70fc aio: check for mu... |
1533 1534 |
if (unlikely(nr > LONG_MAX/sizeof(*iocbpp))) nr = LONG_MAX/sizeof(*iocbpp); |
1da177e4c Linux-2.6.12-rc2 |
1535 1536 1537 1538 1539 |
if (unlikely(!access_ok(VERIFY_READ, iocbpp, (nr*sizeof(*iocbpp))))) return -EFAULT; ctx = lookup_ioctx(ctx_id); if (unlikely(!ctx)) { |
caf4167aa aio: dprintk() ->... |
1540 1541 |
pr_debug("EINVAL: invalid context id "); |
1da177e4c Linux-2.6.12-rc2 |
1542 1543 |
return -EINVAL; } |
9f5b94254 fs: make aio plug |
1544 |
blk_start_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 |
/* * 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; } |
a1c8eae75 aio: kill batch a... |
1562 |
ret = io_submit_one(ctx, user_iocb, &tmp, compat); |
1da177e4c Linux-2.6.12-rc2 |
1563 1564 1565 |
if (ret) break; } |
9f5b94254 fs: make aio plug |
1566 |
blk_finish_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
1567 |
|
723be6e39 aio: percpu ioctx... |
1568 |
percpu_ref_put(&ctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1569 1570 |
return i ? i : ret; } |
9d85cba71 aio: fix the comp... |
1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 |
/* sys_io_submit: * Queue the nr iocbs pointed to by iocbpp for processing. Returns * the number of iocbs queued. May return -EINVAL if the aio_context * specified by ctx_id is invalid, if nr is < 0, if the iocb at * *iocbpp[0] is not properly initialized, if the operation specified * is invalid for the file descriptor in the iocb. May fail with * -EFAULT if any of the data structures point to invalid data. May * fail with -EBADF if the file descriptor specified in the first * iocb is invalid. May fail with -EAGAIN if insufficient resources * are available to queue any iocbs. Will return 0 if nr is 0. Will * fail with -ENOSYS if not implemented. */ SYSCALL_DEFINE3(io_submit, aio_context_t, ctx_id, long, nr, struct iocb __user * __user *, iocbpp) { return do_io_submit(ctx_id, nr, iocbpp, 0); } |
1da177e4c Linux-2.6.12-rc2 |
1588 1589 |
/* lookup_kiocb * Finds a given iocb for cancellation. |
1da177e4c Linux-2.6.12-rc2 |
1590 |
*/ |
04b2fa9f8 fs: split generic... |
1591 1592 |
static struct aio_kiocb * lookup_kiocb(struct kioctx *ctx, struct iocb __user *iocb, u32 key) |
1da177e4c Linux-2.6.12-rc2 |
1593 |
{ |
04b2fa9f8 fs: split generic... |
1594 |
struct aio_kiocb *kiocb; |
d00689af6 [PATCH] aio: repl... |
1595 1596 |
assert_spin_locked(&ctx->ctx_lock); |
8a6608907 aio: kill ki_key |
1597 1598 |
if (key != KIOCB_KEY) return NULL; |
1da177e4c Linux-2.6.12-rc2 |
1599 |
/* TODO: use a hash or array, this sucks. */ |
04b2fa9f8 fs: split generic... |
1600 1601 |
list_for_each_entry(kiocb, &ctx->active_reqs, ki_list) { if (kiocb->ki_user_iocb == iocb) |
1da177e4c Linux-2.6.12-rc2 |
1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 |
return kiocb; } return NULL; } /* sys_io_cancel: * Attempts to cancel an iocb previously passed to io_submit. If * the operation is successfully cancelled, the resulting event is * copied into the memory pointed to by result without being placed * into the completion queue and 0 is returned. May fail with * -EFAULT if any of the data structures pointed to are invalid. * May fail with -EINVAL if aio_context specified by ctx_id is * invalid. May fail with -EAGAIN if the iocb specified was not * cancelled. Will fail with -ENOSYS if not implemented. */ |
002c8976e [CVE-2009-0029] S... |
1617 1618 |
SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb, struct io_event __user *, result) |
1da177e4c Linux-2.6.12-rc2 |
1619 |
{ |
1da177e4c Linux-2.6.12-rc2 |
1620 |
struct kioctx *ctx; |
04b2fa9f8 fs: split generic... |
1621 |
struct aio_kiocb *kiocb; |
1da177e4c Linux-2.6.12-rc2 |
1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 |
u32 key; int ret; ret = get_user(key, &iocb->aio_key); if (unlikely(ret)) return -EFAULT; ctx = lookup_ioctx(ctx_id); if (unlikely(!ctx)) return -EINVAL; spin_lock_irq(&ctx->ctx_lock); |
906b973cf aio: add kiocb_ca... |
1634 |
|
1da177e4c Linux-2.6.12-rc2 |
1635 |
kiocb = lookup_kiocb(ctx, iocb, key); |
906b973cf aio: add kiocb_ca... |
1636 |
if (kiocb) |
d52a8f9ea fs/aio.c: Remove ... |
1637 |
ret = kiocb_cancel(kiocb); |
906b973cf aio: add kiocb_ca... |
1638 1639 |
else ret = -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1640 |
spin_unlock_irq(&ctx->ctx_lock); |
906b973cf aio: add kiocb_ca... |
1641 |
if (!ret) { |
bec68faaf aio: io_cancel() ... |
1642 1643 1644 1645 |
/* * The result argument is no longer used - the io_event is * always delivered via the ring buffer. -EINPROGRESS indicates * cancellation is progress: |
906b973cf aio: add kiocb_ca... |
1646 |
*/ |
bec68faaf aio: io_cancel() ... |
1647 |
ret = -EINPROGRESS; |
906b973cf aio: add kiocb_ca... |
1648 |
} |
1da177e4c Linux-2.6.12-rc2 |
1649 |
|
723be6e39 aio: percpu ioctx... |
1650 |
percpu_ref_put(&ctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1651 1652 1653 1654 1655 1656 |
return ret; } /* io_getevents: * Attempts to read at least min_nr events and up to nr events from |
642b5123a aio: fix wrong su... |
1657 1658 1659 1660 1661 1662 1663 1664 |
* 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 aio: fix io_getev... |
1665 |
* timeout is relative. Will fail with -ENOSYS if not implemented. |
1da177e4c Linux-2.6.12-rc2 |
1666 |
*/ |
002c8976e [CVE-2009-0029] S... |
1667 1668 1669 1670 1671 |
SYSCALL_DEFINE5(io_getevents, aio_context_t, ctx_id, long, min_nr, long, nr, struct io_event __user *, events, struct timespec __user *, timeout) |
1da177e4c Linux-2.6.12-rc2 |
1672 1673 1674 1675 1676 |
{ struct kioctx *ioctx = lookup_ioctx(ctx_id); long ret = -EINVAL; if (likely(ioctx)) { |
2e4102559 aio: remove unnec... |
1677 |
if (likely(min_nr <= nr && min_nr >= 0)) |
1da177e4c Linux-2.6.12-rc2 |
1678 |
ret = read_events(ioctx, min_nr, nr, events, timeout); |
723be6e39 aio: percpu ioctx... |
1679 |
percpu_ref_put(&ioctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1680 |
} |
1da177e4c Linux-2.6.12-rc2 |
1681 1682 |
return ret; } |