Blame view
fs/nfs/pnfs.c
40.7 KB
85e174ba6
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 |
/* * pNFS functions to call and manage layout drivers. * * Copyright (c) 2002 [year of first publication] * The Regents of the University of Michigan * All Rights Reserved * * Dean Hildebrand <dhildebz@umich.edu> * * Permission is granted to use, copy, create derivative works, and * redistribute this software and such derivative works for any purpose, * so long as the name of the University of Michigan is not used in * any advertising or publicity pertaining to the use or distribution * of this software without specific, written prior authorization. If * the above copyright notice or any other identification of the * University of Michigan is included in any copy of any portion of * this software, then the disclaimer below must also be included. * * This software is provided as is, without representation or warranty * of any kind either express or implied, including without limitation * the implied warranties of merchantability, fitness for a particular * purpose, or noninfringement. The Regents of the University of * Michigan shall not be liable for any damages, including special, * indirect, incidental, or consequential damages, with respect to any * claim arising out of or in connection with the use of the software, * even if it has been or is hereafter advised of the possibility of * such damages. */ #include <linux/nfs_fs.h> |
493292ddc
|
31 |
#include <linux/nfs_page.h> |
143cb494c
|
32 |
#include <linux/module.h> |
974cec8ca
|
33 |
#include "internal.h" |
85e174ba6
|
34 |
#include "pnfs.h" |
64419a9b2
|
35 |
#include "iostat.h" |
85e174ba6
|
36 37 |
#define NFSDBG_FACILITY NFSDBG_PNFS |
02c35fca7
|
38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 |
/* Locking: * * pnfs_spinlock: * protects pnfs_modules_tbl. */ static DEFINE_SPINLOCK(pnfs_spinlock); /* * pnfs_modules_tbl holds all pnfs modules */ static LIST_HEAD(pnfs_modules_tbl); /* Return the registered pnfs layout driver module matching given id */ static struct pnfs_layoutdriver_type * find_pnfs_driver_locked(u32 id) { struct pnfs_layoutdriver_type *local; list_for_each_entry(local, &pnfs_modules_tbl, pnfs_tblid) if (local->id == id) goto out; local = NULL; out: dprintk("%s: Searching for id %u, found %p ", __func__, id, local); return local; } |
85e174ba6
|
65 66 67 |
static struct pnfs_layoutdriver_type * find_pnfs_driver(u32 id) { |
02c35fca7
|
68 69 70 71 72 73 |
struct pnfs_layoutdriver_type *local; spin_lock(&pnfs_spinlock); local = find_pnfs_driver_locked(id); spin_unlock(&pnfs_spinlock); return local; |
85e174ba6
|
74 75 76 77 78 |
} void unset_pnfs_layoutdriver(struct nfs_server *nfss) { |
738fd0f36
|
79 80 81 |
if (nfss->pnfs_curr_ld) { if (nfss->pnfs_curr_ld->clear_layoutdriver) nfss->pnfs_curr_ld->clear_layoutdriver(nfss); |
02c35fca7
|
82 |
module_put(nfss->pnfs_curr_ld->owner); |
738fd0f36
|
83 |
} |
85e174ba6
|
84 85 86 87 88 89 90 91 92 93 |
nfss->pnfs_curr_ld = NULL; } /* * Try to set the server's pnfs module to the pnfs layout type specified by id. * Currently only one pNFS layout driver per filesystem is supported. * * @id layout type. Zero (illegal layout type) indicates pNFS not in use. */ void |
738fd0f36
|
94 95 |
set_pnfs_layoutdriver(struct nfs_server *server, const struct nfs_fh *mntfh, u32 id) |
85e174ba6
|
96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 |
{ struct pnfs_layoutdriver_type *ld_type = NULL; if (id == 0) goto out_no_driver; if (!(server->nfs_client->cl_exchange_flags & (EXCHGID4_FLAG_USE_NON_PNFS | EXCHGID4_FLAG_USE_PNFS_MDS))) { printk(KERN_ERR "%s: id %u cl_exchange_flags 0x%x ", __func__, id, server->nfs_client->cl_exchange_flags); goto out_no_driver; } ld_type = find_pnfs_driver(id); if (!ld_type) { request_module("%s-%u", LAYOUT_NFSV4_1_MODULE_PREFIX, id); ld_type = find_pnfs_driver(id); if (!ld_type) { dprintk("%s: No pNFS module found for %u. ", __func__, id); goto out_no_driver; } } |
02c35fca7
|
119 120 121 122 123 |
if (!try_module_get(ld_type->owner)) { dprintk("%s: Could not grab reference on module ", __func__); goto out_no_driver; } |
85e174ba6
|
124 |
server->pnfs_curr_ld = ld_type; |
738fd0f36
|
125 126 127 128 129 130 131 132 |
if (ld_type->set_layoutdriver && ld_type->set_layoutdriver(server, mntfh)) { printk(KERN_ERR "%s: Error initializing pNFS layout driver %u. ", __func__, id); module_put(ld_type->owner); goto out_no_driver; } |
ea8eecdd1
|
133 |
|
85e174ba6
|
134 135 136 137 138 139 140 141 142 |
dprintk("%s: pNFS module for %u set ", __func__, id); return; out_no_driver: dprintk("%s: Using NFSv4 I/O ", __func__); server->pnfs_curr_ld = NULL; } |
02c35fca7
|
143 144 145 146 147 148 149 150 151 152 153 154 |
int pnfs_register_layoutdriver(struct pnfs_layoutdriver_type *ld_type) { int status = -EINVAL; struct pnfs_layoutdriver_type *tmp; if (ld_type->id == 0) { printk(KERN_ERR "%s id 0 is reserved ", __func__); return status; } |
b1f69b754
|
155 156 157 158 159 160 |
if (!ld_type->alloc_lseg || !ld_type->free_lseg) { printk(KERN_ERR "%s Layout driver must provide " "alloc_lseg and free_lseg. ", __func__); return status; } |
02c35fca7
|
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 |
spin_lock(&pnfs_spinlock); tmp = find_pnfs_driver_locked(ld_type->id); if (!tmp) { list_add(&ld_type->pnfs_tblid, &pnfs_modules_tbl); status = 0; dprintk("%s Registering id:%u name:%s ", __func__, ld_type->id, ld_type->name); } else { printk(KERN_ERR "%s Module with id %d already loaded! ", __func__, ld_type->id); } spin_unlock(&pnfs_spinlock); return status; } EXPORT_SYMBOL_GPL(pnfs_register_layoutdriver); void pnfs_unregister_layoutdriver(struct pnfs_layoutdriver_type *ld_type) { dprintk("%s Deregistering id:%u ", __func__, ld_type->id); spin_lock(&pnfs_spinlock); list_del(&ld_type->pnfs_tblid); spin_unlock(&pnfs_spinlock); } EXPORT_SYMBOL_GPL(pnfs_unregister_layoutdriver); |
e5e940170
|
191 |
|
b1f69b754
|
192 193 194 |
/* * pNFS client layout cache */ |
cc6e5340b
|
195 |
/* Need to hold i_lock if caller does not already hold reference */ |
43f1b3da8
|
196 |
void |
cc6e5340b
|
197 |
get_layout_hdr(struct pnfs_layout_hdr *lo) |
e5e940170
|
198 |
{ |
cc6e5340b
|
199 |
atomic_inc(&lo->plh_refcount); |
e5e940170
|
200 |
} |
636fb9c89
|
201 202 203 204 205 206 207 208 209 210 211 212 |
static struct pnfs_layout_hdr * pnfs_alloc_layout_hdr(struct inode *ino, gfp_t gfp_flags) { struct pnfs_layoutdriver_type *ld = NFS_SERVER(ino)->pnfs_curr_ld; return ld->alloc_layout_hdr ? ld->alloc_layout_hdr(ino, gfp_flags) : kzalloc(sizeof(struct pnfs_layout_hdr), gfp_flags); } static void pnfs_free_layout_hdr(struct pnfs_layout_hdr *lo) { struct pnfs_layoutdriver_type *ld = NFS_SERVER(lo->plh_inode)->pnfs_curr_ld; |
9fa407587
|
213 |
put_rpccred(lo->plh_lc_cred); |
636fb9c89
|
214 215 |
return ld->alloc_layout_hdr ? ld->free_layout_hdr(lo) : kfree(lo); } |
e5e940170
|
216 |
static void |
cc6e5340b
|
217 |
destroy_layout_hdr(struct pnfs_layout_hdr *lo) |
e5e940170
|
218 |
{ |
cc6e5340b
|
219 220 221 222 |
dprintk("%s: freeing layout cache %p ", __func__, lo); BUG_ON(!list_empty(&lo->plh_layouts)); NFS_I(lo->plh_inode)->layout = NULL; |
636fb9c89
|
223 |
pnfs_free_layout_hdr(lo); |
cc6e5340b
|
224 |
} |
e5e940170
|
225 |
|
cc6e5340b
|
226 227 228 229 230 |
static void put_layout_hdr_locked(struct pnfs_layout_hdr *lo) { if (atomic_dec_and_test(&lo->plh_refcount)) destroy_layout_hdr(lo); |
e5e940170
|
231 |
} |
b1f69b754
|
232 |
void |
cc6e5340b
|
233 |
put_layout_hdr(struct pnfs_layout_hdr *lo) |
974cec8ca
|
234 |
{ |
cc6e5340b
|
235 236 237 238 239 240 |
struct inode *inode = lo->plh_inode; if (atomic_dec_and_lock(&lo->plh_refcount, &inode->i_lock)) { destroy_layout_hdr(lo); spin_unlock(&inode->i_lock); } |
974cec8ca
|
241 242 243 244 245 |
} static void init_lseg(struct pnfs_layout_hdr *lo, struct pnfs_layout_segment *lseg) { |
566052c53
|
246 |
INIT_LIST_HEAD(&lseg->pls_list); |
a9bae5666
|
247 |
INIT_LIST_HEAD(&lseg->pls_lc_list); |
4541d16c0
|
248 249 250 |
atomic_set(&lseg->pls_refcount, 1); smp_mb(); set_bit(NFS_LSEG_VALID, &lseg->pls_flags); |
566052c53
|
251 |
lseg->pls_layout = lo; |
974cec8ca
|
252 |
} |
4541d16c0
|
253 |
static void free_lseg(struct pnfs_layout_segment *lseg) |
974cec8ca
|
254 |
{ |
b7edfaa19
|
255 |
struct inode *ino = lseg->pls_layout->plh_inode; |
974cec8ca
|
256 |
|
b1f69b754
|
257 |
NFS_SERVER(ino)->pnfs_curr_ld->free_lseg(lseg); |
52fabd731
|
258 |
/* Matched by get_layout_hdr in pnfs_insert_layout */ |
cc6e5340b
|
259 |
put_layout_hdr(NFS_I(ino)->layout); |
974cec8ca
|
260 |
} |
d684d2ae1
|
261 262 263 264 |
static void put_lseg_common(struct pnfs_layout_segment *lseg) { struct inode *inode = lseg->pls_layout->plh_inode; |
d20581aa4
|
265 |
WARN_ON(test_bit(NFS_LSEG_VALID, &lseg->pls_flags)); |
d684d2ae1
|
266 267 268 269 270 271 272 273 |
list_del_init(&lseg->pls_list); if (list_empty(&lseg->pls_layout->plh_segs)) { set_bit(NFS_LAYOUT_DESTROYED, &lseg->pls_layout->plh_flags); /* Matched by initial refcount set in alloc_init_layout_hdr */ put_layout_hdr_locked(lseg->pls_layout); } rpc_wake_up(&NFS_SERVER(inode)->roc_rpcwaitq); } |
bae724ef9
|
274 |
void |
d684d2ae1
|
275 |
put_lseg(struct pnfs_layout_segment *lseg) |
974cec8ca
|
276 |
{ |
d684d2ae1
|
277 278 279 280 |
struct inode *inode; if (!lseg) return; |
4541d16c0
|
281 282 283 284 |
dprintk("%s: lseg %p ref %d valid %d ", __func__, lseg, atomic_read(&lseg->pls_refcount), test_bit(NFS_LSEG_VALID, &lseg->pls_flags)); |
d684d2ae1
|
285 286 287 |
inode = lseg->pls_layout->plh_inode; if (atomic_dec_and_lock(&lseg->pls_refcount, &inode->i_lock)) { LIST_HEAD(free_me); |
4541d16c0
|
288 |
|
d684d2ae1
|
289 290 291 292 |
put_lseg_common(lseg); list_add(&lseg->pls_list, &free_me); spin_unlock(&inode->i_lock); pnfs_free_lseg_list(&free_me); |
4541d16c0
|
293 |
} |
4541d16c0
|
294 |
} |
e0c2b3801
|
295 |
EXPORT_SYMBOL_GPL(put_lseg); |
974cec8ca
|
296 |
|
fb3296eb4
|
297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 |
static inline u64 end_offset(u64 start, u64 len) { u64 end; end = start + len; return end >= start ? end : NFS4_MAX_UINT64; } /* last octet in a range */ static inline u64 last_byte_offset(u64 start, u64 len) { u64 end; BUG_ON(!len); end = start + len; return end > start ? end - 1 : NFS4_MAX_UINT64; } /* * is l2 fully contained in l1? * start1 end1 * [----------------------------------) * start2 end2 * [----------------) */ static inline int lo_seg_contained(struct pnfs_layout_range *l1, struct pnfs_layout_range *l2) { u64 start1 = l1->offset; u64 end1 = end_offset(start1, l1->length); u64 start2 = l2->offset; u64 end2 = end_offset(start2, l2->length); return (start1 <= start2) && (end1 >= end2); } /* * is l1 and l2 intersecting? * start1 end1 * [----------------------------------) * start2 end2 * [----------------) */ static inline int lo_seg_intersecting(struct pnfs_layout_range *l1, struct pnfs_layout_range *l2) { u64 start1 = l1->offset; u64 end1 = end_offset(start1, l1->length); u64 start2 = l2->offset; u64 end2 = end_offset(start2, l2->length); return (end1 == NFS4_MAX_UINT64 || end1 > start2) && (end2 == NFS4_MAX_UINT64 || end2 > start1); } |
4541d16c0
|
355 |
static bool |
778b5502f
|
356 357 |
should_free_lseg(struct pnfs_layout_range *lseg_range, struct pnfs_layout_range *recall_range) |
4541d16c0
|
358 |
{ |
778b5502f
|
359 360 361 |
return (recall_range->iomode == IOMODE_ANY || lseg_range->iomode == recall_range->iomode) && lo_seg_intersecting(lseg_range, recall_range); |
974cec8ca
|
362 |
} |
4541d16c0
|
363 364 365 366 367 368 369 370 371 372 373 |
/* Returns 1 if lseg is removed from list, 0 otherwise */ static int mark_lseg_invalid(struct pnfs_layout_segment *lseg, struct list_head *tmp_list) { int rv = 0; if (test_and_clear_bit(NFS_LSEG_VALID, &lseg->pls_flags)) { /* Remove the reference keeping the lseg in the * list. It will now be removed when all * outstanding io is finished. */ |
d684d2ae1
|
374 375 376 377 378 379 380 381 |
dprintk("%s: lseg %p ref %d ", __func__, lseg, atomic_read(&lseg->pls_refcount)); if (atomic_dec_and_test(&lseg->pls_refcount)) { put_lseg_common(lseg); list_add(&lseg->pls_list, tmp_list); rv = 1; } |
4541d16c0
|
382 383 384 385 386 387 388 |
} return rv; } /* Returns count of number of matching invalid lsegs remaining in list * after call. */ |
43f1b3da8
|
389 |
int |
4541d16c0
|
390 391 |
mark_matching_lsegs_invalid(struct pnfs_layout_hdr *lo, struct list_head *tmp_list, |
778b5502f
|
392 |
struct pnfs_layout_range *recall_range) |
974cec8ca
|
393 394 |
{ struct pnfs_layout_segment *lseg, *next; |
4541d16c0
|
395 |
int invalid = 0, removed = 0; |
974cec8ca
|
396 397 398 |
dprintk("%s:Begin lo %p ", __func__, lo); |
385117224
|
399 400 401 402 403 |
if (list_empty(&lo->plh_segs)) { if (!test_and_set_bit(NFS_LAYOUT_DESTROYED, &lo->plh_flags)) put_layout_hdr_locked(lo); return 0; } |
4541d16c0
|
404 |
list_for_each_entry_safe(lseg, next, &lo->plh_segs, pls_list) |
778b5502f
|
405 406 |
if (!recall_range || should_free_lseg(&lseg->pls_range, recall_range)) { |
4541d16c0
|
407 408 409 410 411 412 413 414 415 416 417 |
dprintk("%s: freeing lseg %p iomode %d " "offset %llu length %llu ", __func__, lseg, lseg->pls_range.iomode, lseg->pls_range.offset, lseg->pls_range.length); invalid++; removed += mark_lseg_invalid(lseg, tmp_list); } dprintk("%s:Return %i ", __func__, invalid - removed); return invalid - removed; |
974cec8ca
|
418 |
} |
f49f9baac
|
419 |
/* note free_me must contain lsegs from a single layout_hdr */ |
43f1b3da8
|
420 |
void |
4541d16c0
|
421 |
pnfs_free_lseg_list(struct list_head *free_me) |
974cec8ca
|
422 |
{ |
4541d16c0
|
423 |
struct pnfs_layout_segment *lseg, *tmp; |
f49f9baac
|
424 425 426 427 428 429 430 |
struct pnfs_layout_hdr *lo; if (list_empty(free_me)) return; lo = list_first_entry(free_me, struct pnfs_layout_segment, pls_list)->pls_layout; |
974cec8ca
|
431 |
|
f49f9baac
|
432 433 434 435 436 437 438 439 |
if (test_bit(NFS_LAYOUT_DESTROYED, &lo->plh_flags)) { struct nfs_client *clp; clp = NFS_SERVER(lo->plh_inode)->nfs_client; spin_lock(&clp->cl_lock); list_del_init(&lo->plh_layouts); spin_unlock(&clp->cl_lock); } |
4541d16c0
|
440 |
list_for_each_entry_safe(lseg, tmp, free_me, pls_list) { |
566052c53
|
441 |
list_del(&lseg->pls_list); |
4541d16c0
|
442 |
free_lseg(lseg); |
974cec8ca
|
443 444 |
} } |
e5e940170
|
445 446 447 448 |
void pnfs_destroy_layout(struct nfs_inode *nfsi) { struct pnfs_layout_hdr *lo; |
974cec8ca
|
449 |
LIST_HEAD(tmp_list); |
e5e940170
|
450 451 452 453 |
spin_lock(&nfsi->vfs_inode.i_lock); lo = nfsi->layout; if (lo) { |
385117224
|
454 |
lo->plh_block_lgets++; /* permanently block new LAYOUTGETs */ |
778b5502f
|
455 |
mark_matching_lsegs_invalid(lo, &tmp_list, NULL); |
e5e940170
|
456 457 |
} spin_unlock(&nfsi->vfs_inode.i_lock); |
974cec8ca
|
458 459 460 461 462 463 464 465 466 467 |
pnfs_free_lseg_list(&tmp_list); } /* * Called by the state manger to remove all layouts established under an * expired lease. */ void pnfs_destroy_all_layouts(struct nfs_client *clp) { |
6382a4413
|
468 |
struct nfs_server *server; |
974cec8ca
|
469 470 |
struct pnfs_layout_hdr *lo; LIST_HEAD(tmp_list); |
c47abcf8f
|
471 472 |
nfs4_deviceid_mark_client_invalid(clp); nfs4_deviceid_purge_client(clp); |
974cec8ca
|
473 |
spin_lock(&clp->cl_lock); |
6382a4413
|
474 475 476 477 478 479 |
rcu_read_lock(); list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) { if (!list_empty(&server->layouts)) list_splice_init(&server->layouts, &tmp_list); } rcu_read_unlock(); |
974cec8ca
|
480 481 482 483 |
spin_unlock(&clp->cl_lock); while (!list_empty(&tmp_list)) { lo = list_entry(tmp_list.next, struct pnfs_layout_hdr, |
b7edfaa19
|
484 |
plh_layouts); |
974cec8ca
|
485 486 |
dprintk("%s freeing layout for inode %lu ", __func__, |
b7edfaa19
|
487 |
lo->plh_inode->i_ino); |
2887fe455
|
488 |
list_del_init(&lo->plh_layouts); |
b7edfaa19
|
489 |
pnfs_destroy_layout(NFS_I(lo->plh_inode)); |
974cec8ca
|
490 |
} |
e5e940170
|
491 |
} |
fd6002e9b
|
492 |
/* update lo->plh_stateid with new if is more recent */ |
43f1b3da8
|
493 494 495 |
void pnfs_set_layout_stateid(struct pnfs_layout_hdr *lo, const nfs4_stateid *new, bool update_barrier) |
b1f69b754
|
496 |
{ |
fd6002e9b
|
497 |
u32 oldseq, newseq; |
b1f69b754
|
498 |
|
fd6002e9b
|
499 500 |
oldseq = be32_to_cpu(lo->plh_stateid.stateid.seqid); newseq = be32_to_cpu(new->stateid.seqid); |
43f1b3da8
|
501 |
if ((int)(newseq - oldseq) > 0) { |
fd6002e9b
|
502 |
memcpy(&lo->plh_stateid, &new->stateid, sizeof(new->stateid)); |
43f1b3da8
|
503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 |
if (update_barrier) { u32 new_barrier = be32_to_cpu(new->stateid.seqid); if ((int)(new_barrier - lo->plh_barrier)) lo->plh_barrier = new_barrier; } else { /* Because of wraparound, we want to keep the barrier * "close" to the current seqids. It needs to be * within 2**31 to count as "behind", so if it * gets too near that limit, give us a litle leeway * and bring it to within 2**30. * NOTE - and yes, this is all unsigned arithmetic. */ if (unlikely((newseq - lo->plh_barrier) > (3 << 29))) lo->plh_barrier = newseq - (1 << 30); } } |
b1f69b754
|
520 |
} |
cf7d63f1f
|
521 522 |
/* lget is set to 1 if called from inside send_layoutget call chain */ static bool |
43f1b3da8
|
523 524 525 526 527 528 |
pnfs_layoutgets_blocked(struct pnfs_layout_hdr *lo, nfs4_stateid *stateid, int lget) { if ((stateid) && (int)(lo->plh_barrier - be32_to_cpu(stateid->stateid.seqid)) >= 0) return true; |
f7e8917a6
|
529 |
return lo->plh_block_lgets || |
385117224
|
530 |
test_bit(NFS_LAYOUT_DESTROYED, &lo->plh_flags) || |
f7e8917a6
|
531 |
test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags) || |
43f1b3da8
|
532 |
(list_empty(&lo->plh_segs) && |
cf7d63f1f
|
533 534 |
(atomic_read(&lo->plh_outstanding) > lget)); } |
fd6002e9b
|
535 536 537 |
int pnfs_choose_layoutget_stateid(nfs4_stateid *dst, struct pnfs_layout_hdr *lo, struct nfs4_state *open_state) |
b1f69b754
|
538 |
{ |
fd6002e9b
|
539 |
int status = 0; |
974cec8ca
|
540 |
|
b1f69b754
|
541 542 |
dprintk("--> %s ", __func__); |
fd6002e9b
|
543 |
spin_lock(&lo->plh_inode->i_lock); |
43f1b3da8
|
544 |
if (pnfs_layoutgets_blocked(lo, NULL, 1)) { |
cf7d63f1f
|
545 546 |
status = -EAGAIN; } else if (list_empty(&lo->plh_segs)) { |
fd6002e9b
|
547 548 549 550 551 552 553 554 555 556 |
int seq; do { seq = read_seqbegin(&open_state->seqlock); memcpy(dst->data, open_state->stateid.data, sizeof(open_state->stateid.data)); } while (read_seqretry(&open_state->seqlock, seq)); } else memcpy(dst->data, lo->plh_stateid.data, sizeof(lo->plh_stateid.data)); spin_unlock(&lo->plh_inode->i_lock); |
b1f69b754
|
557 558 |
dprintk("<-- %s ", __func__); |
fd6002e9b
|
559 |
return status; |
b1f69b754
|
560 561 562 563 564 565 566 567 |
} /* * Get layout from server. * for now, assume that whole file layouts are requested. * arg->offset: 0 * arg->length: all ones */ |
e5e940170
|
568 569 570 |
static struct pnfs_layout_segment * send_layoutget(struct pnfs_layout_hdr *lo, struct nfs_open_context *ctx, |
fb3296eb4
|
571 |
struct pnfs_layout_range *range, |
a75b9df9d
|
572 |
gfp_t gfp_flags) |
e5e940170
|
573 |
{ |
b7edfaa19
|
574 |
struct inode *ino = lo->plh_inode; |
b1f69b754
|
575 576 577 |
struct nfs_server *server = NFS_SERVER(ino); struct nfs4_layoutget *lgp; struct pnfs_layout_segment *lseg = NULL; |
35124a099
|
578 579 580 |
struct page **pages = NULL; int i; u32 max_resp_sz, max_pages; |
b1f69b754
|
581 582 583 |
dprintk("--> %s ", __func__); |
e5e940170
|
584 |
|
b1f69b754
|
585 |
BUG_ON(ctx == NULL); |
a75b9df9d
|
586 |
lgp = kzalloc(sizeof(*lgp), gfp_flags); |
cf7d63f1f
|
587 |
if (lgp == NULL) |
b1f69b754
|
588 |
return NULL; |
35124a099
|
589 590 591 592 |
/* allocate pages for xdr post processing */ max_resp_sz = server->nfs_client->cl_session->fc_attrs.max_resp_sz; max_pages = max_resp_sz >> PAGE_SHIFT; |
a75b9df9d
|
593 |
pages = kzalloc(max_pages * sizeof(struct page *), gfp_flags); |
35124a099
|
594 595 596 597 |
if (!pages) goto out_err_free; for (i = 0; i < max_pages; i++) { |
a75b9df9d
|
598 |
pages[i] = alloc_page(gfp_flags); |
35124a099
|
599 600 601 |
if (!pages[i]) goto out_err_free; } |
fb3296eb4
|
602 603 604 |
lgp->args.minlength = PAGE_CACHE_SIZE; if (lgp->args.minlength > range->length) lgp->args.minlength = range->length; |
b1f69b754
|
605 |
lgp->args.maxcount = PNFS_LAYOUT_MAXSIZE; |
fb3296eb4
|
606 |
lgp->args.range = *range; |
b1f69b754
|
607 608 609 |
lgp->args.type = server->pnfs_curr_ld->id; lgp->args.inode = ino; lgp->args.ctx = get_nfs_open_context(ctx); |
35124a099
|
610 611 |
lgp->args.layout.pages = pages; lgp->args.layout.pglen = max_pages * PAGE_SIZE; |
b1f69b754
|
612 |
lgp->lsegpp = &lseg; |
a75b9df9d
|
613 |
lgp->gfp_flags = gfp_flags; |
b1f69b754
|
614 615 616 617 618 |
/* Synchronously retrieve layout information from server and * store in lseg. */ nfs4_proc_layoutget(lgp); |
974cec8ca
|
619 |
if (!lseg) { |
b1f69b754
|
620 |
/* remember that LAYOUTGET failed and suspend trying */ |
fb3296eb4
|
621 |
set_bit(lo_fail_bit(range->iomode), &lo->plh_flags); |
974cec8ca
|
622 |
} |
35124a099
|
623 624 625 626 627 |
/* free xdr pages */ for (i = 0; i < max_pages; i++) __free_page(pages[i]); kfree(pages); |
974cec8ca
|
628 |
return lseg; |
35124a099
|
629 630 631 632 633 634 635 636 637 638 639 640 641 |
out_err_free: /* free any allocated xdr pages, lgp as it's not used */ if (pages) { for (i = 0; i < max_pages; i++) { if (!pages[i]) break; __free_page(pages[i]); } kfree(pages); } kfree(lgp); return NULL; |
974cec8ca
|
642 |
} |
cbe826036
|
643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 |
/* Initiates a LAYOUTRETURN(FILE) */ int _pnfs_return_layout(struct inode *ino) { struct pnfs_layout_hdr *lo = NULL; struct nfs_inode *nfsi = NFS_I(ino); LIST_HEAD(tmp_list); struct nfs4_layoutreturn *lrp; nfs4_stateid stateid; int status = 0; dprintk("--> %s ", __func__); spin_lock(&ino->i_lock); lo = nfsi->layout; |
a2e1d4f2e
|
659 |
if (!lo) { |
cbe826036
|
660 |
spin_unlock(&ino->i_lock); |
a2e1d4f2e
|
661 662 663 |
dprintk("%s: no layout to return ", __func__); return status; |
cbe826036
|
664 665 666 667 |
} stateid = nfsi->layout->plh_stateid; /* Reference matched in nfs4_layoutreturn_release */ get_layout_hdr(lo); |
ea0ded748
|
668 669 |
mark_matching_lsegs_invalid(lo, &tmp_list, NULL); lo->plh_block_lgets++; |
cbe826036
|
670 671 672 673 674 675 676 677 |
spin_unlock(&ino->i_lock); pnfs_free_lseg_list(&tmp_list); WARN_ON(test_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)); lrp = kzalloc(sizeof(*lrp), GFP_KERNEL); if (unlikely(lrp == NULL)) { status = -ENOMEM; |
9e2dfdb30
|
678 679 |
set_bit(NFS_LAYOUT_RW_FAILED, &lo->plh_flags); set_bit(NFS_LAYOUT_RO_FAILED, &lo->plh_flags); |
1ed3a8539
|
680 |
put_layout_hdr(lo); |
cbe826036
|
681 682 683 684 685 686 |
goto out; } lrp->args.stateid = stateid; lrp->args.layout_type = NFS_SERVER(ino)->pnfs_curr_ld->id; lrp->args.inode = ino; |
a56aaa02b
|
687 |
lrp->args.layout = lo; |
cbe826036
|
688 689 690 691 692 693 694 695 |
lrp->clp = NFS_SERVER(ino)->nfs_client; status = nfs4_proc_layoutreturn(lrp); out: dprintk("<-- %s status: %d ", __func__, status); return status; } |
f7e8917a6
|
696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 |
bool pnfs_roc(struct inode *ino) { struct pnfs_layout_hdr *lo; struct pnfs_layout_segment *lseg, *tmp; LIST_HEAD(tmp_list); bool found = false; spin_lock(&ino->i_lock); lo = NFS_I(ino)->layout; if (!lo || !test_and_clear_bit(NFS_LAYOUT_ROC, &lo->plh_flags) || test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags)) goto out_nolayout; list_for_each_entry_safe(lseg, tmp, &lo->plh_segs, pls_list) if (test_bit(NFS_LSEG_ROC, &lseg->pls_flags)) { mark_lseg_invalid(lseg, &tmp_list); found = true; } if (!found) goto out_nolayout; lo->plh_block_lgets++; get_layout_hdr(lo); /* matched in pnfs_roc_release */ spin_unlock(&ino->i_lock); pnfs_free_lseg_list(&tmp_list); return true; out_nolayout: spin_unlock(&ino->i_lock); return false; } void pnfs_roc_release(struct inode *ino) { struct pnfs_layout_hdr *lo; spin_lock(&ino->i_lock); lo = NFS_I(ino)->layout; lo->plh_block_lgets--; put_layout_hdr_locked(lo); spin_unlock(&ino->i_lock); } void pnfs_roc_set_barrier(struct inode *ino, u32 barrier) { struct pnfs_layout_hdr *lo; spin_lock(&ino->i_lock); lo = NFS_I(ino)->layout; if ((int)(barrier - lo->plh_barrier) > 0) lo->plh_barrier = barrier; spin_unlock(&ino->i_lock); } bool pnfs_roc_drain(struct inode *ino, u32 *barrier) { struct nfs_inode *nfsi = NFS_I(ino); struct pnfs_layout_segment *lseg; bool found = false; spin_lock(&ino->i_lock); list_for_each_entry(lseg, &nfsi->layout->plh_segs, pls_list) if (test_bit(NFS_LSEG_ROC, &lseg->pls_flags)) { found = true; break; } if (!found) { struct pnfs_layout_hdr *lo = nfsi->layout; u32 current_seqid = be32_to_cpu(lo->plh_stateid.stateid.seqid); /* Since close does not return a layout stateid for use as * a barrier, we choose the worst-case barrier. */ *barrier = current_seqid + atomic_read(&lo->plh_outstanding); } spin_unlock(&ino->i_lock); return found; } |
b1f69b754
|
772 773 774 775 776 777 |
/* * Compare two layout segments for sorting into layout cache. * We want to preferentially return RW over RO layouts, so ensure those * are seen first. */ static s64 |
fb3296eb4
|
778 779 |
cmp_layout(struct pnfs_layout_range *l1, struct pnfs_layout_range *l2) |
b1f69b754
|
780 |
{ |
fb3296eb4
|
781 782 783 784 785 786 787 788 789 790 791 |
s64 d; /* high offset > low offset */ d = l1->offset - l2->offset; if (d) return d; /* short length > long length */ d = l2->length - l1->length; if (d) return d; |
b1f69b754
|
792 |
/* read > read/write */ |
fb3296eb4
|
793 |
return (int)(l1->iomode == IOMODE_READ) - (int)(l2->iomode == IOMODE_READ); |
b1f69b754
|
794 |
} |
974cec8ca
|
795 796 797 798 |
static void pnfs_insert_layout(struct pnfs_layout_hdr *lo, struct pnfs_layout_segment *lseg) { |
b1f69b754
|
799 |
struct pnfs_layout_segment *lp; |
b1f69b754
|
800 |
|
974cec8ca
|
801 802 |
dprintk("%s:Begin ", __func__); |
b7edfaa19
|
803 |
assert_spin_locked(&lo->plh_inode->i_lock); |
b7edfaa19
|
804 |
list_for_each_entry(lp, &lo->plh_segs, pls_list) { |
fb3296eb4
|
805 |
if (cmp_layout(&lseg->pls_range, &lp->pls_range) > 0) |
b1f69b754
|
806 |
continue; |
566052c53
|
807 |
list_add_tail(&lseg->pls_list, &lp->pls_list); |
b1f69b754
|
808 809 810 811 |
dprintk("%s: inserted lseg %p " "iomode %d offset %llu length %llu before " "lp %p iomode %d offset %llu length %llu ", |
566052c53
|
812 813 814 815 |
__func__, lseg, lseg->pls_range.iomode, lseg->pls_range.offset, lseg->pls_range.length, lp, lp->pls_range.iomode, lp->pls_range.offset, lp->pls_range.length); |
fb3296eb4
|
816 |
goto out; |
974cec8ca
|
817 |
} |
fb3296eb4
|
818 819 820 821 822 823 824 |
list_add_tail(&lseg->pls_list, &lo->plh_segs); dprintk("%s: inserted lseg %p " "iomode %d offset %llu length %llu at tail ", __func__, lseg, lseg->pls_range.iomode, lseg->pls_range.offset, lseg->pls_range.length); out: |
cc6e5340b
|
825 |
get_layout_hdr(lo); |
974cec8ca
|
826 827 828 |
dprintk("%s:Return ", __func__); |
e5e940170
|
829 830 831 |
} static struct pnfs_layout_hdr * |
9fa407587
|
832 833 834 |
alloc_init_layout_hdr(struct inode *ino, struct nfs_open_context *ctx, gfp_t gfp_flags) |
e5e940170
|
835 836 |
{ struct pnfs_layout_hdr *lo; |
636fb9c89
|
837 |
lo = pnfs_alloc_layout_hdr(ino, gfp_flags); |
e5e940170
|
838 839 |
if (!lo) return NULL; |
cc6e5340b
|
840 |
atomic_set(&lo->plh_refcount, 1); |
b7edfaa19
|
841 842 |
INIT_LIST_HEAD(&lo->plh_layouts); INIT_LIST_HEAD(&lo->plh_segs); |
43f1b3da8
|
843 |
INIT_LIST_HEAD(&lo->plh_bulk_recall); |
b7edfaa19
|
844 |
lo->plh_inode = ino; |
9fa407587
|
845 |
lo->plh_lc_cred = get_rpccred(ctx->state->owner->so_cred); |
e5e940170
|
846 847 848 849 |
return lo; } static struct pnfs_layout_hdr * |
9fa407587
|
850 851 852 |
pnfs_find_alloc_layout(struct inode *ino, struct nfs_open_context *ctx, gfp_t gfp_flags) |
e5e940170
|
853 854 855 856 857 858 859 860 |
{ struct nfs_inode *nfsi = NFS_I(ino); struct pnfs_layout_hdr *new = NULL; dprintk("%s Begin ino=%p layout=%p ", __func__, ino, nfsi->layout); assert_spin_locked(&ino->i_lock); |
4541d16c0
|
861 862 863 864 865 866 |
if (nfsi->layout) { if (test_bit(NFS_LAYOUT_DESTROYED, &nfsi->layout->plh_flags)) return NULL; else return nfsi->layout; } |
e5e940170
|
867 |
spin_unlock(&ino->i_lock); |
9fa407587
|
868 |
new = alloc_init_layout_hdr(ino, ctx, gfp_flags); |
e5e940170
|
869 870 871 872 873 |
spin_lock(&ino->i_lock); if (likely(nfsi->layout == NULL)) /* Won the race? */ nfsi->layout = new; else |
636fb9c89
|
874 |
pnfs_free_layout_hdr(new); |
e5e940170
|
875 876 |
return nfsi->layout; } |
b1f69b754
|
877 878 879 880 881 882 883 884 885 886 887 888 |
/* * iomode matching rules: * iomode lseg match * ----- ----- ----- * ANY READ true * ANY RW true * RW READ false * RW RW true * READ READ true * READ RW true */ static int |
fb3296eb4
|
889 890 |
is_matching_lseg(struct pnfs_layout_range *ls_range, struct pnfs_layout_range *range) |
b1f69b754
|
891 |
{ |
fb3296eb4
|
892 893 894 895 896 897 898 899 900 901 902 |
struct pnfs_layout_range range1; if ((range->iomode == IOMODE_RW && ls_range->iomode != IOMODE_RW) || !lo_seg_intersecting(ls_range, range)) return 0; /* range1 covers only the first byte in the range */ range1 = *range; range1.length = 1; return lo_seg_contained(ls_range, &range1); |
b1f69b754
|
903 904 905 906 907 |
} /* * lookup range in layout */ |
e5e940170
|
908 |
static struct pnfs_layout_segment * |
fb3296eb4
|
909 910 |
pnfs_find_lseg(struct pnfs_layout_hdr *lo, struct pnfs_layout_range *range) |
e5e940170
|
911 |
{ |
b1f69b754
|
912 913 914 915 |
struct pnfs_layout_segment *lseg, *ret = NULL; dprintk("%s:Begin ", __func__); |
b7edfaa19
|
916 917 |
assert_spin_locked(&lo->plh_inode->i_lock); list_for_each_entry(lseg, &lo->plh_segs, pls_list) { |
4541d16c0
|
918 |
if (test_bit(NFS_LSEG_VALID, &lseg->pls_flags) && |
fb3296eb4
|
919 |
is_matching_lseg(&lseg->pls_range, range)) { |
d684d2ae1
|
920 |
ret = get_lseg(lseg); |
b1f69b754
|
921 922 |
break; } |
d771e3a43
|
923 |
if (lseg->pls_range.offset > range->offset) |
b1f69b754
|
924 925 926 927 928 |
break; } dprintk("%s:Return lseg %p ref %d ", |
4541d16c0
|
929 |
__func__, ret, ret ? atomic_read(&ret->pls_refcount) : 0); |
b1f69b754
|
930 |
return ret; |
e5e940170
|
931 932 933 934 935 936 |
} /* * Layout segment is retreived from the server if not cached. * The appropriate layout segment is referenced and returned to the caller. */ |
7c24d9489
|
937 |
struct pnfs_layout_segment * |
e5e940170
|
938 939 |
pnfs_update_layout(struct inode *ino, struct nfs_open_context *ctx, |
fb3296eb4
|
940 941 |
loff_t pos, u64 count, |
a75b9df9d
|
942 943 |
enum pnfs_iomode iomode, gfp_t gfp_flags) |
e5e940170
|
944 |
{ |
fb3296eb4
|
945 946 947 948 949 |
struct pnfs_layout_range arg = { .iomode = iomode, .offset = pos, .length = count, }; |
707ed5fdb
|
950 |
unsigned pg_offset; |
e5e940170
|
951 |
struct nfs_inode *nfsi = NFS_I(ino); |
6382a4413
|
952 953 |
struct nfs_server *server = NFS_SERVER(ino); struct nfs_client *clp = server->nfs_client; |
e5e940170
|
954 955 |
struct pnfs_layout_hdr *lo; struct pnfs_layout_segment *lseg = NULL; |
f49f9baac
|
956 |
bool first = false; |
e5e940170
|
957 958 959 960 |
if (!pnfs_enabled_sb(NFS_SERVER(ino))) return NULL; spin_lock(&ino->i_lock); |
9fa407587
|
961 |
lo = pnfs_find_alloc_layout(ino, ctx, gfp_flags); |
e5e940170
|
962 963 964 965 966 |
if (lo == NULL) { dprintk("%s ERROR: can't get pnfs_layout_hdr ", __func__); goto out_unlock; } |
43f1b3da8
|
967 968 969 970 971 |
/* Do we even need to bother with this? */ if (test_bit(NFS4CLNT_LAYOUTRECALL, &clp->cl_state) || test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags)) { dprintk("%s matches recall, use MDS ", __func__); |
e5e940170
|
972 973 974 975 |
goto out_unlock; } /* if LAYOUTGET already failed once we don't try again */ |
566052c53
|
976 |
if (test_bit(lo_fail_bit(iomode), &nfsi->layout->plh_flags)) |
e5e940170
|
977 |
goto out_unlock; |
568e8c494
|
978 |
/* Check to see if the layout for the given range already exists */ |
fb3296eb4
|
979 |
lseg = pnfs_find_lseg(lo, &arg); |
568e8c494
|
980 981 |
if (lseg) goto out_unlock; |
43f1b3da8
|
982 |
if (pnfs_layoutgets_blocked(lo, NULL, 0)) |
cf7d63f1f
|
983 984 |
goto out_unlock; atomic_inc(&lo->plh_outstanding); |
cc6e5340b
|
985 |
get_layout_hdr(lo); |
f49f9baac
|
986 987 988 989 |
if (list_empty(&lo->plh_segs)) first = true; spin_unlock(&ino->i_lock); if (first) { |
2130ff663
|
990 991 992 993 994 |
/* The lo must be on the clp list if there is any * chance of a CB_LAYOUTRECALL(FILE) coming in. */ spin_lock(&clp->cl_lock); BUG_ON(!list_empty(&lo->plh_layouts)); |
6382a4413
|
995 |
list_add_tail(&lo->plh_layouts, &server->layouts); |
2130ff663
|
996 997 |
spin_unlock(&clp->cl_lock); } |
e5e940170
|
998 |
|
707ed5fdb
|
999 1000 1001 1002 1003 |
pg_offset = arg.offset & ~PAGE_CACHE_MASK; if (pg_offset) { arg.offset -= pg_offset; arg.length += pg_offset; } |
7c24d9489
|
1004 1005 |
if (arg.length != NFS4_MAX_UINT64) arg.length = PAGE_CACHE_ALIGN(arg.length); |
707ed5fdb
|
1006 |
|
fb3296eb4
|
1007 |
lseg = send_layoutget(lo, ctx, &arg, gfp_flags); |
f49f9baac
|
1008 1009 1010 1011 |
if (!lseg && first) { spin_lock(&clp->cl_lock); list_del_init(&lo->plh_layouts); spin_unlock(&clp->cl_lock); |
2130ff663
|
1012 |
} |
cf7d63f1f
|
1013 |
atomic_dec(&lo->plh_outstanding); |
cc6e5340b
|
1014 |
put_layout_hdr(lo); |
e5e940170
|
1015 1016 1017 |
out: dprintk("%s end, state 0x%lx lseg %p ", __func__, |
bf9c1387c
|
1018 |
nfsi->layout ? nfsi->layout->plh_flags : -1, lseg); |
e5e940170
|
1019 1020 1021 1022 1023 |
return lseg; out_unlock: spin_unlock(&ino->i_lock); goto out; } |
7c24d9489
|
1024 |
EXPORT_SYMBOL_GPL(pnfs_update_layout); |
b1f69b754
|
1025 1026 1027 1028 1029 1030 1031 |
int pnfs_layout_process(struct nfs4_layoutget *lgp) { struct pnfs_layout_hdr *lo = NFS_I(lgp->args.inode)->layout; struct nfs4_layoutget_res *res = &lgp->res; struct pnfs_layout_segment *lseg; |
b7edfaa19
|
1032 |
struct inode *ino = lo->plh_inode; |
43f1b3da8
|
1033 |
struct nfs_client *clp = NFS_SERVER(ino)->nfs_client; |
b1f69b754
|
1034 1035 1036 |
int status = 0; /* Inject layout blob into I/O device driver */ |
a75b9df9d
|
1037 |
lseg = NFS_SERVER(ino)->pnfs_curr_ld->alloc_lseg(lo, res, lgp->gfp_flags); |
b1f69b754
|
1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 |
if (!lseg || IS_ERR(lseg)) { if (!lseg) status = -ENOMEM; else status = PTR_ERR(lseg); dprintk("%s: Could not allocate layout: error %d ", __func__, status); goto out; } spin_lock(&ino->i_lock); |
43f1b3da8
|
1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 |
if (test_bit(NFS4CLNT_LAYOUTRECALL, &clp->cl_state) || test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags)) { dprintk("%s forget reply due to recall ", __func__); goto out_forget_reply; } if (pnfs_layoutgets_blocked(lo, &res->stateid, 1)) { dprintk("%s forget reply due to state ", __func__); goto out_forget_reply; } |
b1f69b754
|
1062 |
init_lseg(lo, lseg); |
566052c53
|
1063 |
lseg->pls_range = res->range; |
d684d2ae1
|
1064 |
*lgp->lsegpp = get_lseg(lseg); |
b1f69b754
|
1065 |
pnfs_insert_layout(lo, lseg); |
f7e8917a6
|
1066 1067 1068 1069 |
if (res->return_on_close) { set_bit(NFS_LSEG_ROC, &lseg->pls_flags); set_bit(NFS_LAYOUT_ROC, &lo->plh_flags); } |
b1f69b754
|
1070 |
/* Done processing layoutget. Set the layout stateid */ |
43f1b3da8
|
1071 |
pnfs_set_layout_stateid(lo, &res->stateid, false); |
b1f69b754
|
1072 1073 1074 |
spin_unlock(&ino->i_lock); out: return status; |
43f1b3da8
|
1075 1076 1077 1078 1079 1080 |
out_forget_reply: spin_unlock(&ino->i_lock); lseg->pls_layout = lo; NFS_SERVER(ino)->pnfs_curr_ld->free_lseg(lseg); goto out; |
b1f69b754
|
1081 |
} |
d8007d4dd
|
1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 |
void pnfs_generic_pg_init_read(struct nfs_pageio_descriptor *pgio, struct nfs_page *req) { BUG_ON(pgio->pg_lseg != NULL); pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode, req->wb_context, req_offset(req), req->wb_bytes, IOMODE_READ, GFP_KERNEL); |
e885de1a5
|
1093 1094 |
/* If no lseg, fall back to read through mds */ if (pgio->pg_lseg == NULL) |
1f9453578
|
1095 |
nfs_pageio_reset_read_mds(pgio); |
e885de1a5
|
1096 |
|
d8007d4dd
|
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 |
} EXPORT_SYMBOL_GPL(pnfs_generic_pg_init_read); void pnfs_generic_pg_init_write(struct nfs_pageio_descriptor *pgio, struct nfs_page *req) { BUG_ON(pgio->pg_lseg != NULL); pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode, req->wb_context, req_offset(req), req->wb_bytes, IOMODE_RW, GFP_NOFS); |
e885de1a5
|
1111 1112 |
/* If no lseg, fall back to write through mds */ if (pgio->pg_lseg == NULL) |
1f9453578
|
1113 |
nfs_pageio_reset_write_mds(pgio); |
d8007d4dd
|
1114 1115 |
} EXPORT_SYMBOL_GPL(pnfs_generic_pg_init_write); |
18ad0a9f2
|
1116 |
bool |
1751c3638
|
1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 |
pnfs_pageio_init_read(struct nfs_pageio_descriptor *pgio, struct inode *inode) { struct nfs_server *server = NFS_SERVER(inode); struct pnfs_layoutdriver_type *ld = server->pnfs_curr_ld; if (ld == NULL) return false; nfs_pageio_init(pgio, inode, ld->pg_read_ops, server->rsize, 0); return true; } bool pnfs_pageio_init_write(struct nfs_pageio_descriptor *pgio, struct inode *inode, int ioflags) { struct nfs_server *server = NFS_SERVER(inode); struct pnfs_layoutdriver_type *ld = server->pnfs_curr_ld; if (ld == NULL) return false; nfs_pageio_init(pgio, inode, ld->pg_write_ops, server->wsize, ioflags); return true; } bool |
dfed206b8
|
1141 1142 |
pnfs_generic_pg_test(struct nfs_pageio_descriptor *pgio, struct nfs_page *prev, struct nfs_page *req) |
94ad1c80e
|
1143 |
{ |
d8007d4dd
|
1144 1145 |
if (pgio->pg_lseg == NULL) return nfs_generic_pg_test(pgio, prev, req); |
94ad1c80e
|
1146 |
|
19982ba85
|
1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 |
/* * Test if a nfs_page is fully contained in the pnfs_layout_range. * Note that this test makes several assumptions: * - that the previous nfs_page in the struct nfs_pageio_descriptor * is known to lie within the range. * - that the nfs_page being tested is known to be contiguous with the * previous nfs_page. * - Layout ranges are page aligned, so we only have to test the * start offset of the request. * * Please also note that 'end_offset' is actually the offset of the * first byte that lies outside the pnfs_layout_range. FIXME? * */ return req_offset(req) < end_offset(pgio->pg_lseg->pls_range.offset, pgio->pg_lseg->pls_range.length); |
94ad1c80e
|
1163 |
} |
89a58e32d
|
1164 |
EXPORT_SYMBOL_GPL(pnfs_generic_pg_test); |
94ad1c80e
|
1165 |
|
e2fecb215
|
1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 |
static int pnfs_write_done_resend_to_mds(struct inode *inode, struct list_head *head) { struct nfs_pageio_descriptor pgio; LIST_HEAD(failed); /* Resend all requests through the MDS */ nfs_pageio_init_write_mds(&pgio, inode, FLUSH_STABLE); while (!list_empty(head)) { struct nfs_page *req = nfs_list_entry(head->next); nfs_list_remove_request(req); if (!nfs_pageio_add_request(&pgio, req)) nfs_list_add_request(req, &failed); } nfs_pageio_complete(&pgio); if (!list_empty(&failed)) { /* For some reason our attempt to resend pages. Mark the * overall send request as having failed, and let * nfs_writeback_release_full deal with the error. */ list_move(&failed, head); return -EIO; } return 0; } |
d20581aa4
|
1192 1193 1194 |
/* * Called by non rpc-based layout drivers */ |
8ce160c5e
|
1195 |
void pnfs_ld_write_done(struct nfs_write_data *data) |
44b83799a
|
1196 |
{ |
8ce160c5e
|
1197 |
if (likely(!data->pnfs_error)) { |
d20581aa4
|
1198 1199 |
pnfs_set_layoutcommit(data); data->mds_ops->rpc_call_done(&data->task, data); |
8ce160c5e
|
1200 |
} else { |
8ce160c5e
|
1201 1202 |
dprintk("pnfs write error = %d ", data->pnfs_error); |
fe0fe8358
|
1203 1204 1205 1206 1207 1208 1209 1210 1211 |
if (NFS_SERVER(data->inode)->pnfs_curr_ld->flags & PNFS_LAYOUTRET_ON_ERROR) { /* Don't lo_commit on error, Server will needs to * preform a file recovery. */ clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(data->inode)->flags); pnfs_return_layout(data->inode); } |
e2fecb215
|
1212 |
data->task.tk_status = pnfs_write_done_resend_to_mds(data->inode, &data->pages); |
d20581aa4
|
1213 |
} |
8ce160c5e
|
1214 |
data->mds_ops->rpc_release(data); |
44b83799a
|
1215 |
} |
d20581aa4
|
1216 |
EXPORT_SYMBOL_GPL(pnfs_ld_write_done); |
44b83799a
|
1217 |
|
dce81290e
|
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 |
static void pnfs_write_through_mds(struct nfs_pageio_descriptor *desc, struct nfs_write_data *data) { list_splice_tail_init(&data->pages, &desc->pg_list); if (data->req && list_empty(&data->req->wb_list)) nfs_list_add_request(data->req, &desc->pg_list); nfs_pageio_reset_write_mds(desc); desc->pg_recoalesce = 1; nfs_writedata_release(data); } static enum pnfs_try_status |
0382b7440
|
1231 |
pnfs_try_to_write_data(struct nfs_write_data *wdata, |
dce81290e
|
1232 1233 1234 |
const struct rpc_call_ops *call_ops, struct pnfs_layout_segment *lseg, int how) |
0382b7440
|
1235 1236 1237 1238 1239 1240 |
{ struct inode *inode = wdata->inode; enum pnfs_try_status trypnfs; struct nfs_server *nfss = NFS_SERVER(inode); wdata->mds_ops = call_ops; |
dce81290e
|
1241 |
wdata->lseg = get_lseg(lseg); |
0382b7440
|
1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 |
dprintk("%s: Writing ino:%lu %u@%llu (how %d) ", __func__, inode->i_ino, wdata->args.count, wdata->args.offset, how); trypnfs = nfss->pnfs_curr_ld->write_pagelist(wdata, how); if (trypnfs == PNFS_NOT_ATTEMPTED) { put_lseg(wdata->lseg); wdata->lseg = NULL; } else nfs_inc_stats(inode, NFSIOS_PNFS_WRITE); dprintk("%s End (trypnfs:%d) ", __func__, trypnfs); return trypnfs; } |
dce81290e
|
1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 |
static void pnfs_do_multiple_writes(struct nfs_pageio_descriptor *desc, struct list_head *head, int how) { struct nfs_write_data *data; const struct rpc_call_ops *call_ops = desc->pg_rpc_callops; struct pnfs_layout_segment *lseg = desc->pg_lseg; desc->pg_lseg = NULL; while (!list_empty(head)) { enum pnfs_try_status trypnfs; data = list_entry(head->next, struct nfs_write_data, list); list_del_init(&data->list); trypnfs = pnfs_try_to_write_data(data, call_ops, lseg, how); if (trypnfs == PNFS_NOT_ATTEMPTED) pnfs_write_through_mds(desc, data); } put_lseg(lseg); } int pnfs_generic_pg_writepages(struct nfs_pageio_descriptor *desc) { LIST_HEAD(head); int ret; ret = nfs_generic_flush(desc, &head); if (ret != 0) { put_lseg(desc->pg_lseg); desc->pg_lseg = NULL; return ret; } pnfs_do_multiple_writes(desc, &head, desc->pg_ioflags); return 0; } EXPORT_SYMBOL_GPL(pnfs_generic_pg_writepages); |
62e4a7698
|
1295 1296 1297 1298 1299 1300 1301 1302 |
static void pnfs_ld_handle_read_error(struct nfs_read_data *data) { struct nfs_pageio_descriptor pgio; put_lseg(data->lseg); data->lseg = NULL; dprintk("pnfs write error = %d ", data->pnfs_error); |
fe0fe8358
|
1303 1304 1305 |
if (NFS_SERVER(data->inode)->pnfs_curr_ld->flags & PNFS_LAYOUTRET_ON_ERROR) pnfs_return_layout(data->inode); |
62e4a7698
|
1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 |
nfs_pageio_init_read_mds(&pgio, data->inode); while (!list_empty(&data->pages)) { struct nfs_page *req = nfs_list_entry(data->pages.next); nfs_list_remove_request(req); nfs_pageio_add_request(&pgio, req); } nfs_pageio_complete(&pgio); } |
b1f69b754
|
1317 |
/* |
d20581aa4
|
1318 1319 |
* Called by non rpc-based layout drivers */ |
9b7eecdcf
|
1320 |
void pnfs_ld_read_done(struct nfs_read_data *data) |
d20581aa4
|
1321 |
{ |
9b7eecdcf
|
1322 |
if (likely(!data->pnfs_error)) { |
d20581aa4
|
1323 1324 |
__nfs4_read_done_cb(data); data->mds_ops->rpc_call_done(&data->task, data); |
62e4a7698
|
1325 1326 |
} else pnfs_ld_handle_read_error(data); |
9b7eecdcf
|
1327 |
data->mds_ops->rpc_release(data); |
d20581aa4
|
1328 1329 |
} EXPORT_SYMBOL_GPL(pnfs_ld_read_done); |
493292ddc
|
1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 |
static void pnfs_read_through_mds(struct nfs_pageio_descriptor *desc, struct nfs_read_data *data) { list_splice_tail_init(&data->pages, &desc->pg_list); if (data->req && list_empty(&data->req->wb_list)) nfs_list_add_request(data->req, &desc->pg_list); nfs_pageio_reset_read_mds(desc); desc->pg_recoalesce = 1; nfs_readdata_release(data); } |
d20581aa4
|
1341 |
/* |
64419a9b2
|
1342 1343 |
* Call the appropriate parallel I/O subsystem read function. */ |
493292ddc
|
1344 |
static enum pnfs_try_status |
64419a9b2
|
1345 |
pnfs_try_to_read_data(struct nfs_read_data *rdata, |
493292ddc
|
1346 1347 |
const struct rpc_call_ops *call_ops, struct pnfs_layout_segment *lseg) |
64419a9b2
|
1348 1349 1350 1351 1352 1353 |
{ struct inode *inode = rdata->inode; struct nfs_server *nfss = NFS_SERVER(inode); enum pnfs_try_status trypnfs; rdata->mds_ops = call_ops; |
493292ddc
|
1354 |
rdata->lseg = get_lseg(lseg); |
64419a9b2
|
1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 |
dprintk("%s: Reading ino:%lu %u@%llu ", __func__, inode->i_ino, rdata->args.count, rdata->args.offset); trypnfs = nfss->pnfs_curr_ld->read_pagelist(rdata); if (trypnfs == PNFS_NOT_ATTEMPTED) { put_lseg(rdata->lseg); rdata->lseg = NULL; } else { nfs_inc_stats(inode, NFSIOS_PNFS_READ); } dprintk("%s End (trypnfs:%d) ", __func__, trypnfs); return trypnfs; } |
863a3c6c6
|
1371 |
|
493292ddc
|
1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 |
static void pnfs_do_multiple_reads(struct nfs_pageio_descriptor *desc, struct list_head *head) { struct nfs_read_data *data; const struct rpc_call_ops *call_ops = desc->pg_rpc_callops; struct pnfs_layout_segment *lseg = desc->pg_lseg; desc->pg_lseg = NULL; while (!list_empty(head)) { enum pnfs_try_status trypnfs; data = list_entry(head->next, struct nfs_read_data, list); list_del_init(&data->list); trypnfs = pnfs_try_to_read_data(data, call_ops, lseg); if (trypnfs == PNFS_NOT_ATTEMPTED) pnfs_read_through_mds(desc, data); } put_lseg(lseg); } int pnfs_generic_pg_readpages(struct nfs_pageio_descriptor *desc) { LIST_HEAD(head); int ret; ret = nfs_generic_pagein(desc, &head); if (ret != 0) { put_lseg(desc->pg_lseg); desc->pg_lseg = NULL; return ret; } pnfs_do_multiple_reads(desc, &head); return 0; } EXPORT_SYMBOL_GPL(pnfs_generic_pg_readpages); |
863a3c6c6
|
1409 |
/* |
a9bae5666
|
1410 |
* There can be multiple RW segments. |
863a3c6c6
|
1411 |
*/ |
a9bae5666
|
1412 |
static void pnfs_list_write_lseg(struct inode *inode, struct list_head *listp) |
863a3c6c6
|
1413 |
{ |
a9bae5666
|
1414 |
struct pnfs_layout_segment *lseg; |
863a3c6c6
|
1415 |
|
a9bae5666
|
1416 1417 1418 1419 1420 |
list_for_each_entry(lseg, &NFS_I(inode)->layout->plh_segs, pls_list) { if (lseg->pls_range.iomode == IOMODE_RW && test_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags)) list_add(&lseg->pls_lc_list, listp); } |
863a3c6c6
|
1421 |
} |
1b0ae0687
|
1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 |
void pnfs_set_lo_fail(struct pnfs_layout_segment *lseg) { if (lseg->pls_range.iomode == IOMODE_RW) { dprintk("%s Setting layout IOMODE_RW fail bit ", __func__); set_bit(lo_fail_bit(IOMODE_RW), &lseg->pls_layout->plh_flags); } else { dprintk("%s Setting layout IOMODE_READ fail bit ", __func__); set_bit(lo_fail_bit(IOMODE_READ), &lseg->pls_layout->plh_flags); } } EXPORT_SYMBOL_GPL(pnfs_set_lo_fail); |
863a3c6c6
|
1435 1436 1437 1438 |
void pnfs_set_layoutcommit(struct nfs_write_data *wdata) { struct nfs_inode *nfsi = NFS_I(wdata->inode); |
4b8ee2b82
|
1439 |
loff_t end_pos = wdata->mds_offset + wdata->res.count; |
79a48a1f5
|
1440 |
bool mark_as_dirty = false; |
863a3c6c6
|
1441 1442 1443 |
spin_lock(&nfsi->vfs_inode.i_lock); if (!test_and_set_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) { |
79a48a1f5
|
1444 |
mark_as_dirty = true; |
863a3c6c6
|
1445 1446 1447 |
dprintk("%s: Set layoutcommit for inode %lu ", __func__, wdata->inode->i_ino); } |
a9bae5666
|
1448 1449 1450 1451 |
if (!test_and_set_bit(NFS_LSEG_LAYOUTCOMMIT, &wdata->lseg->pls_flags)) { /* references matched in nfs4_layoutcommit_release */ get_lseg(wdata->lseg); } |
acff58805
|
1452 1453 |
if (end_pos > nfsi->layout->plh_lwb) nfsi->layout->plh_lwb = end_pos; |
863a3c6c6
|
1454 |
spin_unlock(&nfsi->vfs_inode.i_lock); |
acff58805
|
1455 1456 1457 |
dprintk("%s: lseg %p end_pos %llu ", __func__, wdata->lseg, nfsi->layout->plh_lwb); |
79a48a1f5
|
1458 1459 1460 1461 1462 |
/* if pnfs_layoutcommit_inode() runs between inode locks, the next one * will be a noop because NFS_INO_LAYOUTCOMMIT will not be set */ if (mark_as_dirty) mark_inode_dirty_sync(wdata->inode); |
863a3c6c6
|
1463 1464 |
} EXPORT_SYMBOL_GPL(pnfs_set_layoutcommit); |
db29c0890
|
1465 1466 1467 1468 1469 1470 1471 |
void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data) { struct nfs_server *nfss = NFS_SERVER(data->args.inode); if (nfss->pnfs_curr_ld->cleanup_layoutcommit) nfss->pnfs_curr_ld->cleanup_layoutcommit(data); } |
de4b15c7e
|
1472 1473 1474 1475 1476 1477 1478 1479 |
/* * For the LAYOUT4_NFSV4_1_FILES layout type, NFS_DATA_SYNC WRITEs and * NFS_UNSTABLE WRITEs with a COMMIT to data servers must store enough * data to disk to allow the server to recover the data if it crashes. * LAYOUTCOMMIT is only needed when the NFL4_UFLG_COMMIT_THRU_MDS flag * is off, and a COMMIT is sent to a data server, or * if WRITEs to a data server return NFS_DATA_SYNC. */ |
863a3c6c6
|
1480 |
int |
ef3115378
|
1481 |
pnfs_layoutcommit_inode(struct inode *inode, bool sync) |
863a3c6c6
|
1482 1483 1484 |
{ struct nfs4_layoutcommit_data *data; struct nfs_inode *nfsi = NFS_I(inode); |
863a3c6c6
|
1485 1486 1487 1488 1489 |
loff_t end_pos; int status = 0; dprintk("--> %s inode %lu ", __func__, inode->i_ino); |
de4b15c7e
|
1490 1491 |
if (!test_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) return 0; |
863a3c6c6
|
1492 1493 |
/* Note kzalloc ensures data->res.seq_res.sr_slot == NULL */ data = kzalloc(sizeof(*data), GFP_NOFS); |
de4b15c7e
|
1494 |
if (!data) { |
de4b15c7e
|
1495 1496 1497 |
status = -ENOMEM; goto out; } |
863a3c6c6
|
1498 |
|
92407e75c
|
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 |
if (!test_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) goto out_free; if (test_and_set_bit(NFS_INO_LAYOUTCOMMITTING, &nfsi->flags)) { if (!sync) { status = -EAGAIN; goto out_free; } status = wait_on_bit_lock(&nfsi->flags, NFS_INO_LAYOUTCOMMITTING, nfs_wait_bit_killable, TASK_KILLABLE); if (status) goto out_free; } |
a9bae5666
|
1512 |
INIT_LIST_HEAD(&data->lseg_list); |
de4b15c7e
|
1513 |
spin_lock(&inode->i_lock); |
863a3c6c6
|
1514 |
if (!test_and_clear_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) { |
92407e75c
|
1515 |
clear_bit(NFS_INO_LAYOUTCOMMITTING, &nfsi->flags); |
863a3c6c6
|
1516 |
spin_unlock(&inode->i_lock); |
92407e75c
|
1517 1518 |
wake_up_bit(&nfsi->flags, NFS_INO_LAYOUTCOMMITTING); goto out_free; |
863a3c6c6
|
1519 |
} |
a9bae5666
|
1520 1521 |
pnfs_list_write_lseg(inode, &data->lseg_list); |
863a3c6c6
|
1522 |
|
acff58805
|
1523 |
end_pos = nfsi->layout->plh_lwb; |
acff58805
|
1524 |
nfsi->layout->plh_lwb = 0; |
863a3c6c6
|
1525 |
|
de4b15c7e
|
1526 1527 |
memcpy(&data->args.stateid.data, nfsi->layout->plh_stateid.data, sizeof(nfsi->layout->plh_stateid.data)); |
863a3c6c6
|
1528 1529 1530 |
spin_unlock(&inode->i_lock); data->args.inode = inode; |
9fa407587
|
1531 |
data->cred = get_rpccred(nfsi->layout->plh_lc_cred); |
863a3c6c6
|
1532 1533 1534 1535 1536 1537 1538 1539 |
nfs_fattr_init(&data->fattr); data->args.bitmask = NFS_SERVER(inode)->cache_consistency_bitmask; data->res.fattr = &data->fattr; data->args.lastbytewritten = end_pos - 1; data->res.server = NFS_SERVER(inode); status = nfs4_proc_layoutcommit(data, sync); out: |
92407e75c
|
1540 1541 |
if (status) mark_inode_dirty_sync(inode); |
863a3c6c6
|
1542 1543 1544 |
dprintk("<-- %s status %d ", __func__, status); return status; |
92407e75c
|
1545 1546 1547 |
out_free: kfree(data); goto out; |
863a3c6c6
|
1548 |
} |