aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
Diffstat (limited to 'fs')
-rw-r--r--fs/9p/vfs_inode.c41
-rw-r--r--fs/binfmt_elf_fdpic.c13
-rw-r--r--fs/bio.c2
-rw-r--r--fs/btrfs/acl.c12
-rw-r--r--fs/btrfs/extent-tree.c32
-rw-r--r--fs/btrfs/file.c100
-rw-r--r--fs/btrfs/inode.c12
-rw-r--r--fs/btrfs/ordered-data.c2
-rw-r--r--fs/btrfs/relocation.c4
-rw-r--r--fs/btrfs/volumes.c4
-rw-r--r--fs/cifs/cifs_dfs_ref.c3
-rw-r--r--fs/compat_ioctl.c3
-rw-r--r--fs/configfs/symlink.c4
-rw-r--r--fs/ecryptfs/crypto.c4
-rw-r--r--fs/ecryptfs/file.c17
-rw-r--r--fs/ecryptfs/inode.c158
-rw-r--r--fs/ecryptfs/main.c4
-rw-r--r--fs/exofs/inode.c17
-rw-r--r--fs/exofs/pnfs.h10
-rw-r--r--fs/fcntl.c102
-rw-r--r--fs/gfs2/file.c38
-rw-r--r--fs/gfs2/meta_io.c2
-rw-r--r--fs/gfs2/ops_inode.c6
-rw-r--r--fs/gfs2/xattr.c21
-rw-r--r--fs/hppfs/hppfs.c18
-rw-r--r--fs/namei.c23
-rw-r--r--fs/namespace.c14
-rw-r--r--fs/nfs/dir.c1
-rw-r--r--fs/nfsd/vfs.c5
-rw-r--r--fs/notify/inotify/inotify_fsnotify.c2
-rw-r--r--fs/notify/inotify/inotify_user.c4
-rw-r--r--fs/proc/array.c89
-rw-r--r--fs/proc/base.c1
-rw-r--r--fs/proc/task_mmu.c3
-rw-r--r--fs/quota/dquot.c3
-rw-r--r--fs/ramfs/file-nommu.c26
-rw-r--r--fs/reiserfs/inode.c19
-rw-r--r--fs/reiserfs/ioctl.c3
-rw-r--r--fs/reiserfs/xattr.c18
-rw-r--r--fs/reiserfs/xattr_acl.c2
-rw-r--r--fs/ubifs/gc.c96
-rw-r--r--fs/xfs/linux-2.6/xfs_acl.c3
-rw-r--r--fs/xfs/linux-2.6/xfs_super.c14
-rw-r--r--fs/xfs/linux-2.6/xfs_sync.c183
-rw-r--r--fs/xfs/linux-2.6/xfs_sync.h2
-rw-r--r--fs/xfs/linux-2.6/xfs_trace.h1145
-rw-r--r--fs/xfs/quota/xfs_qm_syscalls.c2
-rw-r--r--fs/xfs/xfs_alloc.c44
-rw-r--r--fs/xfs/xfs_dfrag.c106
-rw-r--r--fs/xfs/xfs_iget.c1
-rw-r--r--fs/xfs/xfs_inode.c17
-rw-r--r--fs/xfs/xfs_rtalloc.c2
-rw-r--r--fs/xfs/xfs_vnodeops.c93
53 files changed, 1342 insertions, 1208 deletions
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 18f74ec4dce9..9d03d1ebca6f 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -1001,44 +1001,6 @@ done:
1001} 1001}
1002 1002
1003/** 1003/**
1004 * v9fs_vfs_readlink - read a symlink's location
1005 * @dentry: dentry for symlink
1006 * @buffer: buffer to load symlink location into
1007 * @buflen: length of buffer
1008 *
1009 */
1010
1011static int v9fs_vfs_readlink(struct dentry *dentry, char __user * buffer,
1012 int buflen)
1013{
1014 int retval;
1015 int ret;
1016 char *link = __getname();
1017
1018 if (unlikely(!link))
1019 return -ENOMEM;
1020
1021 if (buflen > PATH_MAX)
1022 buflen = PATH_MAX;
1023
1024 P9_DPRINTK(P9_DEBUG_VFS, " dentry: %s (%p)\n", dentry->d_name.name,
1025 dentry);
1026
1027 retval = v9fs_readlink(dentry, link, buflen);
1028
1029 if (retval > 0) {
1030 if ((ret = copy_to_user(buffer, link, retval)) != 0) {
1031 P9_DPRINTK(P9_DEBUG_ERROR,
1032 "problem copying to user: %d\n", ret);
1033 retval = ret;
1034 }
1035 }
1036
1037 __putname(link);
1038 return retval;
1039}
1040
1041/**
1042 * v9fs_vfs_follow_link - follow a symlink path 1004 * v9fs_vfs_follow_link - follow a symlink path
1043 * @dentry: dentry for symlink 1005 * @dentry: dentry for symlink
1044 * @nd: nameidata 1006 * @nd: nameidata
@@ -1230,7 +1192,6 @@ static const struct inode_operations v9fs_dir_inode_operations_ext = {
1230 .rmdir = v9fs_vfs_rmdir, 1192 .rmdir = v9fs_vfs_rmdir,
1231 .mknod = v9fs_vfs_mknod, 1193 .mknod = v9fs_vfs_mknod,
1232 .rename = v9fs_vfs_rename, 1194 .rename = v9fs_vfs_rename,
1233 .readlink = v9fs_vfs_readlink,
1234 .getattr = v9fs_vfs_getattr, 1195 .getattr = v9fs_vfs_getattr,
1235 .setattr = v9fs_vfs_setattr, 1196 .setattr = v9fs_vfs_setattr,
1236}; 1197};
@@ -1253,7 +1214,7 @@ static const struct inode_operations v9fs_file_inode_operations = {
1253}; 1214};
1254 1215
1255static const struct inode_operations v9fs_symlink_inode_operations = { 1216static const struct inode_operations v9fs_symlink_inode_operations = {
1256 .readlink = v9fs_vfs_readlink, 1217 .readlink = generic_readlink,
1257 .follow_link = v9fs_vfs_follow_link, 1218 .follow_link = v9fs_vfs_follow_link,
1258 .put_link = v9fs_vfs_put_link, 1219 .put_link = v9fs_vfs_put_link,
1259 .getattr = v9fs_vfs_getattr, 1220 .getattr = v9fs_vfs_getattr,
diff --git a/fs/binfmt_elf_fdpic.c b/fs/binfmt_elf_fdpic.c
index 7dc85997e96c..c57d9ce5ff7e 100644
--- a/fs/binfmt_elf_fdpic.c
+++ b/fs/binfmt_elf_fdpic.c
@@ -171,6 +171,9 @@ static int load_elf_fdpic_binary(struct linux_binprm *bprm,
171#ifdef ELF_FDPIC_PLAT_INIT 171#ifdef ELF_FDPIC_PLAT_INIT
172 unsigned long dynaddr; 172 unsigned long dynaddr;
173#endif 173#endif
174#ifndef CONFIG_MMU
175 unsigned long stack_prot;
176#endif
174 struct file *interpreter = NULL; /* to shut gcc up */ 177 struct file *interpreter = NULL; /* to shut gcc up */
175 char *interpreter_name = NULL; 178 char *interpreter_name = NULL;
176 int executable_stack; 179 int executable_stack;
@@ -316,6 +319,8 @@ static int load_elf_fdpic_binary(struct linux_binprm *bprm,
316 * defunct, deceased, etc. after this point we have to exit via 319 * defunct, deceased, etc. after this point we have to exit via
317 * error_kill */ 320 * error_kill */
318 set_personality(PER_LINUX_FDPIC); 321 set_personality(PER_LINUX_FDPIC);
322 if (elf_read_implies_exec(&exec_params.hdr, executable_stack))
323 current->personality |= READ_IMPLIES_EXEC;
319 set_binfmt(&elf_fdpic_format); 324 set_binfmt(&elf_fdpic_format);
320 325
321 current->mm->start_code = 0; 326 current->mm->start_code = 0;
@@ -377,9 +382,13 @@ static int load_elf_fdpic_binary(struct linux_binprm *bprm,
377 if (stack_size < PAGE_SIZE * 2) 382 if (stack_size < PAGE_SIZE * 2)
378 stack_size = PAGE_SIZE * 2; 383 stack_size = PAGE_SIZE * 2;
379 384
385 stack_prot = PROT_READ | PROT_WRITE;
386 if (executable_stack == EXSTACK_ENABLE_X ||
387 (executable_stack == EXSTACK_DEFAULT && VM_STACK_FLAGS & VM_EXEC))
388 stack_prot |= PROT_EXEC;
389
380 down_write(&current->mm->mmap_sem); 390 down_write(&current->mm->mmap_sem);
381 current->mm->start_brk = do_mmap(NULL, 0, stack_size, 391 current->mm->start_brk = do_mmap(NULL, 0, stack_size, stack_prot,
382 PROT_READ | PROT_WRITE | PROT_EXEC,
383 MAP_PRIVATE | MAP_ANONYMOUS | 392 MAP_PRIVATE | MAP_ANONYMOUS |
384 MAP_UNINITIALIZED | MAP_GROWSDOWN, 393 MAP_UNINITIALIZED | MAP_GROWSDOWN,
385 0); 394 0);
diff --git a/fs/bio.c b/fs/bio.c
index 76e6713abf94..12429c9553eb 100644
--- a/fs/bio.c
+++ b/fs/bio.c
@@ -78,7 +78,7 @@ static struct kmem_cache *bio_find_or_create_slab(unsigned int extra_size)
78 78
79 i = 0; 79 i = 0;
80 while (i < bio_slab_nr) { 80 while (i < bio_slab_nr) {
81 struct bio_slab *bslab = &bio_slabs[i]; 81 bslab = &bio_slabs[i];
82 82
83 if (!bslab->slab && entry == -1) 83 if (!bslab->slab && entry == -1)
84 entry = i; 84 entry = i;
diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c
index 2e9e69987a82..54f4798ab46a 100644
--- a/fs/btrfs/acl.c
+++ b/fs/btrfs/acl.c
@@ -112,12 +112,14 @@ static int btrfs_set_acl(struct btrfs_trans_handle *trans,
112 switch (type) { 112 switch (type) {
113 case ACL_TYPE_ACCESS: 113 case ACL_TYPE_ACCESS:
114 mode = inode->i_mode; 114 mode = inode->i_mode;
115 ret = posix_acl_equiv_mode(acl, &mode);
116 if (ret < 0)
117 return ret;
118 ret = 0;
119 inode->i_mode = mode;
120 name = POSIX_ACL_XATTR_ACCESS; 115 name = POSIX_ACL_XATTR_ACCESS;
116 if (acl) {
117 ret = posix_acl_equiv_mode(acl, &mode);
118 if (ret < 0)
119 return ret;
120 inode->i_mode = mode;
121 }
122 ret = 0;
121 break; 123 break;
122 case ACL_TYPE_DEFAULT: 124 case ACL_TYPE_DEFAULT:
123 if (!S_ISDIR(inode->i_mode)) 125 if (!S_ISDIR(inode->i_mode))
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 56e50137d0e6..432a2da4641e 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -83,6 +83,17 @@ static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
83 return (cache->flags & bits) == bits; 83 return (cache->flags & bits) == bits;
84} 84}
85 85
86void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
87{
88 atomic_inc(&cache->count);
89}
90
91void btrfs_put_block_group(struct btrfs_block_group_cache *cache)
92{
93 if (atomic_dec_and_test(&cache->count))
94 kfree(cache);
95}
96
86/* 97/*
87 * this adds the block group to the fs_info rb tree for the block group 98 * this adds the block group to the fs_info rb tree for the block group
88 * cache 99 * cache
@@ -156,7 +167,7 @@ block_group_cache_tree_search(struct btrfs_fs_info *info, u64 bytenr,
156 } 167 }
157 } 168 }
158 if (ret) 169 if (ret)
159 atomic_inc(&ret->count); 170 btrfs_get_block_group(ret);
160 spin_unlock(&info->block_group_cache_lock); 171 spin_unlock(&info->block_group_cache_lock);
161 172
162 return ret; 173 return ret;
@@ -407,6 +418,8 @@ err:
407 418
408 put_caching_control(caching_ctl); 419 put_caching_control(caching_ctl);
409 atomic_dec(&block_group->space_info->caching_threads); 420 atomic_dec(&block_group->space_info->caching_threads);
421 btrfs_put_block_group(block_group);
422
410 return 0; 423 return 0;
411} 424}
412 425
@@ -447,6 +460,7 @@ static int cache_block_group(struct btrfs_block_group_cache *cache)
447 up_write(&fs_info->extent_commit_sem); 460 up_write(&fs_info->extent_commit_sem);
448 461
449 atomic_inc(&cache->space_info->caching_threads); 462 atomic_inc(&cache->space_info->caching_threads);
463 btrfs_get_block_group(cache);
450 464
451 tsk = kthread_run(caching_kthread, cache, "btrfs-cache-%llu\n", 465 tsk = kthread_run(caching_kthread, cache, "btrfs-cache-%llu\n",
452 cache->key.objectid); 466 cache->key.objectid);
@@ -486,12 +500,6 @@ struct btrfs_block_group_cache *btrfs_lookup_block_group(
486 return cache; 500 return cache;
487} 501}
488 502
489void btrfs_put_block_group(struct btrfs_block_group_cache *cache)
490{
491 if (atomic_dec_and_test(&cache->count))
492 kfree(cache);
493}
494
495static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info, 503static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info,
496 u64 flags) 504 u64 flags)
497{ 505{
@@ -2582,7 +2590,7 @@ next_block_group(struct btrfs_root *root,
2582 if (node) { 2590 if (node) {
2583 cache = rb_entry(node, struct btrfs_block_group_cache, 2591 cache = rb_entry(node, struct btrfs_block_group_cache,
2584 cache_node); 2592 cache_node);
2585 atomic_inc(&cache->count); 2593 btrfs_get_block_group(cache);
2586 } else 2594 } else
2587 cache = NULL; 2595 cache = NULL;
2588 spin_unlock(&root->fs_info->block_group_cache_lock); 2596 spin_unlock(&root->fs_info->block_group_cache_lock);
@@ -4227,7 +4235,7 @@ search:
4227 u64 offset; 4235 u64 offset;
4228 int cached; 4236 int cached;
4229 4237
4230 atomic_inc(&block_group->count); 4238 btrfs_get_block_group(block_group);
4231 search_start = block_group->key.objectid; 4239 search_start = block_group->key.objectid;
4232 4240
4233have_block_group: 4241have_block_group:
@@ -4315,7 +4323,7 @@ have_block_group:
4315 4323
4316 btrfs_put_block_group(block_group); 4324 btrfs_put_block_group(block_group);
4317 block_group = last_ptr->block_group; 4325 block_group = last_ptr->block_group;
4318 atomic_inc(&block_group->count); 4326 btrfs_get_block_group(block_group);
4319 spin_unlock(&last_ptr->lock); 4327 spin_unlock(&last_ptr->lock);
4320 spin_unlock(&last_ptr->refill_lock); 4328 spin_unlock(&last_ptr->refill_lock);
4321 4329
@@ -7395,9 +7403,7 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
7395 wait_block_group_cache_done(block_group); 7403 wait_block_group_cache_done(block_group);
7396 7404
7397 btrfs_remove_free_space_cache(block_group); 7405 btrfs_remove_free_space_cache(block_group);
7398 7406 btrfs_put_block_group(block_group);
7399 WARN_ON(atomic_read(&block_group->count) != 1);
7400 kfree(block_group);
7401 7407
7402 spin_lock(&info->block_group_cache_lock); 7408 spin_lock(&info->block_group_cache_lock);
7403 } 7409 }
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index feaa13b105d9..c02033596f02 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -506,7 +506,8 @@ next_slot:
506} 506}
507 507
508static int extent_mergeable(struct extent_buffer *leaf, int slot, 508static int extent_mergeable(struct extent_buffer *leaf, int slot,
509 u64 objectid, u64 bytenr, u64 *start, u64 *end) 509 u64 objectid, u64 bytenr, u64 orig_offset,
510 u64 *start, u64 *end)
510{ 511{
511 struct btrfs_file_extent_item *fi; 512 struct btrfs_file_extent_item *fi;
512 struct btrfs_key key; 513 struct btrfs_key key;
@@ -522,6 +523,7 @@ static int extent_mergeable(struct extent_buffer *leaf, int slot,
522 fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item); 523 fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
523 if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG || 524 if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG ||
524 btrfs_file_extent_disk_bytenr(leaf, fi) != bytenr || 525 btrfs_file_extent_disk_bytenr(leaf, fi) != bytenr ||
526 btrfs_file_extent_offset(leaf, fi) != key.offset - orig_offset ||
525 btrfs_file_extent_compression(leaf, fi) || 527 btrfs_file_extent_compression(leaf, fi) ||
526 btrfs_file_extent_encryption(leaf, fi) || 528 btrfs_file_extent_encryption(leaf, fi) ||
527 btrfs_file_extent_other_encoding(leaf, fi)) 529 btrfs_file_extent_other_encoding(leaf, fi))
@@ -561,6 +563,7 @@ int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
561 u64 split; 563 u64 split;
562 int del_nr = 0; 564 int del_nr = 0;
563 int del_slot = 0; 565 int del_slot = 0;
566 int recow;
564 int ret; 567 int ret;
565 568
566 btrfs_drop_extent_cache(inode, start, end - 1, 0); 569 btrfs_drop_extent_cache(inode, start, end - 1, 0);
@@ -568,6 +571,7 @@ int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
568 path = btrfs_alloc_path(); 571 path = btrfs_alloc_path();
569 BUG_ON(!path); 572 BUG_ON(!path);
570again: 573again:
574 recow = 0;
571 split = start; 575 split = start;
572 key.objectid = inode->i_ino; 576 key.objectid = inode->i_ino;
573 key.type = BTRFS_EXTENT_DATA_KEY; 577 key.type = BTRFS_EXTENT_DATA_KEY;
@@ -591,12 +595,60 @@ again:
591 bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); 595 bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
592 num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi); 596 num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
593 orig_offset = key.offset - btrfs_file_extent_offset(leaf, fi); 597 orig_offset = key.offset - btrfs_file_extent_offset(leaf, fi);
598 memcpy(&new_key, &key, sizeof(new_key));
599
600 if (start == key.offset && end < extent_end) {
601 other_start = 0;
602 other_end = start;
603 if (extent_mergeable(leaf, path->slots[0] - 1,
604 inode->i_ino, bytenr, orig_offset,
605 &other_start, &other_end)) {
606 new_key.offset = end;
607 btrfs_set_item_key_safe(trans, root, path, &new_key);
608 fi = btrfs_item_ptr(leaf, path->slots[0],
609 struct btrfs_file_extent_item);
610 btrfs_set_file_extent_num_bytes(leaf, fi,
611 extent_end - end);
612 btrfs_set_file_extent_offset(leaf, fi,
613 end - orig_offset);
614 fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
615 struct btrfs_file_extent_item);
616 btrfs_set_file_extent_num_bytes(leaf, fi,
617 end - other_start);
618 btrfs_mark_buffer_dirty(leaf);
619 goto out;
620 }
621 }
622
623 if (start > key.offset && end == extent_end) {
624 other_start = end;
625 other_end = 0;
626 if (extent_mergeable(leaf, path->slots[0] + 1,
627 inode->i_ino, bytenr, orig_offset,
628 &other_start, &other_end)) {
629 fi = btrfs_item_ptr(leaf, path->slots[0],
630 struct btrfs_file_extent_item);
631 btrfs_set_file_extent_num_bytes(leaf, fi,
632 start - key.offset);
633 path->slots[0]++;
634 new_key.offset = start;
635 btrfs_set_item_key_safe(trans, root, path, &new_key);
636
637 fi = btrfs_item_ptr(leaf, path->slots[0],
638 struct btrfs_file_extent_item);
639 btrfs_set_file_extent_num_bytes(leaf, fi,
640 other_end - start);
641 btrfs_set_file_extent_offset(leaf, fi,
642 start - orig_offset);
643 btrfs_mark_buffer_dirty(leaf);
644 goto out;
645 }
646 }
594 647
595 while (start > key.offset || end < extent_end) { 648 while (start > key.offset || end < extent_end) {
596 if (key.offset == start) 649 if (key.offset == start)
597 split = end; 650 split = end;
598 651
599 memcpy(&new_key, &key, sizeof(new_key));
600 new_key.offset = split; 652 new_key.offset = split;
601 ret = btrfs_duplicate_item(trans, root, path, &new_key); 653 ret = btrfs_duplicate_item(trans, root, path, &new_key);
602 if (ret == -EAGAIN) { 654 if (ret == -EAGAIN) {
@@ -631,15 +683,18 @@ again:
631 path->slots[0]--; 683 path->slots[0]--;
632 extent_end = end; 684 extent_end = end;
633 } 685 }
686 recow = 1;
634 } 687 }
635 688
636 fi = btrfs_item_ptr(leaf, path->slots[0],
637 struct btrfs_file_extent_item);
638
639 other_start = end; 689 other_start = end;
640 other_end = 0; 690 other_end = 0;
641 if (extent_mergeable(leaf, path->slots[0] + 1, inode->i_ino, 691 if (extent_mergeable(leaf, path->slots[0] + 1,
642 bytenr, &other_start, &other_end)) { 692 inode->i_ino, bytenr, orig_offset,
693 &other_start, &other_end)) {
694 if (recow) {
695 btrfs_release_path(root, path);
696 goto again;
697 }
643 extent_end = other_end; 698 extent_end = other_end;
644 del_slot = path->slots[0] + 1; 699 del_slot = path->slots[0] + 1;
645 del_nr++; 700 del_nr++;
@@ -650,8 +705,13 @@ again:
650 } 705 }
651 other_start = 0; 706 other_start = 0;
652 other_end = start; 707 other_end = start;
653 if (extent_mergeable(leaf, path->slots[0] - 1, inode->i_ino, 708 if (extent_mergeable(leaf, path->slots[0] - 1,
654 bytenr, &other_start, &other_end)) { 709 inode->i_ino, bytenr, orig_offset,
710 &other_start, &other_end)) {
711 if (recow) {
712 btrfs_release_path(root, path);
713 goto again;
714 }
655 key.offset = other_start; 715 key.offset = other_start;
656 del_slot = path->slots[0]; 716 del_slot = path->slots[0];
657 del_nr++; 717 del_nr++;
@@ -660,22 +720,22 @@ again:
660 inode->i_ino, orig_offset); 720 inode->i_ino, orig_offset);
661 BUG_ON(ret); 721 BUG_ON(ret);
662 } 722 }
723 fi = btrfs_item_ptr(leaf, path->slots[0],
724 struct btrfs_file_extent_item);
663 if (del_nr == 0) { 725 if (del_nr == 0) {
664 btrfs_set_file_extent_type(leaf, fi, 726 btrfs_set_file_extent_type(leaf, fi,
665 BTRFS_FILE_EXTENT_REG); 727 BTRFS_FILE_EXTENT_REG);
666 btrfs_mark_buffer_dirty(leaf); 728 btrfs_mark_buffer_dirty(leaf);
667 goto out; 729 } else {
668 } 730 btrfs_set_file_extent_type(leaf, fi,
669 731 BTRFS_FILE_EXTENT_REG);
670 fi = btrfs_item_ptr(leaf, del_slot - 1, 732 btrfs_set_file_extent_num_bytes(leaf, fi,
671 struct btrfs_file_extent_item); 733 extent_end - key.offset);
672 btrfs_set_file_extent_type(leaf, fi, BTRFS_FILE_EXTENT_REG); 734 btrfs_mark_buffer_dirty(leaf);
673 btrfs_set_file_extent_num_bytes(leaf, fi,
674 extent_end - key.offset);
675 btrfs_mark_buffer_dirty(leaf);
676 735
677 ret = btrfs_del_items(trans, root, path, del_slot, del_nr); 736 ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
678 BUG_ON(ret); 737 BUG_ON(ret);
738 }
679out: 739out:
680 btrfs_free_path(path); 740 btrfs_free_path(path);
681 return 0; 741 return 0;
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 5440bab23635..b330e27c2d8b 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -3796,6 +3796,12 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
3796 3796
3797 if (location.type == BTRFS_INODE_ITEM_KEY) { 3797 if (location.type == BTRFS_INODE_ITEM_KEY) {
3798 inode = btrfs_iget(dir->i_sb, &location, root); 3798 inode = btrfs_iget(dir->i_sb, &location, root);
3799 if (unlikely(root->clean_orphans) &&
3800 !(inode->i_sb->s_flags & MS_RDONLY)) {
3801 down_read(&root->fs_info->cleanup_work_sem);
3802 btrfs_orphan_cleanup(root);
3803 up_read(&root->fs_info->cleanup_work_sem);
3804 }
3799 return inode; 3805 return inode;
3800 } 3806 }
3801 3807
@@ -3995,7 +4001,11 @@ skip:
3995 4001
3996 /* Reached end of directory/root. Bump pos past the last item. */ 4002 /* Reached end of directory/root. Bump pos past the last item. */
3997 if (key_type == BTRFS_DIR_INDEX_KEY) 4003 if (key_type == BTRFS_DIR_INDEX_KEY)
3998 filp->f_pos = INT_LIMIT(off_t); 4004 /*
4005 * 32-bit glibc will use getdents64, but then strtol -
4006 * so the last number we can serve is this.
4007 */
4008 filp->f_pos = 0x7fffffff;
3999 else 4009 else
4000 filp->f_pos++; 4010 filp->f_pos++;
4001nopos: 4011nopos:
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index b10a49d4bc6a..5c2a9e78a949 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -626,6 +626,8 @@ int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
626 626
627 if (ordered) 627 if (ordered)
628 offset = entry_end(ordered); 628 offset = entry_end(ordered);
629 else
630 offset = ALIGN(offset, BTRFS_I(inode)->root->sectorsize);
629 631
630 mutex_lock(&tree->mutex); 632 mutex_lock(&tree->mutex);
631 disk_i_size = BTRFS_I(inode)->disk_i_size; 633 disk_i_size = BTRFS_I(inode)->disk_i_size;
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index a9728680eca8..ed3e4a2ec2c8 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -3281,8 +3281,10 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
3281 return -ENOMEM; 3281 return -ENOMEM;
3282 3282
3283 path = btrfs_alloc_path(); 3283 path = btrfs_alloc_path();
3284 if (!path) 3284 if (!path) {
3285 kfree(cluster);
3285 return -ENOMEM; 3286 return -ENOMEM;
3287 }
3286 3288
3287 rc->extents_found = 0; 3289 rc->extents_found = 0;
3288 rc->extents_skipped = 0; 3290 rc->extents_skipped = 0;
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 198cff28766d..220dad5db017 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2649,8 +2649,10 @@ again:
2649 em = lookup_extent_mapping(em_tree, logical, *length); 2649 em = lookup_extent_mapping(em_tree, logical, *length);
2650 read_unlock(&em_tree->lock); 2650 read_unlock(&em_tree->lock);
2651 2651
2652 if (!em && unplug_page) 2652 if (!em && unplug_page) {
2653 kfree(multi);
2653 return 0; 2654 return 0;
2655 }
2654 2656
2655 if (!em) { 2657 if (!em) {
2656 printk(KERN_CRIT "unable to find logical %llu len %llu\n", 2658 printk(KERN_CRIT "unable to find logical %llu len %llu\n",
diff --git a/fs/cifs/cifs_dfs_ref.c b/fs/cifs/cifs_dfs_ref.c
index fea9e898c4ba..b44ce0a0711c 100644
--- a/fs/cifs/cifs_dfs_ref.c
+++ b/fs/cifs/cifs_dfs_ref.c
@@ -269,7 +269,7 @@ static int add_mount_helper(struct vfsmount *newmnt, struct nameidata *nd,
269 int err; 269 int err;
270 270
271 mntget(newmnt); 271 mntget(newmnt);
272 err = do_add_mount(newmnt, &nd->path, nd->path.mnt->mnt_flags, mntlist); 272 err = do_add_mount(newmnt, &nd->path, nd->path.mnt->mnt_flags | MNT_SHRINKABLE, mntlist);
273 switch (err) { 273 switch (err) {
274 case 0: 274 case 0:
275 path_put(&nd->path); 275 path_put(&nd->path);
@@ -371,7 +371,6 @@ cifs_dfs_follow_mountpoint(struct dentry *dentry, struct nameidata *nd)
371 if (IS_ERR(mnt)) 371 if (IS_ERR(mnt))
372 goto out_err; 372 goto out_err;
373 373
374 nd->path.mnt->mnt_flags |= MNT_SHRINKABLE;
375 rc = add_mount_helper(mnt, nd, &cifs_dfs_automount_list); 374 rc = add_mount_helper(mnt, nd, &cifs_dfs_automount_list);
376 375
377out: 376out:
diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c
index 332dd00f0894..c5c45de1a2ee 100644
--- a/fs/compat_ioctl.c
+++ b/fs/compat_ioctl.c
@@ -1005,6 +1005,9 @@ COMPATIBLE_IOCTL(SCSI_IOCTL_SEND_COMMAND)
1005COMPATIBLE_IOCTL(SCSI_IOCTL_PROBE_HOST) 1005COMPATIBLE_IOCTL(SCSI_IOCTL_PROBE_HOST)
1006COMPATIBLE_IOCTL(SCSI_IOCTL_GET_PCI) 1006COMPATIBLE_IOCTL(SCSI_IOCTL_GET_PCI)
1007#endif 1007#endif
1008/* Big V (don't complain on serial console) */
1009IGNORE_IOCTL(VT_OPENQRY)
1010IGNORE_IOCTL(VT_GETMODE)
1008/* Little p (/dev/rtc, /dev/envctrl, etc.) */ 1011/* Little p (/dev/rtc, /dev/envctrl, etc.) */
1009COMPATIBLE_IOCTL(RTC_AIE_ON) 1012COMPATIBLE_IOCTL(RTC_AIE_ON)
1010COMPATIBLE_IOCTL(RTC_AIE_OFF) 1013COMPATIBLE_IOCTL(RTC_AIE_OFF)
diff --git a/fs/configfs/symlink.c b/fs/configfs/symlink.c
index c8afa6b1d91d..32a5f46b1157 100644
--- a/fs/configfs/symlink.c
+++ b/fs/configfs/symlink.c
@@ -121,8 +121,10 @@ static int get_target(const char *symname, struct path *path,
121 ret = -ENOENT; 121 ret = -ENOENT;
122 path_put(path); 122 path_put(path);
123 } 123 }
124 } else 124 } else {
125 ret = -EPERM; 125 ret = -EPERM;
126 path_put(path);
127 }
126 } 128 }
127 129
128 return ret; 130 return ret;
diff --git a/fs/ecryptfs/crypto.c b/fs/ecryptfs/crypto.c
index fbb6e5eed697..7cb0a59f4b9d 100644
--- a/fs/ecryptfs/crypto.c
+++ b/fs/ecryptfs/crypto.c
@@ -1748,7 +1748,7 @@ ecryptfs_process_key_cipher(struct crypto_blkcipher **key_tfm,
1748 char *cipher_name, size_t *key_size) 1748 char *cipher_name, size_t *key_size)
1749{ 1749{
1750 char dummy_key[ECRYPTFS_MAX_KEY_BYTES]; 1750 char dummy_key[ECRYPTFS_MAX_KEY_BYTES];
1751 char *full_alg_name; 1751 char *full_alg_name = NULL;
1752 int rc; 1752 int rc;
1753 1753
1754 *key_tfm = NULL; 1754 *key_tfm = NULL;
@@ -1763,7 +1763,6 @@ ecryptfs_process_key_cipher(struct crypto_blkcipher **key_tfm,
1763 if (rc) 1763 if (rc)
1764 goto out; 1764 goto out;
1765 *key_tfm = crypto_alloc_blkcipher(full_alg_name, 0, CRYPTO_ALG_ASYNC); 1765 *key_tfm = crypto_alloc_blkcipher(full_alg_name, 0, CRYPTO_ALG_ASYNC);
1766 kfree(full_alg_name);
1767 if (IS_ERR(*key_tfm)) { 1766 if (IS_ERR(*key_tfm)) {
1768 rc = PTR_ERR(*key_tfm); 1767 rc = PTR_ERR(*key_tfm);
1769 printk(KERN_ERR "Unable to allocate crypto cipher with name " 1768 printk(KERN_ERR "Unable to allocate crypto cipher with name "
@@ -1786,6 +1785,7 @@ ecryptfs_process_key_cipher(struct crypto_blkcipher **key_tfm,
1786 goto out; 1785 goto out;
1787 } 1786 }
1788out: 1787out:
1788 kfree(full_alg_name);
1789 return rc; 1789 return rc;
1790} 1790}
1791 1791
diff --git a/fs/ecryptfs/file.c b/fs/ecryptfs/file.c
index 9e944057001b..678172b61be2 100644
--- a/fs/ecryptfs/file.c
+++ b/fs/ecryptfs/file.c
@@ -158,7 +158,7 @@ static int ecryptfs_open(struct inode *inode, struct file *file)
158 struct dentry *ecryptfs_dentry = file->f_path.dentry; 158 struct dentry *ecryptfs_dentry = file->f_path.dentry;
159 /* Private value of ecryptfs_dentry allocated in 159 /* Private value of ecryptfs_dentry allocated in
160 * ecryptfs_lookup() */ 160 * ecryptfs_lookup() */
161 struct dentry *lower_dentry = ecryptfs_dentry_to_lower(ecryptfs_dentry); 161 struct dentry *lower_dentry;
162 struct ecryptfs_file_info *file_info; 162 struct ecryptfs_file_info *file_info;
163 163
164 mount_crypt_stat = &ecryptfs_superblock_to_private( 164 mount_crypt_stat = &ecryptfs_superblock_to_private(
@@ -191,13 +191,6 @@ static int ecryptfs_open(struct inode *inode, struct file *file)
191 | ECRYPTFS_ENCRYPTED); 191 | ECRYPTFS_ENCRYPTED);
192 } 192 }
193 mutex_unlock(&crypt_stat->cs_mutex); 193 mutex_unlock(&crypt_stat->cs_mutex);
194 if ((ecryptfs_inode_to_private(inode)->lower_file->f_flags & O_RDONLY)
195 && !(file->f_flags & O_RDONLY)) {
196 rc = -EPERM;
197 printk(KERN_WARNING "%s: Lower persistent file is RO; eCryptfs "
198 "file must hence be opened RO\n", __func__);
199 goto out;
200 }
201 if (!ecryptfs_inode_to_private(inode)->lower_file) { 194 if (!ecryptfs_inode_to_private(inode)->lower_file) {
202 rc = ecryptfs_init_persistent_file(ecryptfs_dentry); 195 rc = ecryptfs_init_persistent_file(ecryptfs_dentry);
203 if (rc) { 196 if (rc) {
@@ -208,6 +201,13 @@ static int ecryptfs_open(struct inode *inode, struct file *file)
208 goto out; 201 goto out;
209 } 202 }
210 } 203 }
204 if ((ecryptfs_inode_to_private(inode)->lower_file->f_flags & O_RDONLY)
205 && !(file->f_flags & O_RDONLY)) {
206 rc = -EPERM;
207 printk(KERN_WARNING "%s: Lower persistent file is RO; eCryptfs "
208 "file must hence be opened RO\n", __func__);
209 goto out;
210 }
211 ecryptfs_set_file_lower( 211 ecryptfs_set_file_lower(
212 file, ecryptfs_inode_to_private(inode)->lower_file); 212 file, ecryptfs_inode_to_private(inode)->lower_file);
213 if (S_ISDIR(ecryptfs_dentry->d_inode->i_mode)) { 213 if (S_ISDIR(ecryptfs_dentry->d_inode->i_mode)) {
@@ -299,7 +299,6 @@ static int ecryptfs_ioctl(struct inode *inode, struct file *file,
299const struct file_operations ecryptfs_dir_fops = { 299const struct file_operations ecryptfs_dir_fops = {
300 .readdir = ecryptfs_readdir, 300 .readdir = ecryptfs_readdir,
301 .ioctl = ecryptfs_ioctl, 301 .ioctl = ecryptfs_ioctl,
302 .mmap = generic_file_mmap,
303 .open = ecryptfs_open, 302 .open = ecryptfs_open,
304 .flush = ecryptfs_flush, 303 .flush = ecryptfs_flush,
305 .release = ecryptfs_release, 304 .release = ecryptfs_release,
diff --git a/fs/ecryptfs/inode.c b/fs/ecryptfs/inode.c
index 429ca0b3ba08..4a430ab4115c 100644
--- a/fs/ecryptfs/inode.c
+++ b/fs/ecryptfs/inode.c
@@ -282,7 +282,8 @@ int ecryptfs_lookup_and_interpose_lower(struct dentry *ecryptfs_dentry,
282 goto out; 282 goto out;
283 } 283 }
284 rc = ecryptfs_interpose(lower_dentry, ecryptfs_dentry, 284 rc = ecryptfs_interpose(lower_dentry, ecryptfs_dentry,
285 ecryptfs_dir_inode->i_sb, 1); 285 ecryptfs_dir_inode->i_sb,
286 ECRYPTFS_INTERPOSE_FLAG_D_ADD);
286 if (rc) { 287 if (rc) {
287 printk(KERN_ERR "%s: Error interposing; rc = [%d]\n", 288 printk(KERN_ERR "%s: Error interposing; rc = [%d]\n",
288 __func__, rc); 289 __func__, rc);
@@ -463,9 +464,6 @@ out_lock:
463 unlock_dir(lower_dir_dentry); 464 unlock_dir(lower_dir_dentry);
464 dput(lower_new_dentry); 465 dput(lower_new_dentry);
465 dput(lower_old_dentry); 466 dput(lower_old_dentry);
466 d_drop(lower_old_dentry);
467 d_drop(new_dentry);
468 d_drop(old_dentry);
469 return rc; 467 return rc;
470} 468}
471 469
@@ -614,6 +612,7 @@ ecryptfs_rename(struct inode *old_dir, struct dentry *old_dentry,
614 struct dentry *lower_new_dentry; 612 struct dentry *lower_new_dentry;
615 struct dentry *lower_old_dir_dentry; 613 struct dentry *lower_old_dir_dentry;
616 struct dentry *lower_new_dir_dentry; 614 struct dentry *lower_new_dir_dentry;
615 struct dentry *trap = NULL;
617 616
618 lower_old_dentry = ecryptfs_dentry_to_lower(old_dentry); 617 lower_old_dentry = ecryptfs_dentry_to_lower(old_dentry);
619 lower_new_dentry = ecryptfs_dentry_to_lower(new_dentry); 618 lower_new_dentry = ecryptfs_dentry_to_lower(new_dentry);
@@ -621,7 +620,17 @@ ecryptfs_rename(struct inode *old_dir, struct dentry *old_dentry,
621 dget(lower_new_dentry); 620 dget(lower_new_dentry);
622 lower_old_dir_dentry = dget_parent(lower_old_dentry); 621 lower_old_dir_dentry = dget_parent(lower_old_dentry);
623 lower_new_dir_dentry = dget_parent(lower_new_dentry); 622 lower_new_dir_dentry = dget_parent(lower_new_dentry);
624 lock_rename(lower_old_dir_dentry, lower_new_dir_dentry); 623 trap = lock_rename(lower_old_dir_dentry, lower_new_dir_dentry);
624 /* source should not be ancestor of target */
625 if (trap == lower_old_dentry) {
626 rc = -EINVAL;
627 goto out_lock;
628 }
629 /* target should not be ancestor of source */
630 if (trap == lower_new_dentry) {
631 rc = -ENOTEMPTY;
632 goto out_lock;
633 }
625 rc = vfs_rename(lower_old_dir_dentry->d_inode, lower_old_dentry, 634 rc = vfs_rename(lower_old_dir_dentry->d_inode, lower_old_dentry,
626 lower_new_dir_dentry->d_inode, lower_new_dentry); 635 lower_new_dir_dentry->d_inode, lower_new_dentry);
627 if (rc) 636 if (rc)
@@ -715,31 +724,31 @@ static void *ecryptfs_follow_link(struct dentry *dentry, struct nameidata *nd)
715 /* Released in ecryptfs_put_link(); only release here on error */ 724 /* Released in ecryptfs_put_link(); only release here on error */
716 buf = kmalloc(len, GFP_KERNEL); 725 buf = kmalloc(len, GFP_KERNEL);
717 if (!buf) { 726 if (!buf) {
718 rc = -ENOMEM; 727 buf = ERR_PTR(-ENOMEM);
719 goto out; 728 goto out;
720 } 729 }
721 old_fs = get_fs(); 730 old_fs = get_fs();
722 set_fs(get_ds()); 731 set_fs(get_ds());
723 rc = dentry->d_inode->i_op->readlink(dentry, (char __user *)buf, len); 732 rc = dentry->d_inode->i_op->readlink(dentry, (char __user *)buf, len);
724 set_fs(old_fs); 733 set_fs(old_fs);
725 if (rc < 0) 734 if (rc < 0) {
726 goto out_free; 735 kfree(buf);
727 else 736 buf = ERR_PTR(rc);
737 } else
728 buf[rc] = '\0'; 738 buf[rc] = '\0';
729 rc = 0;
730 nd_set_link(nd, buf);
731 goto out;
732out_free:
733 kfree(buf);
734out: 739out:
735 return ERR_PTR(rc); 740 nd_set_link(nd, buf);
741 return NULL;
736} 742}
737 743
738static void 744static void
739ecryptfs_put_link(struct dentry *dentry, struct nameidata *nd, void *ptr) 745ecryptfs_put_link(struct dentry *dentry, struct nameidata *nd, void *ptr)
740{ 746{
741 /* Free the char* */ 747 char *buf = nd_get_link(nd);
742 kfree(nd_get_link(nd)); 748 if (!IS_ERR(buf)) {
749 /* Free the char* */
750 kfree(buf);
751 }
743} 752}
744 753
745/** 754/**
@@ -772,18 +781,23 @@ upper_size_to_lower_size(struct ecryptfs_crypt_stat *crypt_stat,
772} 781}
773 782
774/** 783/**
775 * ecryptfs_truncate 784 * truncate_upper
776 * @dentry: The ecryptfs layer dentry 785 * @dentry: The ecryptfs layer dentry
777 * @new_length: The length to expand the file to 786 * @ia: Address of the ecryptfs inode's attributes
787 * @lower_ia: Address of the lower inode's attributes
778 * 788 *
779 * Function to handle truncations modifying the size of the file. Note 789 * Function to handle truncations modifying the size of the file. Note
780 * that the file sizes are interpolated. When expanding, we are simply 790 * that the file sizes are interpolated. When expanding, we are simply
781 * writing strings of 0's out. When truncating, we need to modify the 791 * writing strings of 0's out. When truncating, we truncate the upper
782 * underlying file size according to the page index interpolations. 792 * inode and update the lower_ia according to the page index
793 * interpolations. If ATTR_SIZE is set in lower_ia->ia_valid upon return,
794 * the caller must use lower_ia in a call to notify_change() to perform
795 * the truncation of the lower inode.
783 * 796 *
784 * Returns zero on success; non-zero otherwise 797 * Returns zero on success; non-zero otherwise
785 */ 798 */
786int ecryptfs_truncate(struct dentry *dentry, loff_t new_length) 799static int truncate_upper(struct dentry *dentry, struct iattr *ia,
800 struct iattr *lower_ia)
787{ 801{
788 int rc = 0; 802 int rc = 0;
789 struct inode *inode = dentry->d_inode; 803 struct inode *inode = dentry->d_inode;
@@ -794,8 +808,10 @@ int ecryptfs_truncate(struct dentry *dentry, loff_t new_length)
794 loff_t lower_size_before_truncate; 808 loff_t lower_size_before_truncate;
795 loff_t lower_size_after_truncate; 809 loff_t lower_size_after_truncate;
796 810
797 if (unlikely((new_length == i_size))) 811 if (unlikely((ia->ia_size == i_size))) {
812 lower_ia->ia_valid &= ~ATTR_SIZE;
798 goto out; 813 goto out;
814 }
799 crypt_stat = &ecryptfs_inode_to_private(dentry->d_inode)->crypt_stat; 815 crypt_stat = &ecryptfs_inode_to_private(dentry->d_inode)->crypt_stat;
800 /* Set up a fake ecryptfs file, this is used to interface with 816 /* Set up a fake ecryptfs file, this is used to interface with
801 * the file in the underlying filesystem so that the 817 * the file in the underlying filesystem so that the
@@ -815,28 +831,30 @@ int ecryptfs_truncate(struct dentry *dentry, loff_t new_length)
815 &fake_ecryptfs_file, 831 &fake_ecryptfs_file,
816 ecryptfs_inode_to_private(dentry->d_inode)->lower_file); 832 ecryptfs_inode_to_private(dentry->d_inode)->lower_file);
817 /* Switch on growing or shrinking file */ 833 /* Switch on growing or shrinking file */
818 if (new_length > i_size) { 834 if (ia->ia_size > i_size) {
819 char zero[] = { 0x00 }; 835 char zero[] = { 0x00 };
820 836
837 lower_ia->ia_valid &= ~ATTR_SIZE;
821 /* Write a single 0 at the last position of the file; 838 /* Write a single 0 at the last position of the file;
822 * this triggers code that will fill in 0's throughout 839 * this triggers code that will fill in 0's throughout
823 * the intermediate portion of the previous end of the 840 * the intermediate portion of the previous end of the
824 * file and the new and of the file */ 841 * file and the new and of the file */
825 rc = ecryptfs_write(&fake_ecryptfs_file, zero, 842 rc = ecryptfs_write(&fake_ecryptfs_file, zero,
826 (new_length - 1), 1); 843 (ia->ia_size - 1), 1);
827 } else { /* new_length < i_size_read(inode) */ 844 } else { /* ia->ia_size < i_size_read(inode) */
828 /* We're chopping off all the pages down do the page 845 /* We're chopping off all the pages down to the page
829 * in which new_length is located. Fill in the end of 846 * in which ia->ia_size is located. Fill in the end of
830 * that page from (new_length & ~PAGE_CACHE_MASK) to 847 * that page from (ia->ia_size & ~PAGE_CACHE_MASK) to
831 * PAGE_CACHE_SIZE with zeros. */ 848 * PAGE_CACHE_SIZE with zeros. */
832 size_t num_zeros = (PAGE_CACHE_SIZE 849 size_t num_zeros = (PAGE_CACHE_SIZE
833 - (new_length & ~PAGE_CACHE_MASK)); 850 - (ia->ia_size & ~PAGE_CACHE_MASK));
834 851
835 if (!(crypt_stat->flags & ECRYPTFS_ENCRYPTED)) { 852 if (!(crypt_stat->flags & ECRYPTFS_ENCRYPTED)) {
836 rc = vmtruncate(inode, new_length); 853 rc = vmtruncate(inode, ia->ia_size);
837 if (rc) 854 if (rc)
838 goto out_free; 855 goto out_free;
839 rc = vmtruncate(lower_dentry->d_inode, new_length); 856 lower_ia->ia_size = ia->ia_size;
857 lower_ia->ia_valid |= ATTR_SIZE;
840 goto out_free; 858 goto out_free;
841 } 859 }
842 if (num_zeros) { 860 if (num_zeros) {
@@ -848,7 +866,7 @@ int ecryptfs_truncate(struct dentry *dentry, loff_t new_length)
848 goto out_free; 866 goto out_free;
849 } 867 }
850 rc = ecryptfs_write(&fake_ecryptfs_file, zeros_virt, 868 rc = ecryptfs_write(&fake_ecryptfs_file, zeros_virt,
851 new_length, num_zeros); 869 ia->ia_size, num_zeros);
852 kfree(zeros_virt); 870 kfree(zeros_virt);
853 if (rc) { 871 if (rc) {
854 printk(KERN_ERR "Error attempting to zero out " 872 printk(KERN_ERR "Error attempting to zero out "
@@ -857,7 +875,7 @@ int ecryptfs_truncate(struct dentry *dentry, loff_t new_length)
857 goto out_free; 875 goto out_free;
858 } 876 }
859 } 877 }
860 vmtruncate(inode, new_length); 878 vmtruncate(inode, ia->ia_size);
861 rc = ecryptfs_write_inode_size_to_metadata(inode); 879 rc = ecryptfs_write_inode_size_to_metadata(inode);
862 if (rc) { 880 if (rc) {
863 printk(KERN_ERR "Problem with " 881 printk(KERN_ERR "Problem with "
@@ -870,10 +888,12 @@ int ecryptfs_truncate(struct dentry *dentry, loff_t new_length)
870 lower_size_before_truncate = 888 lower_size_before_truncate =
871 upper_size_to_lower_size(crypt_stat, i_size); 889 upper_size_to_lower_size(crypt_stat, i_size);
872 lower_size_after_truncate = 890 lower_size_after_truncate =
873 upper_size_to_lower_size(crypt_stat, new_length); 891 upper_size_to_lower_size(crypt_stat, ia->ia_size);
874 if (lower_size_after_truncate < lower_size_before_truncate) 892 if (lower_size_after_truncate < lower_size_before_truncate) {
875 vmtruncate(lower_dentry->d_inode, 893 lower_ia->ia_size = lower_size_after_truncate;
876 lower_size_after_truncate); 894 lower_ia->ia_valid |= ATTR_SIZE;
895 } else
896 lower_ia->ia_valid &= ~ATTR_SIZE;
877 } 897 }
878out_free: 898out_free:
879 if (ecryptfs_file_to_private(&fake_ecryptfs_file)) 899 if (ecryptfs_file_to_private(&fake_ecryptfs_file))
@@ -883,6 +903,33 @@ out:
883 return rc; 903 return rc;
884} 904}
885 905
906/**
907 * ecryptfs_truncate
908 * @dentry: The ecryptfs layer dentry
909 * @new_length: The length to expand the file to
910 *
911 * Simple function that handles the truncation of an eCryptfs inode and
912 * its corresponding lower inode.
913 *
914 * Returns zero on success; non-zero otherwise
915 */
916int ecryptfs_truncate(struct dentry *dentry, loff_t new_length)
917{
918 struct iattr ia = { .ia_valid = ATTR_SIZE, .ia_size = new_length };
919 struct iattr lower_ia = { .ia_valid = 0 };
920 int rc;
921
922 rc = truncate_upper(dentry, &ia, &lower_ia);
923 if (!rc && lower_ia.ia_valid & ATTR_SIZE) {
924 struct dentry *lower_dentry = ecryptfs_dentry_to_lower(dentry);
925
926 mutex_lock(&lower_dentry->d_inode->i_mutex);
927 rc = notify_change(lower_dentry, &lower_ia);
928 mutex_unlock(&lower_dentry->d_inode->i_mutex);
929 }
930 return rc;
931}
932
886static int 933static int
887ecryptfs_permission(struct inode *inode, int mask) 934ecryptfs_permission(struct inode *inode, int mask)
888{ 935{
@@ -905,6 +952,7 @@ static int ecryptfs_setattr(struct dentry *dentry, struct iattr *ia)
905{ 952{
906 int rc = 0; 953 int rc = 0;
907 struct dentry *lower_dentry; 954 struct dentry *lower_dentry;
955 struct iattr lower_ia;
908 struct inode *inode; 956 struct inode *inode;
909 struct inode *lower_inode; 957 struct inode *lower_inode;
910 struct ecryptfs_crypt_stat *crypt_stat; 958 struct ecryptfs_crypt_stat *crypt_stat;
@@ -943,15 +991,11 @@ static int ecryptfs_setattr(struct dentry *dentry, struct iattr *ia)
943 } 991 }
944 } 992 }
945 mutex_unlock(&crypt_stat->cs_mutex); 993 mutex_unlock(&crypt_stat->cs_mutex);
994 memcpy(&lower_ia, ia, sizeof(lower_ia));
995 if (ia->ia_valid & ATTR_FILE)
996 lower_ia.ia_file = ecryptfs_file_to_lower(ia->ia_file);
946 if (ia->ia_valid & ATTR_SIZE) { 997 if (ia->ia_valid & ATTR_SIZE) {
947 ecryptfs_printk(KERN_DEBUG, 998 rc = truncate_upper(dentry, ia, &lower_ia);
948 "ia->ia_valid = [0x%x] ATTR_SIZE" " = [0x%x]\n",
949 ia->ia_valid, ATTR_SIZE);
950 rc = ecryptfs_truncate(dentry, ia->ia_size);
951 /* ecryptfs_truncate handles resizing of the lower file */
952 ia->ia_valid &= ~ATTR_SIZE;
953 ecryptfs_printk(KERN_DEBUG, "ia->ia_valid = [%x]\n",
954 ia->ia_valid);
955 if (rc < 0) 999 if (rc < 0)
956 goto out; 1000 goto out;
957 } 1001 }
@@ -960,17 +1004,32 @@ static int ecryptfs_setattr(struct dentry *dentry, struct iattr *ia)
960 * mode change is for clearing setuid/setgid bits. Allow lower fs 1004 * mode change is for clearing setuid/setgid bits. Allow lower fs
961 * to interpret this in its own way. 1005 * to interpret this in its own way.
962 */ 1006 */
963 if (ia->ia_valid & (ATTR_KILL_SUID | ATTR_KILL_SGID)) 1007 if (lower_ia.ia_valid & (ATTR_KILL_SUID | ATTR_KILL_SGID))
964 ia->ia_valid &= ~ATTR_MODE; 1008 lower_ia.ia_valid &= ~ATTR_MODE;
965 1009
966 mutex_lock(&lower_dentry->d_inode->i_mutex); 1010 mutex_lock(&lower_dentry->d_inode->i_mutex);
967 rc = notify_change(lower_dentry, ia); 1011 rc = notify_change(lower_dentry, &lower_ia);
968 mutex_unlock(&lower_dentry->d_inode->i_mutex); 1012 mutex_unlock(&lower_dentry->d_inode->i_mutex);
969out: 1013out:
970 fsstack_copy_attr_all(inode, lower_inode); 1014 fsstack_copy_attr_all(inode, lower_inode);
971 return rc; 1015 return rc;
972} 1016}
973 1017
1018int ecryptfs_getattr(struct vfsmount *mnt, struct dentry *dentry,
1019 struct kstat *stat)
1020{
1021 struct kstat lower_stat;
1022 int rc;
1023
1024 rc = vfs_getattr(ecryptfs_dentry_to_lower_mnt(dentry),
1025 ecryptfs_dentry_to_lower(dentry), &lower_stat);
1026 if (!rc) {
1027 generic_fillattr(dentry->d_inode, stat);
1028 stat->blocks = lower_stat.blocks;
1029 }
1030 return rc;
1031}
1032
974int 1033int
975ecryptfs_setxattr(struct dentry *dentry, const char *name, const void *value, 1034ecryptfs_setxattr(struct dentry *dentry, const char *name, const void *value,
976 size_t size, int flags) 1035 size_t size, int flags)
@@ -1100,6 +1159,7 @@ const struct inode_operations ecryptfs_dir_iops = {
1100const struct inode_operations ecryptfs_main_iops = { 1159const struct inode_operations ecryptfs_main_iops = {
1101 .permission = ecryptfs_permission, 1160 .permission = ecryptfs_permission,
1102 .setattr = ecryptfs_setattr, 1161 .setattr = ecryptfs_setattr,
1162 .getattr = ecryptfs_getattr,
1103 .setxattr = ecryptfs_setxattr, 1163 .setxattr = ecryptfs_setxattr,
1104 .getxattr = ecryptfs_getxattr, 1164 .getxattr = ecryptfs_getxattr,
1105 .listxattr = ecryptfs_listxattr, 1165 .listxattr = ecryptfs_listxattr,
diff --git a/fs/ecryptfs/main.c b/fs/ecryptfs/main.c
index 567bc4b9f70a..ea2f92101dfe 100644
--- a/fs/ecryptfs/main.c
+++ b/fs/ecryptfs/main.c
@@ -585,8 +585,8 @@ out:
585 * with as much information as it can before needing 585 * with as much information as it can before needing
586 * the lower filesystem. 586 * the lower filesystem.
587 * ecryptfs_read_super(): this accesses the lower filesystem and uses 587 * ecryptfs_read_super(): this accesses the lower filesystem and uses
588 * ecryptfs_interpolate to perform most of the linking 588 * ecryptfs_interpose to perform most of the linking
589 * ecryptfs_interpolate(): links the lower filesystem into ecryptfs 589 * ecryptfs_interpose(): links the lower filesystem into ecryptfs (inode.c)
590 */ 590 */
591static int ecryptfs_get_sb(struct file_system_type *fs_type, int flags, 591static int ecryptfs_get_sb(struct file_system_type *fs_type, int flags,
592 const char *dev_name, void *raw_data, 592 const char *dev_name, void *raw_data,
diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c
index 698a8636d39c..2afbcebeda71 100644
--- a/fs/exofs/inode.c
+++ b/fs/exofs/inode.c
@@ -738,13 +738,28 @@ static int exofs_write_begin_export(struct file *file,
738 fsdata); 738 fsdata);
739} 739}
740 740
741static int exofs_write_end(struct file *file, struct address_space *mapping,
742 loff_t pos, unsigned len, unsigned copied,
743 struct page *page, void *fsdata)
744{
745 struct inode *inode = mapping->host;
746 /* According to comment in simple_write_end i_mutex is held */
747 loff_t i_size = inode->i_size;
748 int ret;
749
750 ret = simple_write_end(file, mapping,pos, len, copied, page, fsdata);
751 if (i_size != inode->i_size)
752 mark_inode_dirty(inode);
753 return ret;
754}
755
741const struct address_space_operations exofs_aops = { 756const struct address_space_operations exofs_aops = {
742 .readpage = exofs_readpage, 757 .readpage = exofs_readpage,
743 .readpages = exofs_readpages, 758 .readpages = exofs_readpages,
744 .writepage = exofs_writepage, 759 .writepage = exofs_writepage,
745 .writepages = exofs_writepages, 760 .writepages = exofs_writepages,
746 .write_begin = exofs_write_begin_export, 761 .write_begin = exofs_write_begin_export,
747 .write_end = simple_write_end, 762 .write_end = exofs_write_end,
748}; 763};
749 764
750/****************************************************************************** 765/******************************************************************************
diff --git a/fs/exofs/pnfs.h b/fs/exofs/pnfs.h
index 423033addd1f..c52e9888b8ab 100644
--- a/fs/exofs/pnfs.h
+++ b/fs/exofs/pnfs.h
@@ -15,13 +15,7 @@
15#ifndef __EXOFS_PNFS_H__ 15#ifndef __EXOFS_PNFS_H__
16#define __EXOFS_PNFS_H__ 16#define __EXOFS_PNFS_H__
17 17
18#if defined(CONFIG_PNFS) 18#if ! defined(__PNFS_OSD_XDR_H__)
19
20
21/* FIXME: move this file to: linux/exportfs/pnfs_osd_xdr.h */
22#include "../nfs/objlayout/pnfs_osd_xdr.h"
23
24#else /* defined(CONFIG_PNFS) */
25 19
26enum pnfs_iomode { 20enum pnfs_iomode {
27 IOMODE_READ = 1, 21 IOMODE_READ = 1,
@@ -46,6 +40,6 @@ struct pnfs_osd_data_map {
46 u32 odm_raid_algorithm; 40 u32 odm_raid_algorithm;
47}; 41};
48 42
49#endif /* else defined(CONFIG_PNFS) */ 43#endif /* ! defined(__PNFS_OSD_XDR_H__) */
50 44
51#endif /* __EXOFS_PNFS_H__ */ 45#endif /* __EXOFS_PNFS_H__ */
diff --git a/fs/fcntl.c b/fs/fcntl.c
index 2cf93ec40a67..97e01dc0d95f 100644
--- a/fs/fcntl.c
+++ b/fs/fcntl.c
@@ -618,60 +618,90 @@ static DEFINE_RWLOCK(fasync_lock);
618static struct kmem_cache *fasync_cache __read_mostly; 618static struct kmem_cache *fasync_cache __read_mostly;
619 619
620/* 620/*
621 * fasync_helper() is used by almost all character device drivers 621 * Remove a fasync entry. If successfully removed, return
622 * to set up the fasync queue. It returns negative on error, 0 if it did 622 * positive and clear the FASYNC flag. If no entry exists,
623 * no changes and positive if it added/deleted the entry. 623 * do nothing and return 0.
624 *
625 * NOTE! It is very important that the FASYNC flag always
626 * match the state "is the filp on a fasync list".
627 *
628 * We always take the 'filp->f_lock', in since fasync_lock
629 * needs to be irq-safe.
624 */ 630 */
625int fasync_helper(int fd, struct file * filp, int on, struct fasync_struct **fapp) 631static int fasync_remove_entry(struct file *filp, struct fasync_struct **fapp)
626{ 632{
627 struct fasync_struct *fa, **fp; 633 struct fasync_struct *fa, **fp;
628 struct fasync_struct *new = NULL;
629 int result = 0; 634 int result = 0;
630 635
631 if (on) { 636 spin_lock(&filp->f_lock);
632 new = kmem_cache_alloc(fasync_cache, GFP_KERNEL); 637 write_lock_irq(&fasync_lock);
633 if (!new) 638 for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
634 return -ENOMEM; 639 if (fa->fa_file != filp)
640 continue;
641 *fp = fa->fa_next;
642 kmem_cache_free(fasync_cache, fa);
643 filp->f_flags &= ~FASYNC;
644 result = 1;
645 break;
635 } 646 }
647 write_unlock_irq(&fasync_lock);
648 spin_unlock(&filp->f_lock);
649 return result;
650}
651
652/*
653 * Add a fasync entry. Return negative on error, positive if
654 * added, and zero if did nothing but change an existing one.
655 *
656 * NOTE! It is very important that the FASYNC flag always
657 * match the state "is the filp on a fasync list".
658 */
659static int fasync_add_entry(int fd, struct file *filp, struct fasync_struct **fapp)
660{
661 struct fasync_struct *new, *fa, **fp;
662 int result = 0;
663
664 new = kmem_cache_alloc(fasync_cache, GFP_KERNEL);
665 if (!new)
666 return -ENOMEM;
636 667
637 /*
638 * We need to take f_lock first since it's not an IRQ-safe
639 * lock.
640 */
641 spin_lock(&filp->f_lock); 668 spin_lock(&filp->f_lock);
642 write_lock_irq(&fasync_lock); 669 write_lock_irq(&fasync_lock);
643 for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) { 670 for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
644 if (fa->fa_file == filp) { 671 if (fa->fa_file != filp)
645 if(on) { 672 continue;
646 fa->fa_fd = fd; 673 fa->fa_fd = fd;
647 kmem_cache_free(fasync_cache, new); 674 kmem_cache_free(fasync_cache, new);
648 } else { 675 goto out;
649 *fp = fa->fa_next;
650 kmem_cache_free(fasync_cache, fa);
651 result = 1;
652 }
653 goto out;
654 }
655 } 676 }
656 677
657 if (on) { 678 new->magic = FASYNC_MAGIC;
658 new->magic = FASYNC_MAGIC; 679 new->fa_file = filp;
659 new->fa_file = filp; 680 new->fa_fd = fd;
660 new->fa_fd = fd; 681 new->fa_next = *fapp;
661 new->fa_next = *fapp; 682 *fapp = new;
662 *fapp = new; 683 result = 1;
663 result = 1; 684 filp->f_flags |= FASYNC;
664 } 685
665out: 686out:
666 if (on)
667 filp->f_flags |= FASYNC;
668 else
669 filp->f_flags &= ~FASYNC;
670 write_unlock_irq(&fasync_lock); 687 write_unlock_irq(&fasync_lock);
671 spin_unlock(&filp->f_lock); 688 spin_unlock(&filp->f_lock);
672 return result; 689 return result;
673} 690}
674 691
692/*
693 * fasync_helper() is used by almost all character device drivers
694 * to set up the fasync queue, and for regular files by the file
695 * lease code. It returns negative on error, 0 if it did no changes
696 * and positive if it added/deleted the entry.
697 */
698int fasync_helper(int fd, struct file * filp, int on, struct fasync_struct **fapp)
699{
700 if (!on)
701 return fasync_remove_entry(filp, fapp);
702 return fasync_add_entry(fd, filp, fapp);
703}
704
675EXPORT_SYMBOL(fasync_helper); 705EXPORT_SYMBOL(fasync_helper);
676 706
677void __kill_fasync(struct fasync_struct *fa, int sig, int band) 707void __kill_fasync(struct fasync_struct *fa, int sig, int band)
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index 4eb308aa3234..a6abbae8a278 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -569,6 +569,40 @@ static int gfs2_fsync(struct file *file, struct dentry *dentry, int datasync)
569 return ret; 569 return ret;
570} 570}
571 571
572/**
573 * gfs2_file_aio_write - Perform a write to a file
574 * @iocb: The io context
575 * @iov: The data to write
576 * @nr_segs: Number of @iov segments
577 * @pos: The file position
578 *
579 * We have to do a lock/unlock here to refresh the inode size for
580 * O_APPEND writes, otherwise we can land up writing at the wrong
581 * offset. There is still a race, but provided the app is using its
582 * own file locking, this will make O_APPEND work as expected.
583 *
584 */
585
586static ssize_t gfs2_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
587 unsigned long nr_segs, loff_t pos)
588{
589 struct file *file = iocb->ki_filp;
590
591 if (file->f_flags & O_APPEND) {
592 struct dentry *dentry = file->f_dentry;
593 struct gfs2_inode *ip = GFS2_I(dentry->d_inode);
594 struct gfs2_holder gh;
595 int ret;
596
597 ret = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, 0, &gh);
598 if (ret)
599 return ret;
600 gfs2_glock_dq_uninit(&gh);
601 }
602
603 return generic_file_aio_write(iocb, iov, nr_segs, pos);
604}
605
572#ifdef CONFIG_GFS2_FS_LOCKING_DLM 606#ifdef CONFIG_GFS2_FS_LOCKING_DLM
573 607
574/** 608/**
@@ -711,7 +745,7 @@ const struct file_operations gfs2_file_fops = {
711 .read = do_sync_read, 745 .read = do_sync_read,
712 .aio_read = generic_file_aio_read, 746 .aio_read = generic_file_aio_read,
713 .write = do_sync_write, 747 .write = do_sync_write,
714 .aio_write = generic_file_aio_write, 748 .aio_write = gfs2_file_aio_write,
715 .unlocked_ioctl = gfs2_ioctl, 749 .unlocked_ioctl = gfs2_ioctl,
716 .mmap = gfs2_mmap, 750 .mmap = gfs2_mmap,
717 .open = gfs2_open, 751 .open = gfs2_open,
@@ -741,7 +775,7 @@ const struct file_operations gfs2_file_fops_nolock = {
741 .read = do_sync_read, 775 .read = do_sync_read,
742 .aio_read = generic_file_aio_read, 776 .aio_read = generic_file_aio_read,
743 .write = do_sync_write, 777 .write = do_sync_write,
744 .aio_write = generic_file_aio_write, 778 .aio_write = gfs2_file_aio_write,
745 .unlocked_ioctl = gfs2_ioctl, 779 .unlocked_ioctl = gfs2_ioctl,
746 .mmap = gfs2_mmap, 780 .mmap = gfs2_mmap,
747 .open = gfs2_open, 781 .open = gfs2_open,
diff --git a/fs/gfs2/meta_io.c b/fs/gfs2/meta_io.c
index cb8d7a93d5ec..6f68a5f18eb8 100644
--- a/fs/gfs2/meta_io.c
+++ b/fs/gfs2/meta_io.c
@@ -121,7 +121,7 @@ struct inode *gfs2_aspace_get(struct gfs2_sbd *sdp)
121 if (aspace) { 121 if (aspace) {
122 mapping_set_gfp_mask(aspace->i_mapping, GFP_NOFS); 122 mapping_set_gfp_mask(aspace->i_mapping, GFP_NOFS);
123 aspace->i_mapping->a_ops = &aspace_aops; 123 aspace->i_mapping->a_ops = &aspace_aops;
124 aspace->i_size = ~0ULL; 124 aspace->i_size = MAX_LFS_FILESIZE;
125 ip = GFS2_I(aspace); 125 ip = GFS2_I(aspace);
126 clear_bit(GIF_USER, &ip->i_flags); 126 clear_bit(GIF_USER, &ip->i_flags);
127 insert_inode_hash(aspace); 127 insert_inode_hash(aspace);
diff --git a/fs/gfs2/ops_inode.c b/fs/gfs2/ops_inode.c
index 247436c10deb..78f73ca1ef3e 100644
--- a/fs/gfs2/ops_inode.c
+++ b/fs/gfs2/ops_inode.c
@@ -748,7 +748,7 @@ static int gfs2_rename(struct inode *odir, struct dentry *odentry,
748 struct gfs2_rgrpd *nrgd; 748 struct gfs2_rgrpd *nrgd;
749 unsigned int num_gh; 749 unsigned int num_gh;
750 int dir_rename = 0; 750 int dir_rename = 0;
751 int alloc_required; 751 int alloc_required = 0;
752 unsigned int x; 752 unsigned int x;
753 int error; 753 int error;
754 754
@@ -867,7 +867,9 @@ static int gfs2_rename(struct inode *odir, struct dentry *odentry,
867 goto out_gunlock; 867 goto out_gunlock;
868 } 868 }
869 869
870 alloc_required = error = gfs2_diradd_alloc_required(ndir, &ndentry->d_name); 870 if (nip == NULL)
871 alloc_required = gfs2_diradd_alloc_required(ndir, &ndentry->d_name);
872 error = alloc_required;
871 if (error < 0) 873 if (error < 0)
872 goto out_gunlock; 874 goto out_gunlock;
873 error = 0; 875 error = 0;
diff --git a/fs/gfs2/xattr.c b/fs/gfs2/xattr.c
index 8a04108e0c22..c2ebdf2c01d4 100644
--- a/fs/gfs2/xattr.c
+++ b/fs/gfs2/xattr.c
@@ -1296,6 +1296,7 @@ fail:
1296 1296
1297int gfs2_xattr_acl_chmod(struct gfs2_inode *ip, struct iattr *attr, char *data) 1297int gfs2_xattr_acl_chmod(struct gfs2_inode *ip, struct iattr *attr, char *data)
1298{ 1298{
1299 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
1299 struct gfs2_ea_location el; 1300 struct gfs2_ea_location el;
1300 struct buffer_head *dibh; 1301 struct buffer_head *dibh;
1301 int error; 1302 int error;
@@ -1305,16 +1306,17 @@ int gfs2_xattr_acl_chmod(struct gfs2_inode *ip, struct iattr *attr, char *data)
1305 return error; 1306 return error;
1306 1307
1307 if (GFS2_EA_IS_STUFFED(el.el_ea)) { 1308 if (GFS2_EA_IS_STUFFED(el.el_ea)) {
1308 error = gfs2_trans_begin(GFS2_SB(&ip->i_inode), RES_DINODE + RES_EATTR, 0); 1309 error = gfs2_trans_begin(sdp, RES_DINODE + RES_EATTR, 0);
1309 if (error) 1310 if (error == 0) {
1310 return error; 1311 gfs2_trans_add_bh(ip->i_gl, el.el_bh, 1);
1311 1312 memcpy(GFS2_EA2DATA(el.el_ea), data,
1312 gfs2_trans_add_bh(ip->i_gl, el.el_bh, 1); 1313 GFS2_EA_DATA_LEN(el.el_ea));
1313 memcpy(GFS2_EA2DATA(el.el_ea), data, 1314 }
1314 GFS2_EA_DATA_LEN(el.el_ea)); 1315 } else {
1315 } else
1316 error = ea_acl_chmod_unstuffed(ip, el.el_ea, data); 1316 error = ea_acl_chmod_unstuffed(ip, el.el_ea, data);
1317 }
1317 1318
1319 brelse(el.el_bh);
1318 if (error) 1320 if (error)
1319 return error; 1321 return error;
1320 1322
@@ -1327,8 +1329,7 @@ int gfs2_xattr_acl_chmod(struct gfs2_inode *ip, struct iattr *attr, char *data)
1327 brelse(dibh); 1329 brelse(dibh);
1328 } 1330 }
1329 1331
1330 gfs2_trans_end(GFS2_SB(&ip->i_inode)); 1332 gfs2_trans_end(sdp);
1331
1332 return error; 1333 return error;
1333} 1334}
1334 1335
diff --git a/fs/hppfs/hppfs.c b/fs/hppfs/hppfs.c
index a5089a6dd67a..7239efc690d8 100644
--- a/fs/hppfs/hppfs.c
+++ b/fs/hppfs/hppfs.c
@@ -646,22 +646,27 @@ static const struct super_operations hppfs_sbops = {
646static int hppfs_readlink(struct dentry *dentry, char __user *buffer, 646static int hppfs_readlink(struct dentry *dentry, char __user *buffer,
647 int buflen) 647 int buflen)
648{ 648{
649 struct dentry *proc_dentry; 649 struct dentry *proc_dentry = HPPFS_I(dentry->d_inode)->proc_dentry;
650
651 proc_dentry = HPPFS_I(dentry->d_inode)->proc_dentry;
652 return proc_dentry->d_inode->i_op->readlink(proc_dentry, buffer, 650 return proc_dentry->d_inode->i_op->readlink(proc_dentry, buffer,
653 buflen); 651 buflen);
654} 652}
655 653
656static void *hppfs_follow_link(struct dentry *dentry, struct nameidata *nd) 654static void *hppfs_follow_link(struct dentry *dentry, struct nameidata *nd)
657{ 655{
658 struct dentry *proc_dentry; 656 struct dentry *proc_dentry = HPPFS_I(dentry->d_inode)->proc_dentry;
659
660 proc_dentry = HPPFS_I(dentry->d_inode)->proc_dentry;
661 657
662 return proc_dentry->d_inode->i_op->follow_link(proc_dentry, nd); 658 return proc_dentry->d_inode->i_op->follow_link(proc_dentry, nd);
663} 659}
664 660
661static void hppfs_put_link(struct dentry *dentry, struct nameidata *nd,
662 void *cookie)
663{
664 struct dentry *proc_dentry = HPPFS_I(dentry->d_inode)->proc_dentry;
665
666 if (proc_dentry->d_inode->i_op->put_link)
667 proc_dentry->d_inode->i_op->put_link(proc_dentry, nd, cookie);
668}
669
665static const struct inode_operations hppfs_dir_iops = { 670static const struct inode_operations hppfs_dir_iops = {
666 .lookup = hppfs_lookup, 671 .lookup = hppfs_lookup,
667}; 672};
@@ -669,6 +674,7 @@ static const struct inode_operations hppfs_dir_iops = {
669static const struct inode_operations hppfs_link_iops = { 674static const struct inode_operations hppfs_link_iops = {
670 .readlink = hppfs_readlink, 675 .readlink = hppfs_readlink,
671 .follow_link = hppfs_follow_link, 676 .follow_link = hppfs_follow_link,
677 .put_link = hppfs_put_link,
672}; 678};
673 679
674static struct inode *get_inode(struct super_block *sb, struct dentry *dentry) 680static struct inode *get_inode(struct super_block *sb, struct dentry *dentry)
diff --git a/fs/namei.c b/fs/namei.c
index b55440baf7ab..94a5e60779f9 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -561,6 +561,7 @@ static __always_inline int __do_follow_link(struct path *path, struct nameidata
561 dget(dentry); 561 dget(dentry);
562 } 562 }
563 mntget(path->mnt); 563 mntget(path->mnt);
564 nd->last_type = LAST_BIND;
564 cookie = dentry->d_inode->i_op->follow_link(dentry, nd); 565 cookie = dentry->d_inode->i_op->follow_link(dentry, nd);
565 error = PTR_ERR(cookie); 566 error = PTR_ERR(cookie);
566 if (!IS_ERR(cookie)) { 567 if (!IS_ERR(cookie)) {
@@ -1603,11 +1604,12 @@ struct file *do_filp_open(int dfd, const char *pathname,
1603 struct file *filp; 1604 struct file *filp;
1604 struct nameidata nd; 1605 struct nameidata nd;
1605 int error; 1606 int error;
1606 struct path path, save; 1607 struct path path;
1607 struct dentry *dir; 1608 struct dentry *dir;
1608 int count = 0; 1609 int count = 0;
1609 int will_truncate; 1610 int will_truncate;
1610 int flag = open_to_namei_flags(open_flag); 1611 int flag = open_to_namei_flags(open_flag);
1612 int force_reval = 0;
1611 1613
1612 /* 1614 /*
1613 * O_SYNC is implemented as __O_SYNC|O_DSYNC. As many places only 1615 * O_SYNC is implemented as __O_SYNC|O_DSYNC. As many places only
@@ -1619,7 +1621,7 @@ struct file *do_filp_open(int dfd, const char *pathname,
1619 open_flag |= O_DSYNC; 1621 open_flag |= O_DSYNC;
1620 1622
1621 if (!acc_mode) 1623 if (!acc_mode)
1622 acc_mode = MAY_OPEN | ACC_MODE(flag); 1624 acc_mode = MAY_OPEN | ACC_MODE(open_flag);
1623 1625
1624 /* O_TRUNC implies we need access checks for write permissions */ 1626 /* O_TRUNC implies we need access checks for write permissions */
1625 if (flag & O_TRUNC) 1627 if (flag & O_TRUNC)
@@ -1659,9 +1661,12 @@ struct file *do_filp_open(int dfd, const char *pathname,
1659 /* 1661 /*
1660 * Create - we need to know the parent. 1662 * Create - we need to know the parent.
1661 */ 1663 */
1664reval:
1662 error = path_init(dfd, pathname, LOOKUP_PARENT, &nd); 1665 error = path_init(dfd, pathname, LOOKUP_PARENT, &nd);
1663 if (error) 1666 if (error)
1664 return ERR_PTR(error); 1667 return ERR_PTR(error);
1668 if (force_reval)
1669 nd.flags |= LOOKUP_REVAL;
1665 error = path_walk(pathname, &nd); 1670 error = path_walk(pathname, &nd);
1666 if (error) { 1671 if (error) {
1667 if (nd.root.mnt) 1672 if (nd.root.mnt)
@@ -1853,17 +1858,7 @@ do_link:
1853 error = security_inode_follow_link(path.dentry, &nd); 1858 error = security_inode_follow_link(path.dentry, &nd);
1854 if (error) 1859 if (error)
1855 goto exit_dput; 1860 goto exit_dput;
1856 save = nd.path;
1857 path_get(&save);
1858 error = __do_follow_link(&path, &nd); 1861 error = __do_follow_link(&path, &nd);
1859 if (error == -ESTALE) {
1860 /* nd.path had been dropped */
1861 nd.path = save;
1862 path_get(&nd.path);
1863 nd.flags |= LOOKUP_REVAL;
1864 error = __do_follow_link(&path, &nd);
1865 }
1866 path_put(&save);
1867 path_put(&path); 1862 path_put(&path);
1868 if (error) { 1863 if (error) {
1869 /* Does someone understand code flow here? Or it is only 1864 /* Does someone understand code flow here? Or it is only
@@ -1873,6 +1868,10 @@ do_link:
1873 release_open_intent(&nd); 1868 release_open_intent(&nd);
1874 if (nd.root.mnt) 1869 if (nd.root.mnt)
1875 path_put(&nd.root); 1870 path_put(&nd.root);
1871 if (error == -ESTALE && !force_reval) {
1872 force_reval = 1;
1873 goto reval;
1874 }
1876 return ERR_PTR(error); 1875 return ERR_PTR(error);
1877 } 1876 }
1878 nd.flags &= ~LOOKUP_PARENT; 1877 nd.flags &= ~LOOKUP_PARENT;
diff --git a/fs/namespace.c b/fs/namespace.c
index 7d70d63ceb29..c768f733c8d6 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -965,10 +965,12 @@ EXPORT_SYMBOL(may_umount_tree);
965int may_umount(struct vfsmount *mnt) 965int may_umount(struct vfsmount *mnt)
966{ 966{
967 int ret = 1; 967 int ret = 1;
968 down_read(&namespace_sem);
968 spin_lock(&vfsmount_lock); 969 spin_lock(&vfsmount_lock);
969 if (propagate_mount_busy(mnt, 2)) 970 if (propagate_mount_busy(mnt, 2))
970 ret = 0; 971 ret = 0;
971 spin_unlock(&vfsmount_lock); 972 spin_unlock(&vfsmount_lock);
973 up_read(&namespace_sem);
972 return ret; 974 return ret;
973} 975}
974 976
@@ -1352,12 +1354,12 @@ static int attach_recursive_mnt(struct vfsmount *source_mnt,
1352 if (err) 1354 if (err)
1353 goto out_cleanup_ids; 1355 goto out_cleanup_ids;
1354 1356
1357 spin_lock(&vfsmount_lock);
1358
1355 if (IS_MNT_SHARED(dest_mnt)) { 1359 if (IS_MNT_SHARED(dest_mnt)) {
1356 for (p = source_mnt; p; p = next_mnt(p, source_mnt)) 1360 for (p = source_mnt; p; p = next_mnt(p, source_mnt))
1357 set_mnt_shared(p); 1361 set_mnt_shared(p);
1358 } 1362 }
1359
1360 spin_lock(&vfsmount_lock);
1361 if (parent_path) { 1363 if (parent_path) {
1362 detach_mnt(source_mnt, parent_path); 1364 detach_mnt(source_mnt, parent_path);
1363 attach_mnt(source_mnt, path); 1365 attach_mnt(source_mnt, path);
@@ -1534,8 +1536,12 @@ static int do_remount(struct path *path, int flags, int mnt_flags,
1534 err = change_mount_flags(path->mnt, flags); 1536 err = change_mount_flags(path->mnt, flags);
1535 else 1537 else
1536 err = do_remount_sb(sb, flags, data, 0); 1538 err = do_remount_sb(sb, flags, data, 0);
1537 if (!err) 1539 if (!err) {
1540 spin_lock(&vfsmount_lock);
1541 mnt_flags |= path->mnt->mnt_flags & MNT_PNODE_MASK;
1538 path->mnt->mnt_flags = mnt_flags; 1542 path->mnt->mnt_flags = mnt_flags;
1543 spin_unlock(&vfsmount_lock);
1544 }
1539 up_write(&sb->s_umount); 1545 up_write(&sb->s_umount);
1540 if (!err) { 1546 if (!err) {
1541 security_sb_post_remount(path->mnt, flags, data); 1547 security_sb_post_remount(path->mnt, flags, data);
@@ -1665,6 +1671,8 @@ int do_add_mount(struct vfsmount *newmnt, struct path *path,
1665{ 1671{
1666 int err; 1672 int err;
1667 1673
1674 mnt_flags &= ~(MNT_SHARED | MNT_WRITE_HOLD);
1675
1668 down_write(&namespace_sem); 1676 down_write(&namespace_sem);
1669 /* Something was mounted here while we slept */ 1677 /* Something was mounted here while we slept */
1670 while (d_mountpoint(path->dentry) && 1678 while (d_mountpoint(path->dentry) &&
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 2c5ace4f00a7..3c7f03b669fb 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1615,6 +1615,7 @@ static int nfs_rename(struct inode *old_dir, struct dentry *old_dentry,
1615 goto out; 1615 goto out;
1616 1616
1617 new_dentry = dentry; 1617 new_dentry = dentry;
1618 rehash = NULL;
1618 new_inode = NULL; 1619 new_inode = NULL;
1619 } 1620 }
1620 } 1621 }
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 7c2e337d05af..c194793b642b 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -780,12 +780,9 @@ static inline int nfsd_dosync(struct file *filp, struct dentry *dp,
780 int (*fsync) (struct file *, struct dentry *, int); 780 int (*fsync) (struct file *, struct dentry *, int);
781 int err; 781 int err;
782 782
783 err = filemap_fdatawrite(inode->i_mapping); 783 err = filemap_write_and_wait(inode->i_mapping);
784 if (err == 0 && fop && (fsync = fop->fsync)) 784 if (err == 0 && fop && (fsync = fop->fsync))
785 err = fsync(filp, dp, 0); 785 err = fsync(filp, dp, 0);
786 if (err == 0)
787 err = filemap_fdatawait(inode->i_mapping);
788
789 return err; 786 return err;
790} 787}
791 788
diff --git a/fs/notify/inotify/inotify_fsnotify.c b/fs/notify/inotify/inotify_fsnotify.c
index c9ee67b442e1..1afb0a10229f 100644
--- a/fs/notify/inotify/inotify_fsnotify.c
+++ b/fs/notify/inotify/inotify_fsnotify.c
@@ -121,7 +121,7 @@ static int idr_callback(int id, void *p, void *data)
121 if (warned) 121 if (warned)
122 return 0; 122 return 0;
123 123
124 warned = false; 124 warned = true;
125 entry = p; 125 entry = p;
126 ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry); 126 ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry);
127 127
diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c
index 8271cf05c957..a94e8bd8eb1f 100644
--- a/fs/notify/inotify/inotify_user.c
+++ b/fs/notify/inotify/inotify_user.c
@@ -552,7 +552,7 @@ retry:
552 552
553 spin_lock(&group->inotify_data.idr_lock); 553 spin_lock(&group->inotify_data.idr_lock);
554 ret = idr_get_new_above(&group->inotify_data.idr, &tmp_ientry->fsn_entry, 554 ret = idr_get_new_above(&group->inotify_data.idr, &tmp_ientry->fsn_entry,
555 group->inotify_data.last_wd, 555 group->inotify_data.last_wd+1,
556 &tmp_ientry->wd); 556 &tmp_ientry->wd);
557 spin_unlock(&group->inotify_data.idr_lock); 557 spin_unlock(&group->inotify_data.idr_lock);
558 if (ret) { 558 if (ret) {
@@ -632,7 +632,7 @@ static struct fsnotify_group *inotify_new_group(struct user_struct *user, unsign
632 632
633 spin_lock_init(&group->inotify_data.idr_lock); 633 spin_lock_init(&group->inotify_data.idr_lock);
634 idr_init(&group->inotify_data.idr); 634 idr_init(&group->inotify_data.idr);
635 group->inotify_data.last_wd = 1; 635 group->inotify_data.last_wd = 0;
636 group->inotify_data.user = user; 636 group->inotify_data.user = user;
637 group->inotify_data.fa = NULL; 637 group->inotify_data.fa = NULL;
638 638
diff --git a/fs/proc/array.c b/fs/proc/array.c
index f560325c444f..13b5d0708175 100644
--- a/fs/proc/array.c
+++ b/fs/proc/array.c
@@ -327,94 +327,6 @@ static inline void task_context_switch_counts(struct seq_file *m,
327 p->nivcsw); 327 p->nivcsw);
328} 328}
329 329
330#ifdef CONFIG_MMU
331
332struct stack_stats {
333 struct vm_area_struct *vma;
334 unsigned long startpage;
335 unsigned long usage;
336};
337
338static int stack_usage_pte_range(pmd_t *pmd, unsigned long addr,
339 unsigned long end, struct mm_walk *walk)
340{
341 struct stack_stats *ss = walk->private;
342 struct vm_area_struct *vma = ss->vma;
343 pte_t *pte, ptent;
344 spinlock_t *ptl;
345 int ret = 0;
346
347 pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
348 for (; addr != end; pte++, addr += PAGE_SIZE) {
349 ptent = *pte;
350
351#ifdef CONFIG_STACK_GROWSUP
352 if (pte_present(ptent) || is_swap_pte(ptent))
353 ss->usage = addr - ss->startpage + PAGE_SIZE;
354#else
355 if (pte_present(ptent) || is_swap_pte(ptent)) {
356 ss->usage = ss->startpage - addr + PAGE_SIZE;
357 pte++;
358 ret = 1;
359 break;
360 }
361#endif
362 }
363 pte_unmap_unlock(pte - 1, ptl);
364 cond_resched();
365 return ret;
366}
367
368static inline unsigned long get_stack_usage_in_bytes(struct vm_area_struct *vma,
369 struct task_struct *task)
370{
371 struct stack_stats ss;
372 struct mm_walk stack_walk = {
373 .pmd_entry = stack_usage_pte_range,
374 .mm = vma->vm_mm,
375 .private = &ss,
376 };
377
378 if (!vma->vm_mm || is_vm_hugetlb_page(vma))
379 return 0;
380
381 ss.vma = vma;
382 ss.startpage = task->stack_start & PAGE_MASK;
383 ss.usage = 0;
384
385#ifdef CONFIG_STACK_GROWSUP
386 walk_page_range(KSTK_ESP(task) & PAGE_MASK, vma->vm_end,
387 &stack_walk);
388#else
389 walk_page_range(vma->vm_start, (KSTK_ESP(task) & PAGE_MASK) + PAGE_SIZE,
390 &stack_walk);
391#endif
392 return ss.usage;
393}
394
395static inline void task_show_stack_usage(struct seq_file *m,
396 struct task_struct *task)
397{
398 struct vm_area_struct *vma;
399 struct mm_struct *mm = get_task_mm(task);
400
401 if (mm) {
402 down_read(&mm->mmap_sem);
403 vma = find_vma(mm, task->stack_start);
404 if (vma)
405 seq_printf(m, "Stack usage:\t%lu kB\n",
406 get_stack_usage_in_bytes(vma, task) >> 10);
407
408 up_read(&mm->mmap_sem);
409 mmput(mm);
410 }
411}
412#else
413static void task_show_stack_usage(struct seq_file *m, struct task_struct *task)
414{
415}
416#endif /* CONFIG_MMU */
417
418static void task_cpus_allowed(struct seq_file *m, struct task_struct *task) 330static void task_cpus_allowed(struct seq_file *m, struct task_struct *task)
419{ 331{
420 seq_printf(m, "Cpus_allowed:\t"); 332 seq_printf(m, "Cpus_allowed:\t");
@@ -445,7 +357,6 @@ int proc_pid_status(struct seq_file *m, struct pid_namespace *ns,
445 task_show_regs(m, task); 357 task_show_regs(m, task);
446#endif 358#endif
447 task_context_switch_counts(m, task); 359 task_context_switch_counts(m, task);
448 task_show_stack_usage(m, task);
449 return 0; 360 return 0;
450} 361}
451 362
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 18d5cc62d8ed..e42bbd843ed1 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -1419,7 +1419,6 @@ static void *proc_pid_follow_link(struct dentry *dentry, struct nameidata *nd)
1419 goto out; 1419 goto out;
1420 1420
1421 error = PROC_I(inode)->op.proc_get_link(inode, &nd->path); 1421 error = PROC_I(inode)->op.proc_get_link(inode, &nd->path);
1422 nd->last_type = LAST_BIND;
1423out: 1422out:
1424 return ERR_PTR(error); 1423 return ERR_PTR(error);
1425} 1424}
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 47c03f4336b8..f277c4a111cb 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -361,12 +361,11 @@ static int smaps_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
361 if (!pte_present(ptent)) 361 if (!pte_present(ptent))
362 continue; 362 continue;
363 363
364 mss->resident += PAGE_SIZE;
365
366 page = vm_normal_page(vma, addr, ptent); 364 page = vm_normal_page(vma, addr, ptent);
367 if (!page) 365 if (!page)
368 continue; 366 continue;
369 367
368 mss->resident += PAGE_SIZE;
370 /* Accumulate the size in pages that have been accessed. */ 369 /* Accumulate the size in pages that have been accessed. */
371 if (pte_young(ptent) || PageReferenced(page)) 370 if (pte_young(ptent) || PageReferenced(page))
372 mss->referenced += PAGE_SIZE; 371 mss->referenced += PAGE_SIZE;
diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c
index dea86abdf2e7..3fc62b097bed 100644
--- a/fs/quota/dquot.c
+++ b/fs/quota/dquot.c
@@ -1377,6 +1377,9 @@ static void inode_sub_rsv_space(struct inode *inode, qsize_t number)
1377static qsize_t inode_get_rsv_space(struct inode *inode) 1377static qsize_t inode_get_rsv_space(struct inode *inode)
1378{ 1378{
1379 qsize_t ret; 1379 qsize_t ret;
1380
1381 if (!inode->i_sb->dq_op->get_reserved_space)
1382 return 0;
1380 spin_lock(&inode->i_lock); 1383 spin_lock(&inode->i_lock);
1381 ret = *inode_reserved_space(inode); 1384 ret = *inode_reserved_space(inode);
1382 spin_unlock(&inode->i_lock); 1385 spin_unlock(&inode->i_lock);
diff --git a/fs/ramfs/file-nommu.c b/fs/ramfs/file-nommu.c
index 2efc57173fd7..1739a4aba25f 100644
--- a/fs/ramfs/file-nommu.c
+++ b/fs/ramfs/file-nommu.c
@@ -123,30 +123,6 @@ add_error:
123 123
124/*****************************************************************************/ 124/*****************************************************************************/
125/* 125/*
126 * check that file shrinkage doesn't leave any VMAs dangling in midair
127 */
128static int ramfs_nommu_check_mappings(struct inode *inode,
129 size_t newsize, size_t size)
130{
131 struct vm_area_struct *vma;
132 struct prio_tree_iter iter;
133
134 /* search for VMAs that fall within the dead zone */
135 vma_prio_tree_foreach(vma, &iter, &inode->i_mapping->i_mmap,
136 newsize >> PAGE_SHIFT,
137 (size + PAGE_SIZE - 1) >> PAGE_SHIFT
138 ) {
139 /* found one - only interested if it's shared out of the page
140 * cache */
141 if (vma->vm_flags & VM_SHARED)
142 return -ETXTBSY; /* not quite true, but near enough */
143 }
144
145 return 0;
146}
147
148/*****************************************************************************/
149/*
150 * 126 *
151 */ 127 */
152static int ramfs_nommu_resize(struct inode *inode, loff_t newsize, loff_t size) 128static int ramfs_nommu_resize(struct inode *inode, loff_t newsize, loff_t size)
@@ -164,7 +140,7 @@ static int ramfs_nommu_resize(struct inode *inode, loff_t newsize, loff_t size)
164 140
165 /* check that a decrease in size doesn't cut off any shared mappings */ 141 /* check that a decrease in size doesn't cut off any shared mappings */
166 if (newsize < size) { 142 if (newsize < size) {
167 ret = ramfs_nommu_check_mappings(inode, newsize, size); 143 ret = nommu_shrink_inode_mappings(inode, size, newsize);
168 if (ret < 0) 144 if (ret < 0)
169 return ret; 145 return ret;
170 } 146 }
diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c
index 1150ebb2536f..9087b10209e6 100644
--- a/fs/reiserfs/inode.c
+++ b/fs/reiserfs/inode.c
@@ -3062,13 +3062,14 @@ static ssize_t reiserfs_direct_IO(int rw, struct kiocb *iocb,
3062int reiserfs_setattr(struct dentry *dentry, struct iattr *attr) 3062int reiserfs_setattr(struct dentry *dentry, struct iattr *attr)
3063{ 3063{
3064 struct inode *inode = dentry->d_inode; 3064 struct inode *inode = dentry->d_inode;
3065 int error;
3066 unsigned int ia_valid; 3065 unsigned int ia_valid;
3066 int depth;
3067 int error;
3067 3068
3068 /* must be turned off for recursive notify_change calls */ 3069 /* must be turned off for recursive notify_change calls */
3069 ia_valid = attr->ia_valid &= ~(ATTR_KILL_SUID|ATTR_KILL_SGID); 3070 ia_valid = attr->ia_valid &= ~(ATTR_KILL_SUID|ATTR_KILL_SGID);
3070 3071
3071 reiserfs_write_lock(inode->i_sb); 3072 depth = reiserfs_write_lock_once(inode->i_sb);
3072 if (attr->ia_valid & ATTR_SIZE) { 3073 if (attr->ia_valid & ATTR_SIZE) {
3073 /* version 2 items will be caught by the s_maxbytes check 3074 /* version 2 items will be caught by the s_maxbytes check
3074 ** done for us in vmtruncate 3075 ** done for us in vmtruncate
@@ -3149,8 +3150,17 @@ int reiserfs_setattr(struct dentry *dentry, struct iattr *attr)
3149 journal_end(&th, inode->i_sb, jbegin_count); 3150 journal_end(&th, inode->i_sb, jbegin_count);
3150 } 3151 }
3151 } 3152 }
3152 if (!error) 3153 if (!error) {
3154 /*
3155 * Relax the lock here, as it might truncate the
3156 * inode pages and wait for inode pages locks.
3157 * To release such page lock, the owner needs the
3158 * reiserfs lock
3159 */
3160 reiserfs_write_unlock_once(inode->i_sb, depth);
3153 error = inode_setattr(inode, attr); 3161 error = inode_setattr(inode, attr);
3162 depth = reiserfs_write_lock_once(inode->i_sb);
3163 }
3154 } 3164 }
3155 3165
3156 if (!error && reiserfs_posixacl(inode->i_sb)) { 3166 if (!error && reiserfs_posixacl(inode->i_sb)) {
@@ -3159,7 +3169,8 @@ int reiserfs_setattr(struct dentry *dentry, struct iattr *attr)
3159 } 3169 }
3160 3170
3161 out: 3171 out:
3162 reiserfs_write_unlock(inode->i_sb); 3172 reiserfs_write_unlock_once(inode->i_sb, depth);
3173
3163 return error; 3174 return error;
3164} 3175}
3165 3176
diff --git a/fs/reiserfs/ioctl.c b/fs/reiserfs/ioctl.c
index ace77451ceb1..f53505de0712 100644
--- a/fs/reiserfs/ioctl.c
+++ b/fs/reiserfs/ioctl.c
@@ -104,9 +104,10 @@ setflags_out:
104 err = put_user(inode->i_generation, (int __user *)arg); 104 err = put_user(inode->i_generation, (int __user *)arg);
105 break; 105 break;
106 case REISERFS_IOC_SETVERSION: 106 case REISERFS_IOC_SETVERSION:
107 if (!is_owner_or_cap(inode)) 107 if (!is_owner_or_cap(inode)) {
108 err = -EPERM; 108 err = -EPERM;
109 break; 109 break;
110 }
110 err = mnt_want_write(filp->f_path.mnt); 111 err = mnt_want_write(filp->f_path.mnt);
111 if (err) 112 if (err)
112 break; 113 break;
diff --git a/fs/reiserfs/xattr.c b/fs/reiserfs/xattr.c
index c3b004ee627b..81f09fab8ae4 100644
--- a/fs/reiserfs/xattr.c
+++ b/fs/reiserfs/xattr.c
@@ -452,7 +452,9 @@ static int lookup_and_delete_xattr(struct inode *inode, const char *name)
452 } 452 }
453 453
454 if (dentry->d_inode) { 454 if (dentry->d_inode) {
455 reiserfs_write_lock(inode->i_sb);
455 err = xattr_unlink(xadir->d_inode, dentry); 456 err = xattr_unlink(xadir->d_inode, dentry);
457 reiserfs_write_unlock(inode->i_sb);
456 update_ctime(inode); 458 update_ctime(inode);
457 } 459 }
458 460
@@ -486,17 +488,21 @@ reiserfs_xattr_set_handle(struct reiserfs_transaction_handle *th,
486 if (get_inode_sd_version(inode) == STAT_DATA_V1) 488 if (get_inode_sd_version(inode) == STAT_DATA_V1)
487 return -EOPNOTSUPP; 489 return -EOPNOTSUPP;
488 490
489 if (!buffer)
490 return lookup_and_delete_xattr(inode, name);
491
492 reiserfs_write_unlock(inode->i_sb); 491 reiserfs_write_unlock(inode->i_sb);
492
493 if (!buffer) {
494 err = lookup_and_delete_xattr(inode, name);
495 reiserfs_write_lock(inode->i_sb);
496 return err;
497 }
498
493 dentry = xattr_lookup(inode, name, flags); 499 dentry = xattr_lookup(inode, name, flags);
494 if (IS_ERR(dentry)) { 500 if (IS_ERR(dentry)) {
495 reiserfs_write_lock(inode->i_sb); 501 reiserfs_write_lock(inode->i_sb);
496 return PTR_ERR(dentry); 502 return PTR_ERR(dentry);
497 } 503 }
498 504
499 down_read(&REISERFS_I(inode)->i_xattr_sem); 505 down_write(&REISERFS_I(inode)->i_xattr_sem);
500 506
501 reiserfs_write_lock(inode->i_sb); 507 reiserfs_write_lock(inode->i_sb);
502 508
@@ -554,8 +560,12 @@ reiserfs_xattr_set_handle(struct reiserfs_transaction_handle *th,
554 .ia_size = buffer_size, 560 .ia_size = buffer_size,
555 .ia_valid = ATTR_SIZE | ATTR_CTIME, 561 .ia_valid = ATTR_SIZE | ATTR_CTIME,
556 }; 562 };
563
564 reiserfs_write_unlock(inode->i_sb);
557 mutex_lock_nested(&dentry->d_inode->i_mutex, I_MUTEX_XATTR); 565 mutex_lock_nested(&dentry->d_inode->i_mutex, I_MUTEX_XATTR);
558 down_write(&dentry->d_inode->i_alloc_sem); 566 down_write(&dentry->d_inode->i_alloc_sem);
567 reiserfs_write_lock(inode->i_sb);
568
559 err = reiserfs_setattr(dentry, &newattrs); 569 err = reiserfs_setattr(dentry, &newattrs);
560 up_write(&dentry->d_inode->i_alloc_sem); 570 up_write(&dentry->d_inode->i_alloc_sem);
561 mutex_unlock(&dentry->d_inode->i_mutex); 571 mutex_unlock(&dentry->d_inode->i_mutex);
diff --git a/fs/reiserfs/xattr_acl.c b/fs/reiserfs/xattr_acl.c
index cc32e6ada67b..dd20a7883f0f 100644
--- a/fs/reiserfs/xattr_acl.c
+++ b/fs/reiserfs/xattr_acl.c
@@ -455,7 +455,9 @@ int reiserfs_acl_chmod(struct inode *inode)
455 return 0; 455 return 0;
456 } 456 }
457 457
458 reiserfs_write_unlock(inode->i_sb);
458 acl = reiserfs_get_acl(inode, ACL_TYPE_ACCESS); 459 acl = reiserfs_get_acl(inode, ACL_TYPE_ACCESS);
460 reiserfs_write_lock(inode->i_sb);
459 if (!acl) 461 if (!acl)
460 return 0; 462 return 0;
461 if (IS_ERR(acl)) 463 if (IS_ERR(acl))
diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c
index 618c2701d3a7..e5a3d8e96bb7 100644
--- a/fs/ubifs/gc.c
+++ b/fs/ubifs/gc.c
@@ -54,6 +54,7 @@
54 */ 54 */
55 55
56#include <linux/pagemap.h> 56#include <linux/pagemap.h>
57#include <linux/list_sort.h>
57#include "ubifs.h" 58#include "ubifs.h"
58 59
59/* 60/*
@@ -108,101 +109,6 @@ static int switch_gc_head(struct ubifs_info *c)
108} 109}
109 110
110/** 111/**
111 * list_sort - sort a list.
112 * @priv: private data, passed to @cmp
113 * @head: the list to sort
114 * @cmp: the elements comparison function
115 *
116 * This function has been implemented by Mark J Roberts <mjr@znex.org>. It
117 * implements "merge sort" which has O(nlog(n)) complexity. The list is sorted
118 * in ascending order.
119 *
120 * The comparison function @cmp is supposed to return a negative value if @a is
121 * than @b, and a positive value if @a is greater than @b. If @a and @b are
122 * equivalent, then it does not matter what this function returns.
123 */
124static void list_sort(void *priv, struct list_head *head,
125 int (*cmp)(void *priv, struct list_head *a,
126 struct list_head *b))
127{
128 struct list_head *p, *q, *e, *list, *tail, *oldhead;
129 int insize, nmerges, psize, qsize, i;
130
131 if (list_empty(head))
132 return;
133
134 list = head->next;
135 list_del(head);
136 insize = 1;
137 for (;;) {
138 p = oldhead = list;
139 list = tail = NULL;
140 nmerges = 0;
141
142 while (p) {
143 nmerges++;
144 q = p;
145 psize = 0;
146 for (i = 0; i < insize; i++) {
147 psize++;
148 q = q->next == oldhead ? NULL : q->next;
149 if (!q)
150 break;
151 }
152
153 qsize = insize;
154 while (psize > 0 || (qsize > 0 && q)) {
155 if (!psize) {
156 e = q;
157 q = q->next;
158 qsize--;
159 if (q == oldhead)
160 q = NULL;
161 } else if (!qsize || !q) {
162 e = p;
163 p = p->next;
164 psize--;
165 if (p == oldhead)
166 p = NULL;
167 } else if (cmp(priv, p, q) <= 0) {
168 e = p;
169 p = p->next;
170 psize--;
171 if (p == oldhead)
172 p = NULL;
173 } else {
174 e = q;
175 q = q->next;
176 qsize--;
177 if (q == oldhead)
178 q = NULL;
179 }
180 if (tail)
181 tail->next = e;
182 else
183 list = e;
184 e->prev = tail;
185 tail = e;
186 }
187 p = q;
188 }
189
190 tail->next = list;
191 list->prev = tail;
192
193 if (nmerges <= 1)
194 break;
195
196 insize *= 2;
197 }
198
199 head->next = list;
200 head->prev = list->prev;
201 list->prev->next = head;
202 list->prev = head;
203}
204
205/**
206 * data_nodes_cmp - compare 2 data nodes. 112 * data_nodes_cmp - compare 2 data nodes.
207 * @priv: UBIFS file-system description object 113 * @priv: UBIFS file-system description object
208 * @a: first data node 114 * @a: first data node
diff --git a/fs/xfs/linux-2.6/xfs_acl.c b/fs/xfs/linux-2.6/xfs_acl.c
index 2512125dfa7c..883ca5ab8af5 100644
--- a/fs/xfs/linux-2.6/xfs_acl.c
+++ b/fs/xfs/linux-2.6/xfs_acl.c
@@ -251,8 +251,9 @@ xfs_set_mode(struct inode *inode, mode_t mode)
251 if (mode != inode->i_mode) { 251 if (mode != inode->i_mode) {
252 struct iattr iattr; 252 struct iattr iattr;
253 253
254 iattr.ia_valid = ATTR_MODE; 254 iattr.ia_valid = ATTR_MODE | ATTR_CTIME;
255 iattr.ia_mode = mode; 255 iattr.ia_mode = mode;
256 iattr.ia_ctime = current_fs_time(inode->i_sb);
256 257
257 error = -xfs_setattr(XFS_I(inode), &iattr, XFS_ATTR_NOACL); 258 error = -xfs_setattr(XFS_I(inode), &iattr, XFS_ATTR_NOACL);
258 } 259 }
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 09783cc444ac..77414db10dc2 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -954,16 +954,14 @@ xfs_fs_destroy_inode(
954 ASSERT_ALWAYS(!xfs_iflags_test(ip, XFS_IRECLAIM)); 954 ASSERT_ALWAYS(!xfs_iflags_test(ip, XFS_IRECLAIM));
955 955
956 /* 956 /*
957 * If we have nothing to flush with this inode then complete the 957 * We always use background reclaim here because even if the
958 * teardown now, otherwise delay the flush operation. 958 * inode is clean, it still may be under IO and hence we have
959 * to take the flush lock. The background reclaim path handles
960 * this more efficiently than we can here, so simply let background
961 * reclaim tear down all inodes.
959 */ 962 */
960 if (!xfs_inode_clean(ip)) {
961 xfs_inode_set_reclaim_tag(ip);
962 return;
963 }
964
965out_reclaim: 963out_reclaim:
966 xfs_ireclaim(ip); 964 xfs_inode_set_reclaim_tag(ip);
967} 965}
968 966
969/* 967/*
diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
index 6fed97a8cd3e..1f5e4bb5e970 100644
--- a/fs/xfs/linux-2.6/xfs_sync.c
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -65,7 +65,6 @@ xfs_inode_ag_lookup(
65 * as the tree is sparse and a gang lookup walks to find 65 * as the tree is sparse and a gang lookup walks to find
66 * the number of objects requested. 66 * the number of objects requested.
67 */ 67 */
68 read_lock(&pag->pag_ici_lock);
69 if (tag == XFS_ICI_NO_TAG) { 68 if (tag == XFS_ICI_NO_TAG) {
70 nr_found = radix_tree_gang_lookup(&pag->pag_ici_root, 69 nr_found = radix_tree_gang_lookup(&pag->pag_ici_root,
71 (void **)&ip, *first_index, 1); 70 (void **)&ip, *first_index, 1);
@@ -74,7 +73,7 @@ xfs_inode_ag_lookup(
74 (void **)&ip, *first_index, 1, tag); 73 (void **)&ip, *first_index, 1, tag);
75 } 74 }
76 if (!nr_found) 75 if (!nr_found)
77 goto unlock; 76 return NULL;
78 77
79 /* 78 /*
80 * Update the index for the next lookup. Catch overflows 79 * Update the index for the next lookup. Catch overflows
@@ -84,13 +83,8 @@ xfs_inode_ag_lookup(
84 */ 83 */
85 *first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1); 84 *first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
86 if (*first_index < XFS_INO_TO_AGINO(mp, ip->i_ino)) 85 if (*first_index < XFS_INO_TO_AGINO(mp, ip->i_ino))
87 goto unlock; 86 return NULL;
88
89 return ip; 87 return ip;
90
91unlock:
92 read_unlock(&pag->pag_ici_lock);
93 return NULL;
94} 88}
95 89
96STATIC int 90STATIC int
@@ -100,7 +94,8 @@ xfs_inode_ag_walk(
100 int (*execute)(struct xfs_inode *ip, 94 int (*execute)(struct xfs_inode *ip,
101 struct xfs_perag *pag, int flags), 95 struct xfs_perag *pag, int flags),
102 int flags, 96 int flags,
103 int tag) 97 int tag,
98 int exclusive)
104{ 99{
105 struct xfs_perag *pag = &mp->m_perag[ag]; 100 struct xfs_perag *pag = &mp->m_perag[ag];
106 uint32_t first_index; 101 uint32_t first_index;
@@ -114,10 +109,20 @@ restart:
114 int error = 0; 109 int error = 0;
115 xfs_inode_t *ip; 110 xfs_inode_t *ip;
116 111
112 if (exclusive)
113 write_lock(&pag->pag_ici_lock);
114 else
115 read_lock(&pag->pag_ici_lock);
117 ip = xfs_inode_ag_lookup(mp, pag, &first_index, tag); 116 ip = xfs_inode_ag_lookup(mp, pag, &first_index, tag);
118 if (!ip) 117 if (!ip) {
118 if (exclusive)
119 write_unlock(&pag->pag_ici_lock);
120 else
121 read_unlock(&pag->pag_ici_lock);
119 break; 122 break;
123 }
120 124
125 /* execute releases pag->pag_ici_lock */
121 error = execute(ip, pag, flags); 126 error = execute(ip, pag, flags);
122 if (error == EAGAIN) { 127 if (error == EAGAIN) {
123 skipped++; 128 skipped++;
@@ -125,9 +130,8 @@ restart:
125 } 130 }
126 if (error) 131 if (error)
127 last_error = error; 132 last_error = error;
128 /* 133
129 * bail out if the filesystem is corrupted. 134 /* bail out if the filesystem is corrupted. */
130 */
131 if (error == EFSCORRUPTED) 135 if (error == EFSCORRUPTED)
132 break; 136 break;
133 137
@@ -148,7 +152,8 @@ xfs_inode_ag_iterator(
148 int (*execute)(struct xfs_inode *ip, 152 int (*execute)(struct xfs_inode *ip,
149 struct xfs_perag *pag, int flags), 153 struct xfs_perag *pag, int flags),
150 int flags, 154 int flags,
151 int tag) 155 int tag,
156 int exclusive)
152{ 157{
153 int error = 0; 158 int error = 0;
154 int last_error = 0; 159 int last_error = 0;
@@ -157,7 +162,8 @@ xfs_inode_ag_iterator(
157 for (ag = 0; ag < mp->m_sb.sb_agcount; ag++) { 162 for (ag = 0; ag < mp->m_sb.sb_agcount; ag++) {
158 if (!mp->m_perag[ag].pag_ici_init) 163 if (!mp->m_perag[ag].pag_ici_init)
159 continue; 164 continue;
160 error = xfs_inode_ag_walk(mp, ag, execute, flags, tag); 165 error = xfs_inode_ag_walk(mp, ag, execute, flags, tag,
166 exclusive);
161 if (error) { 167 if (error) {
162 last_error = error; 168 last_error = error;
163 if (error == EFSCORRUPTED) 169 if (error == EFSCORRUPTED)
@@ -174,30 +180,31 @@ xfs_sync_inode_valid(
174 struct xfs_perag *pag) 180 struct xfs_perag *pag)
175{ 181{
176 struct inode *inode = VFS_I(ip); 182 struct inode *inode = VFS_I(ip);
183 int error = EFSCORRUPTED;
177 184
178 /* nothing to sync during shutdown */ 185 /* nothing to sync during shutdown */
179 if (XFS_FORCED_SHUTDOWN(ip->i_mount)) { 186 if (XFS_FORCED_SHUTDOWN(ip->i_mount))
180 read_unlock(&pag->pag_ici_lock); 187 goto out_unlock;
181 return EFSCORRUPTED;
182 }
183 188
184 /* 189 /* avoid new or reclaimable inodes. Leave for reclaim code to flush */
185 * If we can't get a reference on the inode, it must be in reclaim. 190 error = ENOENT;
186 * Leave it for the reclaim code to flush. Also avoid inodes that 191 if (xfs_iflags_test(ip, XFS_INEW | XFS_IRECLAIMABLE | XFS_IRECLAIM))
187 * haven't been fully initialised. 192 goto out_unlock;
188 */
189 if (!igrab(inode)) {
190 read_unlock(&pag->pag_ici_lock);
191 return ENOENT;
192 }
193 read_unlock(&pag->pag_ici_lock);
194 193
195 if (is_bad_inode(inode) || xfs_iflags_test(ip, XFS_INEW)) { 194 /* If we can't grab the inode, it must on it's way to reclaim. */
195 if (!igrab(inode))
196 goto out_unlock;
197
198 if (is_bad_inode(inode)) {
196 IRELE(ip); 199 IRELE(ip);
197 return ENOENT; 200 goto out_unlock;
198 } 201 }
199 202
200 return 0; 203 /* inode is valid */
204 error = 0;
205out_unlock:
206 read_unlock(&pag->pag_ici_lock);
207 return error;
201} 208}
202 209
203STATIC int 210STATIC int
@@ -282,7 +289,7 @@ xfs_sync_data(
282 ASSERT((flags & ~(SYNC_TRYLOCK|SYNC_WAIT)) == 0); 289 ASSERT((flags & ~(SYNC_TRYLOCK|SYNC_WAIT)) == 0);
283 290
284 error = xfs_inode_ag_iterator(mp, xfs_sync_inode_data, flags, 291 error = xfs_inode_ag_iterator(mp, xfs_sync_inode_data, flags,
285 XFS_ICI_NO_TAG); 292 XFS_ICI_NO_TAG, 0);
286 if (error) 293 if (error)
287 return XFS_ERROR(error); 294 return XFS_ERROR(error);
288 295
@@ -304,7 +311,7 @@ xfs_sync_attr(
304 ASSERT((flags & ~SYNC_WAIT) == 0); 311 ASSERT((flags & ~SYNC_WAIT) == 0);
305 312
306 return xfs_inode_ag_iterator(mp, xfs_sync_inode_attr, flags, 313 return xfs_inode_ag_iterator(mp, xfs_sync_inode_attr, flags,
307 XFS_ICI_NO_TAG); 314 XFS_ICI_NO_TAG, 0);
308} 315}
309 316
310STATIC int 317STATIC int
@@ -664,60 +671,6 @@ xfs_syncd_stop(
664 kthread_stop(mp->m_sync_task); 671 kthread_stop(mp->m_sync_task);
665} 672}
666 673
667STATIC int
668xfs_reclaim_inode(
669 xfs_inode_t *ip,
670 int sync_mode)
671{
672 xfs_perag_t *pag = xfs_get_perag(ip->i_mount, ip->i_ino);
673
674 /* The hash lock here protects a thread in xfs_iget_core from
675 * racing with us on linking the inode back with a vnode.
676 * Once we have the XFS_IRECLAIM flag set it will not touch
677 * us.
678 */
679 write_lock(&pag->pag_ici_lock);
680 spin_lock(&ip->i_flags_lock);
681 if (__xfs_iflags_test(ip, XFS_IRECLAIM) ||
682 !__xfs_iflags_test(ip, XFS_IRECLAIMABLE)) {
683 spin_unlock(&ip->i_flags_lock);
684 write_unlock(&pag->pag_ici_lock);
685 return -EAGAIN;
686 }
687 __xfs_iflags_set(ip, XFS_IRECLAIM);
688 spin_unlock(&ip->i_flags_lock);
689 write_unlock(&pag->pag_ici_lock);
690 xfs_put_perag(ip->i_mount, pag);
691
692 /*
693 * If the inode is still dirty, then flush it out. If the inode
694 * is not in the AIL, then it will be OK to flush it delwri as
695 * long as xfs_iflush() does not keep any references to the inode.
696 * We leave that decision up to xfs_iflush() since it has the
697 * knowledge of whether it's OK to simply do a delwri flush of
698 * the inode or whether we need to wait until the inode is
699 * pulled from the AIL.
700 * We get the flush lock regardless, though, just to make sure
701 * we don't free it while it is being flushed.
702 */
703 xfs_ilock(ip, XFS_ILOCK_EXCL);
704 xfs_iflock(ip);
705
706 /*
707 * In the case of a forced shutdown we rely on xfs_iflush() to
708 * wait for the inode to be unpinned before returning an error.
709 */
710 if (!is_bad_inode(VFS_I(ip)) && xfs_iflush(ip, sync_mode) == 0) {
711 /* synchronize with xfs_iflush_done */
712 xfs_iflock(ip);
713 xfs_ifunlock(ip);
714 }
715
716 xfs_iunlock(ip, XFS_ILOCK_EXCL);
717 xfs_ireclaim(ip);
718 return 0;
719}
720
721void 674void
722__xfs_inode_set_reclaim_tag( 675__xfs_inode_set_reclaim_tag(
723 struct xfs_perag *pag, 676 struct xfs_perag *pag,
@@ -760,19 +713,55 @@ __xfs_inode_clear_reclaim_tag(
760} 713}
761 714
762STATIC int 715STATIC int
763xfs_reclaim_inode_now( 716xfs_reclaim_inode(
764 struct xfs_inode *ip, 717 struct xfs_inode *ip,
765 struct xfs_perag *pag, 718 struct xfs_perag *pag,
766 int flags) 719 int sync_mode)
767{ 720{
768 /* ignore if already under reclaim */ 721 /*
769 if (xfs_iflags_test(ip, XFS_IRECLAIM)) { 722 * The radix tree lock here protects a thread in xfs_iget from racing
770 read_unlock(&pag->pag_ici_lock); 723 * with us starting reclaim on the inode. Once we have the
724 * XFS_IRECLAIM flag set it will not touch us.
725 */
726 spin_lock(&ip->i_flags_lock);
727 ASSERT_ALWAYS(__xfs_iflags_test(ip, XFS_IRECLAIMABLE));
728 if (__xfs_iflags_test(ip, XFS_IRECLAIM)) {
729 /* ignore as it is already under reclaim */
730 spin_unlock(&ip->i_flags_lock);
731 write_unlock(&pag->pag_ici_lock);
771 return 0; 732 return 0;
772 } 733 }
773 read_unlock(&pag->pag_ici_lock); 734 __xfs_iflags_set(ip, XFS_IRECLAIM);
735 spin_unlock(&ip->i_flags_lock);
736 write_unlock(&pag->pag_ici_lock);
774 737
775 return xfs_reclaim_inode(ip, flags); 738 /*
739 * If the inode is still dirty, then flush it out. If the inode
740 * is not in the AIL, then it will be OK to flush it delwri as
741 * long as xfs_iflush() does not keep any references to the inode.
742 * We leave that decision up to xfs_iflush() since it has the
743 * knowledge of whether it's OK to simply do a delwri flush of
744 * the inode or whether we need to wait until the inode is
745 * pulled from the AIL.
746 * We get the flush lock regardless, though, just to make sure
747 * we don't free it while it is being flushed.
748 */
749 xfs_ilock(ip, XFS_ILOCK_EXCL);
750 xfs_iflock(ip);
751
752 /*
753 * In the case of a forced shutdown we rely on xfs_iflush() to
754 * wait for the inode to be unpinned before returning an error.
755 */
756 if (!is_bad_inode(VFS_I(ip)) && xfs_iflush(ip, sync_mode) == 0) {
757 /* synchronize with xfs_iflush_done */
758 xfs_iflock(ip);
759 xfs_ifunlock(ip);
760 }
761
762 xfs_iunlock(ip, XFS_ILOCK_EXCL);
763 xfs_ireclaim(ip);
764 return 0;
776} 765}
777 766
778int 767int
@@ -780,6 +769,6 @@ xfs_reclaim_inodes(
780 xfs_mount_t *mp, 769 xfs_mount_t *mp,
781 int mode) 770 int mode)
782{ 771{
783 return xfs_inode_ag_iterator(mp, xfs_reclaim_inode_now, mode, 772 return xfs_inode_ag_iterator(mp, xfs_reclaim_inode, mode,
784 XFS_ICI_RECLAIM_TAG); 773 XFS_ICI_RECLAIM_TAG, 1);
785} 774}
diff --git a/fs/xfs/linux-2.6/xfs_sync.h b/fs/xfs/linux-2.6/xfs_sync.h
index a500b4d91835..ea932b43335d 100644
--- a/fs/xfs/linux-2.6/xfs_sync.h
+++ b/fs/xfs/linux-2.6/xfs_sync.h
@@ -54,6 +54,6 @@ void __xfs_inode_clear_reclaim_tag(struct xfs_mount *mp, struct xfs_perag *pag,
54int xfs_sync_inode_valid(struct xfs_inode *ip, struct xfs_perag *pag); 54int xfs_sync_inode_valid(struct xfs_inode *ip, struct xfs_perag *pag);
55int xfs_inode_ag_iterator(struct xfs_mount *mp, 55int xfs_inode_ag_iterator(struct xfs_mount *mp,
56 int (*execute)(struct xfs_inode *ip, struct xfs_perag *pag, int flags), 56 int (*execute)(struct xfs_inode *ip, struct xfs_perag *pag, int flags),
57 int flags, int tag); 57 int flags, int tag, int write_lock);
58 58
59#endif 59#endif
diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
index c40834bdee58..c22a608321a3 100644
--- a/fs/xfs/linux-2.6/xfs_trace.h
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -33,51 +33,55 @@ struct xfs_dquot;
33struct xlog_ticket; 33struct xlog_ticket;
34struct log; 34struct log;
35 35
36DECLARE_EVENT_CLASS(xfs_attr_list_class,
37 TP_PROTO(struct xfs_attr_list_context *ctx),
38 TP_ARGS(ctx),
39 TP_STRUCT__entry(
40 __field(dev_t, dev)
41 __field(xfs_ino_t, ino)
42 __field(u32, hashval)
43 __field(u32, blkno)
44 __field(u32, offset)
45 __field(void *, alist)
46 __field(int, bufsize)
47 __field(int, count)
48 __field(int, firstu)
49 __field(int, dupcnt)
50 __field(int, flags)
51 ),
52 TP_fast_assign(
53 __entry->dev = VFS_I(ctx->dp)->i_sb->s_dev;
54 __entry->ino = ctx->dp->i_ino;
55 __entry->hashval = ctx->cursor->hashval;
56 __entry->blkno = ctx->cursor->blkno;
57 __entry->offset = ctx->cursor->offset;
58 __entry->alist = ctx->alist;
59 __entry->bufsize = ctx->bufsize;
60 __entry->count = ctx->count;
61 __entry->firstu = ctx->firstu;
62 __entry->flags = ctx->flags;
63 ),
64 TP_printk("dev %d:%d ino 0x%llx cursor h/b/o 0x%x/0x%x/%u dupcnt %u "
65 "alist 0x%p size %u count %u firstu %u flags %d %s",
66 MAJOR(__entry->dev), MINOR(__entry->dev),
67 __entry->ino,
68 __entry->hashval,
69 __entry->blkno,
70 __entry->offset,
71 __entry->dupcnt,
72 __entry->alist,
73 __entry->bufsize,
74 __entry->count,
75 __entry->firstu,
76 __entry->flags,
77 __print_flags(__entry->flags, "|", XFS_ATTR_FLAGS)
78 )
79)
80
36#define DEFINE_ATTR_LIST_EVENT(name) \ 81#define DEFINE_ATTR_LIST_EVENT(name) \
37TRACE_EVENT(name, \ 82DEFINE_EVENT(xfs_attr_list_class, name, \
38 TP_PROTO(struct xfs_attr_list_context *ctx), \ 83 TP_PROTO(struct xfs_attr_list_context *ctx), \
39 TP_ARGS(ctx), \ 84 TP_ARGS(ctx))
40 TP_STRUCT__entry( \
41 __field(dev_t, dev) \
42 __field(xfs_ino_t, ino) \
43 __field(u32, hashval) \
44 __field(u32, blkno) \
45 __field(u32, offset) \
46 __field(void *, alist) \
47 __field(int, bufsize) \
48 __field(int, count) \
49 __field(int, firstu) \
50 __field(int, dupcnt) \
51 __field(int, flags) \
52 ), \
53 TP_fast_assign( \
54 __entry->dev = VFS_I(ctx->dp)->i_sb->s_dev; \
55 __entry->ino = ctx->dp->i_ino; \
56 __entry->hashval = ctx->cursor->hashval; \
57 __entry->blkno = ctx->cursor->blkno; \
58 __entry->offset = ctx->cursor->offset; \
59 __entry->alist = ctx->alist; \
60 __entry->bufsize = ctx->bufsize; \
61 __entry->count = ctx->count; \
62 __entry->firstu = ctx->firstu; \
63 __entry->flags = ctx->flags; \
64 ), \
65 TP_printk("dev %d:%d ino 0x%llx cursor h/b/o 0x%x/0x%x/%u dupcnt %u " \
66 "alist 0x%p size %u count %u firstu %u flags %d %s", \
67 MAJOR(__entry->dev), MINOR(__entry->dev), \
68 __entry->ino, \
69 __entry->hashval, \
70 __entry->blkno, \
71 __entry->offset, \
72 __entry->dupcnt, \
73 __entry->alist, \
74 __entry->bufsize, \
75 __entry->count, \
76 __entry->firstu, \
77 __entry->flags, \
78 __print_flags(__entry->flags, "|", XFS_ATTR_FLAGS) \
79 ) \
80)
81DEFINE_ATTR_LIST_EVENT(xfs_attr_list_sf); 85DEFINE_ATTR_LIST_EVENT(xfs_attr_list_sf);
82DEFINE_ATTR_LIST_EVENT(xfs_attr_list_sf_all); 86DEFINE_ATTR_LIST_EVENT(xfs_attr_list_sf_all);
83DEFINE_ATTR_LIST_EVENT(xfs_attr_list_leaf); 87DEFINE_ATTR_LIST_EVENT(xfs_attr_list_leaf);
@@ -178,91 +182,99 @@ TRACE_EVENT(xfs_iext_insert,
178 (char *)__entry->caller_ip) 182 (char *)__entry->caller_ip)
179); 183);
180 184
185DECLARE_EVENT_CLASS(xfs_bmap_class,
186 TP_PROTO(struct xfs_inode *ip, xfs_extnum_t idx, int state,
187 unsigned long caller_ip),
188 TP_ARGS(ip, idx, state, caller_ip),
189 TP_STRUCT__entry(
190 __field(dev_t, dev)
191 __field(xfs_ino_t, ino)
192 __field(xfs_extnum_t, idx)
193 __field(xfs_fileoff_t, startoff)
194 __field(xfs_fsblock_t, startblock)
195 __field(xfs_filblks_t, blockcount)
196 __field(xfs_exntst_t, state)
197 __field(int, bmap_state)
198 __field(unsigned long, caller_ip)
199 ),
200 TP_fast_assign(
201 struct xfs_ifork *ifp = (state & BMAP_ATTRFORK) ?
202 ip->i_afp : &ip->i_df;
203 struct xfs_bmbt_irec r;
204
205 xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx), &r);
206 __entry->dev = VFS_I(ip)->i_sb->s_dev;
207 __entry->ino = ip->i_ino;
208 __entry->idx = idx;
209 __entry->startoff = r.br_startoff;
210 __entry->startblock = r.br_startblock;
211 __entry->blockcount = r.br_blockcount;
212 __entry->state = r.br_state;
213 __entry->bmap_state = state;
214 __entry->caller_ip = caller_ip;
215 ),
216 TP_printk("dev %d:%d ino 0x%llx state %s idx %ld "
217 "offset %lld block %s count %lld flag %d caller %pf",
218 MAJOR(__entry->dev), MINOR(__entry->dev),
219 __entry->ino,
220 __print_flags(__entry->bmap_state, "|", XFS_BMAP_EXT_FLAGS),
221 (long)__entry->idx,
222 __entry->startoff,
223 xfs_fmtfsblock(__entry->startblock),
224 __entry->blockcount,
225 __entry->state,
226 (char *)__entry->caller_ip)
227)
228
181#define DEFINE_BMAP_EVENT(name) \ 229#define DEFINE_BMAP_EVENT(name) \
182TRACE_EVENT(name, \ 230DEFINE_EVENT(xfs_bmap_class, name, \
183 TP_PROTO(struct xfs_inode *ip, xfs_extnum_t idx, int state, \ 231 TP_PROTO(struct xfs_inode *ip, xfs_extnum_t idx, int state, \
184 unsigned long caller_ip), \ 232 unsigned long caller_ip), \
185 TP_ARGS(ip, idx, state, caller_ip), \ 233 TP_ARGS(ip, idx, state, caller_ip))
186 TP_STRUCT__entry( \
187 __field(dev_t, dev) \
188 __field(xfs_ino_t, ino) \
189 __field(xfs_extnum_t, idx) \
190 __field(xfs_fileoff_t, startoff) \
191 __field(xfs_fsblock_t, startblock) \
192 __field(xfs_filblks_t, blockcount) \
193 __field(xfs_exntst_t, state) \
194 __field(int, bmap_state) \
195 __field(unsigned long, caller_ip) \
196 ), \
197 TP_fast_assign( \
198 struct xfs_ifork *ifp = (state & BMAP_ATTRFORK) ? \
199 ip->i_afp : &ip->i_df; \
200 struct xfs_bmbt_irec r; \
201 \
202 xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx), &r); \
203 __entry->dev = VFS_I(ip)->i_sb->s_dev; \
204 __entry->ino = ip->i_ino; \
205 __entry->idx = idx; \
206 __entry->startoff = r.br_startoff; \
207 __entry->startblock = r.br_startblock; \
208 __entry->blockcount = r.br_blockcount; \
209 __entry->state = r.br_state; \
210 __entry->bmap_state = state; \
211 __entry->caller_ip = caller_ip; \
212 ), \
213 TP_printk("dev %d:%d ino 0x%llx state %s idx %ld " \
214 "offset %lld block %s count %lld flag %d caller %pf", \
215 MAJOR(__entry->dev), MINOR(__entry->dev), \
216 __entry->ino, \
217 __print_flags(__entry->bmap_state, "|", XFS_BMAP_EXT_FLAGS), \
218 (long)__entry->idx, \
219 __entry->startoff, \
220 xfs_fmtfsblock(__entry->startblock), \
221 __entry->blockcount, \
222 __entry->state, \
223 (char *)__entry->caller_ip) \
224)
225
226DEFINE_BMAP_EVENT(xfs_iext_remove); 234DEFINE_BMAP_EVENT(xfs_iext_remove);
227DEFINE_BMAP_EVENT(xfs_bmap_pre_update); 235DEFINE_BMAP_EVENT(xfs_bmap_pre_update);
228DEFINE_BMAP_EVENT(xfs_bmap_post_update); 236DEFINE_BMAP_EVENT(xfs_bmap_post_update);
229DEFINE_BMAP_EVENT(xfs_extlist); 237DEFINE_BMAP_EVENT(xfs_extlist);
230 238
231#define DEFINE_BUF_EVENT(tname) \ 239DECLARE_EVENT_CLASS(xfs_buf_class,
232TRACE_EVENT(tname, \ 240 TP_PROTO(struct xfs_buf *bp, unsigned long caller_ip),
233 TP_PROTO(struct xfs_buf *bp, unsigned long caller_ip), \ 241 TP_ARGS(bp, caller_ip),
234 TP_ARGS(bp, caller_ip), \ 242 TP_STRUCT__entry(
235 TP_STRUCT__entry( \ 243 __field(dev_t, dev)
236 __field(dev_t, dev) \ 244 __field(xfs_daddr_t, bno)
237 __field(xfs_daddr_t, bno) \ 245 __field(size_t, buffer_length)
238 __field(size_t, buffer_length) \ 246 __field(int, hold)
239 __field(int, hold) \ 247 __field(int, pincount)
240 __field(int, pincount) \ 248 __field(unsigned, lockval)
241 __field(unsigned, lockval) \ 249 __field(unsigned, flags)
242 __field(unsigned, flags) \ 250 __field(unsigned long, caller_ip)
243 __field(unsigned long, caller_ip) \ 251 ),
244 ), \ 252 TP_fast_assign(
245 TP_fast_assign( \ 253 __entry->dev = bp->b_target->bt_dev;
246 __entry->dev = bp->b_target->bt_dev; \ 254 __entry->bno = bp->b_bn;
247 __entry->bno = bp->b_bn; \ 255 __entry->buffer_length = bp->b_buffer_length;
248 __entry->buffer_length = bp->b_buffer_length; \ 256 __entry->hold = atomic_read(&bp->b_hold);
249 __entry->hold = atomic_read(&bp->b_hold); \ 257 __entry->pincount = atomic_read(&bp->b_pin_count);
250 __entry->pincount = atomic_read(&bp->b_pin_count); \ 258 __entry->lockval = xfs_buf_lock_value(bp);
251 __entry->lockval = xfs_buf_lock_value(bp); \ 259 __entry->flags = bp->b_flags;
252 __entry->flags = bp->b_flags; \ 260 __entry->caller_ip = caller_ip;
253 __entry->caller_ip = caller_ip; \ 261 ),
254 ), \ 262 TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d "
255 TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d " \ 263 "lock %d flags %s caller %pf",
256 "lock %d flags %s caller %pf", \ 264 MAJOR(__entry->dev), MINOR(__entry->dev),
257 MAJOR(__entry->dev), MINOR(__entry->dev), \ 265 (unsigned long long)__entry->bno,
258 (unsigned long long)__entry->bno, \ 266 __entry->buffer_length,
259 __entry->buffer_length, \ 267 __entry->hold,
260 __entry->hold, \ 268 __entry->pincount,
261 __entry->pincount, \ 269 __entry->lockval,
262 __entry->lockval, \ 270 __print_flags(__entry->flags, "|", XFS_BUF_FLAGS),
263 __print_flags(__entry->flags, "|", XFS_BUF_FLAGS), \ 271 (void *)__entry->caller_ip)
264 (void *)__entry->caller_ip) \
265) 272)
273
274#define DEFINE_BUF_EVENT(name) \
275DEFINE_EVENT(xfs_buf_class, name, \
276 TP_PROTO(struct xfs_buf *bp, unsigned long caller_ip), \
277 TP_ARGS(bp, caller_ip))
266DEFINE_BUF_EVENT(xfs_buf_init); 278DEFINE_BUF_EVENT(xfs_buf_init);
267DEFINE_BUF_EVENT(xfs_buf_free); 279DEFINE_BUF_EVENT(xfs_buf_free);
268DEFINE_BUF_EVENT(xfs_buf_hold); 280DEFINE_BUF_EVENT(xfs_buf_hold);
@@ -299,41 +311,45 @@ DEFINE_BUF_EVENT(xfs_reset_dqcounts);
299DEFINE_BUF_EVENT(xfs_inode_item_push); 311DEFINE_BUF_EVENT(xfs_inode_item_push);
300 312
301/* pass flags explicitly */ 313/* pass flags explicitly */
302#define DEFINE_BUF_FLAGS_EVENT(tname) \ 314DECLARE_EVENT_CLASS(xfs_buf_flags_class,
303TRACE_EVENT(tname, \ 315 TP_PROTO(struct xfs_buf *bp, unsigned flags, unsigned long caller_ip),
304 TP_PROTO(struct xfs_buf *bp, unsigned flags, unsigned long caller_ip), \ 316 TP_ARGS(bp, flags, caller_ip),
305 TP_ARGS(bp, flags, caller_ip), \ 317 TP_STRUCT__entry(
306 TP_STRUCT__entry( \ 318 __field(dev_t, dev)
307 __field(dev_t, dev) \ 319 __field(xfs_daddr_t, bno)
308 __field(xfs_daddr_t, bno) \ 320 __field(size_t, buffer_length)
309 __field(size_t, buffer_length) \ 321 __field(int, hold)
310 __field(int, hold) \ 322 __field(int, pincount)
311 __field(int, pincount) \ 323 __field(unsigned, lockval)
312 __field(unsigned, lockval) \ 324 __field(unsigned, flags)
313 __field(unsigned, flags) \ 325 __field(unsigned long, caller_ip)
314 __field(unsigned long, caller_ip) \ 326 ),
315 ), \ 327 TP_fast_assign(
316 TP_fast_assign( \ 328 __entry->dev = bp->b_target->bt_dev;
317 __entry->dev = bp->b_target->bt_dev; \ 329 __entry->bno = bp->b_bn;
318 __entry->bno = bp->b_bn; \ 330 __entry->buffer_length = bp->b_buffer_length;
319 __entry->buffer_length = bp->b_buffer_length; \ 331 __entry->flags = flags;
320 __entry->flags = flags; \ 332 __entry->hold = atomic_read(&bp->b_hold);
321 __entry->hold = atomic_read(&bp->b_hold); \ 333 __entry->pincount = atomic_read(&bp->b_pin_count);
322 __entry->pincount = atomic_read(&bp->b_pin_count); \ 334 __entry->lockval = xfs_buf_lock_value(bp);
323 __entry->lockval = xfs_buf_lock_value(bp); \ 335 __entry->caller_ip = caller_ip;
324 __entry->caller_ip = caller_ip; \ 336 ),
325 ), \ 337 TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d "
326 TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d " \ 338 "lock %d flags %s caller %pf",
327 "lock %d flags %s caller %pf", \ 339 MAJOR(__entry->dev), MINOR(__entry->dev),
328 MAJOR(__entry->dev), MINOR(__entry->dev), \ 340 (unsigned long long)__entry->bno,
329 (unsigned long long)__entry->bno, \ 341 __entry->buffer_length,
330 __entry->buffer_length, \ 342 __entry->hold,
331 __entry->hold, \ 343 __entry->pincount,
332 __entry->pincount, \ 344 __entry->lockval,
333 __entry->lockval, \ 345 __print_flags(__entry->flags, "|", XFS_BUF_FLAGS),
334 __print_flags(__entry->flags, "|", XFS_BUF_FLAGS), \ 346 (void *)__entry->caller_ip)
335 (void *)__entry->caller_ip) \
336) 347)
348
349#define DEFINE_BUF_FLAGS_EVENT(name) \
350DEFINE_EVENT(xfs_buf_flags_class, name, \
351 TP_PROTO(struct xfs_buf *bp, unsigned flags, unsigned long caller_ip), \
352 TP_ARGS(bp, flags, caller_ip))
337DEFINE_BUF_FLAGS_EVENT(xfs_buf_find); 353DEFINE_BUF_FLAGS_EVENT(xfs_buf_find);
338DEFINE_BUF_FLAGS_EVENT(xfs_buf_get); 354DEFINE_BUF_FLAGS_EVENT(xfs_buf_get);
339DEFINE_BUF_FLAGS_EVENT(xfs_buf_read); 355DEFINE_BUF_FLAGS_EVENT(xfs_buf_read);
@@ -376,55 +392,58 @@ TRACE_EVENT(xfs_buf_ioerror,
376 (void *)__entry->caller_ip) 392 (void *)__entry->caller_ip)
377); 393);
378 394
379#define DEFINE_BUF_ITEM_EVENT(tname) \ 395DECLARE_EVENT_CLASS(xfs_buf_item_class,
380TRACE_EVENT(tname, \ 396 TP_PROTO(struct xfs_buf_log_item *bip),
381 TP_PROTO(struct xfs_buf_log_item *bip), \ 397 TP_ARGS(bip),
382 TP_ARGS(bip), \ 398 TP_STRUCT__entry(
383 TP_STRUCT__entry( \ 399 __field(dev_t, dev)
384 __field(dev_t, dev) \ 400 __field(xfs_daddr_t, buf_bno)
385 __field(xfs_daddr_t, buf_bno) \ 401 __field(size_t, buf_len)
386 __field(size_t, buf_len) \ 402 __field(int, buf_hold)
387 __field(int, buf_hold) \ 403 __field(int, buf_pincount)
388 __field(int, buf_pincount) \ 404 __field(int, buf_lockval)
389 __field(int, buf_lockval) \ 405 __field(unsigned, buf_flags)
390 __field(unsigned, buf_flags) \ 406 __field(unsigned, bli_recur)
391 __field(unsigned, bli_recur) \ 407 __field(int, bli_refcount)
392 __field(int, bli_refcount) \ 408 __field(unsigned, bli_flags)
393 __field(unsigned, bli_flags) \ 409 __field(void *, li_desc)
394 __field(void *, li_desc) \ 410 __field(unsigned, li_flags)
395 __field(unsigned, li_flags) \ 411 ),
396 ), \ 412 TP_fast_assign(
397 TP_fast_assign( \ 413 __entry->dev = bip->bli_buf->b_target->bt_dev;
398 __entry->dev = bip->bli_buf->b_target->bt_dev; \ 414 __entry->bli_flags = bip->bli_flags;
399 __entry->bli_flags = bip->bli_flags; \ 415 __entry->bli_recur = bip->bli_recur;
400 __entry->bli_recur = bip->bli_recur; \ 416 __entry->bli_refcount = atomic_read(&bip->bli_refcount);
401 __entry->bli_refcount = atomic_read(&bip->bli_refcount); \ 417 __entry->buf_bno = bip->bli_buf->b_bn;
402 __entry->buf_bno = bip->bli_buf->b_bn; \ 418 __entry->buf_len = bip->bli_buf->b_buffer_length;
403 __entry->buf_len = bip->bli_buf->b_buffer_length; \ 419 __entry->buf_flags = bip->bli_buf->b_flags;
404 __entry->buf_flags = bip->bli_buf->b_flags; \ 420 __entry->buf_hold = atomic_read(&bip->bli_buf->b_hold);
405 __entry->buf_hold = atomic_read(&bip->bli_buf->b_hold); \ 421 __entry->buf_pincount = atomic_read(&bip->bli_buf->b_pin_count);
406 __entry->buf_pincount = \ 422 __entry->buf_lockval = xfs_buf_lock_value(bip->bli_buf);
407 atomic_read(&bip->bli_buf->b_pin_count); \ 423 __entry->li_desc = bip->bli_item.li_desc;
408 __entry->buf_lockval = xfs_buf_lock_value(bip->bli_buf); \ 424 __entry->li_flags = bip->bli_item.li_flags;
409 __entry->li_desc = bip->bli_item.li_desc; \ 425 ),
410 __entry->li_flags = bip->bli_item.li_flags; \ 426 TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d "
411 ), \ 427 "lock %d flags %s recur %d refcount %d bliflags %s "
412 TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d " \ 428 "lidesc 0x%p liflags %s",
413 "lock %d flags %s recur %d refcount %d bliflags %s " \ 429 MAJOR(__entry->dev), MINOR(__entry->dev),
414 "lidesc 0x%p liflags %s", \ 430 (unsigned long long)__entry->buf_bno,
415 MAJOR(__entry->dev), MINOR(__entry->dev), \ 431 __entry->buf_len,
416 (unsigned long long)__entry->buf_bno, \ 432 __entry->buf_hold,
417 __entry->buf_len, \ 433 __entry->buf_pincount,
418 __entry->buf_hold, \ 434 __entry->buf_lockval,
419 __entry->buf_pincount, \ 435 __print_flags(__entry->buf_flags, "|", XFS_BUF_FLAGS),
420 __entry->buf_lockval, \ 436 __entry->bli_recur,
421 __print_flags(__entry->buf_flags, "|", XFS_BUF_FLAGS), \ 437 __entry->bli_refcount,
422 __entry->bli_recur, \ 438 __print_flags(__entry->bli_flags, "|", XFS_BLI_FLAGS),
423 __entry->bli_refcount, \ 439 __entry->li_desc,
424 __print_flags(__entry->bli_flags, "|", XFS_BLI_FLAGS), \ 440 __print_flags(__entry->li_flags, "|", XFS_LI_FLAGS))
425 __entry->li_desc, \
426 __print_flags(__entry->li_flags, "|", XFS_LI_FLAGS)) \
427) 441)
442
443#define DEFINE_BUF_ITEM_EVENT(name) \
444DEFINE_EVENT(xfs_buf_item_class, name, \
445 TP_PROTO(struct xfs_buf_log_item *bip), \
446 TP_ARGS(bip))
428DEFINE_BUF_ITEM_EVENT(xfs_buf_item_size); 447DEFINE_BUF_ITEM_EVENT(xfs_buf_item_size);
429DEFINE_BUF_ITEM_EVENT(xfs_buf_item_size_stale); 448DEFINE_BUF_ITEM_EVENT(xfs_buf_item_size_stale);
430DEFINE_BUF_ITEM_EVENT(xfs_buf_item_format); 449DEFINE_BUF_ITEM_EVENT(xfs_buf_item_format);
@@ -450,78 +469,90 @@ DEFINE_BUF_ITEM_EVENT(xfs_trans_bhold);
450DEFINE_BUF_ITEM_EVENT(xfs_trans_bhold_release); 469DEFINE_BUF_ITEM_EVENT(xfs_trans_bhold_release);
451DEFINE_BUF_ITEM_EVENT(xfs_trans_binval); 470DEFINE_BUF_ITEM_EVENT(xfs_trans_binval);
452 471
472DECLARE_EVENT_CLASS(xfs_lock_class,
473 TP_PROTO(struct xfs_inode *ip, unsigned lock_flags,
474 unsigned long caller_ip),
475 TP_ARGS(ip, lock_flags, caller_ip),
476 TP_STRUCT__entry(
477 __field(dev_t, dev)
478 __field(xfs_ino_t, ino)
479 __field(int, lock_flags)
480 __field(unsigned long, caller_ip)
481 ),
482 TP_fast_assign(
483 __entry->dev = VFS_I(ip)->i_sb->s_dev;
484 __entry->ino = ip->i_ino;
485 __entry->lock_flags = lock_flags;
486 __entry->caller_ip = caller_ip;
487 ),
488 TP_printk("dev %d:%d ino 0x%llx flags %s caller %pf",
489 MAJOR(__entry->dev), MINOR(__entry->dev),
490 __entry->ino,
491 __print_flags(__entry->lock_flags, "|", XFS_LOCK_FLAGS),
492 (void *)__entry->caller_ip)
493)
494
453#define DEFINE_LOCK_EVENT(name) \ 495#define DEFINE_LOCK_EVENT(name) \
454TRACE_EVENT(name, \ 496DEFINE_EVENT(xfs_lock_class, name, \
455 TP_PROTO(struct xfs_inode *ip, unsigned lock_flags, \ 497 TP_PROTO(struct xfs_inode *ip, unsigned lock_flags, \
456 unsigned long caller_ip), \ 498 unsigned long caller_ip), \
457 TP_ARGS(ip, lock_flags, caller_ip), \ 499 TP_ARGS(ip, lock_flags, caller_ip))
458 TP_STRUCT__entry( \
459 __field(dev_t, dev) \
460 __field(xfs_ino_t, ino) \
461 __field(int, lock_flags) \
462 __field(unsigned long, caller_ip) \
463 ), \
464 TP_fast_assign( \
465 __entry->dev = VFS_I(ip)->i_sb->s_dev; \
466 __entry->ino = ip->i_ino; \
467 __entry->lock_flags = lock_flags; \
468 __entry->caller_ip = caller_ip; \
469 ), \
470 TP_printk("dev %d:%d ino 0x%llx flags %s caller %pf", \
471 MAJOR(__entry->dev), MINOR(__entry->dev), \
472 __entry->ino, \
473 __print_flags(__entry->lock_flags, "|", XFS_LOCK_FLAGS), \
474 (void *)__entry->caller_ip) \
475)
476
477DEFINE_LOCK_EVENT(xfs_ilock); 500DEFINE_LOCK_EVENT(xfs_ilock);
478DEFINE_LOCK_EVENT(xfs_ilock_nowait); 501DEFINE_LOCK_EVENT(xfs_ilock_nowait);
479DEFINE_LOCK_EVENT(xfs_ilock_demote); 502DEFINE_LOCK_EVENT(xfs_ilock_demote);
480DEFINE_LOCK_EVENT(xfs_iunlock); 503DEFINE_LOCK_EVENT(xfs_iunlock);
481 504
505DECLARE_EVENT_CLASS(xfs_iget_class,
506 TP_PROTO(struct xfs_inode *ip),
507 TP_ARGS(ip),
508 TP_STRUCT__entry(
509 __field(dev_t, dev)
510 __field(xfs_ino_t, ino)
511 ),
512 TP_fast_assign(
513 __entry->dev = VFS_I(ip)->i_sb->s_dev;
514 __entry->ino = ip->i_ino;
515 ),
516 TP_printk("dev %d:%d ino 0x%llx",
517 MAJOR(__entry->dev), MINOR(__entry->dev),
518 __entry->ino)
519)
520
482#define DEFINE_IGET_EVENT(name) \ 521#define DEFINE_IGET_EVENT(name) \
483TRACE_EVENT(name, \ 522DEFINE_EVENT(xfs_iget_class, name, \
484 TP_PROTO(struct xfs_inode *ip), \ 523 TP_PROTO(struct xfs_inode *ip), \
485 TP_ARGS(ip), \ 524 TP_ARGS(ip))
486 TP_STRUCT__entry( \
487 __field(dev_t, dev) \
488 __field(xfs_ino_t, ino) \
489 ), \
490 TP_fast_assign( \
491 __entry->dev = VFS_I(ip)->i_sb->s_dev; \
492 __entry->ino = ip->i_ino; \
493 ), \
494 TP_printk("dev %d:%d ino 0x%llx", \
495 MAJOR(__entry->dev), MINOR(__entry->dev), \
496 __entry->ino) \
497)
498DEFINE_IGET_EVENT(xfs_iget_skip); 525DEFINE_IGET_EVENT(xfs_iget_skip);
499DEFINE_IGET_EVENT(xfs_iget_reclaim); 526DEFINE_IGET_EVENT(xfs_iget_reclaim);
500DEFINE_IGET_EVENT(xfs_iget_found); 527DEFINE_IGET_EVENT(xfs_iget_found);
501DEFINE_IGET_EVENT(xfs_iget_alloc); 528DEFINE_IGET_EVENT(xfs_iget_alloc);
502 529
530DECLARE_EVENT_CLASS(xfs_inode_class,
531 TP_PROTO(struct xfs_inode *ip, unsigned long caller_ip),
532 TP_ARGS(ip, caller_ip),
533 TP_STRUCT__entry(
534 __field(dev_t, dev)
535 __field(xfs_ino_t, ino)
536 __field(int, count)
537 __field(unsigned long, caller_ip)
538 ),
539 TP_fast_assign(
540 __entry->dev = VFS_I(ip)->i_sb->s_dev;
541 __entry->ino = ip->i_ino;
542 __entry->count = atomic_read(&VFS_I(ip)->i_count);
543 __entry->caller_ip = caller_ip;
544 ),
545 TP_printk("dev %d:%d ino 0x%llx count %d caller %pf",
546 MAJOR(__entry->dev), MINOR(__entry->dev),
547 __entry->ino,
548 __entry->count,
549 (char *)__entry->caller_ip)
550)
551
503#define DEFINE_INODE_EVENT(name) \ 552#define DEFINE_INODE_EVENT(name) \
504TRACE_EVENT(name, \ 553DEFINE_EVENT(xfs_inode_class, name, \
505 TP_PROTO(struct xfs_inode *ip, unsigned long caller_ip), \ 554 TP_PROTO(struct xfs_inode *ip, unsigned long caller_ip), \
506 TP_ARGS(ip, caller_ip), \ 555 TP_ARGS(ip, caller_ip))
507 TP_STRUCT__entry( \
508 __field(dev_t, dev) \
509 __field(xfs_ino_t, ino) \
510 __field(int, count) \
511 __field(unsigned long, caller_ip) \
512 ), \
513 TP_fast_assign( \
514 __entry->dev = VFS_I(ip)->i_sb->s_dev; \
515 __entry->ino = ip->i_ino; \
516 __entry->count = atomic_read(&VFS_I(ip)->i_count); \
517 __entry->caller_ip = caller_ip; \
518 ), \
519 TP_printk("dev %d:%d ino 0x%llx count %d caller %pf", \
520 MAJOR(__entry->dev), MINOR(__entry->dev), \
521 __entry->ino, \
522 __entry->count, \
523 (char *)__entry->caller_ip) \
524)
525DEFINE_INODE_EVENT(xfs_ihold); 556DEFINE_INODE_EVENT(xfs_ihold);
526DEFINE_INODE_EVENT(xfs_irele); 557DEFINE_INODE_EVENT(xfs_irele);
527/* the old xfs_itrace_entry tracer - to be replaced by s.th. in the VFS */ 558/* the old xfs_itrace_entry tracer - to be replaced by s.th. in the VFS */
@@ -529,55 +560,59 @@ DEFINE_INODE_EVENT(xfs_inode);
529#define xfs_itrace_entry(ip) \ 560#define xfs_itrace_entry(ip) \
530 trace_xfs_inode(ip, _THIS_IP_) 561 trace_xfs_inode(ip, _THIS_IP_)
531 562
532#define DEFINE_DQUOT_EVENT(tname) \ 563DECLARE_EVENT_CLASS(xfs_dquot_class,
533TRACE_EVENT(tname, \ 564 TP_PROTO(struct xfs_dquot *dqp),
534 TP_PROTO(struct xfs_dquot *dqp), \ 565 TP_ARGS(dqp),
535 TP_ARGS(dqp), \ 566 TP_STRUCT__entry(
536 TP_STRUCT__entry( \ 567 __field(dev_t, dev)
537 __field(dev_t, dev) \ 568 __field(__be32, id)
538 __field(__be32, id) \ 569 __field(unsigned, flags)
539 __field(unsigned, flags) \ 570 __field(unsigned, nrefs)
540 __field(unsigned, nrefs) \ 571 __field(unsigned long long, res_bcount)
541 __field(unsigned long long, res_bcount) \ 572 __field(unsigned long long, bcount)
542 __field(unsigned long long, bcount) \ 573 __field(unsigned long long, icount)
543 __field(unsigned long long, icount) \ 574 __field(unsigned long long, blk_hardlimit)
544 __field(unsigned long long, blk_hardlimit) \ 575 __field(unsigned long long, blk_softlimit)
545 __field(unsigned long long, blk_softlimit) \ 576 __field(unsigned long long, ino_hardlimit)
546 __field(unsigned long long, ino_hardlimit) \ 577 __field(unsigned long long, ino_softlimit)
547 __field(unsigned long long, ino_softlimit) \
548 ), \
549 TP_fast_assign( \
550 __entry->dev = dqp->q_mount->m_super->s_dev; \
551 __entry->id = dqp->q_core.d_id; \
552 __entry->flags = dqp->dq_flags; \
553 __entry->nrefs = dqp->q_nrefs; \
554 __entry->res_bcount = dqp->q_res_bcount; \
555 __entry->bcount = be64_to_cpu(dqp->q_core.d_bcount); \
556 __entry->icount = be64_to_cpu(dqp->q_core.d_icount); \
557 __entry->blk_hardlimit = \
558 be64_to_cpu(dqp->q_core.d_blk_hardlimit); \
559 __entry->blk_softlimit = \
560 be64_to_cpu(dqp->q_core.d_blk_softlimit); \
561 __entry->ino_hardlimit = \
562 be64_to_cpu(dqp->q_core.d_ino_hardlimit); \
563 __entry->ino_softlimit = \
564 be64_to_cpu(dqp->q_core.d_ino_softlimit); \
565 ), \ 578 ), \
566 TP_printk("dev %d:%d id 0x%x flags %s nrefs %u res_bc 0x%llx " \ 579 TP_fast_assign(
567 "bcnt 0x%llx [hard 0x%llx | soft 0x%llx] " \ 580 __entry->dev = dqp->q_mount->m_super->s_dev;
568 "icnt 0x%llx [hard 0x%llx | soft 0x%llx]", \ 581 __entry->id = dqp->q_core.d_id;
569 MAJOR(__entry->dev), MINOR(__entry->dev), \ 582 __entry->flags = dqp->dq_flags;
570 be32_to_cpu(__entry->id), \ 583 __entry->nrefs = dqp->q_nrefs;
571 __print_flags(__entry->flags, "|", XFS_DQ_FLAGS), \ 584 __entry->res_bcount = dqp->q_res_bcount;
572 __entry->nrefs, \ 585 __entry->bcount = be64_to_cpu(dqp->q_core.d_bcount);
573 __entry->res_bcount, \ 586 __entry->icount = be64_to_cpu(dqp->q_core.d_icount);
574 __entry->bcount, \ 587 __entry->blk_hardlimit =
575 __entry->blk_hardlimit, \ 588 be64_to_cpu(dqp->q_core.d_blk_hardlimit);
576 __entry->blk_softlimit, \ 589 __entry->blk_softlimit =
577 __entry->icount, \ 590 be64_to_cpu(dqp->q_core.d_blk_softlimit);
578 __entry->ino_hardlimit, \ 591 __entry->ino_hardlimit =
579 __entry->ino_softlimit) \ 592 be64_to_cpu(dqp->q_core.d_ino_hardlimit);
593 __entry->ino_softlimit =
594 be64_to_cpu(dqp->q_core.d_ino_softlimit);
595 ),
596 TP_printk("dev %d:%d id 0x%x flags %s nrefs %u res_bc 0x%llx "
597 "bcnt 0x%llx [hard 0x%llx | soft 0x%llx] "
598 "icnt 0x%llx [hard 0x%llx | soft 0x%llx]",
599 MAJOR(__entry->dev), MINOR(__entry->dev),
600 be32_to_cpu(__entry->id),
601 __print_flags(__entry->flags, "|", XFS_DQ_FLAGS),
602 __entry->nrefs,
603 __entry->res_bcount,
604 __entry->bcount,
605 __entry->blk_hardlimit,
606 __entry->blk_softlimit,
607 __entry->icount,
608 __entry->ino_hardlimit,
609 __entry->ino_softlimit)
580) 610)
611
612#define DEFINE_DQUOT_EVENT(name) \
613DEFINE_EVENT(xfs_dquot_class, name, \
614 TP_PROTO(struct xfs_dquot *dqp), \
615 TP_ARGS(dqp))
581DEFINE_DQUOT_EVENT(xfs_dqadjust); 616DEFINE_DQUOT_EVENT(xfs_dqadjust);
582DEFINE_DQUOT_EVENT(xfs_dqshake_dirty); 617DEFINE_DQUOT_EVENT(xfs_dqshake_dirty);
583DEFINE_DQUOT_EVENT(xfs_dqshake_unlink); 618DEFINE_DQUOT_EVENT(xfs_dqshake_unlink);
@@ -610,72 +645,75 @@ DEFINE_DQUOT_EVENT(xfs_dqflush_done);
610DEFINE_IGET_EVENT(xfs_dquot_dqalloc); 645DEFINE_IGET_EVENT(xfs_dquot_dqalloc);
611DEFINE_IGET_EVENT(xfs_dquot_dqdetach); 646DEFINE_IGET_EVENT(xfs_dquot_dqdetach);
612 647
648DECLARE_EVENT_CLASS(xfs_loggrant_class,
649 TP_PROTO(struct log *log, struct xlog_ticket *tic),
650 TP_ARGS(log, tic),
651 TP_STRUCT__entry(
652 __field(dev_t, dev)
653 __field(unsigned, trans_type)
654 __field(char, ocnt)
655 __field(char, cnt)
656 __field(int, curr_res)
657 __field(int, unit_res)
658 __field(unsigned int, flags)
659 __field(void *, reserve_headq)
660 __field(void *, write_headq)
661 __field(int, grant_reserve_cycle)
662 __field(int, grant_reserve_bytes)
663 __field(int, grant_write_cycle)
664 __field(int, grant_write_bytes)
665 __field(int, curr_cycle)
666 __field(int, curr_block)
667 __field(xfs_lsn_t, tail_lsn)
668 ),
669 TP_fast_assign(
670 __entry->dev = log->l_mp->m_super->s_dev;
671 __entry->trans_type = tic->t_trans_type;
672 __entry->ocnt = tic->t_ocnt;
673 __entry->cnt = tic->t_cnt;
674 __entry->curr_res = tic->t_curr_res;
675 __entry->unit_res = tic->t_unit_res;
676 __entry->flags = tic->t_flags;
677 __entry->reserve_headq = log->l_reserve_headq;
678 __entry->write_headq = log->l_write_headq;
679 __entry->grant_reserve_cycle = log->l_grant_reserve_cycle;
680 __entry->grant_reserve_bytes = log->l_grant_reserve_bytes;
681 __entry->grant_write_cycle = log->l_grant_write_cycle;
682 __entry->grant_write_bytes = log->l_grant_write_bytes;
683 __entry->curr_cycle = log->l_curr_cycle;
684 __entry->curr_block = log->l_curr_block;
685 __entry->tail_lsn = log->l_tail_lsn;
686 ),
687 TP_printk("dev %d:%d type %s t_ocnt %u t_cnt %u t_curr_res %u "
688 "t_unit_res %u t_flags %s reserve_headq 0x%p "
689 "write_headq 0x%p grant_reserve_cycle %d "
690 "grant_reserve_bytes %d grant_write_cycle %d "
691 "grant_write_bytes %d curr_cycle %d curr_block %d "
692 "tail_cycle %d tail_block %d",
693 MAJOR(__entry->dev), MINOR(__entry->dev),
694 __print_symbolic(__entry->trans_type, XFS_TRANS_TYPES),
695 __entry->ocnt,
696 __entry->cnt,
697 __entry->curr_res,
698 __entry->unit_res,
699 __print_flags(__entry->flags, "|", XLOG_TIC_FLAGS),
700 __entry->reserve_headq,
701 __entry->write_headq,
702 __entry->grant_reserve_cycle,
703 __entry->grant_reserve_bytes,
704 __entry->grant_write_cycle,
705 __entry->grant_write_bytes,
706 __entry->curr_cycle,
707 __entry->curr_block,
708 CYCLE_LSN(__entry->tail_lsn),
709 BLOCK_LSN(__entry->tail_lsn)
710 )
711)
613 712
614#define DEFINE_LOGGRANT_EVENT(tname) \ 713#define DEFINE_LOGGRANT_EVENT(name) \
615TRACE_EVENT(tname, \ 714DEFINE_EVENT(xfs_loggrant_class, name, \
616 TP_PROTO(struct log *log, struct xlog_ticket *tic), \ 715 TP_PROTO(struct log *log, struct xlog_ticket *tic), \
617 TP_ARGS(log, tic), \ 716 TP_ARGS(log, tic))
618 TP_STRUCT__entry( \
619 __field(dev_t, dev) \
620 __field(unsigned, trans_type) \
621 __field(char, ocnt) \
622 __field(char, cnt) \
623 __field(int, curr_res) \
624 __field(int, unit_res) \
625 __field(unsigned int, flags) \
626 __field(void *, reserve_headq) \
627 __field(void *, write_headq) \
628 __field(int, grant_reserve_cycle) \
629 __field(int, grant_reserve_bytes) \
630 __field(int, grant_write_cycle) \
631 __field(int, grant_write_bytes) \
632 __field(int, curr_cycle) \
633 __field(int, curr_block) \
634 __field(xfs_lsn_t, tail_lsn) \
635 ), \
636 TP_fast_assign( \
637 __entry->dev = log->l_mp->m_super->s_dev; \
638 __entry->trans_type = tic->t_trans_type; \
639 __entry->ocnt = tic->t_ocnt; \
640 __entry->cnt = tic->t_cnt; \
641 __entry->curr_res = tic->t_curr_res; \
642 __entry->unit_res = tic->t_unit_res; \
643 __entry->flags = tic->t_flags; \
644 __entry->reserve_headq = log->l_reserve_headq; \
645 __entry->write_headq = log->l_write_headq; \
646 __entry->grant_reserve_cycle = log->l_grant_reserve_cycle; \
647 __entry->grant_reserve_bytes = log->l_grant_reserve_bytes; \
648 __entry->grant_write_cycle = log->l_grant_write_cycle; \
649 __entry->grant_write_bytes = log->l_grant_write_bytes; \
650 __entry->curr_cycle = log->l_curr_cycle; \
651 __entry->curr_block = log->l_curr_block; \
652 __entry->tail_lsn = log->l_tail_lsn; \
653 ), \
654 TP_printk("dev %d:%d type %s t_ocnt %u t_cnt %u t_curr_res %u " \
655 "t_unit_res %u t_flags %s reserve_headq 0x%p " \
656 "write_headq 0x%p grant_reserve_cycle %d " \
657 "grant_reserve_bytes %d grant_write_cycle %d " \
658 "grant_write_bytes %d curr_cycle %d curr_block %d " \
659 "tail_cycle %d tail_block %d", \
660 MAJOR(__entry->dev), MINOR(__entry->dev), \
661 __print_symbolic(__entry->trans_type, XFS_TRANS_TYPES), \
662 __entry->ocnt, \
663 __entry->cnt, \
664 __entry->curr_res, \
665 __entry->unit_res, \
666 __print_flags(__entry->flags, "|", XLOG_TIC_FLAGS), \
667 __entry->reserve_headq, \
668 __entry->write_headq, \
669 __entry->grant_reserve_cycle, \
670 __entry->grant_reserve_bytes, \
671 __entry->grant_write_cycle, \
672 __entry->grant_write_bytes, \
673 __entry->curr_cycle, \
674 __entry->curr_block, \
675 CYCLE_LSN(__entry->tail_lsn), \
676 BLOCK_LSN(__entry->tail_lsn) \
677 ) \
678)
679DEFINE_LOGGRANT_EVENT(xfs_log_done_nonperm); 717DEFINE_LOGGRANT_EVENT(xfs_log_done_nonperm);
680DEFINE_LOGGRANT_EVENT(xfs_log_done_perm); 718DEFINE_LOGGRANT_EVENT(xfs_log_done_perm);
681DEFINE_LOGGRANT_EVENT(xfs_log_reserve); 719DEFINE_LOGGRANT_EVENT(xfs_log_reserve);
@@ -815,7 +853,7 @@ TRACE_EVENT(name, \
815 ), \ 853 ), \
816 TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx " \ 854 TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx " \
817 "offset 0x%llx count %zd flags %s " \ 855 "offset 0x%llx count %zd flags %s " \
818 "startoff 0x%llx startblock 0x%llx blockcount 0x%llx", \ 856 "startoff 0x%llx startblock %s blockcount 0x%llx", \
819 MAJOR(__entry->dev), MINOR(__entry->dev), \ 857 MAJOR(__entry->dev), MINOR(__entry->dev), \
820 __entry->ino, \ 858 __entry->ino, \
821 __entry->size, \ 859 __entry->size, \
@@ -824,7 +862,7 @@ TRACE_EVENT(name, \
824 __entry->count, \ 862 __entry->count, \
825 __print_flags(__entry->flags, "|", BMAPI_FLAGS), \ 863 __print_flags(__entry->flags, "|", BMAPI_FLAGS), \
826 __entry->startoff, \ 864 __entry->startoff, \
827 __entry->startblock, \ 865 xfs_fmtfsblock(__entry->startblock), \
828 __entry->blockcount) \ 866 __entry->blockcount) \
829) 867)
830DEFINE_IOMAP_EVENT(xfs_iomap_enter); 868DEFINE_IOMAP_EVENT(xfs_iomap_enter);
@@ -897,28 +935,32 @@ TRACE_EVENT(xfs_itruncate_start,
897 __entry->toss_finish) 935 __entry->toss_finish)
898); 936);
899 937
938DECLARE_EVENT_CLASS(xfs_itrunc_class,
939 TP_PROTO(struct xfs_inode *ip, xfs_fsize_t new_size),
940 TP_ARGS(ip, new_size),
941 TP_STRUCT__entry(
942 __field(dev_t, dev)
943 __field(xfs_ino_t, ino)
944 __field(xfs_fsize_t, size)
945 __field(xfs_fsize_t, new_size)
946 ),
947 TP_fast_assign(
948 __entry->dev = VFS_I(ip)->i_sb->s_dev;
949 __entry->ino = ip->i_ino;
950 __entry->size = ip->i_d.di_size;
951 __entry->new_size = new_size;
952 ),
953 TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx",
954 MAJOR(__entry->dev), MINOR(__entry->dev),
955 __entry->ino,
956 __entry->size,
957 __entry->new_size)
958)
959
900#define DEFINE_ITRUNC_EVENT(name) \ 960#define DEFINE_ITRUNC_EVENT(name) \
901TRACE_EVENT(name, \ 961DEFINE_EVENT(xfs_itrunc_class, name, \
902 TP_PROTO(struct xfs_inode *ip, xfs_fsize_t new_size), \ 962 TP_PROTO(struct xfs_inode *ip, xfs_fsize_t new_size), \
903 TP_ARGS(ip, new_size), \ 963 TP_ARGS(ip, new_size))
904 TP_STRUCT__entry( \
905 __field(dev_t, dev) \
906 __field(xfs_ino_t, ino) \
907 __field(xfs_fsize_t, size) \
908 __field(xfs_fsize_t, new_size) \
909 ), \
910 TP_fast_assign( \
911 __entry->dev = VFS_I(ip)->i_sb->s_dev; \
912 __entry->ino = ip->i_ino; \
913 __entry->size = ip->i_d.di_size; \
914 __entry->new_size = new_size; \
915 ), \
916 TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx", \
917 MAJOR(__entry->dev), MINOR(__entry->dev), \
918 __entry->ino, \
919 __entry->size, \
920 __entry->new_size) \
921)
922DEFINE_ITRUNC_EVENT(xfs_itruncate_finish_start); 964DEFINE_ITRUNC_EVENT(xfs_itruncate_finish_start);
923DEFINE_ITRUNC_EVENT(xfs_itruncate_finish_end); 965DEFINE_ITRUNC_EVENT(xfs_itruncate_finish_end);
924 966
@@ -1037,28 +1079,28 @@ TRACE_EVENT(xfs_alloc_unbusy,
1037 1079
1038TRACE_EVENT(xfs_alloc_busysearch, 1080TRACE_EVENT(xfs_alloc_busysearch,
1039 TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, xfs_agblock_t agbno, 1081 TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, xfs_agblock_t agbno,
1040 xfs_extlen_t len, int found), 1082 xfs_extlen_t len, xfs_lsn_t lsn),
1041 TP_ARGS(mp, agno, agbno, len, found), 1083 TP_ARGS(mp, agno, agbno, len, lsn),
1042 TP_STRUCT__entry( 1084 TP_STRUCT__entry(
1043 __field(dev_t, dev) 1085 __field(dev_t, dev)
1044 __field(xfs_agnumber_t, agno) 1086 __field(xfs_agnumber_t, agno)
1045 __field(xfs_agblock_t, agbno) 1087 __field(xfs_agblock_t, agbno)
1046 __field(xfs_extlen_t, len) 1088 __field(xfs_extlen_t, len)
1047 __field(int, found) 1089 __field(xfs_lsn_t, lsn)
1048 ), 1090 ),
1049 TP_fast_assign( 1091 TP_fast_assign(
1050 __entry->dev = mp->m_super->s_dev; 1092 __entry->dev = mp->m_super->s_dev;
1051 __entry->agno = agno; 1093 __entry->agno = agno;
1052 __entry->agbno = agbno; 1094 __entry->agbno = agbno;
1053 __entry->len = len; 1095 __entry->len = len;
1054 __entry->found = found; 1096 __entry->lsn = lsn;
1055 ), 1097 ),
1056 TP_printk("dev %d:%d agno %u agbno %u len %u %s", 1098 TP_printk("dev %d:%d agno %u agbno %u len %u force lsn 0x%llx",
1057 MAJOR(__entry->dev), MINOR(__entry->dev), 1099 MAJOR(__entry->dev), MINOR(__entry->dev),
1058 __entry->agno, 1100 __entry->agno,
1059 __entry->agbno, 1101 __entry->agbno,
1060 __entry->len, 1102 __entry->len,
1061 __print_symbolic(__entry->found, XFS_BUSY_STATES)) 1103 __entry->lsn)
1062); 1104);
1063 1105
1064TRACE_EVENT(xfs_agf, 1106TRACE_EVENT(xfs_agf,
@@ -1152,77 +1194,80 @@ TRACE_EVENT(xfs_free_extent,
1152 1194
1153); 1195);
1154 1196
1155#define DEFINE_ALLOC_EVENT(name) \ 1197DECLARE_EVENT_CLASS(xfs_alloc_class,
1156TRACE_EVENT(name, \ 1198 TP_PROTO(struct xfs_alloc_arg *args),
1157 TP_PROTO(struct xfs_alloc_arg *args), \ 1199 TP_ARGS(args),
1158 TP_ARGS(args), \ 1200 TP_STRUCT__entry(
1159 TP_STRUCT__entry( \ 1201 __field(dev_t, dev)
1160 __field(dev_t, dev) \ 1202 __field(xfs_agnumber_t, agno)
1161 __field(xfs_agnumber_t, agno) \ 1203 __field(xfs_agblock_t, agbno)
1162 __field(xfs_agblock_t, agbno) \ 1204 __field(xfs_extlen_t, minlen)
1163 __field(xfs_extlen_t, minlen) \ 1205 __field(xfs_extlen_t, maxlen)
1164 __field(xfs_extlen_t, maxlen) \ 1206 __field(xfs_extlen_t, mod)
1165 __field(xfs_extlen_t, mod) \ 1207 __field(xfs_extlen_t, prod)
1166 __field(xfs_extlen_t, prod) \ 1208 __field(xfs_extlen_t, minleft)
1167 __field(xfs_extlen_t, minleft) \ 1209 __field(xfs_extlen_t, total)
1168 __field(xfs_extlen_t, total) \ 1210 __field(xfs_extlen_t, alignment)
1169 __field(xfs_extlen_t, alignment) \ 1211 __field(xfs_extlen_t, minalignslop)
1170 __field(xfs_extlen_t, minalignslop) \ 1212 __field(xfs_extlen_t, len)
1171 __field(xfs_extlen_t, len) \ 1213 __field(short, type)
1172 __field(short, type) \ 1214 __field(short, otype)
1173 __field(short, otype) \ 1215 __field(char, wasdel)
1174 __field(char, wasdel) \ 1216 __field(char, wasfromfl)
1175 __field(char, wasfromfl) \ 1217 __field(char, isfl)
1176 __field(char, isfl) \ 1218 __field(char, userdata)
1177 __field(char, userdata) \ 1219 __field(xfs_fsblock_t, firstblock)
1178 __field(xfs_fsblock_t, firstblock) \ 1220 ),
1179 ), \ 1221 TP_fast_assign(
1180 TP_fast_assign( \ 1222 __entry->dev = args->mp->m_super->s_dev;
1181 __entry->dev = args->mp->m_super->s_dev; \ 1223 __entry->agno = args->agno;
1182 __entry->agno = args->agno; \ 1224 __entry->agbno = args->agbno;
1183 __entry->agbno = args->agbno; \ 1225 __entry->minlen = args->minlen;
1184 __entry->minlen = args->minlen; \ 1226 __entry->maxlen = args->maxlen;
1185 __entry->maxlen = args->maxlen; \ 1227 __entry->mod = args->mod;
1186 __entry->mod = args->mod; \ 1228 __entry->prod = args->prod;
1187 __entry->prod = args->prod; \ 1229 __entry->minleft = args->minleft;
1188 __entry->minleft = args->minleft; \ 1230 __entry->total = args->total;
1189 __entry->total = args->total; \ 1231 __entry->alignment = args->alignment;
1190 __entry->alignment = args->alignment; \ 1232 __entry->minalignslop = args->minalignslop;
1191 __entry->minalignslop = args->minalignslop; \ 1233 __entry->len = args->len;
1192 __entry->len = args->len; \ 1234 __entry->type = args->type;
1193 __entry->type = args->type; \ 1235 __entry->otype = args->otype;
1194 __entry->otype = args->otype; \ 1236 __entry->wasdel = args->wasdel;
1195 __entry->wasdel = args->wasdel; \ 1237 __entry->wasfromfl = args->wasfromfl;
1196 __entry->wasfromfl = args->wasfromfl; \ 1238 __entry->isfl = args->isfl;
1197 __entry->isfl = args->isfl; \ 1239 __entry->userdata = args->userdata;
1198 __entry->userdata = args->userdata; \ 1240 __entry->firstblock = args->firstblock;
1199 __entry->firstblock = args->firstblock; \ 1241 ),
1200 ), \ 1242 TP_printk("dev %d:%d agno %u agbno %u minlen %u maxlen %u mod %u "
1201 TP_printk("dev %d:%d agno %u agbno %u minlen %u maxlen %u mod %u " \ 1243 "prod %u minleft %u total %u alignment %u minalignslop %u "
1202 "prod %u minleft %u total %u alignment %u minalignslop %u " \ 1244 "len %u type %s otype %s wasdel %d wasfromfl %d isfl %d "
1203 "len %u type %s otype %s wasdel %d wasfromfl %d isfl %d " \ 1245 "userdata %d firstblock 0x%llx",
1204 "userdata %d firstblock 0x%llx", \ 1246 MAJOR(__entry->dev), MINOR(__entry->dev),
1205 MAJOR(__entry->dev), MINOR(__entry->dev), \ 1247 __entry->agno,
1206 __entry->agno, \ 1248 __entry->agbno,
1207 __entry->agbno, \ 1249 __entry->minlen,
1208 __entry->minlen, \ 1250 __entry->maxlen,
1209 __entry->maxlen, \ 1251 __entry->mod,
1210 __entry->mod, \ 1252 __entry->prod,
1211 __entry->prod, \ 1253 __entry->minleft,
1212 __entry->minleft, \ 1254 __entry->total,
1213 __entry->total, \ 1255 __entry->alignment,
1214 __entry->alignment, \ 1256 __entry->minalignslop,
1215 __entry->minalignslop, \ 1257 __entry->len,
1216 __entry->len, \ 1258 __print_symbolic(__entry->type, XFS_ALLOC_TYPES),
1217 __print_symbolic(__entry->type, XFS_ALLOC_TYPES), \ 1259 __print_symbolic(__entry->otype, XFS_ALLOC_TYPES),
1218 __print_symbolic(__entry->otype, XFS_ALLOC_TYPES), \ 1260 __entry->wasdel,
1219 __entry->wasdel, \ 1261 __entry->wasfromfl,
1220 __entry->wasfromfl, \ 1262 __entry->isfl,
1221 __entry->isfl, \ 1263 __entry->userdata,
1222 __entry->userdata, \ 1264 __entry->firstblock)
1223 __entry->firstblock) \
1224) 1265)
1225 1266
1267#define DEFINE_ALLOC_EVENT(name) \
1268DEFINE_EVENT(xfs_alloc_class, name, \
1269 TP_PROTO(struct xfs_alloc_arg *args), \
1270 TP_ARGS(args))
1226DEFINE_ALLOC_EVENT(xfs_alloc_exact_done); 1271DEFINE_ALLOC_EVENT(xfs_alloc_exact_done);
1227DEFINE_ALLOC_EVENT(xfs_alloc_exact_error); 1272DEFINE_ALLOC_EVENT(xfs_alloc_exact_error);
1228DEFINE_ALLOC_EVENT(xfs_alloc_near_nominleft); 1273DEFINE_ALLOC_EVENT(xfs_alloc_near_nominleft);
@@ -1245,92 +1290,100 @@ DEFINE_ALLOC_EVENT(xfs_alloc_vextent_noagbp);
1245DEFINE_ALLOC_EVENT(xfs_alloc_vextent_loopfailed); 1290DEFINE_ALLOC_EVENT(xfs_alloc_vextent_loopfailed);
1246DEFINE_ALLOC_EVENT(xfs_alloc_vextent_allfailed); 1291DEFINE_ALLOC_EVENT(xfs_alloc_vextent_allfailed);
1247 1292
1248#define DEFINE_DIR2_TRACE(tname) \ 1293DECLARE_EVENT_CLASS(xfs_dir2_class,
1249TRACE_EVENT(tname, \ 1294 TP_PROTO(struct xfs_da_args *args),
1295 TP_ARGS(args),
1296 TP_STRUCT__entry(
1297 __field(dev_t, dev)
1298 __field(xfs_ino_t, ino)
1299 __dynamic_array(char, name, args->namelen)
1300 __field(int, namelen)
1301 __field(xfs_dahash_t, hashval)
1302 __field(xfs_ino_t, inumber)
1303 __field(int, op_flags)
1304 ),
1305 TP_fast_assign(
1306 __entry->dev = VFS_I(args->dp)->i_sb->s_dev;
1307 __entry->ino = args->dp->i_ino;
1308 if (args->namelen)
1309 memcpy(__get_str(name), args->name, args->namelen);
1310 __entry->namelen = args->namelen;
1311 __entry->hashval = args->hashval;
1312 __entry->inumber = args->inumber;
1313 __entry->op_flags = args->op_flags;
1314 ),
1315 TP_printk("dev %d:%d ino 0x%llx name %.*s namelen %d hashval 0x%x "
1316 "inumber 0x%llx op_flags %s",
1317 MAJOR(__entry->dev), MINOR(__entry->dev),
1318 __entry->ino,
1319 __entry->namelen,
1320 __entry->namelen ? __get_str(name) : NULL,
1321 __entry->namelen,
1322 __entry->hashval,
1323 __entry->inumber,
1324 __print_flags(__entry->op_flags, "|", XFS_DA_OP_FLAGS))
1325)
1326
1327#define DEFINE_DIR2_EVENT(name) \
1328DEFINE_EVENT(xfs_dir2_class, name, \
1250 TP_PROTO(struct xfs_da_args *args), \ 1329 TP_PROTO(struct xfs_da_args *args), \
1251 TP_ARGS(args), \ 1330 TP_ARGS(args))
1252 TP_STRUCT__entry( \ 1331DEFINE_DIR2_EVENT(xfs_dir2_sf_addname);
1253 __field(dev_t, dev) \ 1332DEFINE_DIR2_EVENT(xfs_dir2_sf_create);
1254 __field(xfs_ino_t, ino) \ 1333DEFINE_DIR2_EVENT(xfs_dir2_sf_lookup);
1255 __dynamic_array(char, name, args->namelen) \ 1334DEFINE_DIR2_EVENT(xfs_dir2_sf_replace);
1256 __field(int, namelen) \ 1335DEFINE_DIR2_EVENT(xfs_dir2_sf_removename);
1257 __field(xfs_dahash_t, hashval) \ 1336DEFINE_DIR2_EVENT(xfs_dir2_sf_toino4);
1258 __field(xfs_ino_t, inumber) \ 1337DEFINE_DIR2_EVENT(xfs_dir2_sf_toino8);
1259 __field(int, op_flags) \ 1338DEFINE_DIR2_EVENT(xfs_dir2_sf_to_block);
1260 ), \ 1339DEFINE_DIR2_EVENT(xfs_dir2_block_addname);
1261 TP_fast_assign( \ 1340DEFINE_DIR2_EVENT(xfs_dir2_block_lookup);
1262 __entry->dev = VFS_I(args->dp)->i_sb->s_dev; \ 1341DEFINE_DIR2_EVENT(xfs_dir2_block_replace);
1263 __entry->ino = args->dp->i_ino; \ 1342DEFINE_DIR2_EVENT(xfs_dir2_block_removename);
1264 if (args->namelen) \ 1343DEFINE_DIR2_EVENT(xfs_dir2_block_to_sf);
1265 memcpy(__get_str(name), args->name, args->namelen); \ 1344DEFINE_DIR2_EVENT(xfs_dir2_block_to_leaf);
1266 __entry->namelen = args->namelen; \ 1345DEFINE_DIR2_EVENT(xfs_dir2_leaf_addname);
1267 __entry->hashval = args->hashval; \ 1346DEFINE_DIR2_EVENT(xfs_dir2_leaf_lookup);
1268 __entry->inumber = args->inumber; \ 1347DEFINE_DIR2_EVENT(xfs_dir2_leaf_replace);
1269 __entry->op_flags = args->op_flags; \ 1348DEFINE_DIR2_EVENT(xfs_dir2_leaf_removename);
1270 ), \ 1349DEFINE_DIR2_EVENT(xfs_dir2_leaf_to_block);
1271 TP_printk("dev %d:%d ino 0x%llx name %.*s namelen %d hashval 0x%x " \ 1350DEFINE_DIR2_EVENT(xfs_dir2_leaf_to_node);
1272 "inumber 0x%llx op_flags %s", \ 1351DEFINE_DIR2_EVENT(xfs_dir2_node_addname);
1273 MAJOR(__entry->dev), MINOR(__entry->dev), \ 1352DEFINE_DIR2_EVENT(xfs_dir2_node_lookup);
1274 __entry->ino, \ 1353DEFINE_DIR2_EVENT(xfs_dir2_node_replace);
1275 __entry->namelen, \ 1354DEFINE_DIR2_EVENT(xfs_dir2_node_removename);
1276 __entry->namelen ? __get_str(name) : NULL, \ 1355DEFINE_DIR2_EVENT(xfs_dir2_node_to_leaf);
1277 __entry->namelen, \ 1356
1278 __entry->hashval, \ 1357DECLARE_EVENT_CLASS(xfs_dir2_space_class,
1279 __entry->inumber, \ 1358 TP_PROTO(struct xfs_da_args *args, int idx),
1280 __print_flags(__entry->op_flags, "|", XFS_DA_OP_FLAGS)) \ 1359 TP_ARGS(args, idx),
1360 TP_STRUCT__entry(
1361 __field(dev_t, dev)
1362 __field(xfs_ino_t, ino)
1363 __field(int, op_flags)
1364 __field(int, idx)
1365 ),
1366 TP_fast_assign(
1367 __entry->dev = VFS_I(args->dp)->i_sb->s_dev;
1368 __entry->ino = args->dp->i_ino;
1369 __entry->op_flags = args->op_flags;
1370 __entry->idx = idx;
1371 ),
1372 TP_printk("dev %d:%d ino 0x%llx op_flags %s index %d",
1373 MAJOR(__entry->dev), MINOR(__entry->dev),
1374 __entry->ino,
1375 __print_flags(__entry->op_flags, "|", XFS_DA_OP_FLAGS),
1376 __entry->idx)
1281) 1377)
1282DEFINE_DIR2_TRACE(xfs_dir2_sf_addname);
1283DEFINE_DIR2_TRACE(xfs_dir2_sf_create);
1284DEFINE_DIR2_TRACE(xfs_dir2_sf_lookup);
1285DEFINE_DIR2_TRACE(xfs_dir2_sf_replace);
1286DEFINE_DIR2_TRACE(xfs_dir2_sf_removename);
1287DEFINE_DIR2_TRACE(xfs_dir2_sf_toino4);
1288DEFINE_DIR2_TRACE(xfs_dir2_sf_toino8);
1289DEFINE_DIR2_TRACE(xfs_dir2_sf_to_block);
1290DEFINE_DIR2_TRACE(xfs_dir2_block_addname);
1291DEFINE_DIR2_TRACE(xfs_dir2_block_lookup);
1292DEFINE_DIR2_TRACE(xfs_dir2_block_replace);
1293DEFINE_DIR2_TRACE(xfs_dir2_block_removename);
1294DEFINE_DIR2_TRACE(xfs_dir2_block_to_sf);
1295DEFINE_DIR2_TRACE(xfs_dir2_block_to_leaf);
1296DEFINE_DIR2_TRACE(xfs_dir2_leaf_addname);
1297DEFINE_DIR2_TRACE(xfs_dir2_leaf_lookup);
1298DEFINE_DIR2_TRACE(xfs_dir2_leaf_replace);
1299DEFINE_DIR2_TRACE(xfs_dir2_leaf_removename);
1300DEFINE_DIR2_TRACE(xfs_dir2_leaf_to_block);
1301DEFINE_DIR2_TRACE(xfs_dir2_leaf_to_node);
1302DEFINE_DIR2_TRACE(xfs_dir2_node_addname);
1303DEFINE_DIR2_TRACE(xfs_dir2_node_lookup);
1304DEFINE_DIR2_TRACE(xfs_dir2_node_replace);
1305DEFINE_DIR2_TRACE(xfs_dir2_node_removename);
1306DEFINE_DIR2_TRACE(xfs_dir2_node_to_leaf);
1307 1378
1308#define DEFINE_DIR2_SPACE_TRACE(tname) \ 1379#define DEFINE_DIR2_SPACE_EVENT(name) \
1309TRACE_EVENT(tname, \ 1380DEFINE_EVENT(xfs_dir2_space_class, name, \
1310 TP_PROTO(struct xfs_da_args *args, int idx), \ 1381 TP_PROTO(struct xfs_da_args *args, int idx), \
1311 TP_ARGS(args, idx), \ 1382 TP_ARGS(args, idx))
1312 TP_STRUCT__entry( \ 1383DEFINE_DIR2_SPACE_EVENT(xfs_dir2_leafn_add);
1313 __field(dev_t, dev) \ 1384DEFINE_DIR2_SPACE_EVENT(xfs_dir2_leafn_remove);
1314 __field(xfs_ino_t, ino) \ 1385DEFINE_DIR2_SPACE_EVENT(xfs_dir2_grow_inode);
1315 __field(int, op_flags) \ 1386DEFINE_DIR2_SPACE_EVENT(xfs_dir2_shrink_inode);
1316 __field(int, idx) \
1317 ), \
1318 TP_fast_assign( \
1319 __entry->dev = VFS_I(args->dp)->i_sb->s_dev; \
1320 __entry->ino = args->dp->i_ino; \
1321 __entry->op_flags = args->op_flags; \
1322 __entry->idx = idx; \
1323 ), \
1324 TP_printk("dev %d:%d ino 0x%llx op_flags %s index %d", \
1325 MAJOR(__entry->dev), MINOR(__entry->dev), \
1326 __entry->ino, \
1327 __print_flags(__entry->op_flags, "|", XFS_DA_OP_FLAGS), \
1328 __entry->idx) \
1329)
1330DEFINE_DIR2_SPACE_TRACE(xfs_dir2_leafn_add);
1331DEFINE_DIR2_SPACE_TRACE(xfs_dir2_leafn_remove);
1332DEFINE_DIR2_SPACE_TRACE(xfs_dir2_grow_inode);
1333DEFINE_DIR2_SPACE_TRACE(xfs_dir2_shrink_inode);
1334 1387
1335TRACE_EVENT(xfs_dir2_leafn_moveents, 1388TRACE_EVENT(xfs_dir2_leafn_moveents,
1336 TP_PROTO(struct xfs_da_args *args, int src_idx, int dst_idx, int count), 1389 TP_PROTO(struct xfs_da_args *args, int src_idx, int dst_idx, int count),
diff --git a/fs/xfs/quota/xfs_qm_syscalls.c b/fs/xfs/quota/xfs_qm_syscalls.c
index 71af76fe8a23..873e07e29074 100644
--- a/fs/xfs/quota/xfs_qm_syscalls.c
+++ b/fs/xfs/quota/xfs_qm_syscalls.c
@@ -891,7 +891,7 @@ xfs_qm_dqrele_all_inodes(
891 uint flags) 891 uint flags)
892{ 892{
893 ASSERT(mp->m_quotainfo); 893 ASSERT(mp->m_quotainfo);
894 xfs_inode_ag_iterator(mp, xfs_dqrele_inode, flags, XFS_ICI_NO_TAG); 894 xfs_inode_ag_iterator(mp, xfs_dqrele_inode, flags, XFS_ICI_NO_TAG, 0);
895} 895}
896 896
897/*------------------------------------------------------------------------*/ 897/*------------------------------------------------------------------------*/
diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index a1c65fc6d9c4..275b1f4f9430 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -2563,43 +2563,41 @@ xfs_alloc_search_busy(xfs_trans_t *tp,
2563 xfs_mount_t *mp; 2563 xfs_mount_t *mp;
2564 xfs_perag_busy_t *bsy; 2564 xfs_perag_busy_t *bsy;
2565 xfs_agblock_t uend, bend; 2565 xfs_agblock_t uend, bend;
2566 xfs_lsn_t lsn; 2566 xfs_lsn_t lsn = 0;
2567 int cnt; 2567 int cnt;
2568 2568
2569 mp = tp->t_mountp; 2569 mp = tp->t_mountp;
2570 2570
2571 spin_lock(&mp->m_perag[agno].pagb_lock); 2571 spin_lock(&mp->m_perag[agno].pagb_lock);
2572 cnt = mp->m_perag[agno].pagb_count;
2573 2572
2574 uend = bno + len - 1; 2573 uend = bno + len - 1;
2575 2574
2576 /* search pagb_list for this slot, skipping open slots */ 2575 /*
2577 for (bsy = mp->m_perag[agno].pagb_list; cnt; bsy++) { 2576 * search pagb_list for this slot, skipping open slots. We have to
2577 * search the entire array as there may be multiple overlaps and
2578 * we have to get the most recent LSN for the log force to push out
2579 * all the transactions that span the range.
2580 */
2581 for (cnt = 0; cnt < mp->m_perag[agno].pagb_count; cnt++) {
2582 bsy = &mp->m_perag[agno].pagb_list[cnt];
2583 if (!bsy->busy_tp)
2584 continue;
2578 2585
2579 /* 2586 bend = bsy->busy_start + bsy->busy_length - 1;
2580 * (start1,length1) within (start2, length2) 2587 if (bno > bend || uend < bsy->busy_start)
2581 */ 2588 continue;
2582 if (bsy->busy_tp != NULL) {
2583 bend = bsy->busy_start + bsy->busy_length - 1;
2584 if ((bno > bend) || (uend < bsy->busy_start)) {
2585 cnt--;
2586 } else {
2587 break;
2588 }
2589 }
2590 }
2591 2589
2592 trace_xfs_alloc_busysearch(mp, agno, bno, len, !!cnt); 2590 /* (start1,length1) within (start2, length2) */
2591 if (XFS_LSN_CMP(bsy->busy_tp->t_commit_lsn, lsn) > 0)
2592 lsn = bsy->busy_tp->t_commit_lsn;
2593 }
2594 spin_unlock(&mp->m_perag[agno].pagb_lock);
2595 trace_xfs_alloc_busysearch(tp->t_mountp, agno, bno, len, lsn);
2593 2596
2594 /* 2597 /*
2595 * If a block was found, force the log through the LSN of the 2598 * If a block was found, force the log through the LSN of the
2596 * transaction that freed the block 2599 * transaction that freed the block
2597 */ 2600 */
2598 if (cnt) { 2601 if (lsn)
2599 lsn = bsy->busy_tp->t_commit_lsn;
2600 spin_unlock(&mp->m_perag[agno].pagb_lock);
2601 xfs_log_force(mp, lsn, XFS_LOG_FORCE|XFS_LOG_SYNC); 2602 xfs_log_force(mp, lsn, XFS_LOG_FORCE|XFS_LOG_SYNC);
2602 } else {
2603 spin_unlock(&mp->m_perag[agno].pagb_lock);
2604 }
2605} 2603}
diff --git a/fs/xfs/xfs_dfrag.c b/fs/xfs/xfs_dfrag.c
index d1483a4f71b8..84ca1cf16a1e 100644
--- a/fs/xfs/xfs_dfrag.c
+++ b/fs/xfs/xfs_dfrag.c
@@ -114,10 +114,82 @@ xfs_swapext(
114 return error; 114 return error;
115} 115}
116 116
117/*
118 * We need to check that the format of the data fork in the temporary inode is
119 * valid for the target inode before doing the swap. This is not a problem with
120 * attr1 because of the fixed fork offset, but attr2 has a dynamically sized
121 * data fork depending on the space the attribute fork is taking so we can get
122 * invalid formats on the target inode.
123 *
124 * E.g. target has space for 7 extents in extent format, temp inode only has
125 * space for 6. If we defragment down to 7 extents, then the tmp format is a
126 * btree, but when swapped it needs to be in extent format. Hence we can't just
127 * blindly swap data forks on attr2 filesystems.
128 *
129 * Note that we check the swap in both directions so that we don't end up with
130 * a corrupt temporary inode, either.
131 *
132 * Note that fixing the way xfs_fsr sets up the attribute fork in the source
133 * inode will prevent this situation from occurring, so all we do here is
134 * reject and log the attempt. basically we are putting the responsibility on
135 * userspace to get this right.
136 */
137static int
138xfs_swap_extents_check_format(
139 xfs_inode_t *ip, /* target inode */
140 xfs_inode_t *tip) /* tmp inode */
141{
142
143 /* Should never get a local format */
144 if (ip->i_d.di_format == XFS_DINODE_FMT_LOCAL ||
145 tip->i_d.di_format == XFS_DINODE_FMT_LOCAL)
146 return EINVAL;
147
148 /*
149 * if the target inode has less extents that then temporary inode then
150 * why did userspace call us?
151 */
152 if (ip->i_d.di_nextents < tip->i_d.di_nextents)
153 return EINVAL;
154
155 /*
156 * if the target inode is in extent form and the temp inode is in btree
157 * form then we will end up with the target inode in the wrong format
158 * as we already know there are less extents in the temp inode.
159 */
160 if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
161 tip->i_d.di_format == XFS_DINODE_FMT_BTREE)
162 return EINVAL;
163
164 /* Check temp in extent form to max in target */
165 if (tip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
166 XFS_IFORK_NEXTENTS(tip, XFS_DATA_FORK) > ip->i_df.if_ext_max)
167 return EINVAL;
168
169 /* Check target in extent form to max in temp */
170 if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
171 XFS_IFORK_NEXTENTS(ip, XFS_DATA_FORK) > tip->i_df.if_ext_max)
172 return EINVAL;
173
174 /* Check root block of temp in btree form to max in target */
175 if (tip->i_d.di_format == XFS_DINODE_FMT_BTREE &&
176 XFS_IFORK_BOFF(ip) &&
177 tip->i_df.if_broot_bytes > XFS_IFORK_BOFF(ip))
178 return EINVAL;
179
180 /* Check root block of target in btree form to max in temp */
181 if (ip->i_d.di_format == XFS_DINODE_FMT_BTREE &&
182 XFS_IFORK_BOFF(tip) &&
183 ip->i_df.if_broot_bytes > XFS_IFORK_BOFF(tip))
184 return EINVAL;
185
186 return 0;
187}
188
117int 189int
118xfs_swap_extents( 190xfs_swap_extents(
119 xfs_inode_t *ip, 191 xfs_inode_t *ip, /* target inode */
120 xfs_inode_t *tip, 192 xfs_inode_t *tip, /* tmp inode */
121 xfs_swapext_t *sxp) 193 xfs_swapext_t *sxp)
122{ 194{
123 xfs_mount_t *mp; 195 xfs_mount_t *mp;
@@ -161,13 +233,6 @@ xfs_swap_extents(
161 goto out_unlock; 233 goto out_unlock;
162 } 234 }
163 235
164 /* Should never get a local format */
165 if (ip->i_d.di_format == XFS_DINODE_FMT_LOCAL ||
166 tip->i_d.di_format == XFS_DINODE_FMT_LOCAL) {
167 error = XFS_ERROR(EINVAL);
168 goto out_unlock;
169 }
170
171 if (VN_CACHED(VFS_I(tip)) != 0) { 236 if (VN_CACHED(VFS_I(tip)) != 0) {
172 error = xfs_flushinval_pages(tip, 0, -1, 237 error = xfs_flushinval_pages(tip, 0, -1,
173 FI_REMAPF_LOCKED); 238 FI_REMAPF_LOCKED);
@@ -189,13 +254,12 @@ xfs_swap_extents(
189 goto out_unlock; 254 goto out_unlock;
190 } 255 }
191 256
192 /* 257 /* check inode formats now that data is flushed */
193 * If the target has extended attributes, the tmp file 258 error = xfs_swap_extents_check_format(ip, tip);
194 * must also in order to ensure the correct data fork 259 if (error) {
195 * format. 260 xfs_fs_cmn_err(CE_NOTE, mp,
196 */ 261 "%s: inode 0x%llx format is incompatible for exchanging.",
197 if ( XFS_IFORK_Q(ip) != XFS_IFORK_Q(tip) ) { 262 __FILE__, ip->i_ino);
198 error = XFS_ERROR(EINVAL);
199 goto out_unlock; 263 goto out_unlock;
200 } 264 }
201 265
@@ -276,6 +340,16 @@ xfs_swap_extents(
276 *tifp = *tempifp; /* struct copy */ 340 *tifp = *tempifp; /* struct copy */
277 341
278 /* 342 /*
343 * Fix the in-memory data fork values that are dependent on the fork
344 * offset in the inode. We can't assume they remain the same as attr2
345 * has dynamic fork offsets.
346 */
347 ifp->if_ext_max = XFS_IFORK_SIZE(ip, XFS_DATA_FORK) /
348 (uint)sizeof(xfs_bmbt_rec_t);
349 tifp->if_ext_max = XFS_IFORK_SIZE(tip, XFS_DATA_FORK) /
350 (uint)sizeof(xfs_bmbt_rec_t);
351
352 /*
279 * Fix the on-disk inode values 353 * Fix the on-disk inode values
280 */ 354 */
281 tmp = (__uint64_t)ip->i_d.di_nblocks; 355 tmp = (__uint64_t)ip->i_d.di_nblocks;
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index fa402a6bbbcf..155e798f30a1 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -73,7 +73,6 @@ xfs_inode_alloc(
73 ASSERT(atomic_read(&ip->i_pincount) == 0); 73 ASSERT(atomic_read(&ip->i_pincount) == 0);
74 ASSERT(!spin_is_locked(&ip->i_flags_lock)); 74 ASSERT(!spin_is_locked(&ip->i_flags_lock));
75 ASSERT(completion_done(&ip->i_flush)); 75 ASSERT(completion_done(&ip->i_flush));
76 ASSERT(!rwsem_is_locked(&ip->i_iolock.mr_lock));
77 76
78 mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino); 77 mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino);
79 78
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index ce278b3ae7fc..ef77fd88c8e3 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -2841,8 +2841,8 @@ xfs_iflush(
2841 mp = ip->i_mount; 2841 mp = ip->i_mount;
2842 2842
2843 /* 2843 /*
2844 * If the inode isn't dirty, then just release the inode 2844 * If the inode isn't dirty, then just release the inode flush lock and
2845 * flush lock and do nothing. 2845 * do nothing.
2846 */ 2846 */
2847 if (xfs_inode_clean(ip)) { 2847 if (xfs_inode_clean(ip)) {
2848 xfs_ifunlock(ip); 2848 xfs_ifunlock(ip);
@@ -2868,6 +2868,19 @@ xfs_iflush(
2868 xfs_iunpin_wait(ip); 2868 xfs_iunpin_wait(ip);
2869 2869
2870 /* 2870 /*
2871 * For stale inodes we cannot rely on the backing buffer remaining
2872 * stale in cache for the remaining life of the stale inode and so
2873 * xfs_itobp() below may give us a buffer that no longer contains
2874 * inodes below. We have to check this after ensuring the inode is
2875 * unpinned so that it is safe to reclaim the stale inode after the
2876 * flush call.
2877 */
2878 if (xfs_iflags_test(ip, XFS_ISTALE)) {
2879 xfs_ifunlock(ip);
2880 return 0;
2881 }
2882
2883 /*
2871 * This may have been unpinned because the filesystem is shutting 2884 * This may have been unpinned because the filesystem is shutting
2872 * down forcibly. If that's the case we must not write this inode 2885 * down forcibly. If that's the case we must not write this inode
2873 * to disk, because the log record didn't make it to disk! 2886 * to disk, because the log record didn't make it to disk!
diff --git a/fs/xfs/xfs_rtalloc.c b/fs/xfs/xfs_rtalloc.c
index 9e15a1185362..6be05f756d59 100644
--- a/fs/xfs/xfs_rtalloc.c
+++ b/fs/xfs/xfs_rtalloc.c
@@ -1517,6 +1517,8 @@ xfs_rtfree_range(
1517 */ 1517 */
1518 error = xfs_rtfind_forw(mp, tp, end, mp->m_sb.sb_rextents - 1, 1518 error = xfs_rtfind_forw(mp, tp, end, mp->m_sb.sb_rextents - 1,
1519 &postblock); 1519 &postblock);
1520 if (error)
1521 return error;
1520 /* 1522 /*
1521 * If there are blocks not being freed at the front of the 1523 * If there are blocks not being freed at the front of the
1522 * old extent, add summary data for them to be allocated. 1524 * old extent, add summary data for them to be allocated.
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 6558ffd8d140..6f268756bf36 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -70,7 +70,6 @@ xfs_setattr(
70 uint commit_flags=0; 70 uint commit_flags=0;
71 uid_t uid=0, iuid=0; 71 uid_t uid=0, iuid=0;
72 gid_t gid=0, igid=0; 72 gid_t gid=0, igid=0;
73 int timeflags = 0;
74 struct xfs_dquot *udqp, *gdqp, *olddquot1, *olddquot2; 73 struct xfs_dquot *udqp, *gdqp, *olddquot1, *olddquot2;
75 int need_iolock = 1; 74 int need_iolock = 1;
76 75
@@ -135,16 +134,13 @@ xfs_setattr(
135 if (flags & XFS_ATTR_NOLOCK) 134 if (flags & XFS_ATTR_NOLOCK)
136 need_iolock = 0; 135 need_iolock = 0;
137 if (!(mask & ATTR_SIZE)) { 136 if (!(mask & ATTR_SIZE)) {
138 if ((mask != (ATTR_CTIME|ATTR_ATIME|ATTR_MTIME)) || 137 tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_NOT_SIZE);
139 (mp->m_flags & XFS_MOUNT_WSYNC)) { 138 commit_flags = 0;
140 tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_NOT_SIZE); 139 code = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp),
141 commit_flags = 0; 140 0, 0, 0);
142 if ((code = xfs_trans_reserve(tp, 0, 141 if (code) {
143 XFS_ICHANGE_LOG_RES(mp), 0, 142 lock_flags = 0;
144 0, 0))) { 143 goto error_return;
145 lock_flags = 0;
146 goto error_return;
147 }
148 } 144 }
149 } else { 145 } else {
150 if (DM_EVENT_ENABLED(ip, DM_EVENT_TRUNCATE) && 146 if (DM_EVENT_ENABLED(ip, DM_EVENT_TRUNCATE) &&
@@ -295,15 +291,23 @@ xfs_setattr(
295 * or we are explicitly asked to change it. This handles 291 * or we are explicitly asked to change it. This handles
296 * the semantic difference between truncate() and ftruncate() 292 * the semantic difference between truncate() and ftruncate()
297 * as implemented in the VFS. 293 * as implemented in the VFS.
294 *
295 * The regular truncate() case without ATTR_CTIME and ATTR_MTIME
296 * is a special case where we need to update the times despite
297 * not having these flags set. For all other operations the
298 * VFS set these flags explicitly if it wants a timestamp
299 * update.
298 */ 300 */
299 if (iattr->ia_size != ip->i_size || (mask & ATTR_CTIME)) 301 if (iattr->ia_size != ip->i_size &&
300 timeflags |= XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG; 302 (!(mask & (ATTR_CTIME | ATTR_MTIME)))) {
303 iattr->ia_ctime = iattr->ia_mtime =
304 current_fs_time(inode->i_sb);
305 mask |= ATTR_CTIME | ATTR_MTIME;
306 }
301 307
302 if (iattr->ia_size > ip->i_size) { 308 if (iattr->ia_size > ip->i_size) {
303 ip->i_d.di_size = iattr->ia_size; 309 ip->i_d.di_size = iattr->ia_size;
304 ip->i_size = iattr->ia_size; 310 ip->i_size = iattr->ia_size;
305 if (!(flags & XFS_ATTR_DMI))
306 xfs_ichgtime(ip, XFS_ICHGTIME_CHG);
307 xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE); 311 xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
308 } else if (iattr->ia_size <= ip->i_size || 312 } else if (iattr->ia_size <= ip->i_size ||
309 (iattr->ia_size == 0 && ip->i_d.di_nextents)) { 313 (iattr->ia_size == 0 && ip->i_d.di_nextents)) {
@@ -374,9 +378,6 @@ xfs_setattr(
374 ip->i_d.di_gid = gid; 378 ip->i_d.di_gid = gid;
375 inode->i_gid = gid; 379 inode->i_gid = gid;
376 } 380 }
377
378 xfs_trans_log_inode (tp, ip, XFS_ILOG_CORE);
379 timeflags |= XFS_ICHGTIME_CHG;
380 } 381 }
381 382
382 /* 383 /*
@@ -393,51 +394,37 @@ xfs_setattr(
393 394
394 inode->i_mode &= S_IFMT; 395 inode->i_mode &= S_IFMT;
395 inode->i_mode |= mode & ~S_IFMT; 396 inode->i_mode |= mode & ~S_IFMT;
396
397 xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
398 timeflags |= XFS_ICHGTIME_CHG;
399 } 397 }
400 398
401 /* 399 /*
402 * Change file access or modified times. 400 * Change file access or modified times.
403 */ 401 */
404 if (mask & (ATTR_ATIME|ATTR_MTIME)) { 402 if (mask & ATTR_ATIME) {
405 if (mask & ATTR_ATIME) { 403 inode->i_atime = iattr->ia_atime;
406 inode->i_atime = iattr->ia_atime; 404 ip->i_d.di_atime.t_sec = iattr->ia_atime.tv_sec;
407 ip->i_d.di_atime.t_sec = iattr->ia_atime.tv_sec; 405 ip->i_d.di_atime.t_nsec = iattr->ia_atime.tv_nsec;
408 ip->i_d.di_atime.t_nsec = iattr->ia_atime.tv_nsec; 406 ip->i_update_core = 1;
409 ip->i_update_core = 1;
410 }
411 if (mask & ATTR_MTIME) {
412 inode->i_mtime = iattr->ia_mtime;
413 ip->i_d.di_mtime.t_sec = iattr->ia_mtime.tv_sec;
414 ip->i_d.di_mtime.t_nsec = iattr->ia_mtime.tv_nsec;
415 timeflags &= ~XFS_ICHGTIME_MOD;
416 timeflags |= XFS_ICHGTIME_CHG;
417 }
418 if (tp && (mask & (ATTR_MTIME_SET|ATTR_ATIME_SET)))
419 xfs_trans_log_inode (tp, ip, XFS_ILOG_CORE);
420 } 407 }
421 408 if (mask & ATTR_CTIME) {
422 /*
423 * Change file inode change time only if ATTR_CTIME set
424 * AND we have been called by a DMI function.
425 */
426
427 if ((flags & XFS_ATTR_DMI) && (mask & ATTR_CTIME)) {
428 inode->i_ctime = iattr->ia_ctime; 409 inode->i_ctime = iattr->ia_ctime;
429 ip->i_d.di_ctime.t_sec = iattr->ia_ctime.tv_sec; 410 ip->i_d.di_ctime.t_sec = iattr->ia_ctime.tv_sec;
430 ip->i_d.di_ctime.t_nsec = iattr->ia_ctime.tv_nsec; 411 ip->i_d.di_ctime.t_nsec = iattr->ia_ctime.tv_nsec;
431 ip->i_update_core = 1; 412 ip->i_update_core = 1;
432 timeflags &= ~XFS_ICHGTIME_CHG; 413 }
414 if (mask & ATTR_MTIME) {
415 inode->i_mtime = iattr->ia_mtime;
416 ip->i_d.di_mtime.t_sec = iattr->ia_mtime.tv_sec;
417 ip->i_d.di_mtime.t_nsec = iattr->ia_mtime.tv_nsec;
418 ip->i_update_core = 1;
433 } 419 }
434 420
435 /* 421 /*
436 * Send out timestamp changes that need to be set to the 422 * And finally, log the inode core if any attribute in it
437 * current time. Not done when called by a DMI function. 423 * has been changed.
438 */ 424 */
439 if (timeflags && !(flags & XFS_ATTR_DMI)) 425 if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE|
440 xfs_ichgtime(ip, timeflags); 426 ATTR_ATIME|ATTR_CTIME|ATTR_MTIME))
427 xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
441 428
442 XFS_STATS_INC(xs_ig_attrchg); 429 XFS_STATS_INC(xs_ig_attrchg);
443 430
@@ -452,12 +439,10 @@ xfs_setattr(
452 * mix so this probably isn't worth the trouble to optimize. 439 * mix so this probably isn't worth the trouble to optimize.
453 */ 440 */
454 code = 0; 441 code = 0;
455 if (tp) { 442 if (mp->m_flags & XFS_MOUNT_WSYNC)
456 if (mp->m_flags & XFS_MOUNT_WSYNC) 443 xfs_trans_set_sync(tp);
457 xfs_trans_set_sync(tp);
458 444
459 code = xfs_trans_commit(tp, commit_flags); 445 code = xfs_trans_commit(tp, commit_flags);
460 }
461 446
462 xfs_iunlock(ip, lock_flags); 447 xfs_iunlock(ip, lock_flags);
463 448
generated by cgit v1.2.2 (git 2.25.0) at 2025-12-28 21:07:01 -0500