Blame view
fs/file.c
11.7 KB
1da177e4c
|
1 2 3 4 5 6 7 |
/* * linux/fs/file.c * * Copyright (C) 1998-1999, Stephen Tweedie and Bill Hawes * * Manage the dynamic fd arrays in the process files_struct. */ |
1027abe88
|
8 |
#include <linux/module.h> |
1da177e4c
|
9 10 11 |
#include <linux/fs.h> #include <linux/mm.h> #include <linux/time.h> |
d43c36dc6
|
12 |
#include <linux/sched.h> |
1da177e4c
|
13 14 15 |
#include <linux/slab.h> #include <linux/vmalloc.h> #include <linux/file.h> |
9f3acc314
|
16 |
#include <linux/fdtable.h> |
1da177e4c
|
17 |
#include <linux/bitops.h> |
ab2af1f50
|
18 19 20 21 22 23 24 25 |
#include <linux/interrupt.h> #include <linux/spinlock.h> #include <linux/rcupdate.h> #include <linux/workqueue.h> struct fdtable_defer { spinlock_t lock; struct work_struct wq; |
ab2af1f50
|
26 27 |
struct fdtable *next; }; |
9cfe015aa
|
28 |
int sysctl_nr_open __read_mostly = 1024*1024; |
eceea0b3d
|
29 30 |
int sysctl_nr_open_min = BITS_PER_LONG; int sysctl_nr_open_max = 1024 * 1024; /* raised later */ |
9cfe015aa
|
31 |
|
ab2af1f50
|
32 33 34 35 36 37 38 |
/* * We use this list to defer free fdtables that have vmalloced * sets/arrays. By keeping a per-cpu list, we avoid having to embed * the work_struct in fdtable itself which avoids a 64 byte (i386) increase in * this per-task structure. */ static DEFINE_PER_CPU(struct fdtable_defer, fdtable_defer_list); |
1da177e4c
|
39 |
|
a892e2d7d
|
40 |
static inline void *alloc_fdmem(unsigned int size) |
1da177e4c
|
41 |
{ |
a892e2d7d
|
42 43 44 45 46 47 48 |
void *data; data = kmalloc(size, GFP_KERNEL|__GFP_NOWARN); if (data != NULL) return data; return vmalloc(size); |
1da177e4c
|
49 |
} |
a892e2d7d
|
50 |
static void free_fdmem(void *ptr) |
1da177e4c
|
51 |
{ |
a892e2d7d
|
52 |
is_vmalloc_addr(ptr) ? vfree(ptr) : kfree(ptr); |
1da177e4c
|
53 |
} |
a892e2d7d
|
54 |
static void __free_fdtable(struct fdtable *fdt) |
1da177e4c
|
55 |
{ |
a892e2d7d
|
56 57 58 |
free_fdmem(fdt->fd); free_fdmem(fdt->open_fds); kfree(fdt); |
ab2af1f50
|
59 |
} |
1da177e4c
|
60 |
|
65f27f384
|
61 |
static void free_fdtable_work(struct work_struct *work) |
ab2af1f50
|
62 |
{ |
65f27f384
|
63 64 |
struct fdtable_defer *f = container_of(work, struct fdtable_defer, wq); |
ab2af1f50
|
65 |
struct fdtable *fdt; |
1da177e4c
|
66 |
|
ab2af1f50
|
67 68 69 70 71 72 |
spin_lock_bh(&f->lock); fdt = f->next; f->next = NULL; spin_unlock_bh(&f->lock); while(fdt) { struct fdtable *next = fdt->next; |
a892e2d7d
|
73 74 |
__free_fdtable(fdt); |
ab2af1f50
|
75 76 77 |
fdt = next; } } |
1da177e4c
|
78 |
|
4fd45812c
|
79 |
void free_fdtable_rcu(struct rcu_head *rcu) |
ab2af1f50
|
80 81 |
{ struct fdtable *fdt = container_of(rcu, struct fdtable, rcu); |
ab2af1f50
|
82 |
struct fdtable_defer *fddef; |
1da177e4c
|
83 |
|
ab2af1f50
|
84 |
BUG_ON(!fdt); |
ab2af1f50
|
85 |
|
4fd45812c
|
86 |
if (fdt->max_fds <= NR_OPEN_DEFAULT) { |
ab2af1f50
|
87 |
/* |
4fd45812c
|
88 89 |
* This fdtable is embedded in the files structure and that * structure itself is getting destroyed. |
ab2af1f50
|
90 |
*/ |
4fd45812c
|
91 92 |
kmem_cache_free(files_cachep, container_of(fdt, struct files_struct, fdtab)); |
ab2af1f50
|
93 94 |
return; } |
a892e2d7d
|
95 |
if (!is_vmalloc_addr(fdt->fd) && !is_vmalloc_addr(fdt->open_fds)) { |
ab2af1f50
|
96 |
kfree(fdt->fd); |
5466b456e
|
97 |
kfree(fdt->open_fds); |
ab2af1f50
|
98 |
kfree(fdt); |
1da177e4c
|
99 |
} else { |
ab2af1f50
|
100 101 102 103 |
fddef = &get_cpu_var(fdtable_defer_list); spin_lock(&fddef->lock); fdt->next = fddef->next; fddef->next = fdt; |
593be07ae
|
104 105 |
/* vmallocs are handled from the workqueue context */ schedule_work(&fddef->wq); |
ab2af1f50
|
106 107 |
spin_unlock(&fddef->lock); put_cpu_var(fdtable_defer_list); |
1da177e4c
|
108 |
} |
ab2af1f50
|
109 |
} |
ab2af1f50
|
110 111 112 113 |
/* * Expand the fdset in the files_struct. Called with the files spinlock * held for write. */ |
5466b456e
|
114 |
static void copy_fdtable(struct fdtable *nfdt, struct fdtable *ofdt) |
ab2af1f50
|
115 |
{ |
5466b456e
|
116 |
unsigned int cpy, set; |
ab2af1f50
|
117 |
|
5466b456e
|
118 |
BUG_ON(nfdt->max_fds < ofdt->max_fds); |
5466b456e
|
119 120 121 122 123 124 125 126 127 128 129 130 |
cpy = ofdt->max_fds * sizeof(struct file *); set = (nfdt->max_fds - ofdt->max_fds) * sizeof(struct file *); memcpy(nfdt->fd, ofdt->fd, cpy); memset((char *)(nfdt->fd) + cpy, 0, set); cpy = ofdt->max_fds / BITS_PER_BYTE; set = (nfdt->max_fds - ofdt->max_fds) / BITS_PER_BYTE; memcpy(nfdt->open_fds, ofdt->open_fds, cpy); memset((char *)(nfdt->open_fds) + cpy, 0, set); memcpy(nfdt->close_on_exec, ofdt->close_on_exec, cpy); memset((char *)(nfdt->close_on_exec) + cpy, 0, set); |
1da177e4c
|
131 |
} |
5466b456e
|
132 |
static struct fdtable * alloc_fdtable(unsigned int nr) |
1da177e4c
|
133 |
{ |
5466b456e
|
134 135 |
struct fdtable *fdt; char *data; |
1da177e4c
|
136 |
|
ab2af1f50
|
137 |
/* |
5466b456e
|
138 139 140 141 142 |
* Figure out how many fds we actually want to support in this fdtable. * Allocation steps are keyed to the size of the fdarray, since it * grows far faster than any of the other dynamic data. We try to fit * the fdarray into comfortable page-tuned chunks: starting at 1024B * and growing in powers of two from there on. |
ab2af1f50
|
143 |
*/ |
5466b456e
|
144 145 146 |
nr /= (1024 / sizeof(struct file *)); nr = roundup_pow_of_two(nr + 1); nr *= (1024 / sizeof(struct file *)); |
5c598b342
|
147 148 149 150 151 152 153 154 155 156 |
/* * Note that this can drive nr *below* what we had passed if sysctl_nr_open * had been set lower between the check in expand_files() and here. Deal * with that in caller, it's cheaper that way. * * We make sure that nr remains a multiple of BITS_PER_LONG - otherwise * bitmaps handling below becomes unpleasant, to put it mildly... */ if (unlikely(nr > sysctl_nr_open)) nr = ((sysctl_nr_open - 1) | (BITS_PER_LONG - 1)) + 1; |
bbea9f696
|
157 |
|
5466b456e
|
158 159 |
fdt = kmalloc(sizeof(struct fdtable), GFP_KERNEL); if (!fdt) |
bbea9f696
|
160 |
goto out; |
5466b456e
|
161 162 163 164 165 166 167 168 169 170 171 172 |
fdt->max_fds = nr; data = alloc_fdmem(nr * sizeof(struct file *)); if (!data) goto out_fdt; fdt->fd = (struct file **)data; data = alloc_fdmem(max_t(unsigned int, 2 * nr / BITS_PER_BYTE, L1_CACHE_BYTES)); if (!data) goto out_arr; fdt->open_fds = (fd_set *)data; data += nr / BITS_PER_BYTE; fdt->close_on_exec = (fd_set *)data; |
5466b456e
|
173 |
fdt->next = NULL; |
ab2af1f50
|
174 |
return fdt; |
5466b456e
|
175 176 |
out_arr: |
a892e2d7d
|
177 |
free_fdmem(fdt->fd); |
5466b456e
|
178 |
out_fdt: |
ab2af1f50
|
179 |
kfree(fdt); |
5466b456e
|
180 |
out: |
ab2af1f50
|
181 182 |
return NULL; } |
1da177e4c
|
183 |
|
ab2af1f50
|
184 |
/* |
74d392aaa
|
185 186 187 188 189 |
* Expand the file descriptor table. * This function will allocate a new fdtable and both fd array and fdset, of * the given size. * Return <0 error code on error; 1 on successful completion. * The files->file_lock should be held on entry, and will be held on exit. |
ab2af1f50
|
190 191 192 193 194 |
*/ static int expand_fdtable(struct files_struct *files, int nr) __releases(files->file_lock) __acquires(files->file_lock) { |
74d392aaa
|
195 |
struct fdtable *new_fdt, *cur_fdt; |
ab2af1f50
|
196 197 |
spin_unlock(&files->file_lock); |
74d392aaa
|
198 |
new_fdt = alloc_fdtable(nr); |
ab2af1f50
|
199 |
spin_lock(&files->file_lock); |
74d392aaa
|
200 201 |
if (!new_fdt) return -ENOMEM; |
ab2af1f50
|
202 |
/* |
5c598b342
|
203 204 205 206 |
* extremely unlikely race - sysctl_nr_open decreased between the check in * caller and alloc_fdtable(). Cheaper to catch it here... */ if (unlikely(new_fdt->max_fds <= nr)) { |
a892e2d7d
|
207 |
__free_fdtable(new_fdt); |
5c598b342
|
208 209 210 |
return -EMFILE; } /* |
74d392aaa
|
211 212 |
* Check again since another task may have expanded the fd table while * we dropped the lock |
ab2af1f50
|
213 |
*/ |
74d392aaa
|
214 |
cur_fdt = files_fdtable(files); |
bbea9f696
|
215 |
if (nr >= cur_fdt->max_fds) { |
74d392aaa
|
216 217 218 |
/* Continue as planned */ copy_fdtable(new_fdt, cur_fdt); rcu_assign_pointer(files->fdt, new_fdt); |
4fd45812c
|
219 |
if (cur_fdt->max_fds > NR_OPEN_DEFAULT) |
01b2d93ca
|
220 |
free_fdtable(cur_fdt); |
ab2af1f50
|
221 |
} else { |
74d392aaa
|
222 |
/* Somebody else expanded, so undo our attempt */ |
a892e2d7d
|
223 |
__free_fdtable(new_fdt); |
ab2af1f50
|
224 |
} |
74d392aaa
|
225 |
return 1; |
1da177e4c
|
226 227 228 229 |
} /* * Expand files. |
74d392aaa
|
230 231 232 233 234 |
* This function will expand the file structures, if the requested size exceeds * the current capacity and there is room for expansion. * Return <0 error code on error; 0 when nothing done; 1 when files were * expanded and execution may have blocked. * The files->file_lock should be held on entry, and will be held on exit. |
1da177e4c
|
235 236 237 |
*/ int expand_files(struct files_struct *files, int nr) { |
badf16621
|
238 |
struct fdtable *fdt; |
1da177e4c
|
239 |
|
badf16621
|
240 |
fdt = files_fdtable(files); |
4e1e018ec
|
241 242 243 244 245 |
/* * N.B. For clone tasks sharing a files structure, this test * will limit the total number of files that can be opened. */ |
d554ed895
|
246 |
if (nr >= rlimit(RLIMIT_NOFILE)) |
4e1e018ec
|
247 |
return -EMFILE; |
74d392aaa
|
248 |
/* Do we need to expand? */ |
bbea9f696
|
249 |
if (nr < fdt->max_fds) |
74d392aaa
|
250 |
return 0; |
4e1e018ec
|
251 |
|
74d392aaa
|
252 |
/* Can we expand? */ |
9cfe015aa
|
253 |
if (nr >= sysctl_nr_open) |
74d392aaa
|
254 255 256 257 |
return -EMFILE; /* All good, so we try */ return expand_fdtable(files, nr); |
1da177e4c
|
258 |
} |
ab2af1f50
|
259 |
|
02afc6267
|
260 261 262 263 264 265 266 267 268 269 270 271 272 |
static int count_open_files(struct fdtable *fdt) { int size = fdt->max_fds; int i; /* Find the last open fd */ for (i = size/(8*sizeof(long)); i > 0; ) { if (fdt->open_fds->fds_bits[--i]) break; } i = (i+1) * 8 * sizeof(long); return i; } |
02afc6267
|
273 274 275 276 277 278 279 280 281 282 283 284 285 |
/* * Allocate a new files structure and copy contents from the * passed in files structure. * errorp will be valid only when the returned files_struct is NULL. */ struct files_struct *dup_fd(struct files_struct *oldf, int *errorp) { struct files_struct *newf; struct file **old_fds, **new_fds; int open_files, size, i; struct fdtable *old_fdt, *new_fdt; *errorp = -ENOMEM; |
afbec7fff
|
286 |
newf = kmem_cache_alloc(files_cachep, GFP_KERNEL); |
02afc6267
|
287 288 |
if (!newf) goto out; |
afbec7fff
|
289 290 291 292 293 294 295 296 297 |
atomic_set(&newf->count, 1); spin_lock_init(&newf->file_lock); newf->next_fd = 0; new_fdt = &newf->fdtab; new_fdt->max_fds = NR_OPEN_DEFAULT; new_fdt->close_on_exec = (fd_set *)&newf->close_on_exec_init; new_fdt->open_fds = (fd_set *)&newf->open_fds_init; new_fdt->fd = &newf->fd_array[0]; |
afbec7fff
|
298 |
new_fdt->next = NULL; |
02afc6267
|
299 300 |
spin_lock(&oldf->file_lock); old_fdt = files_fdtable(oldf); |
02afc6267
|
301 302 303 304 |
open_files = count_open_files(old_fdt); /* * Check whether we need to allocate a larger fd array and fd set. |
02afc6267
|
305 |
*/ |
adbecb128
|
306 |
while (unlikely(open_files > new_fdt->max_fds)) { |
02afc6267
|
307 |
spin_unlock(&oldf->file_lock); |
9dec3c4d3
|
308 |
|
a892e2d7d
|
309 310 |
if (new_fdt != &newf->fdtab) __free_fdtable(new_fdt); |
adbecb128
|
311 |
|
9dec3c4d3
|
312 313 314 315 316 317 318 319 |
new_fdt = alloc_fdtable(open_files - 1); if (!new_fdt) { *errorp = -ENOMEM; goto out_release; } /* beyond sysctl_nr_open; nothing to do */ if (unlikely(new_fdt->max_fds < open_files)) { |
a892e2d7d
|
320 |
__free_fdtable(new_fdt); |
9dec3c4d3
|
321 |
*errorp = -EMFILE; |
02afc6267
|
322 |
goto out_release; |
9dec3c4d3
|
323 |
} |
9dec3c4d3
|
324 |
|
02afc6267
|
325 326 327 328 329 330 331 |
/* * Reacquire the oldf lock and a pointer to its fd table * who knows it may have a new bigger fd table. We need * the latest pointer. */ spin_lock(&oldf->file_lock); old_fdt = files_fdtable(oldf); |
adbecb128
|
332 |
open_files = count_open_files(old_fdt); |
02afc6267
|
333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 |
} old_fds = old_fdt->fd; new_fds = new_fdt->fd; memcpy(new_fdt->open_fds->fds_bits, old_fdt->open_fds->fds_bits, open_files/8); memcpy(new_fdt->close_on_exec->fds_bits, old_fdt->close_on_exec->fds_bits, open_files/8); for (i = open_files; i != 0; i--) { struct file *f = *old_fds++; if (f) { get_file(f); } else { /* * The fd may be claimed in the fd bitmap but not yet * instantiated in the files array if a sibling thread * is partway through open(). So make sure that this * fd is available to the new process. */ FD_CLR(open_files - i, new_fdt->open_fds); } rcu_assign_pointer(*new_fds++, f); } spin_unlock(&oldf->file_lock); /* compute the remainder to be cleared */ size = (new_fdt->max_fds - open_files) * sizeof(struct file *); /* This is long word aligned thus could use a optimized version */ memset(new_fds, 0, size); if (new_fdt->max_fds > open_files) { int left = (new_fdt->max_fds-open_files)/8; int start = open_files / (8 * sizeof(unsigned long)); memset(&new_fdt->open_fds->fds_bits[start], 0, left); memset(&new_fdt->close_on_exec->fds_bits[start], 0, left); } |
afbec7fff
|
373 |
rcu_assign_pointer(newf->fdt, new_fdt); |
02afc6267
|
374 375 376 377 378 379 380 |
return newf; out_release: kmem_cache_free(files_cachep, newf); out: return NULL; } |
ab2af1f50
|
381 382 383 384 |
static void __devinit fdtable_defer_list_init(int cpu) { struct fdtable_defer *fddef = &per_cpu(fdtable_defer_list, cpu); spin_lock_init(&fddef->lock); |
65f27f384
|
385 |
INIT_WORK(&fddef->wq, free_fdtable_work); |
ab2af1f50
|
386 387 388 389 390 391 |
fddef->next = NULL; } void __init files_defer_init(void) { int i; |
0a9450227
|
392 |
for_each_possible_cpu(i) |
ab2af1f50
|
393 |
fdtable_defer_list_init(i); |
eceea0b3d
|
394 395 |
sysctl_nr_open_max = min((size_t)INT_MAX, ~(size_t)0/sizeof(void *)) & -BITS_PER_LONG; |
ab2af1f50
|
396 |
} |
f52111b15
|
397 398 399 400 401 402 403 404 405 |
struct files_struct init_files = { .count = ATOMIC_INIT(1), .fdt = &init_files.fdtab, .fdtab = { .max_fds = NR_OPEN_DEFAULT, .fd = &init_files.fd_array[0], .close_on_exec = (fd_set *)&init_files.close_on_exec_init, .open_fds = (fd_set *)&init_files.open_fds_init, |
f52111b15
|
406 407 408 |
}, .file_lock = __SPIN_LOCK_UNLOCKED(init_task.file_lock), }; |
1027abe88
|
409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 |
/* * allocate a file descriptor, mark it busy. */ int alloc_fd(unsigned start, unsigned flags) { struct files_struct *files = current->files; unsigned int fd; int error; struct fdtable *fdt; spin_lock(&files->file_lock); repeat: fdt = files_fdtable(files); fd = start; if (fd < files->next_fd) fd = files->next_fd; if (fd < fdt->max_fds) fd = find_next_zero_bit(fdt->open_fds->fds_bits, fdt->max_fds, fd); error = expand_files(files, fd); if (error < 0) goto out; /* * If we needed to expand the fs array we * might have blocked - try again. */ if (error) goto repeat; if (start <= files->next_fd) files->next_fd = fd + 1; FD_SET(fd, fdt->open_fds); if (flags & O_CLOEXEC) FD_SET(fd, fdt->close_on_exec); else FD_CLR(fd, fdt->close_on_exec); error = fd; #if 1 /* Sanity check */ |
7dc521579
|
453 |
if (rcu_dereference_raw(fdt->fd[fd]) != NULL) { |
1027abe88
|
454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 |
printk(KERN_WARNING "alloc_fd: slot %d not NULL! ", fd); rcu_assign_pointer(fdt->fd[fd], NULL); } #endif out: spin_unlock(&files->file_lock); return error; } int get_unused_fd(void) { return alloc_fd(0, 0); } EXPORT_SYMBOL(get_unused_fd); |