Blame view
fs/aio.c
42.2 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 |
|
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; }; |
4e179bca6 aio: move private... |
76 |
struct kioctx { |
723be6e39 aio: percpu ioctx... |
77 |
struct percpu_ref users; |
36f558890 aio: refcounting ... |
78 |
atomic_t dead; |
4e179bca6 aio: move private... |
79 |
|
e34ecee2a aio: Fix a trinit... |
80 |
struct percpu_ref reqs; |
4e179bca6 aio: move private... |
81 |
unsigned long user_id; |
4e179bca6 aio: move private... |
82 |
|
e1bdd5f27 aio: percpu reqs_... |
83 84 85 86 87 88 89 |
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_... |
90 91 92 93 |
/* * 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 ... |
94 |
* The real limit is nr_events - 1, which will be larger (see |
3e845ce01 aio: change reqs_... |
95 96 |
* aio_setup_ring()) */ |
4e179bca6 aio: move private... |
97 |
unsigned max_reqs; |
58c85dc20 aio: kill struct ... |
98 99 |
/* Size of ringbuffer, in units of struct io_event */ unsigned nr_events; |
4e179bca6 aio: move private... |
100 |
|
58c85dc20 aio: kill struct ... |
101 102 103 104 105 |
unsigned long mmap_base; unsigned long mmap_size; struct page **ring_pages; long nr_pages; |
723be6e39 aio: percpu ioctx... |
106 |
struct work_struct free_work; |
4e23bcaeb aio: give shared ... |
107 |
|
e02ba72aa aio: block io_des... |
108 109 110 111 |
/* * signals when all in-flight requests are done */ struct completion *requests_done; |
4e23bcaeb aio: give shared ... |
112 |
struct { |
34e83fc61 aio: reqs_active ... |
113 114 115 116 117 |
/* * 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_... |
118 119 |
* * We batch accesses to it with a percpu version. |
34e83fc61 aio: reqs_active ... |
120 121 |
*/ atomic_t reqs_available; |
4e23bcaeb aio: give shared ... |
122 123 124 125 126 127 |
} ____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 ... |
128 129 |
struct { struct mutex ring_lock; |
4e23bcaeb aio: give shared ... |
130 131 |
wait_queue_head_t wait; } ____cacheline_aligned_in_smp; |
58c85dc20 aio: kill struct ... |
132 133 134 |
struct { unsigned tail; |
d856f32a8 aio: fix reqs_ava... |
135 |
unsigned completed_events; |
58c85dc20 aio: kill struct ... |
136 |
spinlock_t completion_lock; |
4e23bcaeb aio: give shared ... |
137 |
} ____cacheline_aligned_in_smp; |
58c85dc20 aio: kill struct ... |
138 139 |
struct page *internal_pages[AIO_RING_PAGES]; |
36bc08cc0 fs/aio: Add suppo... |
140 |
struct file *aio_ring_file; |
db446a08c aio: convert the ... |
141 142 |
unsigned id; |
4e179bca6 aio: move private... |
143 |
}; |
1da177e4c Linux-2.6.12-rc2 |
144 |
/*------ sysctl variables----*/ |
d55b5fdaf [PATCH] aio: remo... |
145 146 147 |
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 |
148 |
/*----end sysctl variables---*/ |
e18b890bb [PATCH] slab: rem... |
149 150 |
static struct kmem_cache *kiocb_cachep; static struct kmem_cache *kioctx_cachep; |
1da177e4c Linux-2.6.12-rc2 |
151 |
|
71ad7490c rework aio migrat... |
152 153 154 155 |
static struct vfsmount *aio_mnt; static const struct file_operations aio_ring_fops; static const struct address_space_operations aio_ctx_aops; |
835f252c6 aio: fix uncorren... |
156 157 158 159 160 161 162 163 |
/* Backing dev info for aio fs. * -no dirty page accounting or writeback happens */ static struct backing_dev_info aio_fs_backing_dev_info = { .name = "aiofs", .state = 0, .capabilities = BDI_CAP_NO_ACCT_AND_WRITEBACK | BDI_CAP_MAP_COPY, }; |
71ad7490c rework aio migrat... |
164 165 166 167 168 169 |
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... |
170 171 |
if (IS_ERR(inode)) return ERR_CAST(inode); |
71ad7490c rework aio migrat... |
172 173 174 |
inode->i_mapping->a_ops = &aio_ctx_aops; inode->i_mapping->private_data = ctx; |
835f252c6 aio: fix uncorren... |
175 |
inode->i_mapping->backing_dev_info = &aio_fs_backing_dev_info; |
71ad7490c rework aio migrat... |
176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 |
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... |
193 194 195 196 197 198 199 200 201 |
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, }; |
8dc4379e1 aio: use the macr... |
202 |
return mount_pseudo(fs_type, "aio:", NULL, &ops, AIO_RING_MAGIC); |
71ad7490c rework aio migrat... |
203 |
} |
1da177e4c Linux-2.6.12-rc2 |
204 205 206 207 208 209 |
/* 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... |
210 211 212 213 214 215 216 217 |
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."); |
835f252c6 aio: fix uncorren... |
218 219 |
if (bdi_init(&aio_fs_backing_dev_info)) panic("Failed to init aio fs backing dev info."); |
0a31bd5f2 KMEM_CACHE(): sim... |
220 221 |
kiocb_cachep = KMEM_CACHE(kiocb, SLAB_HWCACHE_ALIGN|SLAB_PANIC); kioctx_cachep = KMEM_CACHE(kioctx,SLAB_HWCACHE_ALIGN|SLAB_PANIC); |
1da177e4c Linux-2.6.12-rc2 |
222 |
|
caf4167aa aio: dprintk() ->... |
223 224 |
pr_debug("sizeof(struct page) = %zu ", sizeof(struct page)); |
1da177e4c Linux-2.6.12-rc2 |
225 226 227 |
return 0; } |
385773e04 aio.c: move EXPOR... |
228 |
__initcall(aio_setup); |
1da177e4c Linux-2.6.12-rc2 |
229 |
|
5e9ae2e5d aio: fix use-afte... |
230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 |
static void put_aio_ring_file(struct kioctx *ctx) { struct file *aio_ring_file = ctx->aio_ring_file; if (aio_ring_file) { truncate_setsize(aio_ring_file->f_inode, 0); /* Prevent further access to the kioctx from migratepages */ spin_lock(&aio_ring_file->f_inode->i_mapping->private_lock); aio_ring_file->f_inode->i_mapping->private_data = NULL; ctx->aio_ring_file = NULL; spin_unlock(&aio_ring_file->f_inode->i_mapping->private_lock); fput(aio_ring_file); } } |
1da177e4c Linux-2.6.12-rc2 |
245 246 |
static void aio_free_ring(struct kioctx *ctx) { |
36bc08cc0 fs/aio: Add suppo... |
247 |
int i; |
1da177e4c Linux-2.6.12-rc2 |
248 |
|
fa8a53c39 aio: v4 ensure ac... |
249 250 251 252 |
/* 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... |
253 |
for (i = 0; i < ctx->nr_pages; i++) { |
8e321fefb aio/migratepages:... |
254 |
struct page *page; |
36bc08cc0 fs/aio: Add suppo... |
255 256 257 |
pr_debug("pid(%d) [%d] page->count=%d ", current->pid, i, page_count(ctx->ring_pages[i])); |
8e321fefb aio/migratepages:... |
258 259 260 261 262 |
page = ctx->ring_pages[i]; if (!page) continue; ctx->ring_pages[i] = NULL; put_page(page); |
36bc08cc0 fs/aio: Add suppo... |
263 |
} |
1da177e4c Linux-2.6.12-rc2 |
264 |
|
ddb8c45ba aio: nullify aio-... |
265 |
if (ctx->ring_pages && ctx->ring_pages != ctx->internal_pages) { |
58c85dc20 aio: kill struct ... |
266 |
kfree(ctx->ring_pages); |
ddb8c45ba aio: nullify aio-... |
267 268 |
ctx->ring_pages = NULL; } |
36bc08cc0 fs/aio: Add suppo... |
269 270 271 272 273 274 275 276 277 278 279 |
} static int aio_ring_mmap(struct file *file, struct vm_area_struct *vma) { vma->vm_ops = &generic_file_vm_ops; return 0; } static const struct file_operations aio_ring_fops = { .mmap = aio_ring_mmap, }; |
0c45355fc aio: fix build wh... |
280 |
#if IS_ENABLED(CONFIG_MIGRATION) |
36bc08cc0 fs/aio: Add suppo... |
281 282 283 |
static int aio_migratepage(struct address_space *mapping, struct page *new, struct page *old, enum migrate_mode mode) { |
5e9ae2e5d aio: fix use-afte... |
284 |
struct kioctx *ctx; |
36bc08cc0 fs/aio: Add suppo... |
285 |
unsigned long flags; |
fa8a53c39 aio: v4 ensure ac... |
286 |
pgoff_t idx; |
36bc08cc0 fs/aio: Add suppo... |
287 |
int rc; |
8e321fefb aio/migratepages:... |
288 |
rc = 0; |
fa8a53c39 aio: v4 ensure ac... |
289 |
/* mapping->private_lock here protects against the kioctx teardown. */ |
8e321fefb aio/migratepages:... |
290 291 |
spin_lock(&mapping->private_lock); ctx = mapping->private_data; |
fa8a53c39 aio: v4 ensure ac... |
292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 |
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:... |
311 312 |
} else rc = -EINVAL; |
8e321fefb aio/migratepages:... |
313 314 |
if (rc != 0) |
fa8a53c39 aio: v4 ensure ac... |
315 |
goto out_unlock; |
8e321fefb aio/migratepages:... |
316 |
|
36bc08cc0 fs/aio: Add suppo... |
317 318 |
/* Writeback must be complete */ BUG_ON(PageWriteback(old)); |
8e321fefb aio/migratepages:... |
319 |
get_page(new); |
36bc08cc0 fs/aio: Add suppo... |
320 |
|
8e321fefb aio/migratepages:... |
321 |
rc = migrate_page_move_mapping(mapping, new, old, NULL, mode, 1); |
36bc08cc0 fs/aio: Add suppo... |
322 |
if (rc != MIGRATEPAGE_SUCCESS) { |
8e321fefb aio/migratepages:... |
323 |
put_page(new); |
fa8a53c39 aio: v4 ensure ac... |
324 |
goto out_unlock; |
36bc08cc0 fs/aio: Add suppo... |
325 |
} |
fa8a53c39 aio: v4 ensure ac... |
326 327 328 |
/* 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... |
329 |
*/ |
fa8a53c39 aio: v4 ensure ac... |
330 331 332 333 334 |
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... |
335 |
|
fa8a53c39 aio: v4 ensure ac... |
336 337 |
/* The old page is no longer accessible. */ put_page(old); |
8e321fefb aio/migratepages:... |
338 |
|
fa8a53c39 aio: v4 ensure ac... |
339 340 341 342 |
out_unlock: mutex_unlock(&ctx->ring_lock); out: spin_unlock(&mapping->private_lock); |
36bc08cc0 fs/aio: Add suppo... |
343 |
return rc; |
1da177e4c Linux-2.6.12-rc2 |
344 |
} |
0c45355fc aio: fix build wh... |
345 |
#endif |
1da177e4c Linux-2.6.12-rc2 |
346 |
|
36bc08cc0 fs/aio: Add suppo... |
347 |
static const struct address_space_operations aio_ctx_aops = { |
835f252c6 aio: fix uncorren... |
348 |
.set_page_dirty = __set_page_dirty_no_writeback, |
0c45355fc aio: fix build wh... |
349 |
#if IS_ENABLED(CONFIG_MIGRATION) |
36bc08cc0 fs/aio: Add suppo... |
350 |
.migratepage = aio_migratepage, |
0c45355fc aio: fix build wh... |
351 |
#endif |
36bc08cc0 fs/aio: Add suppo... |
352 |
}; |
1da177e4c Linux-2.6.12-rc2 |
353 354 355 |
static int aio_setup_ring(struct kioctx *ctx) { struct aio_ring *ring; |
1da177e4c Linux-2.6.12-rc2 |
356 |
unsigned nr_events = ctx->max_reqs; |
41003a7bc aio: remove retry... |
357 |
struct mm_struct *mm = current->mm; |
3dc9acb67 aio: clean up and... |
358 |
unsigned long size, unused; |
1da177e4c Linux-2.6.12-rc2 |
359 |
int nr_pages; |
36bc08cc0 fs/aio: Add suppo... |
360 361 |
int i; struct file *file; |
1da177e4c Linux-2.6.12-rc2 |
362 363 364 365 366 367 |
/* 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 |
368 |
|
36bc08cc0 fs/aio: Add suppo... |
369 |
nr_pages = PFN_UP(size); |
1da177e4c Linux-2.6.12-rc2 |
370 371 |
if (nr_pages < 0) return -EINVAL; |
71ad7490c rework aio migrat... |
372 |
file = aio_private_file(ctx, nr_pages); |
36bc08cc0 fs/aio: Add suppo... |
373 374 |
if (IS_ERR(file)) { ctx->aio_ring_file = NULL; |
fa8a53c39 aio: v4 ensure ac... |
375 |
return -ENOMEM; |
36bc08cc0 fs/aio: Add suppo... |
376 |
} |
3dc9acb67 aio: clean up and... |
377 378 379 380 381 382 383 384 385 386 387 388 389 |
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... |
390 391 392 393 394 395 396 397 398 399 |
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... |
400 |
unlock_page(page); |
3dc9acb67 aio: clean up and... |
401 402 |
ctx->ring_pages[i] = page; |
36bc08cc0 fs/aio: Add suppo... |
403 |
} |
3dc9acb67 aio: clean up and... |
404 |
ctx->nr_pages = i; |
1da177e4c Linux-2.6.12-rc2 |
405 |
|
3dc9acb67 aio: clean up and... |
406 407 |
if (unlikely(i != nr_pages)) { aio_free_ring(ctx); |
fa8a53c39 aio: v4 ensure ac... |
408 |
return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
409 |
} |
58c85dc20 aio: kill struct ... |
410 411 412 |
ctx->mmap_size = nr_pages * PAGE_SIZE; pr_debug("attempting mmap of %lu bytes ", ctx->mmap_size); |
36bc08cc0 fs/aio: Add suppo... |
413 |
|
41003a7bc aio: remove retry... |
414 |
down_write(&mm->mmap_sem); |
36bc08cc0 fs/aio: Add suppo... |
415 416 |
ctx->mmap_base = do_mmap_pgoff(ctx->aio_ring_file, 0, ctx->mmap_size, PROT_READ | PROT_WRITE, |
3dc9acb67 aio: clean up and... |
417 418 |
MAP_SHARED, 0, &unused); up_write(&mm->mmap_sem); |
58c85dc20 aio: kill struct ... |
419 |
if (IS_ERR((void *)ctx->mmap_base)) { |
58c85dc20 aio: kill struct ... |
420 |
ctx->mmap_size = 0; |
1da177e4c Linux-2.6.12-rc2 |
421 |
aio_free_ring(ctx); |
fa8a53c39 aio: v4 ensure ac... |
422 |
return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
423 |
} |
58c85dc20 aio: kill struct ... |
424 425 |
pr_debug("mmap address: 0x%08lx ", ctx->mmap_base); |
d6c355c7d aio: fix race in ... |
426 |
|
58c85dc20 aio: kill struct ... |
427 428 |
ctx->user_id = ctx->mmap_base; ctx->nr_events = nr_events; /* trusted copy */ |
1da177e4c Linux-2.6.12-rc2 |
429 |
|
58c85dc20 aio: kill struct ... |
430 |
ring = kmap_atomic(ctx->ring_pages[0]); |
1da177e4c Linux-2.6.12-rc2 |
431 |
ring->nr = nr_events; /* user copy */ |
db446a08c aio: convert the ... |
432 |
ring->id = ~0U; |
1da177e4c Linux-2.6.12-rc2 |
433 434 435 436 437 |
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... |
438 |
kunmap_atomic(ring); |
58c85dc20 aio: kill struct ... |
439 |
flush_dcache_page(ctx->ring_pages[0]); |
1da177e4c Linux-2.6.12-rc2 |
440 441 442 |
return 0; } |
1da177e4c Linux-2.6.12-rc2 |
443 444 445 |
#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) |
0460fef2a aio: use cancella... |
446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 |
void kiocb_set_cancel_fn(struct kiocb *req, kiocb_cancel_fn *cancel) { 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); |
d52a8f9ea fs/aio.c: Remove ... |
461 |
static int kiocb_cancel(struct kiocb *kiocb) |
906b973cf aio: add kiocb_ca... |
462 |
{ |
0460fef2a aio: use cancella... |
463 |
kiocb_cancel_fn *old, *cancel; |
906b973cf aio: add kiocb_ca... |
464 |
|
0460fef2a aio: use cancella... |
465 466 467 468 469 470 471 472 |
/* * 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 |
473 |
return -EINVAL; |
906b973cf aio: add kiocb_ca... |
474 |
|
0460fef2a aio: use cancella... |
475 476 477 |
old = cancel; cancel = cmpxchg(&kiocb->ki_cancel, old, KIOCB_CANCELLED); } while (cancel != old); |
906b973cf aio: add kiocb_ca... |
478 |
|
57282d8fd aio: Kill ki_users |
479 |
return cancel(kiocb); |
906b973cf aio: add kiocb_ca... |
480 |
} |
e34ecee2a aio: Fix a trinit... |
481 |
static void free_ioctx(struct work_struct *work) |
36f558890 aio: refcounting ... |
482 |
{ |
e34ecee2a aio: Fix a trinit... |
483 |
struct kioctx *ctx = container_of(work, struct kioctx, free_work); |
e1bdd5f27 aio: percpu reqs_... |
484 |
|
e34ecee2a aio: Fix a trinit... |
485 486 |
pr_debug("freeing %p ", ctx); |
e1bdd5f27 aio: percpu reqs_... |
487 |
|
e34ecee2a aio: Fix a trinit... |
488 |
aio_free_ring(ctx); |
e1bdd5f27 aio: percpu reqs_... |
489 |
free_percpu(ctx->cpu); |
9a1049da9 percpu-refcount: ... |
490 491 |
percpu_ref_exit(&ctx->reqs); percpu_ref_exit(&ctx->users); |
36f558890 aio: refcounting ... |
492 493 |
kmem_cache_free(kioctx_cachep, ctx); } |
e34ecee2a aio: Fix a trinit... |
494 495 496 |
static void free_ioctx_reqs(struct percpu_ref *ref) { struct kioctx *ctx = container_of(ref, struct kioctx, reqs); |
e02ba72aa aio: block io_des... |
497 498 499 |
/* At this point we know that there are no any in-flight requests */ if (ctx->requests_done) complete(ctx->requests_done); |
e34ecee2a aio: Fix a trinit... |
500 501 502 |
INIT_WORK(&ctx->free_work, free_ioctx); schedule_work(&ctx->free_work); } |
36f558890 aio: refcounting ... |
503 504 505 506 507 |
/* * 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... |
508 |
static void free_ioctx_users(struct percpu_ref *ref) |
36f558890 aio: refcounting ... |
509 |
{ |
e34ecee2a aio: Fix a trinit... |
510 |
struct kioctx *ctx = container_of(ref, struct kioctx, users); |
36f558890 aio: refcounting ... |
511 512 513 514 515 516 517 518 519 |
struct kiocb *req; spin_lock_irq(&ctx->ctx_lock); while (!list_empty(&ctx->active_reqs)) { req = list_first_entry(&ctx->active_reqs, struct kiocb, ki_list); list_del_init(&req->ki_list); |
d52a8f9ea fs/aio.c: Remove ... |
520 |
kiocb_cancel(req); |
36f558890 aio: refcounting ... |
521 522 523 |
} spin_unlock_irq(&ctx->ctx_lock); |
e34ecee2a aio: Fix a trinit... |
524 525 |
percpu_ref_kill(&ctx->reqs); percpu_ref_put(&ctx->reqs); |
36f558890 aio: refcounting ... |
526 |
} |
db446a08c aio: convert the ... |
527 528 529 530 531 532 533 |
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... |
534 |
table = rcu_dereference_raw(mm->ioctx_table); |
db446a08c aio: convert the ... |
535 536 537 538 539 540 541 542 |
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... |
543 544 545 546 |
/* While kioctx setup is in progress, * we are protected from page migration * changes ring_pages by ->ring_lock. */ |
db446a08c aio: convert the ... |
547 548 549 550 551 552 553 |
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 ... |
554 555 556 557 558 559 560 561 562 563 |
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... |
564 |
old = rcu_dereference_raw(mm->ioctx_table); |
db446a08c aio: convert the ... |
565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 |
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... |
580 581 582 583 584 585 586 587 588 |
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 |
589 590 591 592 593 |
/* 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... |
594 |
struct mm_struct *mm = current->mm; |
1da177e4c Linux-2.6.12-rc2 |
595 |
struct kioctx *ctx; |
e23754f88 aio: don't bother... |
596 |
int err = -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
597 |
|
e1bdd5f27 aio: percpu reqs_... |
598 599 600 601 602 603 604 605 606 607 608 |
/* * 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 |
609 610 611 612 613 614 615 |
/* Prevent overflows */ if ((nr_events > (0x10000000U / sizeof(struct io_event))) || (nr_events > (0x10000000U / sizeof(struct kiocb)))) { pr_debug("ENOMEM: nr_events too high "); return ERR_PTR(-EINVAL); } |
4cd81c3df aio: double aio_m... |
616 |
if (!nr_events || (unsigned long)nr_events > (aio_max_nr * 2UL)) |
1da177e4c Linux-2.6.12-rc2 |
617 |
return ERR_PTR(-EAGAIN); |
c37622296 [PATCH] Transform... |
618 |
ctx = kmem_cache_zalloc(kioctx_cachep, GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
619 620 |
if (!ctx) return ERR_PTR(-ENOMEM); |
1da177e4c Linux-2.6.12-rc2 |
621 |
ctx->max_reqs = nr_events; |
1da177e4c Linux-2.6.12-rc2 |
622 |
|
1da177e4c Linux-2.6.12-rc2 |
623 |
spin_lock_init(&ctx->ctx_lock); |
0460fef2a aio: use cancella... |
624 |
spin_lock_init(&ctx->completion_lock); |
58c85dc20 aio: kill struct ... |
625 |
mutex_init(&ctx->ring_lock); |
fa8a53c39 aio: v4 ensure ac... |
626 627 628 |
/* 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 |
629 630 631 |
init_waitqueue_head(&ctx->wait); INIT_LIST_HEAD(&ctx->active_reqs); |
1da177e4c Linux-2.6.12-rc2 |
632 |
|
2aad2a86f percpu_ref: add P... |
633 |
if (percpu_ref_init(&ctx->users, free_ioctx_users, 0, GFP_KERNEL)) |
fa8a53c39 aio: v4 ensure ac... |
634 |
goto err; |
2aad2a86f percpu_ref: add P... |
635 |
if (percpu_ref_init(&ctx->reqs, free_ioctx_reqs, 0, GFP_KERNEL)) |
fa8a53c39 aio: v4 ensure ac... |
636 |
goto err; |
e1bdd5f27 aio: percpu reqs_... |
637 638 |
ctx->cpu = alloc_percpu(struct kioctx_cpu); if (!ctx->cpu) |
e34ecee2a aio: Fix a trinit... |
639 |
goto err; |
1da177e4c Linux-2.6.12-rc2 |
640 |
|
fa8a53c39 aio: v4 ensure ac... |
641 642 |
err = aio_setup_ring(ctx); if (err < 0) |
e34ecee2a aio: Fix a trinit... |
643 |
goto err; |
e1bdd5f27 aio: percpu reqs_... |
644 |
|
34e83fc61 aio: reqs_active ... |
645 |
atomic_set(&ctx->reqs_available, ctx->nr_events - 1); |
e1bdd5f27 aio: percpu reqs_... |
646 |
ctx->req_batch = (ctx->nr_events - 1) / (num_possible_cpus() * 4); |
6878ea72a aio: be defensive... |
647 648 |
if (ctx->req_batch < 1) ctx->req_batch = 1; |
34e83fc61 aio: reqs_active ... |
649 |
|
1da177e4c Linux-2.6.12-rc2 |
650 |
/* limit the number of system wide aios */ |
9fa1cb397 aio: aio_nr_lock ... |
651 |
spin_lock(&aio_nr_lock); |
4cd81c3df aio: double aio_m... |
652 |
if (aio_nr + nr_events > (aio_max_nr * 2UL) || |
2dd542b7a aio: aio_nr decre... |
653 |
aio_nr + nr_events < aio_nr) { |
9fa1cb397 aio: aio_nr_lock ... |
654 |
spin_unlock(&aio_nr_lock); |
e34ecee2a aio: Fix a trinit... |
655 |
err = -EAGAIN; |
d1b943271 aio: clean up aio... |
656 |
goto err_ctx; |
2dd542b7a aio: aio_nr decre... |
657 658 |
} aio_nr += ctx->max_reqs; |
9fa1cb397 aio: aio_nr_lock ... |
659 |
spin_unlock(&aio_nr_lock); |
1da177e4c Linux-2.6.12-rc2 |
660 |
|
1881686f8 aio: fix kioctx l... |
661 662 |
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... |
663 |
|
da90382c2 aio: fix error ha... |
664 665 |
err = ioctx_add_table(ctx, mm); if (err) |
e34ecee2a aio: Fix a trinit... |
666 |
goto err_cleanup; |
da90382c2 aio: fix error ha... |
667 |
|
fa8a53c39 aio: v4 ensure ac... |
668 669 |
/* Release the ring_lock mutex now that all setup is complete. */ mutex_unlock(&ctx->ring_lock); |
caf4167aa aio: dprintk() ->... |
670 671 |
pr_debug("allocated ioctx %p[%ld]: mm=%p mask=0x%x ", |
58c85dc20 aio: kill struct ... |
672 |
ctx, ctx->user_id, mm, ctx->nr_events); |
1da177e4c Linux-2.6.12-rc2 |
673 |
return ctx; |
e34ecee2a aio: Fix a trinit... |
674 675 |
err_cleanup: aio_nr_sub(ctx->max_reqs); |
d1b943271 aio: clean up aio... |
676 677 |
err_ctx: aio_free_ring(ctx); |
e34ecee2a aio: Fix a trinit... |
678 |
err: |
fa8a53c39 aio: v4 ensure ac... |
679 |
mutex_unlock(&ctx->ring_lock); |
e1bdd5f27 aio: percpu reqs_... |
680 |
free_percpu(ctx->cpu); |
9a1049da9 percpu-refcount: ... |
681 682 |
percpu_ref_exit(&ctx->reqs); percpu_ref_exit(&ctx->users); |
1da177e4c Linux-2.6.12-rc2 |
683 |
kmem_cache_free(kioctx_cachep, ctx); |
caf4167aa aio: dprintk() ->... |
684 685 |
pr_debug("error allocating ioctx %d ", err); |
e23754f88 aio: don't bother... |
686 |
return ERR_PTR(err); |
1da177e4c Linux-2.6.12-rc2 |
687 |
} |
36f558890 aio: refcounting ... |
688 689 690 691 692 |
/* 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... |
693 |
static int kill_ioctx(struct mm_struct *mm, struct kioctx *ctx, |
e02ba72aa aio: block io_des... |
694 |
struct completion *requests_done) |
36f558890 aio: refcounting ... |
695 |
{ |
fa88b6f88 aio: cleanup: fla... |
696 |
struct kioctx_table *table; |
db446a08c aio: convert the ... |
697 |
|
fa88b6f88 aio: cleanup: fla... |
698 699 |
if (atomic_xchg(&ctx->dead, 1)) return -EINVAL; |
db446a08c aio: convert the ... |
700 |
|
db446a08c aio: convert the ... |
701 |
|
fa88b6f88 aio: cleanup: fla... |
702 |
spin_lock(&mm->ioctx_lock); |
855ef0dec aio: kill the mis... |
703 |
table = rcu_dereference_raw(mm->ioctx_table); |
fa88b6f88 aio: cleanup: fla... |
704 705 |
WARN_ON(ctx != table->table[ctx->id]); table->table[ctx->id] = NULL; |
fa88b6f88 aio: cleanup: fla... |
706 |
spin_unlock(&mm->ioctx_lock); |
4fcc712f5 aio: fix io_destr... |
707 |
|
fa88b6f88 aio: cleanup: fla... |
708 709 |
/* percpu_ref_kill() will do the necessary call_rcu() */ wake_up_all(&ctx->wait); |
4fcc712f5 aio: fix io_destr... |
710 |
|
fa88b6f88 aio: cleanup: fla... |
711 712 713 714 715 716 717 718 |
/* * 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... |
719 |
|
fa88b6f88 aio: cleanup: fla... |
720 721 |
if (ctx->mmap_size) vm_munmap(ctx->mmap_base, ctx->mmap_size); |
fb2d44838 aio: report error... |
722 |
|
fa88b6f88 aio: cleanup: fla... |
723 724 725 |
ctx->requests_done = requests_done; percpu_ref_kill(&ctx->users); return 0; |
1da177e4c Linux-2.6.12-rc2 |
726 727 728 729 730 |
} /* wait_on_sync_kiocb: * Waits on the given sync kiocb to complete. */ |
57282d8fd aio: Kill ki_users |
731 |
ssize_t wait_on_sync_kiocb(struct kiocb *req) |
1da177e4c Linux-2.6.12-rc2 |
732 |
{ |
57282d8fd aio: Kill ki_users |
733 |
while (!req->ki_ctx) { |
1da177e4c Linux-2.6.12-rc2 |
734 |
set_current_state(TASK_UNINTERRUPTIBLE); |
57282d8fd aio: Kill ki_users |
735 |
if (req->ki_ctx) |
1da177e4c Linux-2.6.12-rc2 |
736 |
break; |
41d10da37 aio: account I/O ... |
737 |
io_schedule(); |
1da177e4c Linux-2.6.12-rc2 |
738 739 |
} __set_current_state(TASK_RUNNING); |
57282d8fd aio: Kill ki_users |
740 |
return req->ki_user_data; |
1da177e4c Linux-2.6.12-rc2 |
741 |
} |
385773e04 aio.c: move EXPOR... |
742 |
EXPORT_SYMBOL(wait_on_sync_kiocb); |
1da177e4c Linux-2.6.12-rc2 |
743 |
|
36f558890 aio: refcounting ... |
744 745 746 747 748 749 750 |
/* * 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 |
751 |
*/ |
fc9b52cd8 fs: remove fastca... |
752 |
void exit_aio(struct mm_struct *mm) |
1da177e4c Linux-2.6.12-rc2 |
753 |
{ |
4b70ac5fd aio: change exit_... |
754 755 |
struct kioctx_table *table = rcu_dereference_raw(mm->ioctx_table); int i; |
db446a08c aio: convert the ... |
756 |
|
4b70ac5fd aio: change exit_... |
757 758 |
if (!table) return; |
db446a08c aio: convert the ... |
759 |
|
4b70ac5fd aio: change exit_... |
760 761 |
for (i = 0; i < table->nr; ++i) { struct kioctx *ctx = table->table[i]; |
6098b45b3 aio: block exit_a... |
762 763 |
struct completion requests_done = COMPLETION_INITIALIZER_ONSTACK(requests_done); |
abf137dd7 aio: make the loo... |
764 |
|
4b70ac5fd aio: change exit_... |
765 766 |
if (!ctx) continue; |
936af1576 aio: don't bother... |
767 |
/* |
4b70ac5fd aio: change exit_... |
768 769 770 771 772 |
* 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... |
773 |
*/ |
58c85dc20 aio: kill struct ... |
774 |
ctx->mmap_size = 0; |
6098b45b3 aio: block exit_a... |
775 |
kill_ioctx(mm, ctx, &requests_done); |
36f558890 aio: refcounting ... |
776 |
|
6098b45b3 aio: block exit_a... |
777 778 |
/* Wait until all IO for the context are done. */ wait_for_completion(&requests_done); |
1da177e4c Linux-2.6.12-rc2 |
779 |
} |
4b70ac5fd aio: change exit_... |
780 781 782 |
RCU_INIT_POINTER(mm->ioctx_table, NULL); kfree(table); |
1da177e4c Linux-2.6.12-rc2 |
783 |
} |
e1bdd5f27 aio: percpu reqs_... |
784 785 786 |
static void put_reqs_available(struct kioctx *ctx, unsigned nr) { struct kioctx_cpu *kcpu; |
263782c1c aio: protect reqs... |
787 |
unsigned long flags; |
e1bdd5f27 aio: percpu reqs_... |
788 |
|
263782c1c aio: protect reqs... |
789 |
local_irq_save(flags); |
be6fb451a aio: remove no lo... |
790 |
kcpu = this_cpu_ptr(ctx->cpu); |
e1bdd5f27 aio: percpu reqs_... |
791 |
kcpu->reqs_available += nr; |
263782c1c aio: protect reqs... |
792 |
|
e1bdd5f27 aio: percpu reqs_... |
793 794 795 796 |
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... |
797 |
local_irq_restore(flags); |
e1bdd5f27 aio: percpu reqs_... |
798 799 800 801 802 803 |
} static bool get_reqs_available(struct kioctx *ctx) { struct kioctx_cpu *kcpu; bool ret = false; |
263782c1c aio: protect reqs... |
804 |
unsigned long flags; |
e1bdd5f27 aio: percpu reqs_... |
805 |
|
263782c1c aio: protect reqs... |
806 |
local_irq_save(flags); |
be6fb451a aio: remove no lo... |
807 |
kcpu = this_cpu_ptr(ctx->cpu); |
e1bdd5f27 aio: percpu reqs_... |
808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 |
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... |
826 |
local_irq_restore(flags); |
e1bdd5f27 aio: percpu reqs_... |
827 828 |
return ret; } |
d856f32a8 aio: fix reqs_ava... |
829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 |
/* 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 |
890 |
/* aio_get_req |
57282d8fd aio: Kill ki_users |
891 892 |
* Allocate a slot for an aio request. * Returns NULL if no requests are free. |
1da177e4c Linux-2.6.12-rc2 |
893 |
*/ |
a1c8eae75 aio: kill batch a... |
894 |
static inline struct kiocb *aio_get_req(struct kioctx *ctx) |
1da177e4c Linux-2.6.12-rc2 |
895 |
{ |
a1c8eae75 aio: kill batch a... |
896 |
struct kiocb *req; |
d856f32a8 aio: fix reqs_ava... |
897 898 899 900 901 |
if (!get_reqs_available(ctx)) { user_refill_reqs_available(ctx); if (!get_reqs_available(ctx)) return NULL; } |
a1c8eae75 aio: kill batch a... |
902 |
|
0460fef2a aio: use cancella... |
903 |
req = kmem_cache_alloc(kiocb_cachep, GFP_KERNEL|__GFP_ZERO); |
1da177e4c Linux-2.6.12-rc2 |
904 |
if (unlikely(!req)) |
a1c8eae75 aio: kill batch a... |
905 |
goto out_put; |
1da177e4c Linux-2.6.12-rc2 |
906 |
|
e34ecee2a aio: Fix a trinit... |
907 |
percpu_ref_get(&ctx->reqs); |
1da177e4c Linux-2.6.12-rc2 |
908 |
req->ki_ctx = ctx; |
080d676de aio: allocate kio... |
909 |
return req; |
a1c8eae75 aio: kill batch a... |
910 |
out_put: |
e1bdd5f27 aio: percpu reqs_... |
911 |
put_reqs_available(ctx, 1); |
a1c8eae75 aio: kill batch a... |
912 |
return NULL; |
1da177e4c Linux-2.6.12-rc2 |
913 |
} |
11599ebac aio: make aio_put... |
914 |
static void kiocb_free(struct kiocb *req) |
1da177e4c Linux-2.6.12-rc2 |
915 |
{ |
1d98ebfcc aio: do fget() af... |
916 917 |
if (req->ki_filp) fput(req->ki_filp); |
133890103 eventfd: revised ... |
918 919 |
if (req->ki_eventfd != NULL) eventfd_ctx_put(req->ki_eventfd); |
1da177e4c Linux-2.6.12-rc2 |
920 |
kmem_cache_free(kiocb_cachep, req); |
1da177e4c Linux-2.6.12-rc2 |
921 |
} |
d5470b596 fs/aio.c: make 3 ... |
922 |
static struct kioctx *lookup_ioctx(unsigned long ctx_id) |
1da177e4c Linux-2.6.12-rc2 |
923 |
{ |
db446a08c aio: convert the ... |
924 |
struct aio_ring __user *ring = (void __user *)ctx_id; |
abf137dd7 aio: make the loo... |
925 |
struct mm_struct *mm = current->mm; |
65c24491b aio: lookup_ioctx... |
926 |
struct kioctx *ctx, *ret = NULL; |
db446a08c aio: convert the ... |
927 928 929 930 931 |
struct kioctx_table *table; unsigned id; if (get_user(id, &ring->id)) return NULL; |
1da177e4c Linux-2.6.12-rc2 |
932 |
|
abf137dd7 aio: make the loo... |
933 |
rcu_read_lock(); |
db446a08c aio: convert the ... |
934 |
table = rcu_dereference(mm->ioctx_table); |
abf137dd7 aio: make the loo... |
935 |
|
db446a08c aio: convert the ... |
936 937 |
if (!table || id >= table->nr) goto out; |
1da177e4c Linux-2.6.12-rc2 |
938 |
|
db446a08c aio: convert the ... |
939 |
ctx = table->table[id]; |
f30d704fe aio: table lookup... |
940 |
if (ctx && ctx->user_id == ctx_id) { |
db446a08c aio: convert the ... |
941 942 943 944 |
percpu_ref_get(&ctx->users); ret = ctx; } out: |
abf137dd7 aio: make the loo... |
945 |
rcu_read_unlock(); |
65c24491b aio: lookup_ioctx... |
946 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
947 |
} |
1da177e4c Linux-2.6.12-rc2 |
948 949 |
/* aio_complete * Called when the io request on the given iocb is complete. |
1da177e4c Linux-2.6.12-rc2 |
950 |
*/ |
2d68449e8 aio: kill return ... |
951 |
void aio_complete(struct kiocb *iocb, long res, long res2) |
1da177e4c Linux-2.6.12-rc2 |
952 953 |
{ struct kioctx *ctx = iocb->ki_ctx; |
1da177e4c Linux-2.6.12-rc2 |
954 |
struct aio_ring *ring; |
21b40200c aio: use flush_dc... |
955 |
struct io_event *ev_page, *event; |
d856f32a8 aio: fix reqs_ava... |
956 |
unsigned tail, pos, head; |
1da177e4c Linux-2.6.12-rc2 |
957 |
unsigned long flags; |
1da177e4c Linux-2.6.12-rc2 |
958 |
|
20dcae324 [PATCH] aio: remo... |
959 960 961 962 963 964 |
/* * 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 |
965 966 |
*/ if (is_sync_kiocb(iocb)) { |
1da177e4c Linux-2.6.12-rc2 |
967 |
iocb->ki_user_data = res; |
57282d8fd aio: Kill ki_users |
968 969 |
smp_wmb(); iocb->ki_ctx = ERR_PTR(-EXDEV); |
1da177e4c Linux-2.6.12-rc2 |
970 |
wake_up_process(iocb->ki_obj.tsk); |
2d68449e8 aio: kill return ... |
971 |
return; |
1da177e4c Linux-2.6.12-rc2 |
972 |
} |
0460fef2a aio: use cancella... |
973 974 975 976 977 978 979 |
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... |
980 |
|
1da177e4c Linux-2.6.12-rc2 |
981 |
/* |
0460fef2a aio: use cancella... |
982 |
* Add a completion event to the ring buffer. Must be done holding |
4b30f07e7 aio: fix wrong co... |
983 |
* ctx->completion_lock to prevent other code from messing with the tail |
0460fef2a aio: use cancella... |
984 985 986 |
* pointer since we might be called from irq context. */ spin_lock_irqsave(&ctx->completion_lock, flags); |
58c85dc20 aio: kill struct ... |
987 |
tail = ctx->tail; |
21b40200c aio: use flush_dc... |
988 |
pos = tail + AIO_EVENTS_OFFSET; |
58c85dc20 aio: kill struct ... |
989 |
if (++tail >= ctx->nr_events) |
4bf69b2a0 [PATCH] aio: ring... |
990 |
tail = 0; |
1da177e4c Linux-2.6.12-rc2 |
991 |
|
58c85dc20 aio: kill struct ... |
992 |
ev_page = kmap_atomic(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); |
21b40200c aio: use flush_dc... |
993 |
event = ev_page + pos % AIO_EVENTS_PER_PAGE; |
1da177e4c Linux-2.6.12-rc2 |
994 995 996 997 |
event->obj = (u64)(unsigned long)iocb->ki_obj.user; event->data = iocb->ki_user_data; event->res = res; event->res2 = res2; |
21b40200c aio: use flush_dc... |
998 |
kunmap_atomic(ev_page); |
58c85dc20 aio: kill struct ... |
999 |
flush_dcache_page(ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]); |
21b40200c aio: use flush_dc... |
1000 1001 1002 |
pr_debug("%p[%u]: %p: %p %Lx %lx %lx ", |
caf4167aa aio: dprintk() ->... |
1003 1004 |
ctx, tail, iocb, iocb->ki_obj.user, iocb->ki_user_data, res, res2); |
1da177e4c Linux-2.6.12-rc2 |
1005 1006 1007 1008 1009 |
/* 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 ... |
1010 |
ctx->tail = tail; |
1da177e4c Linux-2.6.12-rc2 |
1011 |
|
58c85dc20 aio: kill struct ... |
1012 |
ring = kmap_atomic(ctx->ring_pages[0]); |
d856f32a8 aio: fix reqs_ava... |
1013 |
head = ring->head; |
21b40200c aio: use flush_dc... |
1014 |
ring->tail = tail; |
e8e3c3d66 fs: remove the se... |
1015 |
kunmap_atomic(ring); |
58c85dc20 aio: kill struct ... |
1016 |
flush_dcache_page(ctx->ring_pages[0]); |
1da177e4c Linux-2.6.12-rc2 |
1017 |
|
d856f32a8 aio: fix reqs_ava... |
1018 1019 1020 |
ctx->completed_events++; if (ctx->completed_events > 1) refill_reqs_available(ctx, head, tail); |
0460fef2a aio: use cancella... |
1021 |
spin_unlock_irqrestore(&ctx->completion_lock, flags); |
21b40200c aio: use flush_dc... |
1022 1023 |
pr_debug("added to ring %p at [%u] ", iocb, tail); |
8d1c98b0b eventfd/kaio inte... |
1024 1025 1026 1027 1028 1029 |
/* * 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... |
1030 |
if (iocb->ki_eventfd != NULL) |
8d1c98b0b eventfd/kaio inte... |
1031 |
eventfd_signal(iocb->ki_eventfd, 1); |
1da177e4c Linux-2.6.12-rc2 |
1032 |
/* everything turned out well, dispose of the aiocb. */ |
57282d8fd aio: Kill ki_users |
1033 |
kiocb_free(iocb); |
1da177e4c Linux-2.6.12-rc2 |
1034 |
|
6cb2a2104 aio: bad AIO race... |
1035 1036 1037 1038 1039 1040 1041 |
/* * 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 |
1042 1043 |
if (waitqueue_active(&ctx->wait)) wake_up(&ctx->wait); |
e34ecee2a aio: Fix a trinit... |
1044 |
percpu_ref_put(&ctx->reqs); |
1da177e4c Linux-2.6.12-rc2 |
1045 |
} |
385773e04 aio.c: move EXPOR... |
1046 |
EXPORT_SYMBOL(aio_complete); |
1da177e4c Linux-2.6.12-rc2 |
1047 |
|
2be4e7dee aio: fix some com... |
1048 |
/* aio_read_events_ring |
a31ad380b aio: make aio_rea... |
1049 1050 |
* Pull an event off of the ioctx's event ring. Returns the number of * events fetched |
1da177e4c Linux-2.6.12-rc2 |
1051 |
*/ |
a31ad380b aio: make aio_rea... |
1052 1053 |
static long aio_read_events_ring(struct kioctx *ctx, struct io_event __user *event, long nr) |
1da177e4c Linux-2.6.12-rc2 |
1054 |
{ |
1da177e4c Linux-2.6.12-rc2 |
1055 |
struct aio_ring *ring; |
5ffac122d aio: Don't use ct... |
1056 |
unsigned head, tail, pos; |
a31ad380b aio: make aio_rea... |
1057 1058 |
long ret = 0; int copy_ret; |
58c85dc20 aio: kill struct ... |
1059 |
mutex_lock(&ctx->ring_lock); |
1da177e4c Linux-2.6.12-rc2 |
1060 |
|
fa8a53c39 aio: v4 ensure ac... |
1061 |
/* Access to ->ring_pages here is protected by ctx->ring_lock. */ |
58c85dc20 aio: kill struct ... |
1062 |
ring = kmap_atomic(ctx->ring_pages[0]); |
a31ad380b aio: make aio_rea... |
1063 |
head = ring->head; |
5ffac122d aio: Don't use ct... |
1064 |
tail = ring->tail; |
a31ad380b aio: make aio_rea... |
1065 |
kunmap_atomic(ring); |
2ff396be6 aio: add missing ... |
1066 1067 1068 1069 1070 |
/* * 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... |
1071 1072 |
pr_debug("h%u t%u m%u ", head, tail, ctx->nr_events); |
1da177e4c Linux-2.6.12-rc2 |
1073 |
|
5ffac122d aio: Don't use ct... |
1074 |
if (head == tail) |
1da177e4c Linux-2.6.12-rc2 |
1075 |
goto out; |
edfbbf388 aio: fix kernel m... |
1076 1077 |
head %= ctx->nr_events; tail %= ctx->nr_events; |
a31ad380b aio: make aio_rea... |
1078 1079 1080 1081 |
while (ret < nr) { long avail; struct io_event *ev; struct page *page; |
5ffac122d aio: Don't use ct... |
1082 1083 |
avail = (head <= tail ? tail : ctx->nr_events) - head; if (head == tail) |
a31ad380b aio: make aio_rea... |
1084 1085 1086 1087 1088 1089 1090 |
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 ... |
1091 |
page = ctx->ring_pages[pos / AIO_EVENTS_PER_PAGE]; |
a31ad380b aio: make aio_rea... |
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 |
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 ... |
1106 |
head %= ctx->nr_events; |
1da177e4c Linux-2.6.12-rc2 |
1107 |
} |
1da177e4c Linux-2.6.12-rc2 |
1108 |
|
58c85dc20 aio: kill struct ... |
1109 |
ring = kmap_atomic(ctx->ring_pages[0]); |
a31ad380b aio: make aio_rea... |
1110 |
ring->head = head; |
91d80a84b aio: fix possible... |
1111 |
kunmap_atomic(ring); |
58c85dc20 aio: kill struct ... |
1112 |
flush_dcache_page(ctx->ring_pages[0]); |
a31ad380b aio: make aio_rea... |
1113 |
|
5ffac122d aio: Don't use ct... |
1114 1115 |
pr_debug("%li h%u t%u ", ret, head, tail); |
a31ad380b aio: make aio_rea... |
1116 |
out: |
58c85dc20 aio: kill struct ... |
1117 |
mutex_unlock(&ctx->ring_lock); |
a31ad380b aio: make aio_rea... |
1118 |
|
1da177e4c Linux-2.6.12-rc2 |
1119 1120 |
return ret; } |
a31ad380b aio: make aio_rea... |
1121 1122 |
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 |
1123 |
{ |
a31ad380b aio: make aio_rea... |
1124 |
long ret = aio_read_events_ring(ctx, event + *i, nr - *i); |
1da177e4c Linux-2.6.12-rc2 |
1125 |
|
a31ad380b aio: make aio_rea... |
1126 1127 |
if (ret > 0) *i += ret; |
1da177e4c Linux-2.6.12-rc2 |
1128 |
|
a31ad380b aio: make aio_rea... |
1129 1130 |
if (unlikely(atomic_read(&ctx->dead))) ret = -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1131 |
|
a31ad380b aio: make aio_rea... |
1132 1133 |
if (!*i) *i = ret; |
1da177e4c Linux-2.6.12-rc2 |
1134 |
|
a31ad380b aio: make aio_rea... |
1135 |
return ret < 0 || *i >= min_nr; |
1da177e4c Linux-2.6.12-rc2 |
1136 |
} |
a31ad380b aio: make aio_rea... |
1137 |
static long read_events(struct kioctx *ctx, long min_nr, long nr, |
1da177e4c Linux-2.6.12-rc2 |
1138 1139 1140 |
struct io_event __user *event, struct timespec __user *timeout) { |
a31ad380b aio: make aio_rea... |
1141 1142 |
ktime_t until = { .tv64 = KTIME_MAX }; long ret = 0; |
1da177e4c Linux-2.6.12-rc2 |
1143 |
|
1da177e4c Linux-2.6.12-rc2 |
1144 1145 |
if (timeout) { struct timespec ts; |
a31ad380b aio: make aio_rea... |
1146 |
|
1da177e4c Linux-2.6.12-rc2 |
1147 |
if (unlikely(copy_from_user(&ts, timeout, sizeof(ts)))) |
a31ad380b aio: make aio_rea... |
1148 |
return -EFAULT; |
1da177e4c Linux-2.6.12-rc2 |
1149 |
|
a31ad380b aio: make aio_rea... |
1150 |
until = timespec_to_ktime(ts); |
1da177e4c Linux-2.6.12-rc2 |
1151 |
} |
a31ad380b aio: make aio_rea... |
1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 |
/* * 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. */ wait_event_interruptible_hrtimeout(ctx->wait, aio_read_events(ctx, min_nr, nr, event, &ret), until); |
1da177e4c Linux-2.6.12-rc2 |
1168 |
|
a31ad380b aio: make aio_rea... |
1169 1170 |
if (!ret && signal_pending(current)) ret = -EINTR; |
1da177e4c Linux-2.6.12-rc2 |
1171 |
|
a31ad380b aio: make aio_rea... |
1172 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1173 |
} |
1da177e4c Linux-2.6.12-rc2 |
1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 |
/* 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... |
1187 |
SYSCALL_DEFINE2(io_setup, unsigned, nr_events, aio_context_t __user *, ctxp) |
1da177e4c Linux-2.6.12-rc2 |
1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 |
{ 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... |
1198 1199 1200 1201 |
if (unlikely(ctx || nr_events == 0)) { pr_debug("EINVAL: io_setup: ctx %lu nr_events %u ", ctx, nr_events); |
1da177e4c Linux-2.6.12-rc2 |
1202 1203 1204 1205 1206 1207 1208 |
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... |
1209 |
if (ret) |
e02ba72aa aio: block io_des... |
1210 |
kill_ioctx(current->mm, ioctx, NULL); |
723be6e39 aio: percpu ioctx... |
1211 |
percpu_ref_put(&ioctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1212 1213 1214 1215 1216 1217 1218 1219 1220 |
} 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... |
1221 |
* implemented. May fail with -EINVAL if the context pointed to |
1da177e4c Linux-2.6.12-rc2 |
1222 1223 |
* is invalid. */ |
002c8976e [CVE-2009-0029] S... |
1224 |
SYSCALL_DEFINE1(io_destroy, aio_context_t, ctx) |
1da177e4c Linux-2.6.12-rc2 |
1225 1226 1227 |
{ struct kioctx *ioctx = lookup_ioctx(ctx); if (likely(NULL != ioctx)) { |
e02ba72aa aio: block io_des... |
1228 1229 |
struct completion requests_done = COMPLETION_INITIALIZER_ONSTACK(requests_done); |
fb2d44838 aio: report error... |
1230 |
int ret; |
e02ba72aa aio: block io_des... |
1231 1232 1233 1234 1235 |
/* 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. */ |
fb2d44838 aio: report error... |
1236 |
ret = kill_ioctx(current->mm, ioctx, &requests_done); |
723be6e39 aio: percpu ioctx... |
1237 |
percpu_ref_put(&ioctx->users); |
e02ba72aa aio: block io_des... |
1238 1239 1240 1241 1242 |
/* 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... |
1243 1244 |
if (!ret) wait_for_completion(&requests_done); |
e02ba72aa aio: block io_des... |
1245 |
|
fb2d44838 aio: report error... |
1246 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1247 1248 1249 1250 1251 |
} pr_debug("EINVAL: io_destroy: invalid context id "); return -EINVAL; } |
41ef4eb8e aio: kill ki_retry |
1252 1253 |
typedef ssize_t (aio_rw_op)(struct kiocb *, const struct iovec *, unsigned long, loff_t); |
293bc9822 new methods: ->re... |
1254 |
typedef ssize_t (rw_iter_op)(struct kiocb *, struct iov_iter *); |
41ef4eb8e aio: kill ki_retry |
1255 |
|
8bc92afcf aio: Kill unneede... |
1256 1257 1258 1259 1260 |
static ssize_t aio_setup_vectored_rw(struct kiocb *kiocb, int rw, char __user *buf, unsigned long *nr_segs, struct iovec **iovec, bool compat) |
eed4e51fb [PATCH] Add vecto... |
1261 1262 |
{ ssize_t ret; |
8bc92afcf aio: Kill unneede... |
1263 |
*nr_segs = kiocb->ki_nbytes; |
41ef4eb8e aio: kill ki_retry |
1264 |
|
9d85cba71 aio: fix the comp... |
1265 1266 |
#ifdef CONFIG_COMPAT if (compat) |
41ef4eb8e aio: kill ki_retry |
1267 |
ret = compat_rw_copy_check_uvector(rw, |
8bc92afcf aio: Kill unneede... |
1268 |
(struct compat_iovec __user *)buf, |
00fefb9cf aio: use iovec ar... |
1269 |
*nr_segs, UIO_FASTIOV, *iovec, iovec); |
9d85cba71 aio: fix the comp... |
1270 1271 |
else #endif |
41ef4eb8e aio: kill ki_retry |
1272 |
ret = rw_copy_check_uvector(rw, |
8bc92afcf aio: Kill unneede... |
1273 |
(struct iovec __user *)buf, |
00fefb9cf aio: use iovec ar... |
1274 |
*nr_segs, UIO_FASTIOV, *iovec, iovec); |
eed4e51fb [PATCH] Add vecto... |
1275 |
if (ret < 0) |
41ef4eb8e aio: kill ki_retry |
1276 |
return ret; |
a70b52ec1 vfs: make AIO use... |
1277 |
|
41ef4eb8e aio: kill ki_retry |
1278 |
/* ki_nbytes now reflect bytes instead of segs */ |
eed4e51fb [PATCH] Add vecto... |
1279 |
kiocb->ki_nbytes = ret; |
41ef4eb8e aio: kill ki_retry |
1280 |
return 0; |
eed4e51fb [PATCH] Add vecto... |
1281 |
} |
8bc92afcf aio: Kill unneede... |
1282 1283 1284 1285 |
static ssize_t aio_setup_single_vector(struct kiocb *kiocb, int rw, char __user *buf, unsigned long *nr_segs, struct iovec *iovec) |
eed4e51fb [PATCH] Add vecto... |
1286 |
{ |
8bc92afcf aio: Kill unneede... |
1287 |
if (unlikely(!access_ok(!rw, buf, kiocb->ki_nbytes))) |
41ef4eb8e aio: kill ki_retry |
1288 |
return -EFAULT; |
a70b52ec1 vfs: make AIO use... |
1289 |
|
8bc92afcf aio: Kill unneede... |
1290 1291 1292 |
iovec->iov_base = buf; iovec->iov_len = kiocb->ki_nbytes; *nr_segs = 1; |
eed4e51fb [PATCH] Add vecto... |
1293 1294 |
return 0; } |
1da177e4c Linux-2.6.12-rc2 |
1295 |
/* |
2be4e7dee aio: fix some com... |
1296 1297 |
* aio_run_iocb: * Performs the initial checks and io submission. |
1da177e4c Linux-2.6.12-rc2 |
1298 |
*/ |
8bc92afcf aio: Kill unneede... |
1299 1300 |
static ssize_t aio_run_iocb(struct kiocb *req, unsigned opcode, char __user *buf, bool compat) |
1da177e4c Linux-2.6.12-rc2 |
1301 |
{ |
41ef4eb8e aio: kill ki_retry |
1302 1303 |
struct file *file = req->ki_filp; ssize_t ret; |
8bc92afcf aio: Kill unneede... |
1304 |
unsigned long nr_segs; |
41ef4eb8e aio: kill ki_retry |
1305 1306 1307 |
int rw; fmode_t mode; aio_rw_op *rw_op; |
293bc9822 new methods: ->re... |
1308 |
rw_iter_op *iter_op; |
00fefb9cf aio: use iovec ar... |
1309 |
struct iovec inline_vecs[UIO_FASTIOV], *iovec = inline_vecs; |
293bc9822 new methods: ->re... |
1310 |
struct iov_iter iter; |
1da177e4c Linux-2.6.12-rc2 |
1311 |
|
8bc92afcf aio: Kill unneede... |
1312 |
switch (opcode) { |
1da177e4c Linux-2.6.12-rc2 |
1313 |
case IOCB_CMD_PREAD: |
eed4e51fb [PATCH] Add vecto... |
1314 |
case IOCB_CMD_PREADV: |
41ef4eb8e aio: kill ki_retry |
1315 1316 1317 |
mode = FMODE_READ; rw = READ; rw_op = file->f_op->aio_read; |
293bc9822 new methods: ->re... |
1318 |
iter_op = file->f_op->read_iter; |
41ef4eb8e aio: kill ki_retry |
1319 1320 1321 |
goto rw_common; case IOCB_CMD_PWRITE: |
eed4e51fb [PATCH] Add vecto... |
1322 |
case IOCB_CMD_PWRITEV: |
41ef4eb8e aio: kill ki_retry |
1323 1324 1325 |
mode = FMODE_WRITE; rw = WRITE; rw_op = file->f_op->aio_write; |
293bc9822 new methods: ->re... |
1326 |
iter_op = file->f_op->write_iter; |
41ef4eb8e aio: kill ki_retry |
1327 1328 1329 1330 |
goto rw_common; rw_common: if (unlikely(!(file->f_mode & mode))) return -EBADF; |
293bc9822 new methods: ->re... |
1331 |
if (!rw_op && !iter_op) |
41ef4eb8e aio: kill ki_retry |
1332 |
return -EINVAL; |
8bc92afcf aio: Kill unneede... |
1333 1334 1335 1336 1337 1338 |
ret = (opcode == IOCB_CMD_PREADV || opcode == IOCB_CMD_PWRITEV) ? aio_setup_vectored_rw(req, rw, buf, &nr_segs, &iovec, compat) : aio_setup_single_vector(req, rw, buf, &nr_segs, iovec); |
754320d6e aio: fix potentia... |
1339 1340 |
if (!ret) ret = rw_verify_area(rw, file, &req->ki_pos, req->ki_nbytes); |
8bc92afcf aio: Kill unneede... |
1341 |
if (ret < 0) { |
00fefb9cf aio: use iovec ar... |
1342 |
if (iovec != inline_vecs) |
8bc92afcf aio: Kill unneede... |
1343 |
kfree(iovec); |
41ef4eb8e aio: kill ki_retry |
1344 |
return ret; |
8bc92afcf aio: Kill unneede... |
1345 |
} |
41ef4eb8e aio: kill ki_retry |
1346 1347 |
req->ki_nbytes = ret; |
41ef4eb8e aio: kill ki_retry |
1348 |
|
73a7075e3 aio: Kill aio_rw_... |
1349 1350 1351 1352 1353 1354 1355 1356 1357 |
/* XXX: move/kill - rw_verify_area()? */ /* This matches the pread()/pwrite() logic */ if (req->ki_pos < 0) { ret = -EINVAL; break; } if (rw == WRITE) file_start_write(file); |
293bc9822 new methods: ->re... |
1358 1359 1360 1361 1362 1363 |
if (iter_op) { iov_iter_init(&iter, rw, iovec, nr_segs, req->ki_nbytes); ret = iter_op(req, &iter); } else { ret = rw_op(req, iovec, nr_segs, req->ki_pos); } |
73a7075e3 aio: Kill aio_rw_... |
1364 1365 1366 |
if (rw == WRITE) file_end_write(file); |
1da177e4c Linux-2.6.12-rc2 |
1367 |
break; |
41ef4eb8e aio: kill ki_retry |
1368 |
|
1da177e4c Linux-2.6.12-rc2 |
1369 |
case IOCB_CMD_FDSYNC: |
41ef4eb8e aio: kill ki_retry |
1370 1371 1372 1373 |
if (!file->f_op->aio_fsync) return -EINVAL; ret = file->f_op->aio_fsync(req, 1); |
1da177e4c Linux-2.6.12-rc2 |
1374 |
break; |
41ef4eb8e aio: kill ki_retry |
1375 |
|
1da177e4c Linux-2.6.12-rc2 |
1376 |
case IOCB_CMD_FSYNC: |
41ef4eb8e aio: kill ki_retry |
1377 1378 1379 1380 |
if (!file->f_op->aio_fsync) return -EINVAL; ret = file->f_op->aio_fsync(req, 0); |
1da177e4c Linux-2.6.12-rc2 |
1381 |
break; |
41ef4eb8e aio: kill ki_retry |
1382 |
|
1da177e4c Linux-2.6.12-rc2 |
1383 |
default: |
caf4167aa aio: dprintk() ->... |
1384 1385 |
pr_debug("EINVAL: no operation provided "); |
41ef4eb8e aio: kill ki_retry |
1386 |
return -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1387 |
} |
00fefb9cf aio: use iovec ar... |
1388 |
if (iovec != inline_vecs) |
8bc92afcf aio: Kill unneede... |
1389 |
kfree(iovec); |
41ef4eb8e aio: kill ki_retry |
1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 |
if (ret != -EIOCBQUEUED) { /* * There's no easy way to restart the syscall since other AIO's * may be already running. Just fail this IO with EINTR. */ if (unlikely(ret == -ERESTARTSYS || ret == -ERESTARTNOINTR || ret == -ERESTARTNOHAND || ret == -ERESTART_RESTARTBLOCK)) ret = -EINTR; aio_complete(req, ret, 0); } |
1da177e4c Linux-2.6.12-rc2 |
1401 1402 1403 |
return 0; } |
d5470b596 fs/aio.c: make 3 ... |
1404 |
static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb, |
a1c8eae75 aio: kill batch a... |
1405 |
struct iocb *iocb, bool compat) |
1da177e4c Linux-2.6.12-rc2 |
1406 1407 |
{ struct kiocb *req; |
1da177e4c Linux-2.6.12-rc2 |
1408 1409 1410 |
ssize_t ret; /* enforce forwards compatibility on users */ |
9c3060bed signal/timer/even... |
1411 |
if (unlikely(iocb->aio_reserved1 || iocb->aio_reserved2)) { |
caf4167aa aio: dprintk() ->... |
1412 1413 |
pr_debug("EINVAL: reserve field set "); |
1da177e4c Linux-2.6.12-rc2 |
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 |
return -EINVAL; } /* prevent overflows */ if (unlikely( (iocb->aio_buf != (unsigned long)iocb->aio_buf) || (iocb->aio_nbytes != (size_t)iocb->aio_nbytes) || ((ssize_t)iocb->aio_nbytes < 0) )) { pr_debug("EINVAL: io_submit: overflow check "); return -EINVAL; } |
41ef4eb8e aio: kill ki_retry |
1427 |
req = aio_get_req(ctx); |
1d98ebfcc aio: do fget() af... |
1428 |
if (unlikely(!req)) |
1da177e4c Linux-2.6.12-rc2 |
1429 |
return -EAGAIN; |
1d98ebfcc aio: do fget() af... |
1430 1431 1432 1433 1434 |
req->ki_filp = fget(iocb->aio_fildes); if (unlikely(!req->ki_filp)) { ret = -EBADF; goto out_put_req; |
1da177e4c Linux-2.6.12-rc2 |
1435 |
} |
1d98ebfcc aio: do fget() af... |
1436 |
|
9c3060bed signal/timer/even... |
1437 1438 1439 1440 1441 1442 1443 |
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 ... |
1444 |
req->ki_eventfd = eventfd_ctx_fdget((int) iocb->aio_resfd); |
801678c5a Remove duplicated... |
1445 |
if (IS_ERR(req->ki_eventfd)) { |
9c3060bed signal/timer/even... |
1446 |
ret = PTR_ERR(req->ki_eventfd); |
87c3a86e1 eventfd: remove f... |
1447 |
req->ki_eventfd = NULL; |
9c3060bed signal/timer/even... |
1448 1449 1450 |
goto out_put_req; } } |
1da177e4c Linux-2.6.12-rc2 |
1451 |
|
8a6608907 aio: kill ki_key |
1452 |
ret = put_user(KIOCB_KEY, &user_iocb->aio_key); |
1da177e4c Linux-2.6.12-rc2 |
1453 |
if (unlikely(ret)) { |
caf4167aa aio: dprintk() ->... |
1454 1455 |
pr_debug("EFAULT: aio_key "); |
1da177e4c Linux-2.6.12-rc2 |
1456 1457 1458 1459 1460 1461 |
goto out_put_req; } req->ki_obj.user = user_iocb; req->ki_user_data = iocb->aio_data; req->ki_pos = iocb->aio_offset; |
73a7075e3 aio: Kill aio_rw_... |
1462 |
req->ki_nbytes = iocb->aio_nbytes; |
1da177e4c Linux-2.6.12-rc2 |
1463 |
|
8bc92afcf aio: Kill unneede... |
1464 1465 1466 |
ret = aio_run_iocb(req, iocb->aio_lio_opcode, (char __user *)(unsigned long)iocb->aio_buf, compat); |
1da177e4c Linux-2.6.12-rc2 |
1467 1468 |
if (ret) goto out_put_req; |
1da177e4c Linux-2.6.12-rc2 |
1469 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
1470 |
out_put_req: |
e1bdd5f27 aio: percpu reqs_... |
1471 |
put_reqs_available(ctx, 1); |
e34ecee2a aio: Fix a trinit... |
1472 |
percpu_ref_put(&ctx->reqs); |
57282d8fd aio: Kill ki_users |
1473 |
kiocb_free(req); |
1da177e4c Linux-2.6.12-rc2 |
1474 1475 |
return ret; } |
9d85cba71 aio: fix the comp... |
1476 1477 |
long do_io_submit(aio_context_t ctx_id, long nr, struct iocb __user *__user *iocbpp, bool compat) |
1da177e4c Linux-2.6.12-rc2 |
1478 1479 1480 |
{ struct kioctx *ctx; long ret = 0; |
080d676de aio: allocate kio... |
1481 |
int i = 0; |
9f5b94254 fs: make aio plug |
1482 |
struct blk_plug plug; |
1da177e4c Linux-2.6.12-rc2 |
1483 1484 1485 |
if (unlikely(nr < 0)) return -EINVAL; |
75e1c70fc aio: check for mu... |
1486 1487 |
if (unlikely(nr > LONG_MAX/sizeof(*iocbpp))) nr = LONG_MAX/sizeof(*iocbpp); |
1da177e4c Linux-2.6.12-rc2 |
1488 1489 1490 1491 1492 |
if (unlikely(!access_ok(VERIFY_READ, iocbpp, (nr*sizeof(*iocbpp))))) return -EFAULT; ctx = lookup_ioctx(ctx_id); if (unlikely(!ctx)) { |
caf4167aa aio: dprintk() ->... |
1493 1494 |
pr_debug("EINVAL: invalid context id "); |
1da177e4c Linux-2.6.12-rc2 |
1495 1496 |
return -EINVAL; } |
9f5b94254 fs: make aio plug |
1497 |
blk_start_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 |
/* * 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... |
1515 |
ret = io_submit_one(ctx, user_iocb, &tmp, compat); |
1da177e4c Linux-2.6.12-rc2 |
1516 1517 1518 |
if (ret) break; } |
9f5b94254 fs: make aio plug |
1519 |
blk_finish_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
1520 |
|
723be6e39 aio: percpu ioctx... |
1521 |
percpu_ref_put(&ctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1522 1523 |
return i ? i : ret; } |
9d85cba71 aio: fix the comp... |
1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 |
/* 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 |
1541 1542 |
/* lookup_kiocb * Finds a given iocb for cancellation. |
1da177e4c Linux-2.6.12-rc2 |
1543 |
*/ |
25ee7e383 [PATCH] fs/aio.c:... |
1544 1545 |
static struct kiocb *lookup_kiocb(struct kioctx *ctx, struct iocb __user *iocb, u32 key) |
1da177e4c Linux-2.6.12-rc2 |
1546 1547 |
{ struct list_head *pos; |
d00689af6 [PATCH] aio: repl... |
1548 1549 |
assert_spin_locked(&ctx->ctx_lock); |
8a6608907 aio: kill ki_key |
1550 1551 |
if (key != KIOCB_KEY) return NULL; |
1da177e4c Linux-2.6.12-rc2 |
1552 1553 1554 |
/* TODO: use a hash or array, this sucks. */ list_for_each(pos, &ctx->active_reqs) { struct kiocb *kiocb = list_kiocb(pos); |
8a6608907 aio: kill ki_key |
1555 |
if (kiocb->ki_obj.user == iocb) |
1da177e4c Linux-2.6.12-rc2 |
1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 |
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... |
1571 1572 |
SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb, struct io_event __user *, result) |
1da177e4c Linux-2.6.12-rc2 |
1573 |
{ |
1da177e4c Linux-2.6.12-rc2 |
1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 |
struct kioctx *ctx; struct kiocb *kiocb; u32 key; int ret; ret = get_user(key, &iocb->aio_key); if (unlikely(ret)) return -EFAULT; ctx = lookup_ioctx(ctx_id); if (unlikely(!ctx)) return -EINVAL; spin_lock_irq(&ctx->ctx_lock); |
906b973cf aio: add kiocb_ca... |
1588 |
|
1da177e4c Linux-2.6.12-rc2 |
1589 |
kiocb = lookup_kiocb(ctx, iocb, key); |
906b973cf aio: add kiocb_ca... |
1590 |
if (kiocb) |
d52a8f9ea fs/aio.c: Remove ... |
1591 |
ret = kiocb_cancel(kiocb); |
906b973cf aio: add kiocb_ca... |
1592 1593 |
else ret = -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1594 |
spin_unlock_irq(&ctx->ctx_lock); |
906b973cf aio: add kiocb_ca... |
1595 |
if (!ret) { |
bec68faaf aio: io_cancel() ... |
1596 1597 1598 1599 |
/* * 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... |
1600 |
*/ |
bec68faaf aio: io_cancel() ... |
1601 |
ret = -EINPROGRESS; |
906b973cf aio: add kiocb_ca... |
1602 |
} |
1da177e4c Linux-2.6.12-rc2 |
1603 |
|
723be6e39 aio: percpu ioctx... |
1604 |
percpu_ref_put(&ctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1605 1606 1607 1608 1609 1610 |
return ret; } /* io_getevents: * Attempts to read at least min_nr events and up to nr events from |
642b5123a aio: fix wrong su... |
1611 1612 1613 1614 1615 1616 1617 1618 |
* 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... |
1619 |
* timeout is relative. Will fail with -ENOSYS if not implemented. |
1da177e4c Linux-2.6.12-rc2 |
1620 |
*/ |
002c8976e [CVE-2009-0029] S... |
1621 1622 1623 1624 1625 |
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 |
1626 1627 1628 1629 1630 |
{ struct kioctx *ioctx = lookup_ioctx(ctx_id); long ret = -EINVAL; if (likely(ioctx)) { |
2e4102559 aio: remove unnec... |
1631 |
if (likely(min_nr <= nr && min_nr >= 0)) |
1da177e4c Linux-2.6.12-rc2 |
1632 |
ret = read_events(ioctx, min_nr, nr, events, timeout); |
723be6e39 aio: percpu ioctx... |
1633 |
percpu_ref_put(&ioctx->users); |
1da177e4c Linux-2.6.12-rc2 |
1634 |
} |
1da177e4c Linux-2.6.12-rc2 |
1635 1636 |
return ret; } |