Blame view
fs/aio.c
47.2 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 |
|
174cd4b1e sched/headers: Pr... |
22 |
#include <linux/sched/signal.h> |
1da177e4c Linux-2.6.12-rc2 |
23 24 25 26 |
#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 |
#include <asm/kmap_types.h> |
7c0f6ba68 Replace <asm/uacc... |
44 |
#include <linux/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 |
struct kioctx_table { |
cd21b3400 fs/aio: Use RCU a... |
69 70 71 |
struct rcu_head rcu; unsigned nr; struct kioctx __rcu *table[]; |
db446a08c aio: convert the ... |
72 |
}; |
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; |
076c7c068 fs/aio: Add expli... |
110 111 |
struct rcu_head free_rcu; struct work_struct free_work; /* see free_ioctx() */ |
4e23bcaeb aio: give shared ... |
112 |
|
e02ba72aa aio: block io_des... |
113 114 115 |
/* * signals when all in-flight requests are done */ |
dc48e56d7 aio: fix serial d... |
116 |
struct ctx_rq_wait *rq_wait; |
e02ba72aa aio: block io_des... |
117 |
|
4e23bcaeb aio: give shared ... |
118 |
struct { |
34e83fc61 aio: reqs_active ... |
119 120 121 122 123 |
/* * 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_... |
124 125 |
* * We batch accesses to it with a percpu version. |
34e83fc61 aio: reqs_active ... |
126 127 |
*/ atomic_t reqs_available; |
4e23bcaeb aio: give shared ... |
128 129 130 131 132 133 |
} ____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 ... |
134 135 |
struct { struct mutex ring_lock; |
4e23bcaeb aio: give shared ... |
136 137 |
wait_queue_head_t wait; } ____cacheline_aligned_in_smp; |
58c85dc20 aio: kill struct ... |
138 139 140 |
struct { unsigned tail; |
d856f32a8 aio: fix reqs_ava... |
141 |
unsigned completed_events; |
58c85dc20 aio: kill struct ... |
142 |
spinlock_t completion_lock; |
4e23bcaeb aio: give shared ... |
143 |
} ____cacheline_aligned_in_smp; |
58c85dc20 aio: kill struct ... |
144 145 |
struct page *internal_pages[AIO_RING_PAGES]; |
36bc08cc0 fs/aio: Add suppo... |
146 |
struct file *aio_ring_file; |
db446a08c aio: convert the ... |
147 148 |
unsigned id; |
4e179bca6 aio: move private... |
149 |
}; |
04b2fa9f8 fs: split generic... |
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 |
/* * 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 |
181 |
/*------ sysctl variables----*/ |
d55b5fdaf [PATCH] aio: remo... |
182 183 184 |
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 |
185 |
/*----end sysctl variables---*/ |
e18b890bb [PATCH] slab: rem... |
186 187 |
static struct kmem_cache *kiocb_cachep; static struct kmem_cache *kioctx_cachep; |
1da177e4c Linux-2.6.12-rc2 |
188 |
|
71ad7490c rework aio migrat... |
189 190 191 192 193 194 195 196 197 198 199 |
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... |
200 201 |
if (IS_ERR(inode)) return ERR_CAST(inode); |
71ad7490c rework aio migrat... |
202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 |
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... |
222 223 224 225 226 227 228 229 230 |
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... |
231 232 233 234 235 236 |
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... |
237 |
} |
1da177e4c Linux-2.6.12-rc2 |
238 239 240 241 242 243 |
/* 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... |
244 245 246 247 248 249 250 251 |
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... |
252 |
kiocb_cachep = KMEM_CACHE(aio_kiocb, SLAB_HWCACHE_ALIGN|SLAB_PANIC); |
0a31bd5f2 KMEM_CACHE(): sim... |
253 |
kioctx_cachep = KMEM_CACHE(kioctx,SLAB_HWCACHE_ALIGN|SLAB_PANIC); |
1da177e4c Linux-2.6.12-rc2 |
254 |
|
caf4167aa aio: dprintk() ->... |
255 256 |
pr_debug("sizeof(struct page) = %zu ", sizeof(struct page)); |
1da177e4c Linux-2.6.12-rc2 |
257 258 259 |
return 0; } |
385773e04 aio.c: move EXPOR... |
260 |
__initcall(aio_setup); |
1da177e4c Linux-2.6.12-rc2 |
261 |
|
5e9ae2e5d aio: fix use-afte... |
262 263 264 |
static void put_aio_ring_file(struct kioctx *ctx) { struct file *aio_ring_file = ctx->aio_ring_file; |
de04e7693 fs/aio.c: elimina... |
265 |
struct address_space *i_mapping; |
5e9ae2e5d aio: fix use-afte... |
266 |
if (aio_ring_file) { |
450630975 don't open-code f... |
267 |
truncate_setsize(file_inode(aio_ring_file), 0); |
5e9ae2e5d aio: fix use-afte... |
268 269 |
/* Prevent further access to the kioctx from migratepages */ |
450630975 don't open-code f... |
270 |
i_mapping = aio_ring_file->f_mapping; |
de04e7693 fs/aio.c: elimina... |
271 272 |
spin_lock(&i_mapping->private_lock); i_mapping->private_data = NULL; |
5e9ae2e5d aio: fix use-afte... |
273 |
ctx->aio_ring_file = NULL; |
de04e7693 fs/aio.c: elimina... |
274 |
spin_unlock(&i_mapping->private_lock); |
5e9ae2e5d aio: fix use-afte... |
275 276 277 278 |
fput(aio_ring_file); } } |
1da177e4c Linux-2.6.12-rc2 |
279 280 |
static void aio_free_ring(struct kioctx *ctx) { |
36bc08cc0 fs/aio: Add suppo... |
281 |
int i; |
1da177e4c Linux-2.6.12-rc2 |
282 |
|
fa8a53c39 aio: v4 ensure ac... |
283 284 285 286 |
/* 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... |
287 |
for (i = 0; i < ctx->nr_pages; i++) { |
8e321fefb aio/migratepages:... |
288 |
struct page *page; |
36bc08cc0 fs/aio: Add suppo... |
289 290 291 |
pr_debug("pid(%d) [%d] page->count=%d ", current->pid, i, page_count(ctx->ring_pages[i])); |
8e321fefb aio/migratepages:... |
292 293 294 295 296 |
page = ctx->ring_pages[i]; if (!page) continue; ctx->ring_pages[i] = NULL; put_page(page); |
36bc08cc0 fs/aio: Add suppo... |
297 |
} |
1da177e4c Linux-2.6.12-rc2 |
298 |
|
ddb8c45ba aio: nullify aio-... |
299 |
if (ctx->ring_pages && ctx->ring_pages != ctx->internal_pages) { |
58c85dc20 aio: kill struct ... |
300 |
kfree(ctx->ring_pages); |
ddb8c45ba aio: nullify aio-... |
301 302 |
ctx->ring_pages = NULL; } |
36bc08cc0 fs/aio: Add suppo... |
303 |
} |
5477e70a6 mm: move ->mremap... |
304 |
static int aio_ring_mremap(struct vm_area_struct *vma) |
e4a0d3e72 aio: Make it poss... |
305 |
{ |
5477e70a6 mm: move ->mremap... |
306 |
struct file *file = vma->vm_file; |
e4a0d3e72 aio: Make it poss... |
307 308 |
struct mm_struct *mm = vma->vm_mm; struct kioctx_table *table; |
b2edffdd9 fix mremap() vs. ... |
309 |
int i, res = -EINVAL; |
e4a0d3e72 aio: Make it poss... |
310 311 312 313 314 315 |
spin_lock(&mm->ioctx_lock); rcu_read_lock(); table = rcu_dereference(mm->ioctx_table); for (i = 0; i < table->nr; i++) { struct kioctx *ctx; |
cd21b3400 fs/aio: Use RCU a... |
316 |
ctx = rcu_dereference(table->table[i]); |
e4a0d3e72 aio: Make it poss... |
317 |
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; |
2916ecc0f mm/migrate: new m... |
356 357 358 359 360 361 362 |
/* * 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 aio/migratepages:... |
363 |
rc = 0; |
fa8a53c39 aio: v4 ensure ac... |
364 |
/* mapping->private_lock here protects against the kioctx teardown. */ |
8e321fefb aio/migratepages:... |
365 366 |
spin_lock(&mapping->private_lock); ctx = mapping->private_data; |
fa8a53c39 aio: v4 ensure ac... |
367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 |
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:... |
386 387 |
} else rc = -EINVAL; |
8e321fefb aio/migratepages:... |
388 389 |
if (rc != 0) |
fa8a53c39 aio: v4 ensure ac... |
390 |
goto out_unlock; |
8e321fefb aio/migratepages:... |
391 |
|
36bc08cc0 fs/aio: Add suppo... |
392 393 |
/* Writeback must be complete */ BUG_ON(PageWriteback(old)); |
8e321fefb aio/migratepages:... |
394 |
get_page(new); |
36bc08cc0 fs/aio: Add suppo... |
395 |
|
8e321fefb aio/migratepages:... |
396 |
rc = migrate_page_move_mapping(mapping, new, old, NULL, mode, 1); |
36bc08cc0 fs/aio: Add suppo... |
397 |
if (rc != MIGRATEPAGE_SUCCESS) { |
8e321fefb aio/migratepages:... |
398 |
put_page(new); |
fa8a53c39 aio: v4 ensure ac... |
399 |
goto out_unlock; |
36bc08cc0 fs/aio: Add suppo... |
400 |
} |
fa8a53c39 aio: v4 ensure ac... |
401 402 403 |
/* 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... |
404 |
*/ |
fa8a53c39 aio: v4 ensure ac... |
405 406 407 408 409 |
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... |
410 |
|
fa8a53c39 aio: v4 ensure ac... |
411 412 |
/* The old page is no longer accessible. */ put_page(old); |
8e321fefb aio/migratepages:... |
413 |
|
fa8a53c39 aio: v4 ensure ac... |
414 415 416 417 |
out_unlock: mutex_unlock(&ctx->ring_lock); out: spin_unlock(&mapping->private_lock); |
36bc08cc0 fs/aio: Add suppo... |
418 |
return rc; |
1da177e4c Linux-2.6.12-rc2 |
419 |
} |
0c45355fc aio: fix build wh... |
420 |
#endif |
1da177e4c Linux-2.6.12-rc2 |
421 |
|
36bc08cc0 fs/aio: Add suppo... |
422 |
static const struct address_space_operations aio_ctx_aops = { |
835f252c6 aio: fix uncorren... |
423 |
.set_page_dirty = __set_page_dirty_no_writeback, |
0c45355fc aio: fix build wh... |
424 |
#if IS_ENABLED(CONFIG_MIGRATION) |
36bc08cc0 fs/aio: Add suppo... |
425 |
.migratepage = aio_migratepage, |
0c45355fc aio: fix build wh... |
426 |
#endif |
36bc08cc0 fs/aio: Add suppo... |
427 |
}; |
2a8a98673 fs: aio: fix the ... |
428 |
static int aio_setup_ring(struct kioctx *ctx, unsigned int nr_events) |
1da177e4c Linux-2.6.12-rc2 |
429 430 |
{ struct aio_ring *ring; |
41003a7bc aio: remove retry... |
431 |
struct mm_struct *mm = current->mm; |
3dc9acb67 aio: clean up and... |
432 |
unsigned long size, unused; |
1da177e4c Linux-2.6.12-rc2 |
433 |
int nr_pages; |
36bc08cc0 fs/aio: Add suppo... |
434 435 |
int i; struct file *file; |
1da177e4c Linux-2.6.12-rc2 |
436 437 438 439 440 441 |
/* 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 |
442 |
|
36bc08cc0 fs/aio: Add suppo... |
443 |
nr_pages = PFN_UP(size); |
1da177e4c Linux-2.6.12-rc2 |
444 445 |
if (nr_pages < 0) return -EINVAL; |
71ad7490c rework aio migrat... |
446 |
file = aio_private_file(ctx, nr_pages); |
36bc08cc0 fs/aio: Add suppo... |
447 448 |
if (IS_ERR(file)) { ctx->aio_ring_file = NULL; |
fa8a53c39 aio: v4 ensure ac... |
449 |
return -ENOMEM; |
36bc08cc0 fs/aio: Add suppo... |
450 |
} |
3dc9acb67 aio: clean up and... |
451 452 453 454 455 456 457 458 459 460 461 462 463 |
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... |
464 465 |
for (i = 0; i < nr_pages; i++) { struct page *page; |
450630975 don't open-code f... |
466 |
page = find_or_create_page(file->f_mapping, |
36bc08cc0 fs/aio: Add suppo... |
467 468 469 470 471 472 473 |
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... |
474 |
unlock_page(page); |
3dc9acb67 aio: clean up and... |
475 476 |
ctx->ring_pages[i] = page; |
36bc08cc0 fs/aio: Add suppo... |
477 |
} |
3dc9acb67 aio: clean up and... |
478 |
ctx->nr_pages = i; |
1da177e4c Linux-2.6.12-rc2 |
479 |
|
3dc9acb67 aio: clean up and... |
480 481 |
if (unlikely(i != nr_pages)) { aio_free_ring(ctx); |
fa8a53c39 aio: v4 ensure ac... |
482 |
return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
483 |
} |
58c85dc20 aio: kill struct ... |
484 485 486 |
ctx->mmap_size = nr_pages * PAGE_SIZE; pr_debug("attempting mmap of %lu bytes ", ctx->mmap_size); |
36bc08cc0 fs/aio: Add suppo... |
487 |
|
013373e8b aio: make aio_set... |
488 489 490 491 492 |
if (down_write_killable(&mm->mmap_sem)) { ctx->mmap_size = 0; aio_free_ring(ctx); return -EINTR; } |
36bc08cc0 fs/aio: Add suppo... |
493 494 |
ctx->mmap_base = do_mmap_pgoff(ctx->aio_ring_file, 0, ctx->mmap_size, PROT_READ | PROT_WRITE, |
897ab3e0c userfaultfd: non-... |
495 |
MAP_SHARED, 0, &unused, NULL); |
3dc9acb67 aio: clean up and... |
496 |
up_write(&mm->mmap_sem); |
58c85dc20 aio: kill struct ... |
497 |
if (IS_ERR((void *)ctx->mmap_base)) { |
58c85dc20 aio: kill struct ... |
498 |
ctx->mmap_size = 0; |
1da177e4c Linux-2.6.12-rc2 |
499 |
aio_free_ring(ctx); |
fa8a53c39 aio: v4 ensure ac... |
500 |
return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
501 |
} |
58c85dc20 aio: kill struct ... |
502 503 |
pr_debug("mmap address: 0x%08lx ", ctx->mmap_base); |
d6c355c7d aio: fix race in ... |
504 |
|
58c85dc20 aio: kill struct ... |
505 506 |
ctx->user_id = ctx->mmap_base; ctx->nr_events = nr_events; /* trusted copy */ |
1da177e4c Linux-2.6.12-rc2 |
507 |
|
58c85dc20 aio: kill struct ... |
508 |
ring = kmap_atomic(ctx->ring_pages[0]); |
1da177e4c Linux-2.6.12-rc2 |
509 |
ring->nr = nr_events; /* user copy */ |
db446a08c aio: convert the ... |
510 |
ring->id = ~0U; |
1da177e4c Linux-2.6.12-rc2 |
511 512 513 514 515 |
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... |
516 |
kunmap_atomic(ring); |
58c85dc20 aio: kill struct ... |
517 |
flush_dcache_page(ctx->ring_pages[0]); |
1da177e4c Linux-2.6.12-rc2 |
518 519 520 |
return 0; } |
1da177e4c Linux-2.6.12-rc2 |
521 522 523 |
#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... |
524 |
void kiocb_set_cancel_fn(struct kiocb *iocb, kiocb_cancel_fn *cancel) |
0460fef2a aio: use cancella... |
525 |
{ |
04b2fa9f8 fs: split generic... |
526 |
struct aio_kiocb *req = container_of(iocb, struct aio_kiocb, common); |
0460fef2a aio: use cancella... |
527 528 529 530 531 532 533 534 535 536 537 538 539 |
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... |
540 |
static int kiocb_cancel(struct aio_kiocb *kiocb) |
906b973cf aio: add kiocb_ca... |
541 |
{ |
0460fef2a aio: use cancella... |
542 |
kiocb_cancel_fn *old, *cancel; |
906b973cf aio: add kiocb_ca... |
543 |
|
0460fef2a aio: use cancella... |
544 545 546 547 548 549 550 551 |
/* * 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 |
552 |
return -EINVAL; |
906b973cf aio: add kiocb_ca... |
553 |
|
0460fef2a aio: use cancella... |
554 555 556 |
old = cancel; cancel = cmpxchg(&kiocb->ki_cancel, old, KIOCB_CANCELLED); } while (cancel != old); |
906b973cf aio: add kiocb_ca... |
557 |
|
04b2fa9f8 fs: split generic... |
558 |
return cancel(&kiocb->common); |
906b973cf aio: add kiocb_ca... |
559 |
} |
076c7c068 fs/aio: Add expli... |
560 561 562 563 564 565 |
/* * free_ioctx() should be RCU delayed to synchronize against the RCU * protected lookup_ioctx() and also needs process context to call * aio_free_ring(), so the double bouncing through kioctx->free_rcu and * ->free_work. */ |
e34ecee2a aio: Fix a trinit... |
566 |
static void free_ioctx(struct work_struct *work) |
36f558890 aio: refcounting ... |
567 |
{ |
e34ecee2a aio: Fix a trinit... |
568 |
struct kioctx *ctx = container_of(work, struct kioctx, free_work); |
e1bdd5f27 aio: percpu reqs_... |
569 |
|
e34ecee2a aio: Fix a trinit... |
570 571 |
pr_debug("freeing %p ", ctx); |
e1bdd5f27 aio: percpu reqs_... |
572 |
|
e34ecee2a aio: Fix a trinit... |
573 |
aio_free_ring(ctx); |
e1bdd5f27 aio: percpu reqs_... |
574 |
free_percpu(ctx->cpu); |
9a1049da9 percpu-refcount: ... |
575 576 |
percpu_ref_exit(&ctx->reqs); percpu_ref_exit(&ctx->users); |
36f558890 aio: refcounting ... |
577 578 |
kmem_cache_free(kioctx_cachep, ctx); } |
076c7c068 fs/aio: Add expli... |
579 580 581 582 583 584 585 |
static void free_ioctx_rcufn(struct rcu_head *head) { struct kioctx *ctx = container_of(head, struct kioctx, free_rcu); INIT_WORK(&ctx->free_work, free_ioctx); schedule_work(&ctx->free_work); } |
e34ecee2a aio: Fix a trinit... |
586 587 588 |
static void free_ioctx_reqs(struct percpu_ref *ref) { struct kioctx *ctx = container_of(ref, struct kioctx, reqs); |
e02ba72aa aio: block io_des... |
589 |
/* At this point we know that there are no any in-flight requests */ |
dc48e56d7 aio: fix serial d... |
590 591 |
if (ctx->rq_wait && atomic_dec_and_test(&ctx->rq_wait->count)) complete(&ctx->rq_wait->comp); |
e02ba72aa aio: block io_des... |
592 |
|
076c7c068 fs/aio: Add expli... |
593 594 |
/* Synchronize against RCU protected table->table[] dereferences */ call_rcu(&ctx->free_rcu, free_ioctx_rcufn); |
e34ecee2a aio: Fix a trinit... |
595 |
} |
36f558890 aio: refcounting ... |
596 597 598 599 600 |
/* * 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... |
601 |
static void free_ioctx_users(struct percpu_ref *ref) |
36f558890 aio: refcounting ... |
602 |
{ |
e34ecee2a aio: Fix a trinit... |
603 |
struct kioctx *ctx = container_of(ref, struct kioctx, users); |
04b2fa9f8 fs: split generic... |
604 |
struct aio_kiocb *req; |
36f558890 aio: refcounting ... |
605 606 607 608 609 |
spin_lock_irq(&ctx->ctx_lock); while (!list_empty(&ctx->active_reqs)) { req = list_first_entry(&ctx->active_reqs, |
04b2fa9f8 fs: split generic... |
610 |
struct aio_kiocb, ki_list); |
d52a8f9ea fs/aio.c: Remove ... |
611 |
kiocb_cancel(req); |
6a19487d5 fix io_destroy()/... |
612 |
list_del_init(&req->ki_list); |
36f558890 aio: refcounting ... |
613 614 615 |
} spin_unlock_irq(&ctx->ctx_lock); |
e34ecee2a aio: Fix a trinit... |
616 617 |
percpu_ref_kill(&ctx->reqs); percpu_ref_put(&ctx->reqs); |
36f558890 aio: refcounting ... |
618 |
} |
db446a08c aio: convert the ... |
619 620 621 622 623 624 625 |
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... |
626 |
table = rcu_dereference_raw(mm->ioctx_table); |
db446a08c aio: convert the ... |
627 628 629 630 |
while (1) { if (table) for (i = 0; i < table->nr; i++) |
cd21b3400 fs/aio: Use RCU a... |
631 |
if (!rcu_access_pointer(table->table[i])) { |
db446a08c aio: convert the ... |
632 |
ctx->id = i; |
cd21b3400 fs/aio: Use RCU a... |
633 |
rcu_assign_pointer(table->table[i], ctx); |
db446a08c aio: convert the ... |
634 |
spin_unlock(&mm->ioctx_lock); |
fa8a53c39 aio: v4 ensure ac... |
635 636 637 638 |
/* While kioctx setup is in progress, * we are protected from page migration * changes ring_pages by ->ring_lock. */ |
db446a08c aio: convert the ... |
639 640 641 642 643 644 645 |
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 ... |
646 647 648 649 650 651 652 653 654 655 |
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... |
656 |
old = rcu_dereference_raw(mm->ioctx_table); |
db446a08c aio: convert the ... |
657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 |
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... |
672 673 674 675 676 677 678 679 680 |
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 |
681 682 683 684 685 |
/* 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... |
686 |
struct mm_struct *mm = current->mm; |
1da177e4c Linux-2.6.12-rc2 |
687 |
struct kioctx *ctx; |
e23754f88 aio: don't bother... |
688 |
int err = -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
689 |
|
e1bdd5f27 aio: percpu reqs_... |
690 |
/* |
2a8a98673 fs: aio: fix the ... |
691 692 693 694 695 696 |
* 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 aio: percpu reqs_... |
697 698 699 700 701 702 703 704 705 706 |
* 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 |
707 |
/* Prevent overflows */ |
08397acdd ioctx_alloc(): re... |
708 |
if (nr_events > (0x10000000U / sizeof(struct io_event))) { |
1da177e4c Linux-2.6.12-rc2 |
709 710 711 712 |
pr_debug("ENOMEM: nr_events too high "); return ERR_PTR(-EINVAL); } |
2a8a98673 fs: aio: fix the ... |
713 |
if (!nr_events || (unsigned long)max_reqs > aio_max_nr) |
1da177e4c Linux-2.6.12-rc2 |
714 |
return ERR_PTR(-EAGAIN); |
c37622296 [PATCH] Transform... |
715 |
ctx = kmem_cache_zalloc(kioctx_cachep, GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
716 717 |
if (!ctx) return ERR_PTR(-ENOMEM); |
2a8a98673 fs: aio: fix the ... |
718 |
ctx->max_reqs = max_reqs; |
1da177e4c Linux-2.6.12-rc2 |
719 |
|
1da177e4c Linux-2.6.12-rc2 |
720 |
spin_lock_init(&ctx->ctx_lock); |
0460fef2a aio: use cancella... |
721 |
spin_lock_init(&ctx->completion_lock); |
58c85dc20 aio: kill struct ... |
722 |
mutex_init(&ctx->ring_lock); |
fa8a53c39 aio: v4 ensure ac... |
723 724 725 |
/* 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 |
726 727 728 |
init_waitqueue_head(&ctx->wait); INIT_LIST_HEAD(&ctx->active_reqs); |
1da177e4c Linux-2.6.12-rc2 |
729 |
|
2aad2a86f percpu_ref: add P... |
730 |
if (percpu_ref_init(&ctx->users, free_ioctx_users, 0, GFP_KERNEL)) |
fa8a53c39 aio: v4 ensure ac... |
731 |
goto err; |
2aad2a86f percpu_ref: add P... |
732 |
if (percpu_ref_init(&ctx->reqs, free_ioctx_reqs, 0, GFP_KERNEL)) |
fa8a53c39 aio: v4 ensure ac... |
733 |
goto err; |
e1bdd5f27 aio: percpu reqs_... |
734 735 |
ctx->cpu = alloc_percpu(struct kioctx_cpu); if (!ctx->cpu) |
e34ecee2a aio: Fix a trinit... |
736 |
goto err; |
1da177e4c Linux-2.6.12-rc2 |
737 |
|
2a8a98673 fs: aio: fix the ... |
738 |
err = aio_setup_ring(ctx, nr_events); |
fa8a53c39 aio: v4 ensure ac... |
739 |
if (err < 0) |
e34ecee2a aio: Fix a trinit... |
740 |
goto err; |
e1bdd5f27 aio: percpu reqs_... |
741 |
|
34e83fc61 aio: reqs_active ... |
742 |
atomic_set(&ctx->reqs_available, ctx->nr_events - 1); |
e1bdd5f27 aio: percpu reqs_... |
743 |
ctx->req_batch = (ctx->nr_events - 1) / (num_possible_cpus() * 4); |
6878ea72a aio: be defensive... |
744 745 |
if (ctx->req_batch < 1) ctx->req_batch = 1; |
34e83fc61 aio: reqs_active ... |
746 |
|
1da177e4c Linux-2.6.12-rc2 |
747 |
/* limit the number of system wide aios */ |
9fa1cb397 aio: aio_nr_lock ... |
748 |
spin_lock(&aio_nr_lock); |
2a8a98673 fs: aio: fix the ... |
749 750 |
if (aio_nr + ctx->max_reqs > aio_max_nr || aio_nr + ctx->max_reqs < aio_nr) { |
9fa1cb397 aio: aio_nr_lock ... |
751 |
spin_unlock(&aio_nr_lock); |
e34ecee2a aio: Fix a trinit... |
752 |
err = -EAGAIN; |
d1b943271 aio: clean up aio... |
753 |
goto err_ctx; |
2dd542b7a aio: aio_nr decre... |
754 755 |
} aio_nr += ctx->max_reqs; |
9fa1cb397 aio: aio_nr_lock ... |
756 |
spin_unlock(&aio_nr_lock); |
1da177e4c Linux-2.6.12-rc2 |
757 |
|
1881686f8 aio: fix kioctx l... |
758 759 |
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... |
760 |
|
da90382c2 aio: fix error ha... |
761 762 |
err = ioctx_add_table(ctx, mm); if (err) |
e34ecee2a aio: Fix a trinit... |
763 |
goto err_cleanup; |
da90382c2 aio: fix error ha... |
764 |
|
fa8a53c39 aio: v4 ensure ac... |
765 766 |
/* Release the ring_lock mutex now that all setup is complete. */ mutex_unlock(&ctx->ring_lock); |
caf4167aa aio: dprintk() ->... |
767 768 |
pr_debug("allocated ioctx %p[%ld]: mm=%p mask=0x%x ", |
58c85dc20 aio: kill struct ... |
769 |
ctx, ctx->user_id, mm, ctx->nr_events); |
1da177e4c Linux-2.6.12-rc2 |
770 |
return ctx; |
e34ecee2a aio: Fix a trinit... |
771 772 |
err_cleanup: aio_nr_sub(ctx->max_reqs); |
d1b943271 aio: clean up aio... |
773 |
err_ctx: |
deeb8525f ioctx_alloc(): fi... |
774 775 776 |
atomic_set(&ctx->dead, 1); if (ctx->mmap_size) vm_munmap(ctx->mmap_base, ctx->mmap_size); |
d1b943271 aio: clean up aio... |
777 |
aio_free_ring(ctx); |
e34ecee2a aio: Fix a trinit... |
778 |
err: |
fa8a53c39 aio: v4 ensure ac... |
779 |
mutex_unlock(&ctx->ring_lock); |
e1bdd5f27 aio: percpu reqs_... |
780 |
free_percpu(ctx->cpu); |
9a1049da9 percpu-refcount: ... |
781 782 |
percpu_ref_exit(&ctx->reqs); percpu_ref_exit(&ctx->users); |
1da177e4c Linux-2.6.12-rc2 |
783 |
kmem_cache_free(kioctx_cachep, ctx); |
caf4167aa aio: dprintk() ->... |
784 785 |
pr_debug("error allocating ioctx %d ", err); |
e23754f88 aio: don't bother... |
786 |
return ERR_PTR(err); |
1da177e4c Linux-2.6.12-rc2 |
787 |
} |
36f558890 aio: refcounting ... |
788 789 790 791 792 |
/* 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... |
793 |
static int kill_ioctx(struct mm_struct *mm, struct kioctx *ctx, |
dc48e56d7 aio: fix serial d... |
794 |
struct ctx_rq_wait *wait) |
36f558890 aio: refcounting ... |
795 |
{ |
fa88b6f88 aio: cleanup: fla... |
796 |
struct kioctx_table *table; |
db446a08c aio: convert the ... |
797 |
|
b2edffdd9 fix mremap() vs. ... |
798 799 800 |
spin_lock(&mm->ioctx_lock); if (atomic_xchg(&ctx->dead, 1)) { spin_unlock(&mm->ioctx_lock); |
fa88b6f88 aio: cleanup: fla... |
801 |
return -EINVAL; |
b2edffdd9 fix mremap() vs. ... |
802 |
} |
db446a08c aio: convert the ... |
803 |
|
855ef0dec aio: kill the mis... |
804 |
table = rcu_dereference_raw(mm->ioctx_table); |
cd21b3400 fs/aio: Use RCU a... |
805 806 |
WARN_ON(ctx != rcu_access_pointer(table->table[ctx->id])); RCU_INIT_POINTER(table->table[ctx->id], NULL); |
fa88b6f88 aio: cleanup: fla... |
807 |
spin_unlock(&mm->ioctx_lock); |
4fcc712f5 aio: fix io_destr... |
808 |
|
076c7c068 fs/aio: Add expli... |
809 |
/* free_ioctx_reqs() will do the necessary RCU synchronization */ |
fa88b6f88 aio: cleanup: fla... |
810 |
wake_up_all(&ctx->wait); |
4fcc712f5 aio: fix io_destr... |
811 |
|
fa88b6f88 aio: cleanup: fla... |
812 813 814 815 816 817 818 819 |
/* * 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... |
820 |
|
fa88b6f88 aio: cleanup: fla... |
821 822 |
if (ctx->mmap_size) vm_munmap(ctx->mmap_base, ctx->mmap_size); |
fb2d44838 aio: report error... |
823 |
|
dc48e56d7 aio: fix serial d... |
824 |
ctx->rq_wait = wait; |
fa88b6f88 aio: cleanup: fla... |
825 826 |
percpu_ref_kill(&ctx->users); return 0; |
1da177e4c Linux-2.6.12-rc2 |
827 |
} |
36f558890 aio: refcounting ... |
828 829 830 831 832 833 834 |
/* * 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 |
835 |
*/ |
fc9b52cd8 fs: remove fastca... |
836 |
void exit_aio(struct mm_struct *mm) |
1da177e4c Linux-2.6.12-rc2 |
837 |
{ |
4b70ac5fd aio: change exit_... |
838 |
struct kioctx_table *table = rcu_dereference_raw(mm->ioctx_table); |
dc48e56d7 aio: fix serial d... |
839 840 |
struct ctx_rq_wait wait; int i, skipped; |
db446a08c aio: convert the ... |
841 |
|
4b70ac5fd aio: change exit_... |
842 843 |
if (!table) return; |
db446a08c aio: convert the ... |
844 |
|
dc48e56d7 aio: fix serial d... |
845 846 847 848 |
atomic_set(&wait.count, table->nr); init_completion(&wait.comp); skipped = 0; |
4b70ac5fd aio: change exit_... |
849 |
for (i = 0; i < table->nr; ++i) { |
cd21b3400 fs/aio: Use RCU a... |
850 851 |
struct kioctx *ctx = rcu_dereference_protected(table->table[i], true); |
abf137dd7 aio: make the loo... |
852 |
|
dc48e56d7 aio: fix serial d... |
853 854 |
if (!ctx) { skipped++; |
4b70ac5fd aio: change exit_... |
855 |
continue; |
dc48e56d7 aio: fix serial d... |
856 |
} |
936af1576 aio: don't bother... |
857 |
/* |
4b70ac5fd aio: change exit_... |
858 859 860 861 862 |
* 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... |
863 |
*/ |
58c85dc20 aio: kill struct ... |
864 |
ctx->mmap_size = 0; |
dc48e56d7 aio: fix serial d... |
865 866 |
kill_ioctx(mm, ctx, &wait); } |
36f558890 aio: refcounting ... |
867 |
|
dc48e56d7 aio: fix serial d... |
868 |
if (!atomic_sub_and_test(skipped, &wait.count)) { |
6098b45b3 aio: block exit_a... |
869 |
/* Wait until all IO for the context are done. */ |
dc48e56d7 aio: fix serial d... |
870 |
wait_for_completion(&wait.comp); |
1da177e4c Linux-2.6.12-rc2 |
871 |
} |
4b70ac5fd aio: change exit_... |
872 873 874 |
RCU_INIT_POINTER(mm->ioctx_table, NULL); kfree(table); |
1da177e4c Linux-2.6.12-rc2 |
875 |
} |
e1bdd5f27 aio: percpu reqs_... |
876 877 878 |
static void put_reqs_available(struct kioctx *ctx, unsigned nr) { struct kioctx_cpu *kcpu; |
263782c1c aio: protect reqs... |
879 |
unsigned long flags; |
e1bdd5f27 aio: percpu reqs_... |
880 |
|
263782c1c aio: protect reqs... |
881 |
local_irq_save(flags); |
be6fb451a aio: remove no lo... |
882 |
kcpu = this_cpu_ptr(ctx->cpu); |
e1bdd5f27 aio: percpu reqs_... |
883 |
kcpu->reqs_available += nr; |
263782c1c aio: protect reqs... |
884 |
|
e1bdd5f27 aio: percpu reqs_... |
885 886 887 888 |
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... |
889 |
local_irq_restore(flags); |
e1bdd5f27 aio: percpu reqs_... |
890 891 892 893 894 895 |
} static bool get_reqs_available(struct kioctx *ctx) { struct kioctx_cpu *kcpu; bool ret = false; |
263782c1c aio: protect reqs... |
896 |
unsigned long flags; |
e1bdd5f27 aio: percpu reqs_... |
897 |
|
263782c1c aio: protect reqs... |
898 |
local_irq_save(flags); |
be6fb451a aio: remove no lo... |
899 |
kcpu = this_cpu_ptr(ctx->cpu); |
e1bdd5f27 aio: percpu reqs_... |
900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 |
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... |
918 |
local_irq_restore(flags); |
e1bdd5f27 aio: percpu reqs_... |
919 920 |
return ret; } |
d856f32a8 aio: fix reqs_ava... |
921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 |
/* 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 |
982 |
/* aio_get_req |
57282d8fd aio: Kill ki_users |
983 984 |
* Allocate a slot for an aio request. * Returns NULL if no requests are free. |
1da177e4c Linux-2.6.12-rc2 |
985 |
*/ |
04b2fa9f8 fs: split generic... |
986 |
static inline struct aio_kiocb *aio_get_req(struct kioctx *ctx) |
1da177e4c Linux-2.6.12-rc2 |
987 |
{ |
04b2fa9f8 fs: split generic... |
988 |
struct aio_kiocb *req; |
a1c8eae75 aio: kill batch a... |
989 |
|
d856f32a8 aio: fix reqs_ava... |
990 991 992 993 994 |
if (!get_reqs_available(ctx)) { user_refill_reqs_available(ctx); if (!get_reqs_available(ctx)) return NULL; } |
a1c8eae75 aio: kill batch a... |
995 |
|
0460fef2a aio: use cancella... |
996 |
req = kmem_cache_alloc(kiocb_cachep, GFP_KERNEL|__GFP_ZERO); |
1da177e4c Linux-2.6.12-rc2 |
997 |
if (unlikely(!req)) |
a1c8eae75 aio: kill batch a... |
998 |
goto out_put; |
1da177e4c Linux-2.6.12-rc2 |
999 |
|
e34ecee2a aio: Fix a trinit... |
1000 |
percpu_ref_get(&ctx->reqs); |
1da177e4c Linux-2.6.12-rc2 |
1001 |
req->ki_ctx = ctx; |
080d676de aio: allocate kio... |
1002 |
return req; |
a1c8eae75 aio: kill batch a... |
1003 |
out_put: |
e1bdd5f27 aio: percpu reqs_... |
1004 |
put_reqs_available(ctx, 1); |
a1c8eae75 aio: kill batch a... |
1005 |
return NULL; |
1da177e4c Linux-2.6.12-rc2 |
1006 |
} |
04b2fa9f8 fs: split generic... |
1007 |
static void kiocb_free(struct aio_kiocb *req) |
1da177e4c Linux-2.6.12-rc2 |
1008 |
{ |
04b2fa9f8 fs: split generic... |
1009 1010 |
if (req->common.ki_filp) fput(req->common.ki_filp); |
133890103 eventfd: revised ... |
1011 1012 |
if (req->ki_eventfd != NULL) eventfd_ctx_put(req->ki_eventfd); |
1da177e4c Linux-2.6.12-rc2 |
1013 |
kmem_cache_free(kiocb_cachep, req); |
1da177e4c Linux-2.6.12-rc2 |
1014 |
} |
d5470b596 fs/aio.c: make 3 ... |
1015 |
static struct kioctx *lookup_ioctx(unsigned long ctx_id) |
1da177e4c Linux-2.6.12-rc2 |
1016 |
{ |
db446a08c aio: convert the ... |
1017 |
struct aio_ring __user *ring = (void __user *)ctx_id; |
abf137dd7 aio: make the loo... |
1018 |
struct mm_struct *mm = current->mm; |
65c24491b aio: lookup_ioctx... |
1019 |
struct kioctx *ctx, *ret = NULL; |
db446a08c aio: convert the ... |
1020 1021 1022 1023 1024 |
struct kioctx_table *table; unsigned id; if (get_user(id, &ring->id)) return NULL; |
1da177e4c Linux-2.6.12-rc2 |
1025 |
|
abf137dd7 aio: make the loo... |
1026 |
rcu_read_lock(); |
db446a08c aio: convert the ... |
1027 |
table = rcu_dereference(mm->ioctx_table); |
abf137dd7 aio: make the loo... |
1028 |
|
db446a08c aio: convert the ... |
1029 1030 |
if (!table || id >= table->nr) goto out; |
1da177e4c Linux-2.6.12-rc2 |
1031 |
|
cd21b3400 fs/aio: Use RCU a... |
1032 |
ctx = rcu_dereference(table->table[id]); |
f30d704fe aio: table lookup... |
1033 |
if (ctx && ctx->user_id == ctx_id) { |
fbcede36b aio: fix io_destr... |
1034 1035 |
if (percpu_ref_tryget_live(&ctx->users)) ret = ctx; |
db446a08c aio: convert the ... |
1036 1037 |
} out: |
abf137dd7 aio: make the loo... |
1038 |
rcu_read_unlock(); |
65c24491b aio: lookup_ioctx... |
1039 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1040 |
} |
1da177e4c Linux-2.6.12-rc2 |
1041 1042 |
/* aio_complete * Called when the io request on the given iocb is complete. |
1da177e4c Linux-2.6.12-rc2 |
1043 |
*/ |
04b2fa9f8 fs: split generic... |
1044 |
static void aio_complete(struct kiocb *kiocb, long res, long res2) |
1da177e4c Linux-2.6.12-rc2 |
1045 |
{ |
04b2fa9f8 fs: split generic... |
1046 |
struct aio_kiocb *iocb = container_of(kiocb, struct aio_kiocb, common); |
1da177e4c Linux-2.6.12-rc2 |
1047 |
struct kioctx *ctx = iocb->ki_ctx; |
1da177e4c Linux-2.6.12-rc2 |
1048 |
struct aio_ring *ring; |
21b40200c aio: use flush_dc... |
1049 |
struct io_event *ev_page, *event; |
d856f32a8 aio: fix reqs_ava... |
1050 |
unsigned tail, pos, head; |
1da177e4c Linux-2.6.12-rc2 |
1051 |
unsigned long flags; |
1da177e4c Linux-2.6.12-rc2 |
1052 |
|
70fe2f481 aio: fix freeze p... |
1053 1054 1055 1056 1057 1058 1059 |
if (kiocb->ki_flags & IOCB_WRITE) { struct file *file = kiocb->ki_filp; /* * Tell lockdep we inherited freeze protection from submission * thread. */ |
a12f1ae61 aio: fix lock dep... |
1060 1061 |
if (S_ISREG(file_inode(file)->i_mode)) __sb_writers_acquired(file_inode(file)->i_sb, SB_FREEZE_WRITE); |
70fe2f481 aio: fix freeze p... |
1062 1063 |
file_end_write(file); } |
20dcae324 [PATCH] aio: remo... |
1064 1065 1066 1067 1068 1069 |
/* * 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 |
1070 |
*/ |
04b2fa9f8 fs: split generic... |
1071 |
BUG_ON(is_sync_kiocb(kiocb)); |
1da177e4c Linux-2.6.12-rc2 |
1072 |
|
0460fef2a aio: use cancella... |
1073 1074 1075 1076 1077 1078 1079 |
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... |
1080 |
|
1da177e4c Linux-2.6.12-rc2 |
1081 |
/* |
0460fef2a aio: use cancella... |
1082 |
* Add a completion event to the ring buffer. Must be done holding |
4b30f07e7 aio: fix wrong co... |
1083 |
* ctx->completion_lock to prevent other code from messing with the tail |
0460fef2a aio: use cancella... |
1084 1085 1086 |
* pointer since we might be called from irq context. */ spin_lock_irqsave(&ctx->completion_lock, flags); |
58c85dc20 aio: kill struct ... |
1087 |
tail = ctx->tail; |
21b40200c aio: use flush_dc... |
1088 |
pos = tail + AIO_EVENTS_OFFSET; |
58c85dc20 aio: kill struct ... |
1089 |
if (++tail >= ctx->nr_events) |
4bf69b2a0 [PATCH] aio: ring... |
1090 |
tail = 0; |
1da177e4c Linux-2.6.12-rc2 |
1091 |
|
58c85dc20 aio: kill struct ... |
1092 |
ev_page = kmap_atomic(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); |
21b40200c aio: use flush_dc... |
1093 |
event = ev_page + pos % AIO_EVENTS_PER_PAGE; |
04b2fa9f8 fs: split generic... |
1094 |
event->obj = (u64)(unsigned long)iocb->ki_user_iocb; |
1da177e4c Linux-2.6.12-rc2 |
1095 1096 1097 |
event->data = iocb->ki_user_data; event->res = res; event->res2 = res2; |
21b40200c aio: use flush_dc... |
1098 |
kunmap_atomic(ev_page); |
58c85dc20 aio: kill struct ... |
1099 |
flush_dcache_page(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); |
21b40200c aio: use flush_dc... |
1100 1101 1102 |
pr_debug("%p[%u]: %p: %p %Lx %lx %lx ", |
04b2fa9f8 fs: split generic... |
1103 |
ctx, tail, iocb, iocb->ki_user_iocb, iocb->ki_user_data, |
caf4167aa aio: dprintk() ->... |
1104 |
res, res2); |
1da177e4c Linux-2.6.12-rc2 |
1105 1106 1107 1108 1109 |
/* 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 ... |
1110 |
ctx->tail = tail; |
1da177e4c Linux-2.6.12-rc2 |
1111 |
|
58c85dc20 aio: kill struct ... |
1112 |
ring = kmap_atomic(ctx->ring_pages[0]); |
d856f32a8 aio: fix reqs_ava... |
1113 |
head = ring->head; |
21b40200c aio: use flush_dc... |
1114 |
ring->tail = tail; |
e8e3c3d66 fs: remove the se... |
1115 |
kunmap_atomic(ring); |
58c85dc20 aio: kill struct ... |
1116 |
flush_dcache_page(ctx->ring_pages[0]); |
1da177e4c Linux-2.6.12-rc2 |
1117 |
|
d856f32a8 aio: fix reqs_ava... |
1118 1119 1120 |
ctx->completed_events++; if (ctx->completed_events > 1) refill_reqs_available(ctx, head, tail); |
0460fef2a aio: use cancella... |
1121 |
spin_unlock_irqrestore(&ctx->completion_lock, flags); |
21b40200c aio: use flush_dc... |
1122 1123 |
pr_debug("added to ring %p at [%u] ", iocb, tail); |
8d1c98b0b eventfd/kaio inte... |
1124 1125 1126 1127 1128 1129 |
/* * 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... |
1130 |
if (iocb->ki_eventfd != NULL) |
8d1c98b0b eventfd/kaio inte... |
1131 |
eventfd_signal(iocb->ki_eventfd, 1); |
1da177e4c Linux-2.6.12-rc2 |
1132 |
/* everything turned out well, dispose of the aiocb. */ |
57282d8fd aio: Kill ki_users |
1133 |
kiocb_free(iocb); |
1da177e4c Linux-2.6.12-rc2 |
1134 |
|
6cb2a2104 aio: bad AIO race... |
1135 1136 1137 1138 1139 1140 1141 |
/* * 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 |
1142 1143 |
if (waitqueue_active(&ctx->wait)) wake_up(&ctx->wait); |
e34ecee2a aio: Fix a trinit... |
1144 |
percpu_ref_put(&ctx->reqs); |
1da177e4c Linux-2.6.12-rc2 |
1145 |
} |
2be4e7dee aio: fix some com... |
1146 |
/* aio_read_events_ring |
a31ad380b aio: make aio_rea... |
1147 1148 |
* Pull an event off of the ioctx's event ring. Returns the number of * events fetched |
1da177e4c Linux-2.6.12-rc2 |
1149 |
*/ |
a31ad380b aio: make aio_rea... |
1150 1151 |
static long aio_read_events_ring(struct kioctx *ctx, struct io_event __user *event, long nr) |
1da177e4c Linux-2.6.12-rc2 |
1152 |
{ |
1da177e4c Linux-2.6.12-rc2 |
1153 |
struct aio_ring *ring; |
5ffac122d aio: Don't use ct... |
1154 |
unsigned head, tail, pos; |
a31ad380b aio: make aio_rea... |
1155 1156 |
long ret = 0; int copy_ret; |
9c9ce763b aio: annotate aio... |
1157 1158 1159 1160 1161 1162 1163 |
/* * 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 ... |
1164 |
mutex_lock(&ctx->ring_lock); |
1da177e4c Linux-2.6.12-rc2 |
1165 |
|
fa8a53c39 aio: v4 ensure ac... |
1166 |
/* Access to ->ring_pages here is protected by ctx->ring_lock. */ |
58c85dc20 aio: kill struct ... |
1167 |
ring = kmap_atomic(ctx->ring_pages[0]); |
a31ad380b aio: make aio_rea... |
1168 |
head = ring->head; |
5ffac122d aio: Don't use ct... |
1169 |
tail = ring->tail; |
a31ad380b aio: make aio_rea... |
1170 |
kunmap_atomic(ring); |
2ff396be6 aio: add missing ... |
1171 1172 1173 1174 1175 |
/* * 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... |
1176 1177 |
pr_debug("h%u t%u m%u ", head, tail, ctx->nr_events); |
1da177e4c Linux-2.6.12-rc2 |
1178 |
|
5ffac122d aio: Don't use ct... |
1179 |
if (head == tail) |
1da177e4c Linux-2.6.12-rc2 |
1180 |
goto out; |
edfbbf388 aio: fix kernel m... |
1181 1182 |
head %= ctx->nr_events; tail %= ctx->nr_events; |
a31ad380b aio: make aio_rea... |
1183 1184 1185 1186 |
while (ret < nr) { long avail; struct io_event *ev; struct page *page; |
5ffac122d aio: Don't use ct... |
1187 1188 |
avail = (head <= tail ? tail : ctx->nr_events) - head; if (head == tail) |
a31ad380b aio: make aio_rea... |
1189 1190 1191 1192 1193 1194 1195 |
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 ... |
1196 |
page = ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]; |
a31ad380b aio: make aio_rea... |
1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 |
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 ... |
1211 |
head %= ctx->nr_events; |
1da177e4c Linux-2.6.12-rc2 |
1212 |
} |
1da177e4c Linux-2.6.12-rc2 |
1213 |
|
58c85dc20 aio: kill struct ... |
1214 |
ring = kmap_atomic(ctx->ring_pages[0]); |
a31ad380b aio: make aio_rea... |
1215 |
ring->head = head; |
91d80a84b aio: fix possible... |
1216 |
kunmap_atomic(ring); |
58c85dc20 aio: kill struct ... |
1217 |
flush_dcache_page(ctx->ring_pages[0]); |
a31ad380b aio: make aio_rea... |
1218 |
|
5ffac122d aio: Don't use ct... |
1219 1220 |
pr_debug("%li h%u t%u ", ret, head, tail); |
a31ad380b aio: make aio_rea... |
1221 |
out: |
58c85dc20 aio: kill struct ... |
1222 |
mutex_unlock(&ctx->ring_lock); |
a31ad380b aio: make aio_rea... |
1223 |
|
1da177e4c Linux-2.6.12-rc2 |
1224 1225 |
return ret; } |
a31ad380b aio: make aio_rea... |
1226 1227 |
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 |
1228 |
{ |
a31ad380b aio: make aio_rea... |
1229 |
long ret = aio_read_events_ring(ctx, event + *i, nr - *i); |
1da177e4c Linux-2.6.12-rc2 |
1230 |
|
a31ad380b aio: make aio_rea... |
1231 1232 |
if (ret > 0) *i += ret; |
1da177e4c Linux-2.6.12-rc2 |
1233 |
|
a31ad380b aio: make aio_rea... |
1234 1235 |
if (unlikely(atomic_read(&ctx->dead))) ret = -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1236 |
|
a31ad380b aio: make aio_rea... |
1237 1238 |
if (!*i) *i = ret; |
1da177e4c Linux-2.6.12-rc2 |
1239 |
|
a31ad380b aio: make aio_rea... |
1240 |
return ret < 0 || *i >= min_nr; |
1da177e4c Linux-2.6.12-rc2 |
1241 |
} |
a31ad380b aio: make aio_rea... |
1242 |
static long read_events(struct kioctx *ctx, long min_nr, long nr, |
1da177e4c Linux-2.6.12-rc2 |
1243 1244 1245 |
struct io_event __user *event, struct timespec __user *timeout) { |
2456e8553 ktime: Get rid of... |
1246 |
ktime_t until = KTIME_MAX; |
a31ad380b aio: make aio_rea... |
1247 |
long ret = 0; |
1da177e4c Linux-2.6.12-rc2 |
1248 |
|
1da177e4c Linux-2.6.12-rc2 |
1249 1250 |
if (timeout) { struct timespec ts; |
a31ad380b aio: make aio_rea... |
1251 |
|
1da177e4c Linux-2.6.12-rc2 |
1252 |
if (unlikely(copy_from_user(&ts, timeout, sizeof(ts)))) |
a31ad380b aio: make aio_rea... |
1253 |
return -EFAULT; |
1da177e4c Linux-2.6.12-rc2 |
1254 |
|
a31ad380b aio: make aio_rea... |
1255 |
until = timespec_to_ktime(ts); |
1da177e4c Linux-2.6.12-rc2 |
1256 |
} |
a31ad380b aio: make aio_rea... |
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 |
/* * 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 ktime: Get rid of... |
1271 |
if (until == 0) |
5f785de58 aio: Skip timer f... |
1272 1273 1274 1275 1276 |
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 |
1277 |
|
a31ad380b aio: make aio_rea... |
1278 1279 |
if (!ret && signal_pending(current)) ret = -EINTR; |
1da177e4c Linux-2.6.12-rc2 |
1280 |
|
a31ad380b aio: make aio_rea... |
1281 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1282 |
} |
1da177e4c Linux-2.6.12-rc2 |
1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 |
/* 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... |
1296 |
SYSCALL_DEFINE2(io_setup, unsigned, nr_events, aio_context_t __user *, ctxp) |
1da177e4c Linux-2.6.12-rc2 |
1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 |
{ 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... |
1307 |
if (unlikely(ctx || nr_events == 0)) { |
acd88d4e1 fs/aio.c: Remove ... |
1308 1309 |
pr_debug("EINVAL: ctx %lu nr_events %u ", |
d55b5fdaf [PATCH] aio: remo... |
1310 |
ctx, nr_events); |
1da177e4c Linux-2.6.12-rc2 |
1311 1312 1313 1314 1315 1316 1317 |
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... |
1318 |
if (ret) |
e02ba72aa aio: block io_des... |
1319 |
kill_ioctx(current->mm, ioctx, NULL); |
723be6e39 aio: percpu ioctx... |
1320 |
percpu_ref_put(&ioctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1321 1322 1323 1324 1325 |
} out: return ret; } |
c00d2c7e8 move aio compat t... |
1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 |
#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 Linux-2.6.12-rc2 |
1359 1360 1361 |
/* 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... |
1362 |
* implemented. May fail with -EINVAL if the context pointed to |
1da177e4c Linux-2.6.12-rc2 |
1363 1364 |
* is invalid. */ |
002c8976e [CVE-2009-0029] S... |
1365 |
SYSCALL_DEFINE1(io_destroy, aio_context_t, ctx) |
1da177e4c Linux-2.6.12-rc2 |
1366 1367 1368 |
{ struct kioctx *ioctx = lookup_ioctx(ctx); if (likely(NULL != ioctx)) { |
dc48e56d7 aio: fix serial d... |
1369 |
struct ctx_rq_wait wait; |
fb2d44838 aio: report error... |
1370 |
int ret; |
e02ba72aa aio: block io_des... |
1371 |
|
dc48e56d7 aio: fix serial d... |
1372 1373 |
init_completion(&wait.comp); atomic_set(&wait.count, 1); |
e02ba72aa aio: block io_des... |
1374 1375 1376 1377 |
/* 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... |
1378 |
ret = kill_ioctx(current->mm, ioctx, &wait); |
723be6e39 aio: percpu ioctx... |
1379 |
percpu_ref_put(&ioctx->users); |
e02ba72aa aio: block io_des... |
1380 1381 1382 1383 1384 |
/* 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... |
1385 |
if (!ret) |
dc48e56d7 aio: fix serial d... |
1386 |
wait_for_completion(&wait.comp); |
e02ba72aa aio: block io_des... |
1387 |
|
fb2d44838 aio: report error... |
1388 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1389 |
} |
acd88d4e1 fs/aio.c: Remove ... |
1390 1391 |
pr_debug("EINVAL: invalid context id "); |
1da177e4c Linux-2.6.12-rc2 |
1392 1393 |
return -EINVAL; } |
89319d31d fs: remove aio_ru... |
1394 1395 |
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... |
1396 |
{ |
89319d31d fs: remove aio_ru... |
1397 1398 1399 1400 1401 1402 1403 1404 |
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... |
1405 1406 |
#ifdef CONFIG_COMPAT if (compat) |
89319d31d fs: remove aio_ru... |
1407 1408 |
return compat_import_iovec(rw, buf, len, UIO_FASTIOV, iovec, iter); |
9d85cba71 aio: fix the comp... |
1409 |
#endif |
89319d31d fs: remove aio_ru... |
1410 |
return import_iovec(rw, buf, len, UIO_FASTIOV, iovec, iter); |
eed4e51fb [PATCH] Add vecto... |
1411 |
} |
89319d31d fs: remove aio_ru... |
1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 |
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 |
1435 |
{ |
41ef4eb8e aio: kill ki_retry |
1436 |
struct file *file = req->ki_filp; |
00fefb9cf aio: use iovec ar... |
1437 |
struct iovec inline_vecs[UIO_FASTIOV], *iovec = inline_vecs; |
293bc9822 new methods: ->re... |
1438 |
struct iov_iter iter; |
89319d31d fs: remove aio_ru... |
1439 |
ssize_t ret; |
1da177e4c Linux-2.6.12-rc2 |
1440 |
|
89319d31d fs: remove aio_ru... |
1441 1442 1443 1444 |
if (unlikely(!(file->f_mode & FMODE_READ))) return -EBADF; if (unlikely(!file->f_op->read_iter)) return -EINVAL; |
73a7075e3 aio: Kill aio_rw_... |
1445 |
|
89319d31d fs: remove aio_ru... |
1446 1447 1448 1449 1450 |
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) |
bb7462b6f vfs: use helpers ... |
1451 |
ret = aio_ret(req, call_read_iter(file, req, &iter)); |
89319d31d fs: remove aio_ru... |
1452 1453 1454 |
kfree(iovec); return ret; } |
73a7075e3 aio: Kill aio_rw_... |
1455 |
|
89319d31d fs: remove aio_ru... |
1456 1457 1458 1459 1460 1461 1462 |
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 |
1463 |
|
89319d31d fs: remove aio_ru... |
1464 1465 1466 |
if (unlikely(!(file->f_mode & FMODE_WRITE))) return -EBADF; if (unlikely(!file->f_op->write_iter)) |
41ef4eb8e aio: kill ki_retry |
1467 |
return -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1468 |
|
89319d31d fs: remove aio_ru... |
1469 1470 1471 1472 1473 |
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... |
1474 |
req->ki_flags |= IOCB_WRITE; |
89319d31d fs: remove aio_ru... |
1475 |
file_start_write(file); |
bb7462b6f vfs: use helpers ... |
1476 |
ret = aio_ret(req, call_write_iter(file, req, &iter)); |
70fe2f481 aio: fix freeze p... |
1477 1478 1479 1480 1481 |
/* * 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. */ |
a12f1ae61 aio: fix lock dep... |
1482 1483 |
if (S_ISREG(file_inode(file)->i_mode)) __sb_writers_release(file_inode(file)->i_sb, SB_FREEZE_WRITE); |
41ef4eb8e aio: kill ki_retry |
1484 |
} |
89319d31d fs: remove aio_ru... |
1485 1486 |
kfree(iovec); return ret; |
1da177e4c Linux-2.6.12-rc2 |
1487 |
} |
d5470b596 fs/aio.c: make 3 ... |
1488 |
static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb, |
a1c8eae75 aio: kill batch a... |
1489 |
struct iocb *iocb, bool compat) |
1da177e4c Linux-2.6.12-rc2 |
1490 |
{ |
04b2fa9f8 fs: split generic... |
1491 |
struct aio_kiocb *req; |
89319d31d fs: remove aio_ru... |
1492 |
struct file *file; |
1da177e4c Linux-2.6.12-rc2 |
1493 1494 1495 |
ssize_t ret; /* enforce forwards compatibility on users */ |
9830f4be1 fs: Use RWF_* fla... |
1496 |
if (unlikely(iocb->aio_reserved2)) { |
caf4167aa aio: dprintk() ->... |
1497 1498 |
pr_debug("EINVAL: reserve field set "); |
1da177e4c Linux-2.6.12-rc2 |
1499 1500 1501 1502 1503 1504 1505 1506 1507 |
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 ... |
1508 1509 |
pr_debug("EINVAL: overflow check "); |
1da177e4c Linux-2.6.12-rc2 |
1510 1511 |
return -EINVAL; } |
41ef4eb8e aio: kill ki_retry |
1512 |
req = aio_get_req(ctx); |
1d98ebfcc aio: do fget() af... |
1513 |
if (unlikely(!req)) |
1da177e4c Linux-2.6.12-rc2 |
1514 |
return -EAGAIN; |
1d98ebfcc aio: do fget() af... |
1515 |
|
89319d31d fs: remove aio_ru... |
1516 |
req->common.ki_filp = file = fget(iocb->aio_fildes); |
04b2fa9f8 fs: split generic... |
1517 |
if (unlikely(!req->common.ki_filp)) { |
1d98ebfcc aio: do fget() af... |
1518 1519 |
ret = -EBADF; goto out_put_req; |
1da177e4c Linux-2.6.12-rc2 |
1520 |
} |
04b2fa9f8 fs: split generic... |
1521 1522 |
req->common.ki_pos = iocb->aio_offset; req->common.ki_complete = aio_complete; |
2ba48ce51 mirror O_APPEND a... |
1523 |
req->common.ki_flags = iocb_flags(req->common.ki_filp); |
45d06cf70 fs: add O_DIRECT ... |
1524 |
req->common.ki_hint = file_write_hint(file); |
1d98ebfcc aio: do fget() af... |
1525 |
|
9c3060bed signal/timer/even... |
1526 1527 1528 1529 1530 1531 1532 |
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 ... |
1533 |
req->ki_eventfd = eventfd_ctx_fdget((int) iocb->aio_resfd); |
801678c5a Remove duplicated... |
1534 |
if (IS_ERR(req->ki_eventfd)) { |
9c3060bed signal/timer/even... |
1535 |
ret = PTR_ERR(req->ki_eventfd); |
87c3a86e1 eventfd: remove f... |
1536 |
req->ki_eventfd = NULL; |
9c3060bed signal/timer/even... |
1537 1538 |
goto out_put_req; } |
04b2fa9f8 fs: split generic... |
1539 1540 |
req->common.ki_flags |= IOCB_EVENTFD; |
9c3060bed signal/timer/even... |
1541 |
} |
1da177e4c Linux-2.6.12-rc2 |
1542 |
|
9830f4be1 fs: Use RWF_* fla... |
1543 1544 1545 1546 1547 1548 |
ret = kiocb_set_rw_flags(&req->common, iocb->aio_rw_flags); if (unlikely(ret)) { pr_debug("EINVAL: aio_rw_flags "); goto out_put_req; } |
8a6608907 aio: kill ki_key |
1549 |
ret = put_user(KIOCB_KEY, &user_iocb->aio_key); |
1da177e4c Linux-2.6.12-rc2 |
1550 |
if (unlikely(ret)) { |
caf4167aa aio: dprintk() ->... |
1551 1552 |
pr_debug("EFAULT: aio_key "); |
1da177e4c Linux-2.6.12-rc2 |
1553 1554 |
goto out_put_req; } |
04b2fa9f8 fs: split generic... |
1555 |
req->ki_user_iocb = user_iocb; |
1da177e4c Linux-2.6.12-rc2 |
1556 |
req->ki_user_data = iocb->aio_data; |
1da177e4c Linux-2.6.12-rc2 |
1557 |
|
89319d31d fs: remove aio_ru... |
1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 |
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 |
1579 |
|
89319d31d fs: remove aio_ru... |
1580 1581 |
if (ret && ret != -EIOCBQUEUED) goto out_put_req; |
1da177e4c Linux-2.6.12-rc2 |
1582 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
1583 |
out_put_req: |
e1bdd5f27 aio: percpu reqs_... |
1584 |
put_reqs_available(ctx, 1); |
e34ecee2a aio: Fix a trinit... |
1585 |
percpu_ref_put(&ctx->reqs); |
57282d8fd aio: Kill ki_users |
1586 |
kiocb_free(req); |
1da177e4c Linux-2.6.12-rc2 |
1587 1588 |
return ret; } |
c00d2c7e8 move aio compat t... |
1589 1590 |
static long do_io_submit(aio_context_t ctx_id, long nr, struct iocb __user *__user *iocbpp, bool compat) |
1da177e4c Linux-2.6.12-rc2 |
1591 1592 1593 |
{ struct kioctx *ctx; long ret = 0; |
080d676de aio: allocate kio... |
1594 |
int i = 0; |
9f5b94254 fs: make aio plug |
1595 |
struct blk_plug plug; |
1da177e4c Linux-2.6.12-rc2 |
1596 1597 1598 |
if (unlikely(nr < 0)) return -EINVAL; |
75e1c70fc aio: check for mu... |
1599 1600 |
if (unlikely(nr > LONG_MAX/sizeof(*iocbpp))) nr = LONG_MAX/sizeof(*iocbpp); |
1da177e4c Linux-2.6.12-rc2 |
1601 1602 1603 1604 1605 |
if (unlikely(!access_ok(VERIFY_READ, iocbpp, (nr*sizeof(*iocbpp))))) return -EFAULT; ctx = lookup_ioctx(ctx_id); if (unlikely(!ctx)) { |
caf4167aa aio: dprintk() ->... |
1606 1607 |
pr_debug("EINVAL: invalid context id "); |
1da177e4c Linux-2.6.12-rc2 |
1608 1609 |
return -EINVAL; } |
9f5b94254 fs: make aio plug |
1610 |
blk_start_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 |
/* * 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... |
1628 |
ret = io_submit_one(ctx, user_iocb, &tmp, compat); |
1da177e4c Linux-2.6.12-rc2 |
1629 1630 1631 |
if (ret) break; } |
9f5b94254 fs: make aio plug |
1632 |
blk_finish_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
1633 |
|
723be6e39 aio: percpu ioctx... |
1634 |
percpu_ref_put(&ctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1635 1636 |
return i ? i : ret; } |
9d85cba71 aio: fix the comp... |
1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 |
/* 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); } |
c00d2c7e8 move aio compat t... |
1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 |
#ifdef CONFIG_COMPAT static inline long copy_iocb(long nr, u32 __user *ptr32, struct iocb __user * __user *ptr64) { compat_uptr_t uptr; int i; for (i = 0; i < nr; ++i) { if (get_user(uptr, ptr32 + i)) return -EFAULT; if (put_user(compat_ptr(uptr), ptr64 + i)) return -EFAULT; } return 0; } #define MAX_AIO_SUBMITS (PAGE_SIZE/sizeof(struct iocb *)) COMPAT_SYSCALL_DEFINE3(io_submit, compat_aio_context_t, ctx_id, int, nr, u32 __user *, iocb) { struct iocb __user * __user *iocb64; long ret; if (unlikely(nr < 0)) return -EINVAL; if (nr > MAX_AIO_SUBMITS) nr = MAX_AIO_SUBMITS; iocb64 = compat_alloc_user_space(nr * sizeof(*iocb64)); ret = copy_iocb(nr, iocb, iocb64); if (!ret) ret = do_io_submit(ctx_id, nr, iocb64, 1); return ret; } #endif |
1da177e4c Linux-2.6.12-rc2 |
1691 1692 |
/* lookup_kiocb * Finds a given iocb for cancellation. |
1da177e4c Linux-2.6.12-rc2 |
1693 |
*/ |
04b2fa9f8 fs: split generic... |
1694 1695 |
static struct aio_kiocb * lookup_kiocb(struct kioctx *ctx, struct iocb __user *iocb, u32 key) |
1da177e4c Linux-2.6.12-rc2 |
1696 |
{ |
04b2fa9f8 fs: split generic... |
1697 |
struct aio_kiocb *kiocb; |
d00689af6 [PATCH] aio: repl... |
1698 1699 |
assert_spin_locked(&ctx->ctx_lock); |
8a6608907 aio: kill ki_key |
1700 1701 |
if (key != KIOCB_KEY) return NULL; |
1da177e4c Linux-2.6.12-rc2 |
1702 |
/* TODO: use a hash or array, this sucks. */ |
04b2fa9f8 fs: split generic... |
1703 1704 |
list_for_each_entry(kiocb, &ctx->active_reqs, ki_list) { if (kiocb->ki_user_iocb == iocb) |
1da177e4c Linux-2.6.12-rc2 |
1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 |
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... |
1720 1721 |
SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb, struct io_event __user *, result) |
1da177e4c Linux-2.6.12-rc2 |
1722 |
{ |
1da177e4c Linux-2.6.12-rc2 |
1723 |
struct kioctx *ctx; |
04b2fa9f8 fs: split generic... |
1724 |
struct aio_kiocb *kiocb; |
1da177e4c Linux-2.6.12-rc2 |
1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 |
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... |
1737 |
|
1da177e4c Linux-2.6.12-rc2 |
1738 |
kiocb = lookup_kiocb(ctx, iocb, key); |
906b973cf aio: add kiocb_ca... |
1739 |
if (kiocb) |
d52a8f9ea fs/aio.c: Remove ... |
1740 |
ret = kiocb_cancel(kiocb); |
906b973cf aio: add kiocb_ca... |
1741 1742 |
else ret = -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1743 |
spin_unlock_irq(&ctx->ctx_lock); |
906b973cf aio: add kiocb_ca... |
1744 |
if (!ret) { |
bec68faaf aio: io_cancel() ... |
1745 1746 1747 1748 |
/* * 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... |
1749 |
*/ |
bec68faaf aio: io_cancel() ... |
1750 |
ret = -EINPROGRESS; |
906b973cf aio: add kiocb_ca... |
1751 |
} |
1da177e4c Linux-2.6.12-rc2 |
1752 |
|
723be6e39 aio: percpu ioctx... |
1753 |
percpu_ref_put(&ctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1754 1755 1756 1757 1758 1759 |
return ret; } /* io_getevents: * Attempts to read at least min_nr events and up to nr events from |
642b5123a aio: fix wrong su... |
1760 1761 1762 1763 1764 1765 1766 1767 |
* 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... |
1768 |
* timeout is relative. Will fail with -ENOSYS if not implemented. |
1da177e4c Linux-2.6.12-rc2 |
1769 |
*/ |
002c8976e [CVE-2009-0029] S... |
1770 1771 1772 1773 1774 |
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 |
1775 1776 1777 1778 1779 |
{ struct kioctx *ioctx = lookup_ioctx(ctx_id); long ret = -EINVAL; if (likely(ioctx)) { |
2e4102559 aio: remove unnec... |
1780 |
if (likely(min_nr <= nr && min_nr >= 0)) |
1da177e4c Linux-2.6.12-rc2 |
1781 |
ret = read_events(ioctx, min_nr, nr, events, timeout); |
723be6e39 aio: percpu ioctx... |
1782 |
percpu_ref_put(&ioctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1783 |
} |
1da177e4c Linux-2.6.12-rc2 |
1784 1785 |
return ret; } |
c00d2c7e8 move aio compat t... |
1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 |
#ifdef CONFIG_COMPAT COMPAT_SYSCALL_DEFINE5(io_getevents, compat_aio_context_t, ctx_id, compat_long_t, min_nr, compat_long_t, nr, struct io_event __user *, events, struct compat_timespec __user *, timeout) { struct timespec t; struct timespec __user *ut = NULL; if (timeout) { if (compat_get_timespec(&t, timeout)) return -EFAULT; ut = compat_alloc_user_space(sizeof(*ut)); if (copy_to_user(ut, &t, sizeof(t))) return -EFAULT; } return sys_io_getevents(ctx_id, min_nr, nr, events, ut); } #endif |