1// SPDX-License-Identifier: GPL-2.0-only 2/* 3 * linux/fs/nfs/inode.c 4 * 5 * Copyright (C) 1992 Rick Sladkey 6 * 7 * nfs inode and superblock handling functions 8 * 9 * Modularised by Alan Cox <alan@lxorguk.ukuu.org.uk>, while hacking some 10 * experimental NFS changes. Modularisation taken straight from SYS5 fs. 11 * 12 * Change to nfs_read_super() to permit NFS mounts to multi-homed hosts. 13 * J.S.Peatfield@damtp.cam.ac.uk 14 * 15 */ 16 17#include <linux/module.h> 18#include <linux/init.h> 19#include <linux/sched/signal.h> 20#include <linux/time.h> 21#include <linux/kernel.h> 22#include <linux/mm.h> 23#include <linux/string.h> 24#include <linux/stat.h> 25#include <linux/errno.h> 26#include <linux/unistd.h> 27#include <linux/sunrpc/clnt.h> 28#include <linux/sunrpc/stats.h> 29#include <linux/sunrpc/metrics.h> 30#include <linux/nfs_fs.h> 31#include <linux/nfs_mount.h> 32#include <linux/nfs4_mount.h> 33#include <linux/lockd/bind.h> 34#include <linux/seq_file.h> 35#include <linux/mount.h> 36#include <linux/vfs.h> 37#include <linux/inet.h> 38#include <linux/nfs_xdr.h> 39#include <linux/slab.h> 40#include <linux/compat.h> 41#include <linux/freezer.h> 42#include <linux/uaccess.h> 43#include <linux/iversion.h> 44 45#include "nfs4_fs.h" 46#include "callback.h" 47#include "delegation.h" 48#include "iostat.h" 49#include "internal.h" 50#include "fscache.h" 51#include "pnfs.h" 52#include "nfs.h" 53#include "netns.h" 54#include "sysfs.h" 55 56#include "nfstrace.h" 57 58#define NFSDBG_FACILITY NFSDBG_VFS 59 60#define NFS_64_BIT_INODE_NUMBERS_ENABLED 1 61 62/* Default is to see 64-bit inode numbers */ 63static bool enable_ino64 = NFS_64_BIT_INODE_NUMBERS_ENABLED; 64 65static int nfs_update_inode(struct inode *, struct nfs_fattr *); 66 67static struct kmem_cache * nfs_inode_cachep; 68 69static inline unsigned long 70nfs_fattr_to_ino_t(struct nfs_fattr *fattr) 71{ 72 return nfs_fileid_to_ino_t(fattr->fileid); 73} 74 75static int nfs_wait_killable(int mode) 76{ 77 freezable_schedule_unsafe(); 78 if (signal_pending_state(mode, current)) 79 return -ERESTARTSYS; 80 return 0; 81} 82 83int nfs_wait_bit_killable(struct wait_bit_key *key, int mode) 84{ 85 return nfs_wait_killable(mode); 86} 87EXPORT_SYMBOL_GPL(nfs_wait_bit_killable); 88 89/** 90 * nfs_compat_user_ino64 - returns the user-visible inode number 91 * @fileid: 64-bit fileid 92 * 93 * This function returns a 32-bit inode number if the boot parameter 94 * nfs.enable_ino64 is zero. 95 */ 96u64 nfs_compat_user_ino64(u64 fileid) 97{ 98#ifdef CONFIG_COMPAT 99 compat_ulong_t ino; 100#else 101 unsigned long ino; 102#endif 103 104 if (enable_ino64) 105 return fileid; 106 ino = fileid; 107 if (sizeof(ino) < sizeof(fileid)) 108 ino ^= fileid >> (sizeof(fileid)-sizeof(ino)) * 8; 109 return ino; 110} 111 112int nfs_drop_inode(struct inode *inode) 113{ 114 return NFS_STALE(inode) || generic_drop_inode(inode); 115} 116EXPORT_SYMBOL_GPL(nfs_drop_inode); 117 118void nfs_clear_inode(struct inode *inode) 119{ 120 /* 121 * The following should never happen... 122 */ 123 WARN_ON_ONCE(nfs_have_writebacks(inode)); 124 WARN_ON_ONCE(!list_empty(&NFS_I(inode)->open_files)); 125 nfs_zap_acl_cache(inode); 126 nfs_access_zap_cache(inode); 127 nfs_fscache_clear_inode(inode); 128} 129EXPORT_SYMBOL_GPL(nfs_clear_inode); 130 131void nfs_evict_inode(struct inode *inode) 132{ 133 truncate_inode_pages_final(&inode->i_data); 134 clear_inode(inode); 135 nfs_clear_inode(inode); 136} 137 138int nfs_sync_inode(struct inode *inode) 139{ 140 inode_dio_wait(inode); 141 return nfs_wb_all(inode); 142} 143EXPORT_SYMBOL_GPL(nfs_sync_inode); 144 145/** 146 * nfs_sync_mapping - helper to flush all mmapped dirty data to disk 147 * @mapping: pointer to struct address_space 148 */ 149int nfs_sync_mapping(struct address_space *mapping) 150{ 151 int ret = 0; 152 153 if (mapping->nrpages != 0) { 154 unmap_mapping_range(mapping, 0, 0, 0); 155 ret = nfs_wb_all(mapping->host); 156 } 157 return ret; 158} 159 160static int nfs_attribute_timeout(struct inode *inode) 161{ 162 struct nfs_inode *nfsi = NFS_I(inode); 163 164 return !time_in_range_open(jiffies, nfsi->read_cache_jiffies, nfsi->read_cache_jiffies + nfsi->attrtimeo); 165} 166 167static bool nfs_check_cache_invalid_delegated(struct inode *inode, unsigned long flags) 168{ 169 unsigned long cache_validity = READ_ONCE(NFS_I(inode)->cache_validity); 170 171 /* Special case for the pagecache or access cache */ 172 if (flags == NFS_INO_REVAL_PAGECACHE && 173 !(cache_validity & NFS_INO_REVAL_FORCED)) 174 return false; 175 return (cache_validity & flags) != 0; 176} 177 178static bool nfs_check_cache_invalid_not_delegated(struct inode *inode, unsigned long flags) 179{ 180 unsigned long cache_validity = READ_ONCE(NFS_I(inode)->cache_validity); 181 182 if ((cache_validity & flags) != 0) 183 return true; 184 if (nfs_attribute_timeout(inode)) 185 return true; 186 return false; 187} 188 189bool nfs_check_cache_invalid(struct inode *inode, unsigned long flags) 190{ 191 if (NFS_PROTO(inode)->have_delegation(inode, FMODE_READ)) 192 return nfs_check_cache_invalid_delegated(inode, flags); 193 194 return nfs_check_cache_invalid_not_delegated(inode, flags); 195} 196EXPORT_SYMBOL_GPL(nfs_check_cache_invalid); 197 198#ifdef CONFIG_NFS_V4_2 199static bool nfs_has_xattr_cache(const struct nfs_inode *nfsi) 200{ 201 return nfsi->xattr_cache != NULL; 202} 203#else 204static bool nfs_has_xattr_cache(const struct nfs_inode *nfsi) 205{ 206 return false; 207} 208#endif 209 210static void nfs_set_cache_invalid(struct inode *inode, unsigned long flags) 211{ 212 struct nfs_inode *nfsi = NFS_I(inode); 213 bool have_delegation = NFS_PROTO(inode)->have_delegation(inode, FMODE_READ); 214 215 if (have_delegation) { 216 if (!(flags & NFS_INO_REVAL_FORCED)) 217 flags &= ~NFS_INO_INVALID_OTHER; 218 flags &= ~(NFS_INO_INVALID_CHANGE 219 | NFS_INO_INVALID_SIZE 220 | NFS_INO_REVAL_PAGECACHE 221 | NFS_INO_INVALID_XATTR); 222 } else if (flags & NFS_INO_REVAL_PAGECACHE) 223 flags |= NFS_INO_INVALID_CHANGE | NFS_INO_INVALID_SIZE; 224 225 if (!nfs_has_xattr_cache(nfsi)) 226 flags &= ~NFS_INO_INVALID_XATTR; 227 if (inode->i_mapping->nrpages == 0) 228 flags &= ~(NFS_INO_INVALID_DATA|NFS_INO_DATA_INVAL_DEFER); 229 nfsi->cache_validity |= flags; 230 if (flags & NFS_INO_INVALID_DATA) 231 nfs_fscache_invalidate(inode); 232} 233 234/* 235 * Invalidate the local caches 236 */ 237static void nfs_zap_caches_locked(struct inode *inode) 238{ 239 struct nfs_inode *nfsi = NFS_I(inode); 240 int mode = inode->i_mode; 241 242 nfs_inc_stats(inode, NFSIOS_ATTRINVALIDATE); 243 244 nfsi->attrtimeo = NFS_MINATTRTIMEO(inode); 245 nfsi->attrtimeo_timestamp = jiffies; 246 247 memset(NFS_I(inode)->cookieverf, 0, sizeof(NFS_I(inode)->cookieverf)); 248 if (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)) { 249 nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR 250 | NFS_INO_INVALID_DATA 251 | NFS_INO_INVALID_ACCESS 252 | NFS_INO_INVALID_ACL 253 | NFS_INO_INVALID_XATTR 254 | NFS_INO_REVAL_PAGECACHE); 255 } else 256 nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR 257 | NFS_INO_INVALID_ACCESS 258 | NFS_INO_INVALID_ACL 259 | NFS_INO_INVALID_XATTR 260 | NFS_INO_REVAL_PAGECACHE); 261 nfs_zap_label_cache_locked(nfsi); 262} 263 264void nfs_zap_caches(struct inode *inode) 265{ 266 spin_lock(&inode->i_lock); 267 nfs_zap_caches_locked(inode); 268 spin_unlock(&inode->i_lock); 269} 270 271void nfs_zap_mapping(struct inode *inode, struct address_space *mapping) 272{ 273 if (mapping->nrpages != 0) { 274 spin_lock(&inode->i_lock); 275 nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA); 276 spin_unlock(&inode->i_lock); 277 } 278} 279 280void nfs_zap_acl_cache(struct inode *inode) 281{ 282 void (*clear_acl_cache)(struct inode *); 283 284 clear_acl_cache = NFS_PROTO(inode)->clear_acl_cache; 285 if (clear_acl_cache != NULL) 286 clear_acl_cache(inode); 287 spin_lock(&inode->i_lock); 288 NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_ACL; 289 spin_unlock(&inode->i_lock); 290} 291EXPORT_SYMBOL_GPL(nfs_zap_acl_cache); 292 293void nfs_invalidate_atime(struct inode *inode) 294{ 295 spin_lock(&inode->i_lock); 296 nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATIME); 297 spin_unlock(&inode->i_lock); 298} 299EXPORT_SYMBOL_GPL(nfs_invalidate_atime); 300 301/* 302 * Invalidate, but do not unhash, the inode. 303 * NB: must be called with inode->i_lock held! 304 */ 305static void nfs_set_inode_stale_locked(struct inode *inode) 306{ 307 set_bit(NFS_INO_STALE, &NFS_I(inode)->flags); 308 nfs_zap_caches_locked(inode); 309 trace_nfs_set_inode_stale(inode); 310} 311 312void nfs_set_inode_stale(struct inode *inode) 313{ 314 spin_lock(&inode->i_lock); 315 nfs_set_inode_stale_locked(inode); 316 spin_unlock(&inode->i_lock); 317} 318 319struct nfs_find_desc { 320 struct nfs_fh *fh; 321 struct nfs_fattr *fattr; 322}; 323 324/* 325 * In NFSv3 we can have 64bit inode numbers. In order to support 326 * this, and re-exported directories (also seen in NFSv2) 327 * we are forced to allow 2 different inodes to have the same 328 * i_ino. 329 */ 330static int 331nfs_find_actor(struct inode *inode, void *opaque) 332{ 333 struct nfs_find_desc *desc = (struct nfs_find_desc *)opaque; 334 struct nfs_fh *fh = desc->fh; 335 struct nfs_fattr *fattr = desc->fattr; 336 337 if (NFS_FILEID(inode) != fattr->fileid) 338 return 0; 339 if (inode_wrong_type(inode, fattr->mode)) 340 return 0; 341 if (nfs_compare_fh(NFS_FH(inode), fh)) 342 return 0; 343 if (is_bad_inode(inode) || NFS_STALE(inode)) 344 return 0; 345 return 1; 346} 347 348static int 349nfs_init_locked(struct inode *inode, void *opaque) 350{ 351 struct nfs_find_desc *desc = (struct nfs_find_desc *)opaque; 352 struct nfs_fattr *fattr = desc->fattr; 353 354 set_nfs_fileid(inode, fattr->fileid); 355 inode->i_mode = fattr->mode; 356 nfs_copy_fh(NFS_FH(inode), desc->fh); 357 return 0; 358} 359 360#ifdef CONFIG_NFS_V4_SECURITY_LABEL 361static void nfs_clear_label_invalid(struct inode *inode) 362{ 363 spin_lock(&inode->i_lock); 364 NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_LABEL; 365 spin_unlock(&inode->i_lock); 366} 367 368void nfs_setsecurity(struct inode *inode, struct nfs_fattr *fattr, 369 struct nfs4_label *label) 370{ 371 int error; 372 373 if (label == NULL) 374 return; 375 376 if ((fattr->valid & NFS_ATTR_FATTR_V4_SECURITY_LABEL) && inode->i_security) { 377 error = security_inode_notifysecctx(inode, label->label, 378 label->len); 379 if (error) 380 printk(KERN_ERR "%s() %s %d " 381 "security_inode_notifysecctx() %d\n", 382 __func__, 383 (char *)label->label, 384 label->len, error); 385 nfs_clear_label_invalid(inode); 386 } 387} 388 389struct nfs4_label *nfs4_label_alloc(struct nfs_server *server, gfp_t flags) 390{ 391 struct nfs4_label *label = NULL; 392 int minor_version = server->nfs_client->cl_minorversion; 393 394 if (minor_version < 2) 395 return label; 396 397 if (!(server->caps & NFS_CAP_SECURITY_LABEL)) 398 return label; 399 400 label = kzalloc(sizeof(struct nfs4_label), flags); 401 if (label == NULL) 402 return ERR_PTR(-ENOMEM); 403 404 label->label = kzalloc(NFS4_MAXLABELLEN, flags); 405 if (label->label == NULL) { 406 kfree(label); 407 return ERR_PTR(-ENOMEM); 408 } 409 label->len = NFS4_MAXLABELLEN; 410 411 return label; 412} 413EXPORT_SYMBOL_GPL(nfs4_label_alloc); 414#else 415void nfs_setsecurity(struct inode *inode, struct nfs_fattr *fattr, 416 struct nfs4_label *label) 417{ 418} 419#endif 420EXPORT_SYMBOL_GPL(nfs_setsecurity); 421 422/* Search for inode identified by fh, fileid and i_mode in inode cache. */ 423struct inode * 424nfs_ilookup(struct super_block *sb, struct nfs_fattr *fattr, struct nfs_fh *fh) 425{ 426 struct nfs_find_desc desc = { 427 .fh = fh, 428 .fattr = fattr, 429 }; 430 struct inode *inode; 431 unsigned long hash; 432 433 if (!(fattr->valid & NFS_ATTR_FATTR_FILEID) || 434 !(fattr->valid & NFS_ATTR_FATTR_TYPE)) 435 return NULL; 436 437 hash = nfs_fattr_to_ino_t(fattr); 438 inode = ilookup5(sb, hash, nfs_find_actor, &desc); 439 440 dprintk("%s: returning %p\n", __func__, inode); 441 return inode; 442} 443 444/* 445 * This is our front-end to iget that looks up inodes by file handle 446 * instead of inode number. 447 */ 448struct inode * 449nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr, struct nfs4_label *label) 450{ 451 struct nfs_find_desc desc = { 452 .fh = fh, 453 .fattr = fattr 454 }; 455 struct inode *inode = ERR_PTR(-ENOENT); 456 unsigned long hash; 457 458 nfs_attr_check_mountpoint(sb, fattr); 459 460 if (nfs_attr_use_mounted_on_fileid(fattr)) 461 fattr->fileid = fattr->mounted_on_fileid; 462 else if ((fattr->valid & NFS_ATTR_FATTR_FILEID) == 0) 463 goto out_no_inode; 464 if ((fattr->valid & NFS_ATTR_FATTR_TYPE) == 0) 465 goto out_no_inode; 466 467 hash = nfs_fattr_to_ino_t(fattr); 468 469 inode = iget5_locked(sb, hash, nfs_find_actor, nfs_init_locked, &desc); 470 if (inode == NULL) { 471 inode = ERR_PTR(-ENOMEM); 472 goto out_no_inode; 473 } 474 475 if (inode->i_state & I_NEW) { 476 struct nfs_inode *nfsi = NFS_I(inode); 477 unsigned long now = jiffies; 478 479 /* We set i_ino for the few things that still rely on it, 480 * such as stat(2) */ 481 inode->i_ino = hash; 482 483 /* We can't support update_atime(), since the server will reset it */ 484 inode->i_flags |= S_NOATIME|S_NOCMTIME; 485 inode->i_mode = fattr->mode; 486 nfsi->cache_validity = 0; 487 if ((fattr->valid & NFS_ATTR_FATTR_MODE) == 0 488 && nfs_server_capable(inode, NFS_CAP_MODE)) 489 nfs_set_cache_invalid(inode, NFS_INO_INVALID_OTHER); 490 /* Why so? Because we want revalidate for devices/FIFOs, and 491 * that's precisely what we have in nfs_file_inode_operations. 492 */ 493 inode->i_op = NFS_SB(sb)->nfs_client->rpc_ops->file_inode_ops; 494 if (S_ISREG(inode->i_mode)) { 495 inode->i_fop = NFS_SB(sb)->nfs_client->rpc_ops->file_ops; 496 inode->i_data.a_ops = &nfs_file_aops; 497 } else if (S_ISDIR(inode->i_mode)) { 498 inode->i_op = NFS_SB(sb)->nfs_client->rpc_ops->dir_inode_ops; 499 inode->i_fop = &nfs_dir_operations; 500 inode->i_data.a_ops = &nfs_dir_aops; 501 /* Deal with crossing mountpoints */ 502 if (fattr->valid & NFS_ATTR_FATTR_MOUNTPOINT || 503 fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL) { 504 if (fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL) 505 inode->i_op = &nfs_referral_inode_operations; 506 else 507 inode->i_op = &nfs_mountpoint_inode_operations; 508 inode->i_fop = NULL; 509 inode->i_flags |= S_AUTOMOUNT; 510 } 511 } else if (S_ISLNK(inode->i_mode)) { 512 inode->i_op = &nfs_symlink_inode_operations; 513 inode_nohighmem(inode); 514 } else 515 init_special_inode(inode, inode->i_mode, fattr->rdev); 516 517 memset(&inode->i_atime, 0, sizeof(inode->i_atime)); 518 memset(&inode->i_mtime, 0, sizeof(inode->i_mtime)); 519 memset(&inode->i_ctime, 0, sizeof(inode->i_ctime)); 520 inode_set_iversion_raw(inode, 0); 521 inode->i_size = 0; 522 clear_nlink(inode); 523 inode->i_uid = make_kuid(&init_user_ns, -2); 524 inode->i_gid = make_kgid(&init_user_ns, -2); 525 inode->i_blocks = 0; 526 memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf)); 527 nfsi->write_io = 0; 528 nfsi->read_io = 0; 529 530 nfsi->read_cache_jiffies = fattr->time_start; 531 nfsi->attr_gencount = fattr->gencount; 532 if (fattr->valid & NFS_ATTR_FATTR_ATIME) 533 inode->i_atime = fattr->atime; 534 else if (nfs_server_capable(inode, NFS_CAP_ATIME)) 535 nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATIME); 536 if (fattr->valid & NFS_ATTR_FATTR_MTIME) 537 inode->i_mtime = fattr->mtime; 538 else if (nfs_server_capable(inode, NFS_CAP_MTIME)) 539 nfs_set_cache_invalid(inode, NFS_INO_INVALID_MTIME); 540 if (fattr->valid & NFS_ATTR_FATTR_CTIME) 541 inode->i_ctime = fattr->ctime; 542 else if (nfs_server_capable(inode, NFS_CAP_CTIME)) 543 nfs_set_cache_invalid(inode, NFS_INO_INVALID_CTIME); 544 if (fattr->valid & NFS_ATTR_FATTR_CHANGE) 545 inode_set_iversion_raw(inode, fattr->change_attr); 546 else 547 nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE); 548 if (fattr->valid & NFS_ATTR_FATTR_SIZE) 549 inode->i_size = nfs_size_to_loff_t(fattr->size); 550 else 551 nfs_set_cache_invalid(inode, NFS_INO_INVALID_SIZE); 552 if (fattr->valid & NFS_ATTR_FATTR_NLINK) 553 set_nlink(inode, fattr->nlink); 554 else if (nfs_server_capable(inode, NFS_CAP_NLINK)) 555 nfs_set_cache_invalid(inode, NFS_INO_INVALID_OTHER); 556 if (fattr->valid & NFS_ATTR_FATTR_OWNER) 557 inode->i_uid = fattr->uid; 558 else if (nfs_server_capable(inode, NFS_CAP_OWNER)) 559 nfs_set_cache_invalid(inode, NFS_INO_INVALID_OTHER); 560 if (fattr->valid & NFS_ATTR_FATTR_GROUP) 561 inode->i_gid = fattr->gid; 562 else if (nfs_server_capable(inode, NFS_CAP_OWNER_GROUP)) 563 nfs_set_cache_invalid(inode, NFS_INO_INVALID_OTHER); 564 if (nfs_server_capable(inode, NFS_CAP_XATTR)) 565 nfs_set_cache_invalid(inode, NFS_INO_INVALID_XATTR); 566 if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED) 567 inode->i_blocks = fattr->du.nfs2.blocks; 568 if (fattr->valid & NFS_ATTR_FATTR_SPACE_USED) { 569 /* 570 * report the blocks in 512byte units 571 */ 572 inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used); 573 } 574 575 if (nfsi->cache_validity != 0) 576 nfsi->cache_validity |= NFS_INO_REVAL_FORCED; 577 578 nfs_setsecurity(inode, fattr, label); 579 580 nfsi->attrtimeo = NFS_MINATTRTIMEO(inode); 581 nfsi->attrtimeo_timestamp = now; 582 nfsi->access_cache = RB_ROOT; 583 584 nfs_fscache_init_inode(inode); 585 586 unlock_new_inode(inode); 587 } else { 588 int err = nfs_refresh_inode(inode, fattr); 589 if (err < 0) { 590 iput(inode); 591 inode = ERR_PTR(err); 592 goto out_no_inode; 593 } 594 } 595 dprintk("NFS: nfs_fhget(%s/%Lu fh_crc=0x%08x ct=%d)\n", 596 inode->i_sb->s_id, 597 (unsigned long long)NFS_FILEID(inode), 598 nfs_display_fhandle_hash(fh), 599 atomic_read(&inode->i_count)); 600 601out: 602 return inode; 603 604out_no_inode: 605 dprintk("nfs_fhget: iget failed with error %ld\n", PTR_ERR(inode)); 606 goto out; 607} 608EXPORT_SYMBOL_GPL(nfs_fhget); 609 610#define NFS_VALID_ATTRS (ATTR_MODE|ATTR_UID|ATTR_GID|ATTR_SIZE|ATTR_ATIME|ATTR_ATIME_SET|ATTR_MTIME|ATTR_MTIME_SET|ATTR_FILE|ATTR_OPEN) 611 612int 613nfs_setattr(struct dentry *dentry, struct iattr *attr) 614{ 615 struct inode *inode = d_inode(dentry); 616 struct nfs_fattr *fattr; 617 int error = 0; 618 619 nfs_inc_stats(inode, NFSIOS_VFSSETATTR); 620 621 /* skip mode change if it's just for clearing setuid/setgid */ 622 if (attr->ia_valid & (ATTR_KILL_SUID | ATTR_KILL_SGID)) 623 attr->ia_valid &= ~ATTR_MODE; 624 625 if (attr->ia_valid & ATTR_SIZE) { 626 BUG_ON(!S_ISREG(inode->i_mode)); 627 628 error = inode_newsize_ok(inode, attr->ia_size); 629 if (error) 630 return error; 631 632 if (attr->ia_size == i_size_read(inode)) 633 attr->ia_valid &= ~ATTR_SIZE; 634 } 635 636 /* Optimization: if the end result is no change, don't RPC */ 637 attr->ia_valid &= NFS_VALID_ATTRS; 638 if ((attr->ia_valid & ~(ATTR_FILE|ATTR_OPEN)) == 0) 639 return 0; 640 641 trace_nfs_setattr_enter(inode); 642 643 /* Write all dirty data */ 644 if (S_ISREG(inode->i_mode)) 645 nfs_sync_inode(inode); 646 647 fattr = nfs_alloc_fattr(); 648 if (fattr == NULL) { 649 error = -ENOMEM; 650 goto out; 651 } 652 653 error = NFS_PROTO(inode)->setattr(dentry, fattr, attr); 654 if (error == 0) 655 error = nfs_refresh_inode(inode, fattr); 656 nfs_free_fattr(fattr); 657out: 658 trace_nfs_setattr_exit(inode, error); 659 return error; 660} 661EXPORT_SYMBOL_GPL(nfs_setattr); 662 663/** 664 * nfs_vmtruncate - unmap mappings "freed" by truncate() syscall 665 * @inode: inode of the file used 666 * @offset: file offset to start truncating 667 * 668 * This is a copy of the common vmtruncate, but with the locking 669 * corrected to take into account the fact that NFS requires 670 * inode->i_size to be updated under the inode->i_lock. 671 * Note: must be called with inode->i_lock held! 672 */ 673static int nfs_vmtruncate(struct inode * inode, loff_t offset) 674{ 675 int err; 676 677 err = inode_newsize_ok(inode, offset); 678 if (err) 679 goto out; 680 681 i_size_write(inode, offset); 682 /* Optimisation */ 683 if (offset == 0) 684 NFS_I(inode)->cache_validity &= ~(NFS_INO_INVALID_DATA | 685 NFS_INO_DATA_INVAL_DEFER); 686 NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_SIZE; 687 688 spin_unlock(&inode->i_lock); 689 truncate_pagecache(inode, offset); 690 spin_lock(&inode->i_lock); 691out: 692 return err; 693} 694 695/** 696 * nfs_setattr_update_inode - Update inode metadata after a setattr call. 697 * @inode: pointer to struct inode 698 * @attr: pointer to struct iattr 699 * @fattr: pointer to struct nfs_fattr 700 * 701 * Note: we do this in the *proc.c in order to ensure that 702 * it works for things like exclusive creates too. 703 */ 704void nfs_setattr_update_inode(struct inode *inode, struct iattr *attr, 705 struct nfs_fattr *fattr) 706{ 707 /* Barrier: bump the attribute generation count. */ 708 nfs_fattr_set_barrier(fattr); 709 710 spin_lock(&inode->i_lock); 711 NFS_I(inode)->attr_gencount = fattr->gencount; 712 if ((attr->ia_valid & ATTR_SIZE) != 0) { 713 nfs_set_cache_invalid(inode, NFS_INO_INVALID_MTIME); 714 nfs_inc_stats(inode, NFSIOS_SETATTRTRUNC); 715 nfs_vmtruncate(inode, attr->ia_size); 716 } 717 if ((attr->ia_valid & (ATTR_MODE|ATTR_UID|ATTR_GID)) != 0) { 718 NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_CTIME; 719 if ((attr->ia_valid & ATTR_MODE) != 0) { 720 int mode = attr->ia_mode & S_IALLUGO; 721 mode |= inode->i_mode & ~S_IALLUGO; 722 inode->i_mode = mode; 723 } 724 if ((attr->ia_valid & ATTR_UID) != 0) 725 inode->i_uid = attr->ia_uid; 726 if ((attr->ia_valid & ATTR_GID) != 0) 727 inode->i_gid = attr->ia_gid; 728 if (fattr->valid & NFS_ATTR_FATTR_CTIME) 729 inode->i_ctime = fattr->ctime; 730 else 731 nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE 732 | NFS_INO_INVALID_CTIME); 733 nfs_set_cache_invalid(inode, NFS_INO_INVALID_ACCESS 734 | NFS_INO_INVALID_ACL); 735 } 736 if (attr->ia_valid & (ATTR_ATIME_SET|ATTR_ATIME)) { 737 NFS_I(inode)->cache_validity &= ~(NFS_INO_INVALID_ATIME 738 | NFS_INO_INVALID_CTIME); 739 if (fattr->valid & NFS_ATTR_FATTR_ATIME) 740 inode->i_atime = fattr->atime; 741 else if (attr->ia_valid & ATTR_ATIME_SET) 742 inode->i_atime = attr->ia_atime; 743 else 744 nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATIME); 745 746 if (fattr->valid & NFS_ATTR_FATTR_CTIME) 747 inode->i_ctime = fattr->ctime; 748 else 749 nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE 750 | NFS_INO_INVALID_CTIME); 751 } 752 if (attr->ia_valid & (ATTR_MTIME_SET|ATTR_MTIME)) { 753 NFS_I(inode)->cache_validity &= ~(NFS_INO_INVALID_MTIME 754 | NFS_INO_INVALID_CTIME); 755 if (fattr->valid & NFS_ATTR_FATTR_MTIME) 756 inode->i_mtime = fattr->mtime; 757 else if (attr->ia_valid & ATTR_MTIME_SET) 758 inode->i_mtime = attr->ia_mtime; 759 else 760 nfs_set_cache_invalid(inode, NFS_INO_INVALID_MTIME); 761 762 if (fattr->valid & NFS_ATTR_FATTR_CTIME) 763 inode->i_ctime = fattr->ctime; 764 else 765 nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE 766 | NFS_INO_INVALID_CTIME); 767 } 768 if (fattr->valid) 769 nfs_update_inode(inode, fattr); 770 spin_unlock(&inode->i_lock); 771} 772EXPORT_SYMBOL_GPL(nfs_setattr_update_inode); 773 774static void nfs_readdirplus_parent_cache_miss(struct dentry *dentry) 775{ 776 struct dentry *parent; 777 778 if (!nfs_server_capable(d_inode(dentry), NFS_CAP_READDIRPLUS)) 779 return; 780 parent = dget_parent(dentry); 781 nfs_force_use_readdirplus(d_inode(parent)); 782 dput(parent); 783} 784 785static void nfs_readdirplus_parent_cache_hit(struct dentry *dentry) 786{ 787 struct dentry *parent; 788 789 if (!nfs_server_capable(d_inode(dentry), NFS_CAP_READDIRPLUS)) 790 return; 791 parent = dget_parent(dentry); 792 nfs_advise_use_readdirplus(d_inode(parent)); 793 dput(parent); 794} 795 796static bool nfs_need_revalidate_inode(struct inode *inode) 797{ 798 if (NFS_I(inode)->cache_validity & 799 (NFS_INO_INVALID_ATTR|NFS_INO_INVALID_LABEL)) 800 return true; 801 if (nfs_attribute_cache_expired(inode)) 802 return true; 803 return false; 804} 805 806int nfs_getattr(const struct path *path, struct kstat *stat, 807 u32 request_mask, unsigned int query_flags) 808{ 809 struct inode *inode = d_inode(path->dentry); 810 struct nfs_server *server = NFS_SERVER(inode); 811 unsigned long cache_validity; 812 int err = 0; 813 bool force_sync = query_flags & AT_STATX_FORCE_SYNC; 814 bool do_update = false; 815 816 trace_nfs_getattr_enter(inode); 817 818 if ((query_flags & AT_STATX_DONT_SYNC) && !force_sync) { 819 nfs_readdirplus_parent_cache_hit(path->dentry); 820 goto out_no_update; 821 } 822 823 /* Flush out writes to the server in order to update c/mtime. */ 824 if ((request_mask & (STATX_CTIME | STATX_MTIME)) && 825 S_ISREG(inode->i_mode)) 826 filemap_write_and_wait(inode->i_mapping); 827 828 /* 829 * We may force a getattr if the user cares about atime. 830 * 831 * Note that we only have to check the vfsmount flags here: 832 * - NFS always sets S_NOATIME by so checking it would give a 833 * bogus result 834 * - NFS never sets SB_NOATIME or SB_NODIRATIME so there is 835 * no point in checking those. 836 */ 837 if ((path->mnt->mnt_flags & MNT_NOATIME) || 838 ((path->mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode))) 839 request_mask &= ~STATX_ATIME; 840 841 /* Is the user requesting attributes that might need revalidation? */ 842 if (!(request_mask & (STATX_MODE|STATX_NLINK|STATX_ATIME|STATX_CTIME| 843 STATX_MTIME|STATX_UID|STATX_GID| 844 STATX_SIZE|STATX_BLOCKS))) 845 goto out_no_revalidate; 846 847 /* Check whether the cached attributes are stale */ 848 do_update |= force_sync || nfs_attribute_cache_expired(inode); 849 cache_validity = READ_ONCE(NFS_I(inode)->cache_validity); 850 do_update |= cache_validity & 851 (NFS_INO_INVALID_ATTR|NFS_INO_INVALID_LABEL); 852 if (request_mask & STATX_ATIME) 853 do_update |= cache_validity & NFS_INO_INVALID_ATIME; 854 if (request_mask & (STATX_CTIME|STATX_MTIME)) 855 do_update |= cache_validity & NFS_INO_REVAL_PAGECACHE; 856 if (request_mask & STATX_BLOCKS) 857 do_update |= cache_validity & NFS_INO_INVALID_BLOCKS; 858 if (do_update) { 859 /* Update the attribute cache */ 860 if (!(server->flags & NFS_MOUNT_NOAC)) 861 nfs_readdirplus_parent_cache_miss(path->dentry); 862 else 863 nfs_readdirplus_parent_cache_hit(path->dentry); 864 err = __nfs_revalidate_inode(server, inode); 865 if (err) 866 goto out; 867 } else 868 nfs_readdirplus_parent_cache_hit(path->dentry); 869out_no_revalidate: 870 /* Only return attributes that were revalidated. */ 871 stat->result_mask &= request_mask; 872out_no_update: 873 generic_fillattr(inode, stat); 874 stat->ino = nfs_compat_user_ino64(NFS_FILEID(inode)); 875 if (S_ISDIR(inode->i_mode)) 876 stat->blksize = NFS_SERVER(inode)->dtsize; 877out: 878 trace_nfs_getattr_exit(inode, err); 879 return err; 880} 881EXPORT_SYMBOL_GPL(nfs_getattr); 882 883static void nfs_init_lock_context(struct nfs_lock_context *l_ctx) 884{ 885 refcount_set(&l_ctx->count, 1); 886 l_ctx->lockowner = current->files; 887 INIT_LIST_HEAD(&l_ctx->list); 888 atomic_set(&l_ctx->io_count, 0); 889} 890 891static struct nfs_lock_context *__nfs_find_lock_context(struct nfs_open_context *ctx) 892{ 893 struct nfs_lock_context *pos; 894 895 list_for_each_entry_rcu(pos, &ctx->lock_context.list, list) { 896 if (pos->lockowner != current->files) 897 continue; 898 if (refcount_inc_not_zero(&pos->count)) 899 return pos; 900 } 901 return NULL; 902} 903 904struct nfs_lock_context *nfs_get_lock_context(struct nfs_open_context *ctx) 905{ 906 struct nfs_lock_context *res, *new = NULL; 907 struct inode *inode = d_inode(ctx->dentry); 908 909 rcu_read_lock(); 910 res = __nfs_find_lock_context(ctx); 911 rcu_read_unlock(); 912 if (res == NULL) { 913 new = kmalloc(sizeof(*new), GFP_KERNEL); 914 if (new == NULL) 915 return ERR_PTR(-ENOMEM); 916 nfs_init_lock_context(new); 917 spin_lock(&inode->i_lock); 918 res = __nfs_find_lock_context(ctx); 919 if (res == NULL) { 920 new->open_context = get_nfs_open_context(ctx); 921 if (new->open_context) { 922 list_add_tail_rcu(&new->list, 923 &ctx->lock_context.list); 924 res = new; 925 new = NULL; 926 } else 927 res = ERR_PTR(-EBADF); 928 } 929 spin_unlock(&inode->i_lock); 930 kfree(new); 931 } 932 return res; 933} 934EXPORT_SYMBOL_GPL(nfs_get_lock_context); 935 936void nfs_put_lock_context(struct nfs_lock_context *l_ctx) 937{ 938 struct nfs_open_context *ctx = l_ctx->open_context; 939 struct inode *inode = d_inode(ctx->dentry); 940 941 if (!refcount_dec_and_lock(&l_ctx->count, &inode->i_lock)) 942 return; 943 list_del_rcu(&l_ctx->list); 944 spin_unlock(&inode->i_lock); 945 put_nfs_open_context(ctx); 946 kfree_rcu(l_ctx, rcu_head); 947} 948EXPORT_SYMBOL_GPL(nfs_put_lock_context); 949 950/** 951 * nfs_close_context - Common close_context() routine NFSv2/v3 952 * @ctx: pointer to context 953 * @is_sync: is this a synchronous close 954 * 955 * Ensure that the attributes are up to date if we're mounted 956 * with close-to-open semantics and we have cached data that will 957 * need to be revalidated on open. 958 */ 959void nfs_close_context(struct nfs_open_context *ctx, int is_sync) 960{ 961 struct nfs_inode *nfsi; 962 struct inode *inode; 963 struct nfs_server *server; 964 965 if (!(ctx->mode & FMODE_WRITE)) 966 return; 967 if (!is_sync) 968 return; 969 inode = d_inode(ctx->dentry); 970 if (NFS_PROTO(inode)->have_delegation(inode, FMODE_READ)) 971 return; 972 nfsi = NFS_I(inode); 973 if (inode->i_mapping->nrpages == 0) 974 return; 975 if (nfsi->cache_validity & NFS_INO_INVALID_DATA) 976 return; 977 if (!list_empty(&nfsi->open_files)) 978 return; 979 server = NFS_SERVER(inode); 980 if (server->flags & NFS_MOUNT_NOCTO) 981 return; 982 nfs_revalidate_inode(server, inode); 983} 984EXPORT_SYMBOL_GPL(nfs_close_context); 985 986struct nfs_open_context *alloc_nfs_open_context(struct dentry *dentry, 987 fmode_t f_mode, 988 struct file *filp) 989{ 990 struct nfs_open_context *ctx; 991 992 ctx = kmalloc(sizeof(*ctx), GFP_KERNEL); 993 if (!ctx) 994 return ERR_PTR(-ENOMEM); 995 nfs_sb_active(dentry->d_sb); 996 ctx->dentry = dget(dentry); 997 if (filp) 998 ctx->cred = get_cred(filp->f_cred); 999 else 1000 ctx->cred = get_current_cred(); 1001 ctx->ll_cred = NULL; 1002 ctx->state = NULL; 1003 ctx->mode = f_mode; 1004 ctx->flags = 0; 1005 ctx->error = 0; 1006 ctx->flock_owner = (fl_owner_t)filp; 1007 nfs_init_lock_context(&ctx->lock_context); 1008 ctx->lock_context.open_context = ctx; 1009 INIT_LIST_HEAD(&ctx->list); 1010 ctx->mdsthreshold = NULL; 1011 return ctx; 1012} 1013EXPORT_SYMBOL_GPL(alloc_nfs_open_context); 1014 1015struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx) 1016{ 1017 if (ctx != NULL && refcount_inc_not_zero(&ctx->lock_context.count)) 1018 return ctx; 1019 return NULL; 1020} 1021EXPORT_SYMBOL_GPL(get_nfs_open_context); 1022 1023static void __put_nfs_open_context(struct nfs_open_context *ctx, int is_sync) 1024{ 1025 struct inode *inode = d_inode(ctx->dentry); 1026 struct super_block *sb = ctx->dentry->d_sb; 1027 1028 if (!refcount_dec_and_test(&ctx->lock_context.count)) 1029 return; 1030 if (!list_empty(&ctx->list)) { 1031 spin_lock(&inode->i_lock); 1032 list_del_rcu(&ctx->list); 1033 spin_unlock(&inode->i_lock); 1034 } 1035 if (inode != NULL) 1036 NFS_PROTO(inode)->close_context(ctx, is_sync); 1037 put_cred(ctx->cred); 1038 dput(ctx->dentry); 1039 nfs_sb_deactive(sb); 1040 put_rpccred(ctx->ll_cred); 1041 kfree(ctx->mdsthreshold); 1042 kfree_rcu(ctx, rcu_head); 1043} 1044 1045void put_nfs_open_context(struct nfs_open_context *ctx) 1046{ 1047 __put_nfs_open_context(ctx, 0); 1048} 1049EXPORT_SYMBOL_GPL(put_nfs_open_context); 1050 1051static void put_nfs_open_context_sync(struct nfs_open_context *ctx) 1052{ 1053 __put_nfs_open_context(ctx, 1); 1054} 1055 1056/* 1057 * Ensure that mmap has a recent RPC credential for use when writing out 1058 * shared pages 1059 */ 1060void nfs_inode_attach_open_context(struct nfs_open_context *ctx) 1061{ 1062 struct inode *inode = d_inode(ctx->dentry); 1063 struct nfs_inode *nfsi = NFS_I(inode); 1064 1065 spin_lock(&inode->i_lock); 1066 if (list_empty(&nfsi->open_files) && 1067 (nfsi->cache_validity & NFS_INO_DATA_INVAL_DEFER)) 1068 nfsi->cache_validity |= NFS_INO_INVALID_DATA | 1069 NFS_INO_REVAL_FORCED; 1070 list_add_tail_rcu(&ctx->list, &nfsi->open_files); 1071 spin_unlock(&inode->i_lock); 1072} 1073EXPORT_SYMBOL_GPL(nfs_inode_attach_open_context); 1074 1075void nfs_file_set_open_context(struct file *filp, struct nfs_open_context *ctx) 1076{ 1077 filp->private_data = get_nfs_open_context(ctx); 1078 set_bit(NFS_CONTEXT_FILE_OPEN, &ctx->flags); 1079 if (list_empty(&ctx->list)) 1080 nfs_inode_attach_open_context(ctx); 1081} 1082EXPORT_SYMBOL_GPL(nfs_file_set_open_context); 1083 1084/* 1085 * Given an inode, search for an open context with the desired characteristics 1086 */ 1087struct nfs_open_context *nfs_find_open_context(struct inode *inode, const struct cred *cred, fmode_t mode) 1088{ 1089 struct nfs_inode *nfsi = NFS_I(inode); 1090 struct nfs_open_context *pos, *ctx = NULL; 1091 1092 rcu_read_lock(); 1093 list_for_each_entry_rcu(pos, &nfsi->open_files, list) { 1094 if (cred != NULL && cred_fscmp(pos->cred, cred) != 0) 1095 continue; 1096 if ((pos->mode & (FMODE_READ|FMODE_WRITE)) != mode) 1097 continue; 1098 if (!test_bit(NFS_CONTEXT_FILE_OPEN, &pos->flags)) 1099 continue; 1100 ctx = get_nfs_open_context(pos); 1101 if (ctx) 1102 break; 1103 } 1104 rcu_read_unlock(); 1105 return ctx; 1106} 1107 1108void nfs_file_clear_open_context(struct file *filp) 1109{ 1110 struct nfs_open_context *ctx = nfs_file_open_context(filp); 1111 1112 if (ctx) { 1113 struct inode *inode = d_inode(ctx->dentry); 1114 1115 clear_bit(NFS_CONTEXT_FILE_OPEN, &ctx->flags); 1116 /* 1117 * We fatal error on write before. Try to writeback 1118 * every page again. 1119 */ 1120 if (ctx->error < 0) 1121 invalidate_inode_pages2(inode->i_mapping); 1122 filp->private_data = NULL; 1123 put_nfs_open_context_sync(ctx); 1124 } 1125} 1126 1127/* 1128 * These allocate and release file read/write context information. 1129 */ 1130int nfs_open(struct inode *inode, struct file *filp) 1131{ 1132 struct nfs_open_context *ctx; 1133 1134 ctx = alloc_nfs_open_context(file_dentry(filp), filp->f_mode, filp); 1135 if (IS_ERR(ctx)) 1136 return PTR_ERR(ctx); 1137 nfs_file_set_open_context(filp, ctx); 1138 put_nfs_open_context(ctx); 1139 nfs_fscache_open_file(inode, filp); 1140 return 0; 1141} 1142 1143/* 1144 * This function is called whenever some part of NFS notices that 1145 * the cached attributes have to be refreshed. 1146 */ 1147int 1148__nfs_revalidate_inode(struct nfs_server *server, struct inode *inode) 1149{ 1150 int status = -ESTALE; 1151 struct nfs4_label *label = NULL; 1152 struct nfs_fattr *fattr = NULL; 1153 struct nfs_inode *nfsi = NFS_I(inode); 1154 1155 dfprintk(PAGECACHE, "NFS: revalidating (%s/%Lu)\n", 1156 inode->i_sb->s_id, (unsigned long long)NFS_FILEID(inode)); 1157 1158 trace_nfs_revalidate_inode_enter(inode); 1159 1160 if (is_bad_inode(inode)) 1161 goto out; 1162 if (NFS_STALE(inode)) 1163 goto out; 1164 1165 /* pNFS: Attributes aren't updated until we layoutcommit */ 1166 if (S_ISREG(inode->i_mode)) { 1167 status = pnfs_sync_inode(inode, false); 1168 if (status) 1169 goto out; 1170 } 1171 1172 status = -ENOMEM; 1173 fattr = nfs_alloc_fattr(); 1174 if (fattr == NULL) 1175 goto out; 1176 1177 nfs_inc_stats(inode, NFSIOS_INODEREVALIDATE); 1178 1179 label = nfs4_label_alloc(NFS_SERVER(inode), GFP_KERNEL); 1180 if (IS_ERR(label)) { 1181 status = PTR_ERR(label); 1182 goto out; 1183 } 1184 1185 status = NFS_PROTO(inode)->getattr(server, NFS_FH(inode), fattr, 1186 label, inode); 1187 if (status != 0) { 1188 dfprintk(PAGECACHE, "nfs_revalidate_inode: (%s/%Lu) getattr failed, error=%d\n", 1189 inode->i_sb->s_id, 1190 (unsigned long long)NFS_FILEID(inode), status); 1191 switch (status) { 1192 case -ETIMEDOUT: 1193 /* A soft timeout occurred. Use cached information? */ 1194 if (server->flags & NFS_MOUNT_SOFTREVAL) 1195 status = 0; 1196 break; 1197 case -ESTALE: 1198 if (!S_ISDIR(inode->i_mode)) 1199 nfs_set_inode_stale(inode); 1200 else 1201 nfs_zap_caches(inode); 1202 } 1203 goto err_out; 1204 } 1205 1206 status = nfs_refresh_inode(inode, fattr); 1207 if (status) { 1208 dfprintk(PAGECACHE, "nfs_revalidate_inode: (%s/%Lu) refresh failed, error=%d\n", 1209 inode->i_sb->s_id, 1210 (unsigned long long)NFS_FILEID(inode), status); 1211 goto err_out; 1212 } 1213 1214 if (nfsi->cache_validity & NFS_INO_INVALID_ACL) 1215 nfs_zap_acl_cache(inode); 1216 1217 nfs_setsecurity(inode, fattr, label); 1218 1219 dfprintk(PAGECACHE, "NFS: (%s/%Lu) revalidation complete\n", 1220 inode->i_sb->s_id, 1221 (unsigned long long)NFS_FILEID(inode)); 1222 1223err_out: 1224 nfs4_label_free(label); 1225out: 1226 nfs_free_fattr(fattr); 1227 trace_nfs_revalidate_inode_exit(inode, status); 1228 return status; 1229} 1230 1231int nfs_attribute_cache_expired(struct inode *inode) 1232{ 1233 if (nfs_have_delegated_attributes(inode)) 1234 return 0; 1235 return nfs_attribute_timeout(inode); 1236} 1237 1238/** 1239 * nfs_revalidate_inode - Revalidate the inode attributes 1240 * @server: pointer to nfs_server struct 1241 * @inode: pointer to inode struct 1242 * 1243 * Updates inode attribute information by retrieving the data from the server. 1244 */ 1245int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode) 1246{ 1247 if (!nfs_need_revalidate_inode(inode)) 1248 return NFS_STALE(inode) ? -ESTALE : 0; 1249 return __nfs_revalidate_inode(server, inode); 1250} 1251EXPORT_SYMBOL_GPL(nfs_revalidate_inode); 1252 1253static int nfs_invalidate_mapping(struct inode *inode, struct address_space *mapping) 1254{ 1255 struct nfs_inode *nfsi = NFS_I(inode); 1256 int ret; 1257 1258 if (mapping->nrpages != 0) { 1259 if (S_ISREG(inode->i_mode)) { 1260 ret = nfs_sync_mapping(mapping); 1261 if (ret < 0) 1262 return ret; 1263 } 1264 ret = invalidate_inode_pages2(mapping); 1265 if (ret < 0) 1266 return ret; 1267 } 1268 if (S_ISDIR(inode->i_mode)) { 1269 spin_lock(&inode->i_lock); 1270 memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf)); 1271 spin_unlock(&inode->i_lock); 1272 } 1273 nfs_inc_stats(inode, NFSIOS_DATAINVALIDATE); 1274 nfs_fscache_wait_on_invalidate(inode); 1275 1276 dfprintk(PAGECACHE, "NFS: (%s/%Lu) data cache invalidated\n", 1277 inode->i_sb->s_id, 1278 (unsigned long long)NFS_FILEID(inode)); 1279 return 0; 1280} 1281 1282bool nfs_mapping_need_revalidate_inode(struct inode *inode) 1283{ 1284 return nfs_check_cache_invalid(inode, NFS_INO_REVAL_PAGECACHE) || 1285 NFS_STALE(inode); 1286} 1287 1288int nfs_revalidate_mapping_rcu(struct inode *inode) 1289{ 1290 struct nfs_inode *nfsi = NFS_I(inode); 1291 unsigned long *bitlock = &nfsi->flags; 1292 int ret = 0; 1293 1294 if (IS_SWAPFILE(inode)) 1295 goto out; 1296 if (nfs_mapping_need_revalidate_inode(inode)) { 1297 ret = -ECHILD; 1298 goto out; 1299 } 1300 spin_lock(&inode->i_lock); 1301 if (test_bit(NFS_INO_INVALIDATING, bitlock) || 1302 (nfsi->cache_validity & NFS_INO_INVALID_DATA)) 1303 ret = -ECHILD; 1304 spin_unlock(&inode->i_lock); 1305out: 1306 return ret; 1307} 1308 1309/** 1310 * nfs_revalidate_mapping - Revalidate the pagecache 1311 * @inode: pointer to host inode 1312 * @mapping: pointer to mapping 1313 */ 1314int nfs_revalidate_mapping(struct inode *inode, 1315 struct address_space *mapping) 1316{ 1317 struct nfs_inode *nfsi = NFS_I(inode); 1318 unsigned long *bitlock = &nfsi->flags; 1319 int ret = 0; 1320 1321 /* swapfiles are not supposed to be shared. */ 1322 if (IS_SWAPFILE(inode)) 1323 goto out; 1324 1325 if (nfs_mapping_need_revalidate_inode(inode)) { 1326 ret = __nfs_revalidate_inode(NFS_SERVER(inode), inode); 1327 if (ret < 0) 1328 goto out; 1329 } 1330 1331 /* 1332 * We must clear NFS_INO_INVALID_DATA first to ensure that 1333 * invalidations that come in while we're shooting down the mappings 1334 * are respected. But, that leaves a race window where one revalidator 1335 * can clear the flag, and then another checks it before the mapping 1336 * gets invalidated. Fix that by serializing access to this part of 1337 * the function. 1338 * 1339 * At the same time, we need to allow other tasks to see whether we 1340 * might be in the middle of invalidating the pages, so we only set 1341 * the bit lock here if it looks like we're going to be doing that. 1342 */ 1343 for (;;) { 1344 ret = wait_on_bit_action(bitlock, NFS_INO_INVALIDATING, 1345 nfs_wait_bit_killable, TASK_KILLABLE); 1346 if (ret) 1347 goto out; 1348 spin_lock(&inode->i_lock); 1349 if (test_bit(NFS_INO_INVALIDATING, bitlock)) { 1350 spin_unlock(&inode->i_lock); 1351 continue; 1352 } 1353 if (nfsi->cache_validity & NFS_INO_INVALID_DATA) 1354 break; 1355 spin_unlock(&inode->i_lock); 1356 goto out; 1357 } 1358 1359 set_bit(NFS_INO_INVALIDATING, bitlock); 1360 smp_wmb(); 1361 nfsi->cache_validity &= ~(NFS_INO_INVALID_DATA| 1362 NFS_INO_DATA_INVAL_DEFER); 1363 spin_unlock(&inode->i_lock); 1364 trace_nfs_invalidate_mapping_enter(inode); 1365 ret = nfs_invalidate_mapping(inode, mapping); 1366 trace_nfs_invalidate_mapping_exit(inode, ret); 1367 1368 clear_bit_unlock(NFS_INO_INVALIDATING, bitlock); 1369 smp_mb__after_atomic(); 1370 wake_up_bit(bitlock, NFS_INO_INVALIDATING); 1371out: 1372 return ret; 1373} 1374 1375static bool nfs_file_has_writers(struct nfs_inode *nfsi) 1376{ 1377 struct inode *inode = &nfsi->vfs_inode; 1378 1379 if (!S_ISREG(inode->i_mode)) 1380 return false; 1381 if (list_empty(&nfsi->open_files)) 1382 return false; 1383 return inode_is_open_for_write(inode); 1384} 1385 1386static bool nfs_file_has_buffered_writers(struct nfs_inode *nfsi) 1387{ 1388 return nfs_file_has_writers(nfsi) && nfs_file_io_is_buffered(nfsi); 1389} 1390 1391static void nfs_wcc_update_inode(struct inode *inode, struct nfs_fattr *fattr) 1392{ 1393 struct timespec64 ts; 1394 1395 if ((fattr->valid & NFS_ATTR_FATTR_PRECHANGE) 1396 && (fattr->valid & NFS_ATTR_FATTR_CHANGE) 1397 && inode_eq_iversion_raw(inode, fattr->pre_change_attr)) { 1398 inode_set_iversion_raw(inode, fattr->change_attr); 1399 if (S_ISDIR(inode->i_mode)) 1400 nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA); 1401 else if (nfs_server_capable(inode, NFS_CAP_XATTR)) 1402 nfs_set_cache_invalid(inode, NFS_INO_INVALID_XATTR); 1403 } 1404 /* If we have atomic WCC data, we may update some attributes */ 1405 ts = inode->i_ctime; 1406 if ((fattr->valid & NFS_ATTR_FATTR_PRECTIME) 1407 && (fattr->valid & NFS_ATTR_FATTR_CTIME) 1408 && timespec64_equal(&ts, &fattr->pre_ctime)) { 1409 inode->i_ctime = fattr->ctime; 1410 } 1411 1412 ts = inode->i_mtime; 1413 if ((fattr->valid & NFS_ATTR_FATTR_PREMTIME) 1414 && (fattr->valid & NFS_ATTR_FATTR_MTIME) 1415 && timespec64_equal(&ts, &fattr->pre_mtime)) { 1416 inode->i_mtime = fattr->mtime; 1417 if (S_ISDIR(inode->i_mode)) 1418 nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA); 1419 } 1420 if ((fattr->valid & NFS_ATTR_FATTR_PRESIZE) 1421 && (fattr->valid & NFS_ATTR_FATTR_SIZE) 1422 && i_size_read(inode) == nfs_size_to_loff_t(fattr->pre_size) 1423 && !nfs_have_writebacks(inode)) { 1424 i_size_write(inode, nfs_size_to_loff_t(fattr->size)); 1425 } 1426} 1427 1428/** 1429 * nfs_check_inode_attributes - verify consistency of the inode attribute cache 1430 * @inode: pointer to inode 1431 * @fattr: updated attributes 1432 * 1433 * Verifies the attribute cache. If we have just changed the attributes, 1434 * so that fattr carries weak cache consistency data, then it may 1435 * also update the ctime/mtime/change_attribute. 1436 */ 1437static int nfs_check_inode_attributes(struct inode *inode, struct nfs_fattr *fattr) 1438{ 1439 struct nfs_inode *nfsi = NFS_I(inode); 1440 loff_t cur_size, new_isize; 1441 unsigned long invalid = 0; 1442 struct timespec64 ts; 1443 1444 if (NFS_PROTO(inode)->have_delegation(inode, FMODE_READ)) 1445 return 0; 1446 1447 if (!(fattr->valid & NFS_ATTR_FATTR_FILEID)) { 1448 /* Only a mounted-on-fileid? Just exit */ 1449 if (fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) 1450 return 0; 1451 /* Has the inode gone and changed behind our back? */ 1452 } else if (nfsi->fileid != fattr->fileid) { 1453 /* Is this perhaps the mounted-on fileid? */ 1454 if ((fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) && 1455 nfsi->fileid == fattr->mounted_on_fileid) 1456 return 0; 1457 return -ESTALE; 1458 } 1459 if ((fattr->valid & NFS_ATTR_FATTR_TYPE) && inode_wrong_type(inode, fattr->mode)) 1460 return -ESTALE; 1461 1462 1463 if (!nfs_file_has_buffered_writers(nfsi)) { 1464 /* Verify a few of the more important attributes */ 1465 if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 && !inode_eq_iversion_raw(inode, fattr->change_attr)) 1466 invalid |= NFS_INO_INVALID_CHANGE 1467 | NFS_INO_REVAL_PAGECACHE; 1468 1469 ts = inode->i_mtime; 1470 if ((fattr->valid & NFS_ATTR_FATTR_MTIME) && !timespec64_equal(&ts, &fattr->mtime)) 1471 invalid |= NFS_INO_INVALID_MTIME; 1472 1473 ts = inode->i_ctime; 1474 if ((fattr->valid & NFS_ATTR_FATTR_CTIME) && !timespec64_equal(&ts, &fattr->ctime)) 1475 invalid |= NFS_INO_INVALID_CTIME; 1476 1477 if (fattr->valid & NFS_ATTR_FATTR_SIZE) { 1478 cur_size = i_size_read(inode); 1479 new_isize = nfs_size_to_loff_t(fattr->size); 1480 if (cur_size != new_isize) 1481 invalid |= NFS_INO_INVALID_SIZE 1482 | NFS_INO_REVAL_PAGECACHE; 1483 } 1484 } 1485 1486 /* Have any file permissions changed? */ 1487 if ((fattr->valid & NFS_ATTR_FATTR_MODE) && (inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)) 1488 invalid |= NFS_INO_INVALID_ACCESS 1489 | NFS_INO_INVALID_ACL 1490 | NFS_INO_INVALID_OTHER; 1491 if ((fattr->valid & NFS_ATTR_FATTR_OWNER) && !uid_eq(inode->i_uid, fattr->uid)) 1492 invalid |= NFS_INO_INVALID_ACCESS 1493 | NFS_INO_INVALID_ACL 1494 | NFS_INO_INVALID_OTHER; 1495 if ((fattr->valid & NFS_ATTR_FATTR_GROUP) && !gid_eq(inode->i_gid, fattr->gid)) 1496 invalid |= NFS_INO_INVALID_ACCESS 1497 | NFS_INO_INVALID_ACL 1498 | NFS_INO_INVALID_OTHER; 1499 1500 /* Has the link count changed? */ 1501 if ((fattr->valid & NFS_ATTR_FATTR_NLINK) && inode->i_nlink != fattr->nlink) 1502 invalid |= NFS_INO_INVALID_OTHER; 1503 1504 ts = inode->i_atime; 1505 if ((fattr->valid & NFS_ATTR_FATTR_ATIME) && !timespec64_equal(&ts, &fattr->atime)) 1506 invalid |= NFS_INO_INVALID_ATIME; 1507 1508 if (invalid != 0) 1509 nfs_set_cache_invalid(inode, invalid); 1510 1511 nfsi->read_cache_jiffies = fattr->time_start; 1512 return 0; 1513} 1514 1515static atomic_long_t nfs_attr_generation_counter; 1516 1517static unsigned long nfs_read_attr_generation_counter(void) 1518{ 1519 return atomic_long_read(&nfs_attr_generation_counter); 1520} 1521 1522unsigned long nfs_inc_attr_generation_counter(void) 1523{ 1524 return atomic_long_inc_return(&nfs_attr_generation_counter); 1525} 1526EXPORT_SYMBOL_GPL(nfs_inc_attr_generation_counter); 1527 1528void nfs_fattr_init(struct nfs_fattr *fattr) 1529{ 1530 fattr->valid = 0; 1531 fattr->time_start = jiffies; 1532 fattr->gencount = nfs_inc_attr_generation_counter(); 1533 fattr->owner_name = NULL; 1534 fattr->group_name = NULL; 1535} 1536EXPORT_SYMBOL_GPL(nfs_fattr_init); 1537 1538/** 1539 * nfs_fattr_set_barrier 1540 * @fattr: attributes 1541 * 1542 * Used to set a barrier after an attribute was updated. This 1543 * barrier ensures that older attributes from RPC calls that may 1544 * have raced with our update cannot clobber these new values. 1545 * Note that you are still responsible for ensuring that other 1546 * operations which change the attribute on the server do not 1547 * collide. 1548 */ 1549void nfs_fattr_set_barrier(struct nfs_fattr *fattr) 1550{ 1551 fattr->gencount = nfs_inc_attr_generation_counter(); 1552} 1553 1554struct nfs_fattr *nfs_alloc_fattr(void) 1555{ 1556 struct nfs_fattr *fattr; 1557 1558 fattr = kmalloc(sizeof(*fattr), GFP_NOFS); 1559 if (fattr != NULL) 1560 nfs_fattr_init(fattr); 1561 return fattr; 1562} 1563EXPORT_SYMBOL_GPL(nfs_alloc_fattr); 1564 1565struct nfs_fh *nfs_alloc_fhandle(void) 1566{ 1567 struct nfs_fh *fh; 1568 1569 fh = kmalloc(sizeof(struct nfs_fh), GFP_NOFS); 1570 if (fh != NULL) 1571 fh->size = 0; 1572 return fh; 1573} 1574EXPORT_SYMBOL_GPL(nfs_alloc_fhandle); 1575 1576#ifdef NFS_DEBUG 1577/* 1578 * _nfs_display_fhandle_hash - calculate the crc32 hash for the filehandle 1579 * in the same way that wireshark does 1580 * 1581 * @fh: file handle 1582 * 1583 * For debugging only. 1584 */ 1585u32 _nfs_display_fhandle_hash(const struct nfs_fh *fh) 1586{ 1587 /* wireshark uses 32-bit AUTODIN crc and does a bitwise 1588 * not on the result */ 1589 return nfs_fhandle_hash(fh); 1590} 1591EXPORT_SYMBOL_GPL(_nfs_display_fhandle_hash); 1592 1593/* 1594 * _nfs_display_fhandle - display an NFS file handle on the console 1595 * 1596 * @fh: file handle to display 1597 * @caption: display caption 1598 * 1599 * For debugging only. 1600 */ 1601void _nfs_display_fhandle(const struct nfs_fh *fh, const char *caption) 1602{ 1603 unsigned short i; 1604 1605 if (fh == NULL || fh->size == 0) { 1606 printk(KERN_DEFAULT "%s at %p is empty\n", caption, fh); 1607 return; 1608 } 1609 1610 printk(KERN_DEFAULT "%s at %p is %u bytes, crc: 0x%08x:\n", 1611 caption, fh, fh->size, _nfs_display_fhandle_hash(fh)); 1612 for (i = 0; i < fh->size; i += 16) { 1613 __be32 *pos = (__be32 *)&fh->data[i]; 1614 1615 switch ((fh->size - i - 1) >> 2) { 1616 case 0: 1617 printk(KERN_DEFAULT " %08x\n", 1618 be32_to_cpup(pos)); 1619 break; 1620 case 1: 1621 printk(KERN_DEFAULT " %08x %08x\n", 1622 be32_to_cpup(pos), be32_to_cpup(pos + 1)); 1623 break; 1624 case 2: 1625 printk(KERN_DEFAULT " %08x %08x %08x\n", 1626 be32_to_cpup(pos), be32_to_cpup(pos + 1), 1627 be32_to_cpup(pos + 2)); 1628 break; 1629 default: 1630 printk(KERN_DEFAULT " %08x %08x %08x %08x\n", 1631 be32_to_cpup(pos), be32_to_cpup(pos + 1), 1632 be32_to_cpup(pos + 2), be32_to_cpup(pos + 3)); 1633 } 1634 } 1635} 1636EXPORT_SYMBOL_GPL(_nfs_display_fhandle); 1637#endif 1638 1639/** 1640 * nfs_inode_attrs_need_update - check if the inode attributes need updating 1641 * @inode: pointer to inode 1642 * @fattr: attributes 1643 * 1644 * Attempt to divine whether or not an RPC call reply carrying stale 1645 * attributes got scheduled after another call carrying updated ones. 1646 * 1647 * To do so, the function first assumes that a more recent ctime means 1648 * that the attributes in fattr are newer, however it also attempt to 1649 * catch the case where ctime either didn't change, or went backwards 1650 * (if someone reset the clock on the server) by looking at whether 1651 * or not this RPC call was started after the inode was last updated. 1652 * Note also the check for wraparound of 'attr_gencount' 1653 * 1654 * The function returns 'true' if it thinks the attributes in 'fattr' are 1655 * more recent than the ones cached in the inode. 1656 * 1657 */ 1658static int nfs_inode_attrs_need_update(const struct inode *inode, const struct nfs_fattr *fattr) 1659{ 1660 unsigned long attr_gencount = NFS_I(inode)->attr_gencount; 1661 1662 return (long)(fattr->gencount - attr_gencount) > 0 || 1663 (long)(attr_gencount - nfs_read_attr_generation_counter()) > 0; 1664} 1665 1666static int nfs_refresh_inode_locked(struct inode *inode, struct nfs_fattr *fattr) 1667{ 1668 int ret; 1669 1670 trace_nfs_refresh_inode_enter(inode); 1671 1672 if (nfs_inode_attrs_need_update(inode, fattr)) 1673 ret = nfs_update_inode(inode, fattr); 1674 else 1675 ret = nfs_check_inode_attributes(inode, fattr); 1676 1677 trace_nfs_refresh_inode_exit(inode, ret); 1678 return ret; 1679} 1680 1681/** 1682 * nfs_refresh_inode - try to update the inode attribute cache 1683 * @inode: pointer to inode 1684 * @fattr: updated attributes 1685 * 1686 * Check that an RPC call that returned attributes has not overlapped with 1687 * other recent updates of the inode metadata, then decide whether it is 1688 * safe to do a full update of the inode attributes, or whether just to 1689 * call nfs_check_inode_attributes. 1690 */ 1691int nfs_refresh_inode(struct inode *inode, struct nfs_fattr *fattr) 1692{ 1693 int status; 1694 1695 if ((fattr->valid & NFS_ATTR_FATTR) == 0) 1696 return 0; 1697 spin_lock(&inode->i_lock); 1698 status = nfs_refresh_inode_locked(inode, fattr); 1699 spin_unlock(&inode->i_lock); 1700 1701 return status; 1702} 1703EXPORT_SYMBOL_GPL(nfs_refresh_inode); 1704 1705static int nfs_post_op_update_inode_locked(struct inode *inode, 1706 struct nfs_fattr *fattr, unsigned int invalid) 1707{ 1708 if (S_ISDIR(inode->i_mode)) 1709 invalid |= NFS_INO_INVALID_DATA; 1710 nfs_set_cache_invalid(inode, invalid); 1711 if ((fattr->valid & NFS_ATTR_FATTR) == 0) 1712 return 0; 1713 return nfs_refresh_inode_locked(inode, fattr); 1714} 1715 1716/** 1717 * nfs_post_op_update_inode - try to update the inode attribute cache 1718 * @inode: pointer to inode 1719 * @fattr: updated attributes 1720 * 1721 * After an operation that has changed the inode metadata, mark the 1722 * attribute cache as being invalid, then try to update it. 1723 * 1724 * NB: if the server didn't return any post op attributes, this 1725 * function will force the retrieval of attributes before the next 1726 * NFS request. Thus it should be used only for operations that 1727 * are expected to change one or more attributes, to avoid 1728 * unnecessary NFS requests and trips through nfs_update_inode(). 1729 */ 1730int nfs_post_op_update_inode(struct inode *inode, struct nfs_fattr *fattr) 1731{ 1732 int status; 1733 1734 spin_lock(&inode->i_lock); 1735 nfs_fattr_set_barrier(fattr); 1736 status = nfs_post_op_update_inode_locked(inode, fattr, 1737 NFS_INO_INVALID_CHANGE 1738 | NFS_INO_INVALID_CTIME 1739 | NFS_INO_REVAL_FORCED); 1740 spin_unlock(&inode->i_lock); 1741 1742 return status; 1743} 1744EXPORT_SYMBOL_GPL(nfs_post_op_update_inode); 1745 1746/** 1747 * nfs_post_op_update_inode_force_wcc_locked - update the inode attribute cache 1748 * @inode: pointer to inode 1749 * @fattr: updated attributes 1750 * 1751 * After an operation that has changed the inode metadata, mark the 1752 * attribute cache as being invalid, then try to update it. Fake up 1753 * weak cache consistency data, if none exist. 1754 * 1755 * This function is mainly designed to be used by the ->write_done() functions. 1756 */ 1757int nfs_post_op_update_inode_force_wcc_locked(struct inode *inode, struct nfs_fattr *fattr) 1758{ 1759 int status; 1760 1761 /* Don't do a WCC update if these attributes are already stale */ 1762 if ((fattr->valid & NFS_ATTR_FATTR) == 0 || 1763 !nfs_inode_attrs_need_update(inode, fattr)) { 1764 fattr->valid &= ~(NFS_ATTR_FATTR_PRECHANGE 1765 | NFS_ATTR_FATTR_PRESIZE 1766 | NFS_ATTR_FATTR_PREMTIME 1767 | NFS_ATTR_FATTR_PRECTIME); 1768 goto out_noforce; 1769 } 1770 if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 && 1771 (fattr->valid & NFS_ATTR_FATTR_PRECHANGE) == 0) { 1772 fattr->pre_change_attr = inode_peek_iversion_raw(inode); 1773 fattr->valid |= NFS_ATTR_FATTR_PRECHANGE; 1774 } 1775 if ((fattr->valid & NFS_ATTR_FATTR_CTIME) != 0 && 1776 (fattr->valid & NFS_ATTR_FATTR_PRECTIME) == 0) { 1777 fattr->pre_ctime = inode->i_ctime; 1778 fattr->valid |= NFS_ATTR_FATTR_PRECTIME; 1779 } 1780 if ((fattr->valid & NFS_ATTR_FATTR_MTIME) != 0 && 1781 (fattr->valid & NFS_ATTR_FATTR_PREMTIME) == 0) { 1782 fattr->pre_mtime = inode->i_mtime; 1783 fattr->valid |= NFS_ATTR_FATTR_PREMTIME; 1784 } 1785 if ((fattr->valid & NFS_ATTR_FATTR_SIZE) != 0 && 1786 (fattr->valid & NFS_ATTR_FATTR_PRESIZE) == 0) { 1787 fattr->pre_size = i_size_read(inode); 1788 fattr->valid |= NFS_ATTR_FATTR_PRESIZE; 1789 } 1790out_noforce: 1791 status = nfs_post_op_update_inode_locked(inode, fattr, 1792 NFS_INO_INVALID_CHANGE 1793 | NFS_INO_INVALID_CTIME 1794 | NFS_INO_INVALID_MTIME 1795 | NFS_INO_INVALID_BLOCKS); 1796 return status; 1797} 1798 1799/** 1800 * nfs_post_op_update_inode_force_wcc - try to update the inode attribute cache 1801 * @inode: pointer to inode 1802 * @fattr: updated attributes 1803 * 1804 * After an operation that has changed the inode metadata, mark the 1805 * attribute cache as being invalid, then try to update it. Fake up 1806 * weak cache consistency data, if none exist. 1807 * 1808 * This function is mainly designed to be used by the ->write_done() functions. 1809 */ 1810int nfs_post_op_update_inode_force_wcc(struct inode *inode, struct nfs_fattr *fattr) 1811{ 1812 int status; 1813 1814 spin_lock(&inode->i_lock); 1815 nfs_fattr_set_barrier(fattr); 1816 status = nfs_post_op_update_inode_force_wcc_locked(inode, fattr); 1817 spin_unlock(&inode->i_lock); 1818 return status; 1819} 1820EXPORT_SYMBOL_GPL(nfs_post_op_update_inode_force_wcc); 1821 1822 1823/* 1824 * Many nfs protocol calls return the new file attributes after 1825 * an operation. Here we update the inode to reflect the state 1826 * of the server's inode. 1827 * 1828 * This is a bit tricky because we have to make sure all dirty pages 1829 * have been sent off to the server before calling invalidate_inode_pages. 1830 * To make sure no other process adds more write requests while we try 1831 * our best to flush them, we make them sleep during the attribute refresh. 1832 * 1833 * A very similar scenario holds for the dir cache. 1834 */ 1835static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr) 1836{ 1837 struct nfs_server *server; 1838 struct nfs_inode *nfsi = NFS_I(inode); 1839 loff_t cur_isize, new_isize; 1840 unsigned long invalid = 0; 1841 unsigned long now = jiffies; 1842 unsigned long save_cache_validity; 1843 bool have_writers = nfs_file_has_buffered_writers(nfsi); 1844 bool cache_revalidated = true; 1845 bool attr_changed = false; 1846 bool have_delegation; 1847 1848 dfprintk(VFS, "NFS: %s(%s/%lu fh_crc=0x%08x ct=%d info=0x%x)\n", 1849 __func__, inode->i_sb->s_id, inode->i_ino, 1850 nfs_display_fhandle_hash(NFS_FH(inode)), 1851 atomic_read(&inode->i_count), fattr->valid); 1852 1853 if (!(fattr->valid & NFS_ATTR_FATTR_FILEID)) { 1854 /* Only a mounted-on-fileid? Just exit */ 1855 if (fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) 1856 return 0; 1857 /* Has the inode gone and changed behind our back? */ 1858 } else if (nfsi->fileid != fattr->fileid) { 1859 /* Is this perhaps the mounted-on fileid? */ 1860 if ((fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) && 1861 nfsi->fileid == fattr->mounted_on_fileid) 1862 return 0; 1863 printk(KERN_ERR "NFS: server %s error: fileid changed\n" 1864 "fsid %s: expected fileid 0x%Lx, got 0x%Lx\n", 1865 NFS_SERVER(inode)->nfs_client->cl_hostname, 1866 inode->i_sb->s_id, (long long)nfsi->fileid, 1867 (long long)fattr->fileid); 1868 goto out_err; 1869 } 1870 1871 /* 1872 * Make sure the inode's type hasn't changed. 1873 */ 1874 if ((fattr->valid & NFS_ATTR_FATTR_TYPE) && inode_wrong_type(inode, fattr->mode)) { 1875 /* 1876 * Big trouble! The inode has become a different object. 1877 */ 1878 printk(KERN_DEBUG "NFS: %s: inode %lu mode changed, %07o to %07o\n", 1879 __func__, inode->i_ino, inode->i_mode, fattr->mode); 1880 goto out_err; 1881 } 1882 1883 server = NFS_SERVER(inode); 1884 /* Update the fsid? */ 1885 if (S_ISDIR(inode->i_mode) && (fattr->valid & NFS_ATTR_FATTR_FSID) && 1886 !nfs_fsid_equal(&server->fsid, &fattr->fsid) && 1887 !IS_AUTOMOUNT(inode)) 1888 server->fsid = fattr->fsid; 1889 1890 /* Save the delegation state before clearing cache_validity */ 1891 have_delegation = nfs_have_delegated_attributes(inode); 1892 1893 /* 1894 * Update the read time so we don't revalidate too often. 1895 */ 1896 nfsi->read_cache_jiffies = fattr->time_start; 1897 1898 save_cache_validity = nfsi->cache_validity; 1899 nfsi->cache_validity &= ~(NFS_INO_INVALID_ATTR 1900 | NFS_INO_INVALID_ATIME 1901 | NFS_INO_REVAL_FORCED 1902 | NFS_INO_REVAL_PAGECACHE 1903 | NFS_INO_INVALID_BLOCKS); 1904 1905 /* Do atomic weak cache consistency updates */ 1906 nfs_wcc_update_inode(inode, fattr); 1907 1908 if (pnfs_layoutcommit_outstanding(inode)) { 1909 nfsi->cache_validity |= 1910 save_cache_validity & 1911 (NFS_INO_INVALID_CHANGE | NFS_INO_INVALID_CTIME | 1912 NFS_INO_INVALID_MTIME | NFS_INO_INVALID_SIZE | 1913 NFS_INO_REVAL_FORCED); 1914 cache_revalidated = false; 1915 } 1916 1917 /* More cache consistency checks */ 1918 if (fattr->valid & NFS_ATTR_FATTR_CHANGE) { 1919 if (!inode_eq_iversion_raw(inode, fattr->change_attr)) { 1920 /* Could it be a race with writeback? */ 1921 if (!(have_writers || have_delegation)) { 1922 invalid |= NFS_INO_INVALID_DATA 1923 | NFS_INO_INVALID_ACCESS 1924 | NFS_INO_INVALID_ACL 1925 | NFS_INO_INVALID_XATTR; 1926 /* Force revalidate of all attributes */ 1927 save_cache_validity |= NFS_INO_INVALID_CTIME 1928 | NFS_INO_INVALID_MTIME 1929 | NFS_INO_INVALID_SIZE 1930 | NFS_INO_INVALID_OTHER; 1931 if (S_ISDIR(inode->i_mode)) 1932 nfs_force_lookup_revalidate(inode); 1933 dprintk("NFS: change_attr change on server for file %s/%ld\n", 1934 inode->i_sb->s_id, 1935 inode->i_ino); 1936 } else if (!have_delegation) 1937 nfsi->cache_validity |= NFS_INO_DATA_INVAL_DEFER; 1938 inode_set_iversion_raw(inode, fattr->change_attr); 1939 attr_changed = true; 1940 } 1941 } else { 1942 nfsi->cache_validity |= save_cache_validity & 1943 (NFS_INO_INVALID_CHANGE 1944 | NFS_INO_REVAL_PAGECACHE 1945 | NFS_INO_REVAL_FORCED); 1946 cache_revalidated = false; 1947 } 1948 1949 if (fattr->valid & NFS_ATTR_FATTR_MTIME) { 1950 inode->i_mtime = fattr->mtime; 1951 } else if (server->caps & NFS_CAP_MTIME) { 1952 nfsi->cache_validity |= save_cache_validity & 1953 (NFS_INO_INVALID_MTIME 1954 | NFS_INO_REVAL_FORCED); 1955 cache_revalidated = false; 1956 } 1957 1958 if (fattr->valid & NFS_ATTR_FATTR_CTIME) { 1959 inode->i_ctime = fattr->ctime; 1960 } else if (server->caps & NFS_CAP_CTIME) { 1961 nfsi->cache_validity |= save_cache_validity & 1962 (NFS_INO_INVALID_CTIME 1963 | NFS_INO_REVAL_FORCED); 1964 cache_revalidated = false; 1965 } 1966 1967 /* Check if our cached file size is stale */ 1968 if (fattr->valid & NFS_ATTR_FATTR_SIZE) { 1969 new_isize = nfs_size_to_loff_t(fattr->size); 1970 cur_isize = i_size_read(inode); 1971 if (new_isize != cur_isize && !have_delegation) { 1972 /* Do we perhaps have any outstanding writes, or has 1973 * the file grown beyond our last write? */ 1974 if (!nfs_have_writebacks(inode) || new_isize > cur_isize) { 1975 i_size_write(inode, new_isize); 1976 if (!have_writers) 1977 invalid |= NFS_INO_INVALID_DATA; 1978 attr_changed = true; 1979 } 1980 dprintk("NFS: isize change on server for file %s/%ld " 1981 "(%Ld to %Ld)\n", 1982 inode->i_sb->s_id, 1983 inode->i_ino, 1984 (long long)cur_isize, 1985 (long long)new_isize); 1986 } 1987 } else { 1988 nfsi->cache_validity |= save_cache_validity & 1989 (NFS_INO_INVALID_SIZE 1990 | NFS_INO_REVAL_PAGECACHE 1991 | NFS_INO_REVAL_FORCED); 1992 cache_revalidated = false; 1993 } 1994 1995 1996 if (fattr->valid & NFS_ATTR_FATTR_ATIME) 1997 inode->i_atime = fattr->atime; 1998 else if (server->caps & NFS_CAP_ATIME) { 1999 nfsi->cache_validity |= save_cache_validity & 2000 (NFS_INO_INVALID_ATIME 2001 | NFS_INO_REVAL_FORCED); 2002 cache_revalidated = false; 2003 } 2004 2005 if (fattr->valid & NFS_ATTR_FATTR_MODE) { 2006 if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)) { 2007 umode_t newmode = inode->i_mode & S_IFMT; 2008 newmode |= fattr->mode & S_IALLUGO; 2009 inode->i_mode = newmode; 2010 invalid |= NFS_INO_INVALID_ACCESS 2011 | NFS_INO_INVALID_ACL; 2012 attr_changed = true; 2013 } 2014 } else if (server->caps & NFS_CAP_MODE) { 2015 nfsi->cache_validity |= save_cache_validity & 2016 (NFS_INO_INVALID_OTHER 2017 | NFS_INO_REVAL_FORCED); 2018 cache_revalidated = false; 2019 } 2020 2021 if (fattr->valid & NFS_ATTR_FATTR_OWNER) { 2022 if (!uid_eq(inode->i_uid, fattr->uid)) { 2023 invalid |= NFS_INO_INVALID_ACCESS 2024 | NFS_INO_INVALID_ACL; 2025 inode->i_uid = fattr->uid; 2026 attr_changed = true; 2027 } 2028 } else if (server->caps & NFS_CAP_OWNER) { 2029 nfsi->cache_validity |= save_cache_validity & 2030 (NFS_INO_INVALID_OTHER 2031 | NFS_INO_REVAL_FORCED); 2032 cache_revalidated = false; 2033 } 2034 2035 if (fattr->valid & NFS_ATTR_FATTR_GROUP) { 2036 if (!gid_eq(inode->i_gid, fattr->gid)) { 2037 invalid |= NFS_INO_INVALID_ACCESS 2038 | NFS_INO_INVALID_ACL; 2039 inode->i_gid = fattr->gid; 2040 attr_changed = true; 2041 } 2042 } else if (server->caps & NFS_CAP_OWNER_GROUP) { 2043 nfsi->cache_validity |= save_cache_validity & 2044 (NFS_INO_INVALID_OTHER 2045 | NFS_INO_REVAL_FORCED); 2046 cache_revalidated = false; 2047 } 2048 2049 if (fattr->valid & NFS_ATTR_FATTR_NLINK) { 2050 if (inode->i_nlink != fattr->nlink) { 2051 if (S_ISDIR(inode->i_mode)) 2052 invalid |= NFS_INO_INVALID_DATA; 2053 set_nlink(inode, fattr->nlink); 2054 attr_changed = true; 2055 } 2056 } else if (server->caps & NFS_CAP_NLINK) { 2057 nfsi->cache_validity |= save_cache_validity & 2058 (NFS_INO_INVALID_OTHER 2059 | NFS_INO_REVAL_FORCED); 2060 cache_revalidated = false; 2061 } 2062 2063 if (fattr->valid & NFS_ATTR_FATTR_SPACE_USED) { 2064 /* 2065 * report the blocks in 512byte units 2066 */ 2067 inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used); 2068 } else if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED) 2069 inode->i_blocks = fattr->du.nfs2.blocks; 2070 else { 2071 nfsi->cache_validity |= save_cache_validity & 2072 (NFS_INO_INVALID_BLOCKS 2073 | NFS_INO_REVAL_FORCED); 2074 cache_revalidated = false; 2075 } 2076 2077 /* Update attrtimeo value if we're out of the unstable period */ 2078 if (attr_changed) { 2079 invalid &= ~NFS_INO_INVALID_ATTR; 2080 nfs_inc_stats(inode, NFSIOS_ATTRINVALIDATE); 2081 nfsi->attrtimeo = NFS_MINATTRTIMEO(inode); 2082 nfsi->attrtimeo_timestamp = now; 2083 /* Set barrier to be more recent than all outstanding updates */ 2084 nfsi->attr_gencount = nfs_inc_attr_generation_counter(); 2085 } else { 2086 if (cache_revalidated) { 2087 if (!time_in_range_open(now, nfsi->attrtimeo_timestamp, 2088 nfsi->attrtimeo_timestamp + nfsi->attrtimeo)) { 2089 nfsi->attrtimeo <<= 1; 2090 if (nfsi->attrtimeo > NFS_MAXATTRTIMEO(inode)) 2091 nfsi->attrtimeo = NFS_MAXATTRTIMEO(inode); 2092 } 2093 nfsi->attrtimeo_timestamp = now; 2094 } 2095 /* Set the barrier to be more recent than this fattr */ 2096 if ((long)(fattr->gencount - nfsi->attr_gencount) > 0) 2097 nfsi->attr_gencount = fattr->gencount; 2098 } 2099 2100 /* Don't invalidate the data if we were to blame */ 2101 if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) 2102 || S_ISLNK(inode->i_mode))) 2103 invalid &= ~NFS_INO_INVALID_DATA; 2104 nfs_set_cache_invalid(inode, invalid); 2105 2106 return 0; 2107 out_err: 2108 /* 2109 * No need to worry about unhashing the dentry, as the 2110 * lookup validation will know that the inode is bad. 2111 * (But we fall through to invalidate the caches.) 2112 */ 2113 nfs_set_inode_stale_locked(inode); 2114 return -ESTALE; 2115} 2116 2117struct inode *nfs_alloc_inode(struct super_block *sb) 2118{ 2119 struct nfs_inode *nfsi; 2120 nfsi = kmem_cache_alloc(nfs_inode_cachep, GFP_KERNEL); 2121 if (!nfsi) 2122 return NULL; 2123 nfsi->flags = 0UL; 2124 nfsi->cache_validity = 0UL; 2125#if IS_ENABLED(CONFIG_NFS_V4) 2126 nfsi->nfs4_acl = NULL; 2127#endif /* CONFIG_NFS_V4 */ 2128#ifdef CONFIG_NFS_V4_2 2129 nfsi->xattr_cache = NULL; 2130#endif 2131 return &nfsi->vfs_inode; 2132} 2133EXPORT_SYMBOL_GPL(nfs_alloc_inode); 2134 2135void nfs_free_inode(struct inode *inode) 2136{ 2137 kmem_cache_free(nfs_inode_cachep, NFS_I(inode)); 2138} 2139EXPORT_SYMBOL_GPL(nfs_free_inode); 2140 2141static inline void nfs4_init_once(struct nfs_inode *nfsi) 2142{ 2143#if IS_ENABLED(CONFIG_NFS_V4) 2144 INIT_LIST_HEAD(&nfsi->open_states); 2145 nfsi->delegation = NULL; 2146 init_rwsem(&nfsi->rwsem); 2147 nfsi->layout = NULL; 2148#endif 2149} 2150 2151static void init_once(void *foo) 2152{ 2153 struct nfs_inode *nfsi = (struct nfs_inode *) foo; 2154 2155 inode_init_once(&nfsi->vfs_inode); 2156 INIT_LIST_HEAD(&nfsi->open_files); 2157 INIT_LIST_HEAD(&nfsi->access_cache_entry_lru); 2158 INIT_LIST_HEAD(&nfsi->access_cache_inode_lru); 2159 INIT_LIST_HEAD(&nfsi->commit_info.list); 2160 atomic_long_set(&nfsi->nrequests, 0); 2161 atomic_long_set(&nfsi->commit_info.ncommit, 0); 2162 atomic_set(&nfsi->commit_info.rpcs_out, 0); 2163 init_rwsem(&nfsi->rmdir_sem); 2164 mutex_init(&nfsi->commit_mutex); 2165 nfs4_init_once(nfsi); 2166 nfsi->cache_change_attribute = 0; 2167} 2168 2169static int __init nfs_init_inodecache(void) 2170{ 2171 nfs_inode_cachep = kmem_cache_create("nfs_inode_cache", 2172 sizeof(struct nfs_inode), 2173 0, (SLAB_RECLAIM_ACCOUNT| 2174 SLAB_MEM_SPREAD|SLAB_ACCOUNT), 2175 init_once); 2176 if (nfs_inode_cachep == NULL) 2177 return -ENOMEM; 2178 2179 return 0; 2180} 2181 2182static void nfs_destroy_inodecache(void) 2183{ 2184 /* 2185 * Make sure all delayed rcu free inodes are flushed before we 2186 * destroy cache. 2187 */ 2188 rcu_barrier(); 2189 kmem_cache_destroy(nfs_inode_cachep); 2190} 2191 2192struct workqueue_struct *nfsiod_workqueue; 2193EXPORT_SYMBOL_GPL(nfsiod_workqueue); 2194 2195/* 2196 * start up the nfsiod workqueue 2197 */ 2198static int nfsiod_start(void) 2199{ 2200 struct workqueue_struct *wq; 2201 dprintk("RPC: creating workqueue nfsiod\n"); 2202 wq = alloc_workqueue("nfsiod", WQ_MEM_RECLAIM | WQ_UNBOUND, 0); 2203 if (wq == NULL) 2204 return -ENOMEM; 2205 nfsiod_workqueue = wq; 2206 return 0; 2207} 2208 2209/* 2210 * Destroy the nfsiod workqueue 2211 */ 2212static void nfsiod_stop(void) 2213{ 2214 struct workqueue_struct *wq; 2215 2216 wq = nfsiod_workqueue; 2217 if (wq == NULL) 2218 return; 2219 nfsiod_workqueue = NULL; 2220 destroy_workqueue(wq); 2221} 2222 2223unsigned int nfs_net_id; 2224EXPORT_SYMBOL_GPL(nfs_net_id); 2225 2226static int nfs_net_init(struct net *net) 2227{ 2228 struct nfs_net *nn = net_generic(net, nfs_net_id); 2229 2230 nfs_clients_init(net); 2231 2232 if (!rpc_proc_register(net, &nn->rpcstats)) { 2233 nfs_clients_exit(net); 2234 return -ENOMEM; 2235 } 2236 2237 return nfs_fs_proc_net_init(net); 2238} 2239 2240static void nfs_net_exit(struct net *net) 2241{ 2242 rpc_proc_unregister(net, "nfs"); 2243 nfs_fs_proc_net_exit(net); 2244 nfs_clients_exit(net); 2245} 2246 2247static struct pernet_operations nfs_net_ops = { 2248 .init = nfs_net_init, 2249 .exit = nfs_net_exit, 2250 .id = &nfs_net_id, 2251 .size = sizeof(struct nfs_net), 2252}; 2253 2254/* 2255 * Initialize NFS 2256 */ 2257static int __init init_nfs_fs(void) 2258{ 2259 int err; 2260 2261 err = nfs_sysfs_init(); 2262 if (err < 0) 2263 goto out10; 2264 2265 err = register_pernet_subsys(&nfs_net_ops); 2266 if (err < 0) 2267 goto out9; 2268 2269 err = nfs_fscache_register(); 2270 if (err < 0) 2271 goto out8; 2272 2273 err = nfsiod_start(); 2274 if (err) 2275 goto out7; 2276 2277 err = nfs_fs_proc_init(); 2278 if (err) 2279 goto out6; 2280 2281 err = nfs_init_nfspagecache(); 2282 if (err) 2283 goto out5; 2284 2285 err = nfs_init_inodecache(); 2286 if (err) 2287 goto out4; 2288 2289 err = nfs_init_readpagecache(); 2290 if (err) 2291 goto out3; 2292 2293 err = nfs_init_writepagecache(); 2294 if (err) 2295 goto out2; 2296 2297 err = nfs_init_directcache(); 2298 if (err) 2299 goto out1; 2300 2301 err = register_nfs_fs(); 2302 if (err) 2303 goto out0; 2304 2305 return 0; 2306out0: 2307 nfs_destroy_directcache(); 2308out1: 2309 nfs_destroy_writepagecache(); 2310out2: 2311 nfs_destroy_readpagecache(); 2312out3: 2313 nfs_destroy_inodecache(); 2314out4: 2315 nfs_destroy_nfspagecache(); 2316out5: 2317 nfs_fs_proc_exit(); 2318out6: 2319 nfsiod_stop(); 2320out7: 2321 nfs_fscache_unregister(); 2322out8: 2323 unregister_pernet_subsys(&nfs_net_ops); 2324out9: 2325 nfs_sysfs_exit(); 2326out10: 2327 return err; 2328} 2329 2330static void __exit exit_nfs_fs(void) 2331{ 2332 nfs_destroy_directcache(); 2333 nfs_destroy_writepagecache(); 2334 nfs_destroy_readpagecache(); 2335 nfs_destroy_inodecache(); 2336 nfs_destroy_nfspagecache(); 2337 nfs_fscache_unregister(); 2338 unregister_pernet_subsys(&nfs_net_ops); 2339 unregister_nfs_fs(); 2340 nfs_fs_proc_exit(); 2341 nfsiod_stop(); 2342 nfs_sysfs_exit(); 2343} 2344 2345/* Not quite true; I just maintain it */ 2346MODULE_AUTHOR("Olaf Kirch <okir@monad.swb.de>"); 2347MODULE_LICENSE("GPL"); 2348module_param(enable_ino64, bool, 0644); 2349 2350module_init(init_nfs_fs) 2351module_exit(exit_nfs_fs) 2352