aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
Diffstat (limited to 'fs')
-rw-r--r--fs/9p/v9fs_vfs.h1
-rw-r--r--fs/9p/vfs_file.c4
-rw-r--r--fs/9p/vfs_inode.c2
-rw-r--r--fs/aio.c1
-rw-r--r--fs/autofs4/root.c6
-rw-r--r--fs/binfmt_elf.c7
-rw-r--r--fs/binfmt_elf_fdpic.c17
-rw-r--r--fs/char_dev.c2
-rw-r--r--fs/cifs/CHANGES6
-rw-r--r--fs/cifs/cifspdu.h10
-rw-r--r--fs/cifs/inode.c21
-rw-r--r--fs/cifs/readdir.c4
-rw-r--r--fs/compat_ioctl.c9
-rw-r--r--fs/configfs/dir.c27
-rw-r--r--fs/ecryptfs/dentry.c15
-rw-r--r--fs/exec.c18
-rw-r--r--fs/ext3/inode.c85
-rw-r--r--fs/ext3/xattr.c3
-rw-r--r--fs/ext4/inode.c85
-rw-r--r--fs/fuse/dir.c5
-rw-r--r--fs/fuse/fuse_i.h5
-rw-r--r--fs/fuse/inode.c2
-rw-r--r--fs/hostfs/hostfs_kern.c25
-rw-r--r--fs/nfs/dir.c3
-rw-r--r--fs/nfs/direct.c11
-rw-r--r--fs/nfs/inode.c9
-rw-r--r--fs/nfs/super.c4
-rw-r--r--fs/nfs/sysctl.c8
-rw-r--r--fs/nfs/write.c301
-rw-r--r--fs/nfsd/nfs3xdr.c6
-rw-r--r--fs/nfsd/nfs4acl.c2
-rw-r--r--fs/nfsd/nfs4state.c6
-rw-r--r--fs/nfsd/nfsfh.c1
-rw-r--r--fs/ocfs2/aops.c26
-rw-r--r--fs/ocfs2/cluster/heartbeat.c50
-rw-r--r--fs/ocfs2/cluster/heartbeat.h2
-rw-r--r--fs/ocfs2/cluster/tcp.c13
-rw-r--r--fs/ocfs2/dlm/dlmdomain.c8
-rw-r--r--fs/ocfs2/dlm/dlmmaster.c102
-rw-r--r--fs/ocfs2/dlm/dlmthread.c10
-rw-r--r--fs/ocfs2/heartbeat.c15
-rw-r--r--fs/partitions/Kconfig3
-rw-r--r--fs/partitions/check.c2
-rw-r--r--fs/proc/Makefile3
-rw-r--r--fs/proc/base.c21
-rw-r--r--fs/proc/internal.h4
-rw-r--r--fs/proc/root.c2
-rw-r--r--fs/reiserfs/item_ops.c2
-rw-r--r--fs/reiserfs/xattr.c92
-rw-r--r--fs/smbfs/request.c1
-rw-r--r--fs/splice.c109
-rw-r--r--fs/sysfs/file.c64
-rw-r--r--fs/sysfs/inode.c10
-rw-r--r--fs/ufs/balloc.c86
-rw-r--r--fs/ufs/ialloc.c5
-rw-r--r--fs/ufs/inode.c65
-rw-r--r--fs/ufs/truncate.c38
-rw-r--r--fs/xfs/linux-2.6/xfs_buf.c4
58 files changed, 768 insertions, 680 deletions
diff --git a/fs/9p/v9fs_vfs.h b/fs/9p/v9fs_vfs.h
index 8ada4c5c5d70..6a82d39dc498 100644
--- a/fs/9p/v9fs_vfs.h
+++ b/fs/9p/v9fs_vfs.h
@@ -40,7 +40,6 @@
40extern struct file_system_type v9fs_fs_type; 40extern struct file_system_type v9fs_fs_type;
41extern const struct address_space_operations v9fs_addr_operations; 41extern const struct address_space_operations v9fs_addr_operations;
42extern const struct file_operations v9fs_file_operations; 42extern const struct file_operations v9fs_file_operations;
43extern const struct file_operations v9fs_cached_file_operations;
44extern const struct file_operations v9fs_dir_operations; 43extern const struct file_operations v9fs_dir_operations;
45extern struct dentry_operations v9fs_dentry_operations; 44extern struct dentry_operations v9fs_dentry_operations;
46extern struct dentry_operations v9fs_cached_dentry_operations; 45extern struct dentry_operations v9fs_cached_dentry_operations;
diff --git a/fs/9p/vfs_file.c b/fs/9p/vfs_file.c
index 653dfa5b2531..c7b677253843 100644
--- a/fs/9p/vfs_file.c
+++ b/fs/9p/vfs_file.c
@@ -42,6 +42,8 @@
42#include "v9fs_vfs.h" 42#include "v9fs_vfs.h"
43#include "fid.h" 43#include "fid.h"
44 44
45static const struct file_operations v9fs_cached_file_operations;
46
45/** 47/**
46 * v9fs_file_open - open a file (or directory) 48 * v9fs_file_open - open a file (or directory)
47 * @inode: inode to be opened 49 * @inode: inode to be opened
@@ -245,7 +247,7 @@ v9fs_file_write(struct file *filp, const char __user * data,
245 return total; 247 return total;
246} 248}
247 249
248const struct file_operations v9fs_cached_file_operations = { 250static const struct file_operations v9fs_cached_file_operations = {
249 .llseek = generic_file_llseek, 251 .llseek = generic_file_llseek,
250 .read = do_sync_read, 252 .read = do_sync_read,
251 .aio_read = generic_file_aio_read, 253 .aio_read = generic_file_aio_read,
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 124a085d1f2e..b01b0a457932 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -415,7 +415,7 @@ static int v9fs_remove(struct inode *dir, struct dentry *file, int rmdir)
415 file_inode = file->d_inode; 415 file_inode = file->d_inode;
416 sb = file_inode->i_sb; 416 sb = file_inode->i_sb;
417 v9ses = v9fs_inode2v9ses(file_inode); 417 v9ses = v9fs_inode2v9ses(file_inode);
418 v9fid = v9fs_fid_lookup(file); 418 v9fid = v9fs_fid_clone(file);
419 if(IS_ERR(v9fid)) 419 if(IS_ERR(v9fid))
420 return PTR_ERR(v9fid); 420 return PTR_ERR(v9fid);
421 421
diff --git a/fs/aio.c b/fs/aio.c
index 0b4ee0a5c83e..e4598d6d49dd 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -136,7 +136,6 @@ static int aio_setup_ring(struct kioctx *ctx)
136 0); 136 0);
137 if (IS_ERR((void *)info->mmap_base)) { 137 if (IS_ERR((void *)info->mmap_base)) {
138 up_write(&ctx->mm->mmap_sem); 138 up_write(&ctx->mm->mmap_sem);
139 printk("mmap err: %ld\n", -info->mmap_base);
140 info->mmap_size = 0; 139 info->mmap_size = 0;
141 aio_free_ring(ctx); 140 aio_free_ring(ctx);
142 return -EAGAIN; 141 return -EAGAIN;
diff --git a/fs/autofs4/root.c b/fs/autofs4/root.c
index b4631046867e..d0e9b3a3905d 100644
--- a/fs/autofs4/root.c
+++ b/fs/autofs4/root.c
@@ -470,9 +470,6 @@ void autofs4_dentry_release(struct dentry *de)
470 if (inf) { 470 if (inf) {
471 struct autofs_sb_info *sbi = autofs4_sbi(de->d_sb); 471 struct autofs_sb_info *sbi = autofs4_sbi(de->d_sb);
472 472
473 inf->dentry = NULL;
474 inf->inode = NULL;
475
476 if (sbi) { 473 if (sbi) {
477 spin_lock(&sbi->rehash_lock); 474 spin_lock(&sbi->rehash_lock);
478 if (!list_empty(&inf->rehash)) 475 if (!list_empty(&inf->rehash))
@@ -480,6 +477,9 @@ void autofs4_dentry_release(struct dentry *de)
480 spin_unlock(&sbi->rehash_lock); 477 spin_unlock(&sbi->rehash_lock);
481 } 478 }
482 479
480 inf->dentry = NULL;
481 inf->inode = NULL;
482
483 autofs4_free_ino(inf); 483 autofs4_free_ino(inf);
484 } 484 }
485} 485}
diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
index 51db1182b27e..9cc4f0a8aaae 100644
--- a/fs/binfmt_elf.c
+++ b/fs/binfmt_elf.c
@@ -507,7 +507,7 @@ out:
507#define INTERPRETER_ELF 2 507#define INTERPRETER_ELF 2
508 508
509#ifndef STACK_RND_MASK 509#ifndef STACK_RND_MASK
510#define STACK_RND_MASK 0x7ff /* with 4K pages 8MB of VA */ 510#define STACK_RND_MASK (0x7ff >> (PAGE_SHIFT - 12)) /* 8MB of VA */
511#endif 511#endif
512 512
513static unsigned long randomize_stack_top(unsigned long stack_top) 513static unsigned long randomize_stack_top(unsigned long stack_top)
@@ -1704,7 +1704,10 @@ static int elf_core_dump(long signr, struct pt_regs *regs, struct file *file)
1704 DUMP_SEEK(PAGE_SIZE); 1704 DUMP_SEEK(PAGE_SIZE);
1705 } else { 1705 } else {
1706 if (page == ZERO_PAGE(addr)) { 1706 if (page == ZERO_PAGE(addr)) {
1707 DUMP_SEEK(PAGE_SIZE); 1707 if (!dump_seek(file, PAGE_SIZE)) {
1708 page_cache_release(page);
1709 goto end_coredump;
1710 }
1708 } else { 1711 } else {
1709 void *kaddr; 1712 void *kaddr;
1710 flush_cache_page(vma, addr, 1713 flush_cache_page(vma, addr,
diff --git a/fs/binfmt_elf_fdpic.c b/fs/binfmt_elf_fdpic.c
index 5810aa1339fd..f3ddca4a387b 100644
--- a/fs/binfmt_elf_fdpic.c
+++ b/fs/binfmt_elf_fdpic.c
@@ -179,6 +179,8 @@ static int load_elf_fdpic_binary(struct linux_binprm *bprm,
179 int executable_stack; 179 int executable_stack;
180 int retval, i; 180 int retval, i;
181 181
182 kdebug("____ LOAD %d ____", current->pid);
183
182 memset(&exec_params, 0, sizeof(exec_params)); 184 memset(&exec_params, 0, sizeof(exec_params));
183 memset(&interp_params, 0, sizeof(interp_params)); 185 memset(&interp_params, 0, sizeof(interp_params));
184 186
@@ -941,8 +943,11 @@ static int elf_fdpic_map_file_constdisp_on_uclinux(
941 943
942 if (mm) { 944 if (mm) {
943 if (phdr->p_flags & PF_X) { 945 if (phdr->p_flags & PF_X) {
944 mm->start_code = seg->addr; 946 if (!mm->start_code) {
945 mm->end_code = seg->addr + phdr->p_memsz; 947 mm->start_code = seg->addr;
948 mm->end_code = seg->addr +
949 phdr->p_memsz;
950 }
946 } else if (!mm->start_data) { 951 } else if (!mm->start_data) {
947 mm->start_data = seg->addr; 952 mm->start_data = seg->addr;
948#ifndef CONFIG_MMU 953#ifndef CONFIG_MMU
@@ -1123,8 +1128,10 @@ static int elf_fdpic_map_file_by_direct_mmap(struct elf_fdpic_params *params,
1123 1128
1124 if (mm) { 1129 if (mm) {
1125 if (phdr->p_flags & PF_X) { 1130 if (phdr->p_flags & PF_X) {
1126 mm->start_code = maddr; 1131 if (!mm->start_code) {
1127 mm->end_code = maddr + phdr->p_memsz; 1132 mm->start_code = maddr;
1133 mm->end_code = maddr + phdr->p_memsz;
1134 }
1128 } else if (!mm->start_data) { 1135 } else if (!mm->start_data) {
1129 mm->start_data = maddr; 1136 mm->start_data = maddr;
1130 mm->end_data = maddr + phdr->p_memsz; 1137 mm->end_data = maddr + phdr->p_memsz;
@@ -1473,8 +1480,8 @@ static int elf_fdpic_dump_segments(struct file *file, struct mm_struct *mm,
1473 DUMP_SEEK(file->f_pos + PAGE_SIZE); 1480 DUMP_SEEK(file->f_pos + PAGE_SIZE);
1474 } 1481 }
1475 else if (page == ZERO_PAGE(addr)) { 1482 else if (page == ZERO_PAGE(addr)) {
1476 DUMP_SEEK(file->f_pos + PAGE_SIZE);
1477 page_cache_release(page); 1483 page_cache_release(page);
1484 DUMP_SEEK(file->f_pos + PAGE_SIZE);
1478 } 1485 }
1479 else { 1486 else {
1480 void *kaddr; 1487 void *kaddr;
diff --git a/fs/char_dev.c b/fs/char_dev.c
index 78ced721554d..164a45cdaf5f 100644
--- a/fs/char_dev.c
+++ b/fs/char_dev.c
@@ -109,8 +109,6 @@ __register_chrdev_region(unsigned int major, unsigned int baseminor,
109 /* temporary */ 109 /* temporary */
110 if (major == 0) { 110 if (major == 0) {
111 for (i = ARRAY_SIZE(chrdevs)-1; i > 0; i--) { 111 for (i = ARRAY_SIZE(chrdevs)-1; i > 0; i--) {
112 if (is_lanana_major(i))
113 continue;
114 if (chrdevs[i] == NULL) 112 if (chrdevs[i] == NULL)
115 break; 113 break;
116 } 114 }
diff --git a/fs/cifs/CHANGES b/fs/cifs/CHANGES
index 6247628bdaed..5d1f4873d701 100644
--- a/fs/cifs/CHANGES
+++ b/fs/cifs/CHANGES
@@ -4,6 +4,12 @@ Fix mtime bouncing around from local idea of last write times to remote time.
4Fix hang (in i_size_read) when simultaneous size update of same remote file 4Fix hang (in i_size_read) when simultaneous size update of same remote file
5on smp system corrupts sequence number. Do not reread unnecessarily partial page 5on smp system corrupts sequence number. Do not reread unnecessarily partial page
6(which we are about to overwrite anyway) when writing out file opened rw. 6(which we are about to overwrite anyway) when writing out file opened rw.
7When DOS attribute of file on non-Unix server's file changes on the server side
8from read-only back to read-write, reflect this change in default file mode
9(we had been leaving a file's mode read-only until the inode were reloaded).
10Allow setting of attribute back to ATTR_NORMAL (removing readonly dos attribute
11when archive dos attribute not set and we are changing mode back to writeable
12on server which does not support the Unix Extensions).
7 13
8Version 1.47 14Version 1.47
9------------ 15------------
diff --git a/fs/cifs/cifspdu.h b/fs/cifs/cifspdu.h
index 0efdf35aab2c..4d8948e8762c 100644
--- a/fs/cifs/cifspdu.h
+++ b/fs/cifs/cifspdu.h
@@ -220,7 +220,7 @@
220 */ 220 */
221#define CIFS_NO_HANDLE 0xFFFF 221#define CIFS_NO_HANDLE 0xFFFF
222 222
223#define NO_CHANGE_64 0xFFFFFFFFFFFFFFFFULL 223#define NO_CHANGE_64 cpu_to_le64(0xFFFFFFFFFFFFFFFFULL)
224#define NO_CHANGE_32 0xFFFFFFFFUL 224#define NO_CHANGE_32 0xFFFFFFFFUL
225 225
226/* IPC$ in ASCII */ 226/* IPC$ in ASCII */
@@ -1887,7 +1887,13 @@ typedef struct {
1887 calls including posix open 1887 calls including posix open
1888 and posix unlink */ 1888 and posix unlink */
1889#ifdef CONFIG_CIFS_POSIX 1889#ifdef CONFIG_CIFS_POSIX
1890#define CIFS_UNIX_CAP_MASK 0x0000003b 1890/* Can not set pathnames cap yet until we send new posix create SMB since
1891 otherwise server can treat such handles opened with older ntcreatex
1892 (by a new client which knows how to send posix path ops)
1893 as non-posix handles (can affect write behavior with byte range locks.
1894 We can add back in POSIX_PATH_OPS cap when Posix Create/Mkdir finished */
1895/* #define CIFS_UNIX_CAP_MASK 0x0000003b */
1896#define CIFS_UNIX_CAP_MASK 0x0000001b
1891#else 1897#else
1892#define CIFS_UNIX_CAP_MASK 0x00000013 1898#define CIFS_UNIX_CAP_MASK 0x00000013
1893#endif /* CONFIG_CIFS_POSIX */ 1899#endif /* CONFIG_CIFS_POSIX */
diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c
index 86b9dbbd8441..f414526e476a 100644
--- a/fs/cifs/inode.c
+++ b/fs/cifs/inode.c
@@ -494,6 +494,12 @@ int cifs_get_inode_info(struct inode **pinode,
494 mode e.g. 555 */ 494 mode e.g. 555 */
495 if (cifsInfo->cifsAttrs & ATTR_READONLY) 495 if (cifsInfo->cifsAttrs & ATTR_READONLY)
496 inode->i_mode &= ~(S_IWUGO); 496 inode->i_mode &= ~(S_IWUGO);
497 else if ((inode->i_mode & S_IWUGO) == 0)
498 /* the ATTR_READONLY flag may have been */
499 /* changed on server -- set any w bits */
500 /* allowed by mnt_file_mode */
501 inode->i_mode |= (S_IWUGO &
502 cifs_sb->mnt_file_mode);
497 /* BB add code here - 503 /* BB add code here -
498 validate if device or weird share or device type? */ 504 validate if device or weird share or device type? */
499 } 505 }
@@ -1190,6 +1196,7 @@ int cifs_setattr(struct dentry *direntry, struct iattr *attrs)
1190 struct cifsFileInfo *open_file = NULL; 1196 struct cifsFileInfo *open_file = NULL;
1191 FILE_BASIC_INFO time_buf; 1197 FILE_BASIC_INFO time_buf;
1192 int set_time = FALSE; 1198 int set_time = FALSE;
1199 int set_dosattr = FALSE;
1193 __u64 mode = 0xFFFFFFFFFFFFFFFFULL; 1200 __u64 mode = 0xFFFFFFFFFFFFFFFFULL;
1194 __u64 uid = 0xFFFFFFFFFFFFFFFFULL; 1201 __u64 uid = 0xFFFFFFFFFFFFFFFFULL;
1195 __u64 gid = 0xFFFFFFFFFFFFFFFFULL; 1202 __u64 gid = 0xFFFFFFFFFFFFFFFFULL;
@@ -1326,15 +1333,23 @@ int cifs_setattr(struct dentry *direntry, struct iattr *attrs)
1326 else if (attrs->ia_valid & ATTR_MODE) { 1333 else if (attrs->ia_valid & ATTR_MODE) {
1327 rc = 0; 1334 rc = 0;
1328 if ((mode & S_IWUGO) == 0) /* not writeable */ { 1335 if ((mode & S_IWUGO) == 0) /* not writeable */ {
1329 if ((cifsInode->cifsAttrs & ATTR_READONLY) == 0) 1336 if ((cifsInode->cifsAttrs & ATTR_READONLY) == 0) {
1337 set_dosattr = TRUE;
1330 time_buf.Attributes = 1338 time_buf.Attributes =
1331 cpu_to_le32(cifsInode->cifsAttrs | 1339 cpu_to_le32(cifsInode->cifsAttrs |
1332 ATTR_READONLY); 1340 ATTR_READONLY);
1341 }
1333 } else if ((mode & S_IWUGO) == S_IWUGO) { 1342 } else if ((mode & S_IWUGO) == S_IWUGO) {
1334 if (cifsInode->cifsAttrs & ATTR_READONLY) 1343 if (cifsInode->cifsAttrs & ATTR_READONLY) {
1344 set_dosattr = TRUE;
1335 time_buf.Attributes = 1345 time_buf.Attributes =
1336 cpu_to_le32(cifsInode->cifsAttrs & 1346 cpu_to_le32(cifsInode->cifsAttrs &
1337 (~ATTR_READONLY)); 1347 (~ATTR_READONLY));
1348 /* Windows ignores set to zero */
1349 if(time_buf.Attributes == 0)
1350 time_buf.Attributes |=
1351 cpu_to_le32(ATTR_NORMAL);
1352 }
1338 } 1353 }
1339 /* BB to be implemented - 1354 /* BB to be implemented -
1340 via Windows security descriptors or streams */ 1355 via Windows security descriptors or streams */
@@ -1372,7 +1387,7 @@ int cifs_setattr(struct dentry *direntry, struct iattr *attrs)
1372 } else 1387 } else
1373 time_buf.ChangeTime = 0; 1388 time_buf.ChangeTime = 0;
1374 1389
1375 if (set_time || time_buf.Attributes) { 1390 if (set_time || set_dosattr) {
1376 time_buf.CreationTime = 0; /* do not change */ 1391 time_buf.CreationTime = 0; /* do not change */
1377 /* In the future we should experiment - try setting timestamps 1392 /* In the future we should experiment - try setting timestamps
1378 via Handle (SetFileInfo) instead of by path */ 1393 via Handle (SetFileInfo) instead of by path */
diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c
index 44cfb528797d..2a374d5215ab 100644
--- a/fs/cifs/readdir.c
+++ b/fs/cifs/readdir.c
@@ -219,6 +219,10 @@ static void fill_in_inode(struct inode *tmp_inode, int new_buf_type,
219 tmp_inode->i_mode |= S_IFREG; 219 tmp_inode->i_mode |= S_IFREG;
220 if (attr & ATTR_READONLY) 220 if (attr & ATTR_READONLY)
221 tmp_inode->i_mode &= ~(S_IWUGO); 221 tmp_inode->i_mode &= ~(S_IWUGO);
222 else if ((tmp_inode->i_mode & S_IWUGO) == 0)
223 /* the ATTR_READONLY flag may have been changed on */
224 /* server -- set any w bits allowed by mnt_file_mode */
225 tmp_inode->i_mode |= (S_IWUGO & cifs_sb->mnt_file_mode);
222 } /* could add code here - to validate if device or weird share type? */ 226 } /* could add code here - to validate if device or weird share type? */
223 227
224 /* can not fill in nlink here as in qpathinfo version and Unx search */ 228 /* can not fill in nlink here as in qpathinfo version and Unx search */
diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c
index c81c958b3e1d..8b1c5d8bf4ef 100644
--- a/fs/compat_ioctl.c
+++ b/fs/compat_ioctl.c
@@ -2553,11 +2553,15 @@ HANDLE_IOCTL(I2C_RDWR, do_i2c_rdwr_ioctl)
2553HANDLE_IOCTL(I2C_SMBUS, do_i2c_smbus_ioctl) 2553HANDLE_IOCTL(I2C_SMBUS, do_i2c_smbus_ioctl)
2554/* wireless */ 2554/* wireless */
2555HANDLE_IOCTL(SIOCGIWRANGE, do_wireless_ioctl) 2555HANDLE_IOCTL(SIOCGIWRANGE, do_wireless_ioctl)
2556HANDLE_IOCTL(SIOCGIWPRIV, do_wireless_ioctl)
2557HANDLE_IOCTL(SIOCGIWSTATS, do_wireless_ioctl)
2556HANDLE_IOCTL(SIOCSIWSPY, do_wireless_ioctl) 2558HANDLE_IOCTL(SIOCSIWSPY, do_wireless_ioctl)
2557HANDLE_IOCTL(SIOCGIWSPY, do_wireless_ioctl) 2559HANDLE_IOCTL(SIOCGIWSPY, do_wireless_ioctl)
2558HANDLE_IOCTL(SIOCSIWTHRSPY, do_wireless_ioctl) 2560HANDLE_IOCTL(SIOCSIWTHRSPY, do_wireless_ioctl)
2559HANDLE_IOCTL(SIOCGIWTHRSPY, do_wireless_ioctl) 2561HANDLE_IOCTL(SIOCGIWTHRSPY, do_wireless_ioctl)
2562HANDLE_IOCTL(SIOCSIWMLME, do_wireless_ioctl)
2560HANDLE_IOCTL(SIOCGIWAPLIST, do_wireless_ioctl) 2563HANDLE_IOCTL(SIOCGIWAPLIST, do_wireless_ioctl)
2564HANDLE_IOCTL(SIOCSIWSCAN, do_wireless_ioctl)
2561HANDLE_IOCTL(SIOCGIWSCAN, do_wireless_ioctl) 2565HANDLE_IOCTL(SIOCGIWSCAN, do_wireless_ioctl)
2562HANDLE_IOCTL(SIOCSIWESSID, do_wireless_ioctl) 2566HANDLE_IOCTL(SIOCSIWESSID, do_wireless_ioctl)
2563HANDLE_IOCTL(SIOCGIWESSID, do_wireless_ioctl) 2567HANDLE_IOCTL(SIOCGIWESSID, do_wireless_ioctl)
@@ -2565,6 +2569,11 @@ HANDLE_IOCTL(SIOCSIWNICKN, do_wireless_ioctl)
2565HANDLE_IOCTL(SIOCGIWNICKN, do_wireless_ioctl) 2569HANDLE_IOCTL(SIOCGIWNICKN, do_wireless_ioctl)
2566HANDLE_IOCTL(SIOCSIWENCODE, do_wireless_ioctl) 2570HANDLE_IOCTL(SIOCSIWENCODE, do_wireless_ioctl)
2567HANDLE_IOCTL(SIOCGIWENCODE, do_wireless_ioctl) 2571HANDLE_IOCTL(SIOCGIWENCODE, do_wireless_ioctl)
2572HANDLE_IOCTL(SIOCSIWGENIE, do_wireless_ioctl)
2573HANDLE_IOCTL(SIOCGIWGENIE, do_wireless_ioctl)
2574HANDLE_IOCTL(SIOCSIWENCODEEXT, do_wireless_ioctl)
2575HANDLE_IOCTL(SIOCGIWENCODEEXT, do_wireless_ioctl)
2576HANDLE_IOCTL(SIOCSIWPMKSA, do_wireless_ioctl)
2568HANDLE_IOCTL(SIOCSIFBR, old_bridge_ioctl) 2577HANDLE_IOCTL(SIOCSIFBR, old_bridge_ioctl)
2569HANDLE_IOCTL(SIOCGIFBR, old_bridge_ioctl) 2578HANDLE_IOCTL(SIOCGIFBR, old_bridge_ioctl)
2570HANDLE_IOCTL(RTC_IRQP_READ32, rtc_ioctl) 2579HANDLE_IOCTL(RTC_IRQP_READ32, rtc_ioctl)
diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c
index 34750d5e4ff2..5e6e37e58f36 100644
--- a/fs/configfs/dir.c
+++ b/fs/configfs/dir.c
@@ -1141,25 +1141,22 @@ int configfs_register_subsystem(struct configfs_subsystem *subsys)
1141 1141
1142 err = -ENOMEM; 1142 err = -ENOMEM;
1143 dentry = d_alloc(configfs_sb->s_root, &name); 1143 dentry = d_alloc(configfs_sb->s_root, &name);
1144 if (!dentry) 1144 if (dentry) {
1145 goto out_release; 1145 d_add(dentry, NULL);
1146
1147 d_add(dentry, NULL);
1148 1146
1149 err = configfs_attach_group(sd->s_element, &group->cg_item, 1147 err = configfs_attach_group(sd->s_element, &group->cg_item,
1150 dentry); 1148 dentry);
1151 if (!err) 1149 if (err) {
1152 dentry = NULL; 1150 d_delete(dentry);
1153 else 1151 dput(dentry);
1154 d_delete(dentry); 1152 }
1153 }
1155 1154
1156 mutex_unlock(&configfs_sb->s_root->d_inode->i_mutex); 1155 mutex_unlock(&configfs_sb->s_root->d_inode->i_mutex);
1157 1156
1158 if (dentry) { 1157 if (err) {
1159 dput(dentry); 1158 unlink_group(group);
1160out_release: 1159 configfs_release_fs();
1161 unlink_group(group);
1162 configfs_release_fs();
1163 } 1160 }
1164 1161
1165 return err; 1162 return err;
diff --git a/fs/ecryptfs/dentry.c b/fs/ecryptfs/dentry.c
index 329efcd3d8c9..cb20b964419f 100644
--- a/fs/ecryptfs/dentry.c
+++ b/fs/ecryptfs/dentry.c
@@ -78,18 +78,13 @@ struct kmem_cache *ecryptfs_dentry_info_cache;
78 */ 78 */
79static void ecryptfs_d_release(struct dentry *dentry) 79static void ecryptfs_d_release(struct dentry *dentry)
80{ 80{
81 struct dentry *lower_dentry; 81 if (ecryptfs_dentry_to_private(dentry)) {
82 82 if (ecryptfs_dentry_to_lower(dentry)) {
83 lower_dentry = ecryptfs_dentry_to_lower(dentry); 83 mntput(ecryptfs_dentry_to_lower_mnt(dentry));
84 if (ecryptfs_dentry_to_private(dentry)) 84 dput(ecryptfs_dentry_to_lower(dentry));
85 }
85 kmem_cache_free(ecryptfs_dentry_info_cache, 86 kmem_cache_free(ecryptfs_dentry_info_cache,
86 ecryptfs_dentry_to_private(dentry)); 87 ecryptfs_dentry_to_private(dentry));
87 if (lower_dentry) {
88 struct vfsmount *lower_mnt =
89 ecryptfs_dentry_to_lower_mnt(dentry);
90
91 mntput(lower_mnt);
92 dput(lower_dentry);
93 } 88 }
94 return; 89 return;
95} 90}
diff --git a/fs/exec.c b/fs/exec.c
index 7e36c6f6f538..3155e915307a 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -1244,13 +1244,17 @@ EXPORT_SYMBOL(set_binfmt);
1244 * name into corename, which must have space for at least 1244 * name into corename, which must have space for at least
1245 * CORENAME_MAX_SIZE bytes plus one byte for the zero terminator. 1245 * CORENAME_MAX_SIZE bytes plus one byte for the zero terminator.
1246 */ 1246 */
1247static void format_corename(char *corename, const char *pattern, long signr) 1247static int format_corename(char *corename, const char *pattern, long signr)
1248{ 1248{
1249 const char *pat_ptr = pattern; 1249 const char *pat_ptr = pattern;
1250 char *out_ptr = corename; 1250 char *out_ptr = corename;
1251 char *const out_end = corename + CORENAME_MAX_SIZE; 1251 char *const out_end = corename + CORENAME_MAX_SIZE;
1252 int rc; 1252 int rc;
1253 int pid_in_pattern = 0; 1253 int pid_in_pattern = 0;
1254 int ispipe = 0;
1255
1256 if (*pattern == '|')
1257 ispipe = 1;
1254 1258
1255 /* Repeat as long as we have more pattern to process and more output 1259 /* Repeat as long as we have more pattern to process and more output
1256 space */ 1260 space */
@@ -1341,8 +1345,8 @@ static void format_corename(char *corename, const char *pattern, long signr)
1341 * 1345 *
1342 * If core_pattern does not include a %p (as is the default) 1346 * If core_pattern does not include a %p (as is the default)
1343 * and core_uses_pid is set, then .%pid will be appended to 1347 * and core_uses_pid is set, then .%pid will be appended to
1344 * the filename */ 1348 * the filename. Do not do this for piped commands. */
1345 if (!pid_in_pattern 1349 if (!ispipe && !pid_in_pattern
1346 && (core_uses_pid || atomic_read(&current->mm->mm_users) != 1)) { 1350 && (core_uses_pid || atomic_read(&current->mm->mm_users) != 1)) {
1347 rc = snprintf(out_ptr, out_end - out_ptr, 1351 rc = snprintf(out_ptr, out_end - out_ptr,
1348 ".%d", current->tgid); 1352 ".%d", current->tgid);
@@ -1350,8 +1354,9 @@ static void format_corename(char *corename, const char *pattern, long signr)
1350 goto out; 1354 goto out;
1351 out_ptr += rc; 1355 out_ptr += rc;
1352 } 1356 }
1353 out: 1357out:
1354 *out_ptr = 0; 1358 *out_ptr = 0;
1359 return ispipe;
1355} 1360}
1356 1361
1357static void zap_process(struct task_struct *start) 1362static void zap_process(struct task_struct *start)
@@ -1502,16 +1507,15 @@ int do_coredump(long signr, int exit_code, struct pt_regs * regs)
1502 * uses lock_kernel() 1507 * uses lock_kernel()
1503 */ 1508 */
1504 lock_kernel(); 1509 lock_kernel();
1505 format_corename(corename, core_pattern, signr); 1510 ispipe = format_corename(corename, core_pattern, signr);
1506 unlock_kernel(); 1511 unlock_kernel();
1507 if (corename[0] == '|') { 1512 if (ispipe) {
1508 /* SIGPIPE can happen, but it's just never processed */ 1513 /* SIGPIPE can happen, but it's just never processed */
1509 if(call_usermodehelper_pipe(corename+1, NULL, NULL, &file)) { 1514 if(call_usermodehelper_pipe(corename+1, NULL, NULL, &file)) {
1510 printk(KERN_INFO "Core dump to %s pipe failed\n", 1515 printk(KERN_INFO "Core dump to %s pipe failed\n",
1511 corename); 1516 corename);
1512 goto fail_unlock; 1517 goto fail_unlock;
1513 } 1518 }
1514 ispipe = 1;
1515 } else 1519 } else
1516 file = filp_open(corename, 1520 file = filp_open(corename,
1517 O_CREAT | 2 | O_NOFOLLOW | O_LARGEFILE | flag, 1521 O_CREAT | 2 | O_NOFOLLOW | O_LARGEFILE | flag,
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index 8a824f4ce5c6..a5b150f7e8a2 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -1148,102 +1148,37 @@ static int do_journal_get_write_access(handle_t *handle,
1148 return ext3_journal_get_write_access(handle, bh); 1148 return ext3_journal_get_write_access(handle, bh);
1149} 1149}
1150 1150
1151/*
1152 * The idea of this helper function is following:
1153 * if prepare_write has allocated some blocks, but not all of them, the
1154 * transaction must include the content of the newly allocated blocks.
1155 * This content is expected to be set to zeroes by block_prepare_write().
1156 * 2006/10/14 SAW
1157 */
1158static int ext3_prepare_failure(struct file *file, struct page *page,
1159 unsigned from, unsigned to)
1160{
1161 struct address_space *mapping;
1162 struct buffer_head *bh, *head, *next;
1163 unsigned block_start, block_end;
1164 unsigned blocksize;
1165 int ret;
1166 handle_t *handle = ext3_journal_current_handle();
1167
1168 mapping = page->mapping;
1169 if (ext3_should_writeback_data(mapping->host)) {
1170 /* optimization: no constraints about data */
1171skip:
1172 return ext3_journal_stop(handle);
1173 }
1174
1175 head = page_buffers(page);
1176 blocksize = head->b_size;
1177 for ( bh = head, block_start = 0;
1178 bh != head || !block_start;
1179 block_start = block_end, bh = next)
1180 {
1181 next = bh->b_this_page;
1182 block_end = block_start + blocksize;
1183 if (block_end <= from)
1184 continue;
1185 if (block_start >= to) {
1186 block_start = to;
1187 break;
1188 }
1189 if (!buffer_mapped(bh))
1190 /* prepare_write failed on this bh */
1191 break;
1192 if (ext3_should_journal_data(mapping->host)) {
1193 ret = do_journal_get_write_access(handle, bh);
1194 if (ret) {
1195 ext3_journal_stop(handle);
1196 return ret;
1197 }
1198 }
1199 /*
1200 * block_start here becomes the first block where the current iteration
1201 * of prepare_write failed.
1202 */
1203 }
1204 if (block_start <= from)
1205 goto skip;
1206
1207 /* commit allocated and zeroed buffers */
1208 return mapping->a_ops->commit_write(file, page, from, block_start);
1209}
1210
1211static int ext3_prepare_write(struct file *file, struct page *page, 1151static int ext3_prepare_write(struct file *file, struct page *page,
1212 unsigned from, unsigned to) 1152 unsigned from, unsigned to)
1213{ 1153{
1214 struct inode *inode = page->mapping->host; 1154 struct inode *inode = page->mapping->host;
1215 int ret, ret2; 1155 int ret, needed_blocks = ext3_writepage_trans_blocks(inode);
1216 int needed_blocks = ext3_writepage_trans_blocks(inode);
1217 handle_t *handle; 1156 handle_t *handle;
1218 int retries = 0; 1157 int retries = 0;
1219 1158
1220retry: 1159retry:
1221 handle = ext3_journal_start(inode, needed_blocks); 1160 handle = ext3_journal_start(inode, needed_blocks);
1222 if (IS_ERR(handle)) 1161 if (IS_ERR(handle)) {
1223 return PTR_ERR(handle); 1162 ret = PTR_ERR(handle);
1163 goto out;
1164 }
1224 if (test_opt(inode->i_sb, NOBH) && ext3_should_writeback_data(inode)) 1165 if (test_opt(inode->i_sb, NOBH) && ext3_should_writeback_data(inode))
1225 ret = nobh_prepare_write(page, from, to, ext3_get_block); 1166 ret = nobh_prepare_write(page, from, to, ext3_get_block);
1226 else 1167 else
1227 ret = block_prepare_write(page, from, to, ext3_get_block); 1168 ret = block_prepare_write(page, from, to, ext3_get_block);
1228 if (ret) 1169 if (ret)
1229 goto failure; 1170 goto prepare_write_failed;
1230 1171
1231 if (ext3_should_journal_data(inode)) { 1172 if (ext3_should_journal_data(inode)) {
1232 ret = walk_page_buffers(handle, page_buffers(page), 1173 ret = walk_page_buffers(handle, page_buffers(page),
1233 from, to, NULL, do_journal_get_write_access); 1174 from, to, NULL, do_journal_get_write_access);
1234 if (ret)
1235 /* fatal error, just put the handle and return */
1236 journal_stop(handle);
1237 } 1175 }
1238 return ret; 1176prepare_write_failed:
1239 1177 if (ret)
1240failure: 1178 ext3_journal_stop(handle);
1241 ret2 = ext3_prepare_failure(file, page, from, to);
1242 if (ret2 < 0)
1243 return ret2;
1244 if (ret == -ENOSPC && ext3_should_retry_alloc(inode->i_sb, &retries)) 1179 if (ret == -ENOSPC && ext3_should_retry_alloc(inode->i_sb, &retries))
1245 goto retry; 1180 goto retry;
1246 /* retry number exceeded, or other error like -EDQUOT */ 1181out:
1247 return ret; 1182 return ret;
1248} 1183}
1249 1184
diff --git a/fs/ext3/xattr.c b/fs/ext3/xattr.c
index 12f7dda1232c..f58cbb26323e 100644
--- a/fs/ext3/xattr.c
+++ b/fs/ext3/xattr.c
@@ -495,7 +495,8 @@ ext3_xattr_release_block(handle_t *handle, struct inode *inode,
495 BHDR(bh)->h_refcount = cpu_to_le32( 495 BHDR(bh)->h_refcount = cpu_to_le32(
496 le32_to_cpu(BHDR(bh)->h_refcount) - 1); 496 le32_to_cpu(BHDR(bh)->h_refcount) - 1);
497 error = ext3_journal_dirty_metadata(handle, bh); 497 error = ext3_journal_dirty_metadata(handle, bh);
498 handle->h_sync = 1; 498 if (IS_SYNC(inode))
499 handle->h_sync = 1;
499 DQUOT_FREE_BLOCK(inode, 1); 500 DQUOT_FREE_BLOCK(inode, 1);
500 ea_bdebug(bh, "refcount now=%d; releasing", 501 ea_bdebug(bh, "refcount now=%d; releasing",
501 le32_to_cpu(BHDR(bh)->h_refcount)); 502 le32_to_cpu(BHDR(bh)->h_refcount));
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index fbff4b9e122a..810b6d6474bf 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1147,102 +1147,37 @@ static int do_journal_get_write_access(handle_t *handle,
1147 return ext4_journal_get_write_access(handle, bh); 1147 return ext4_journal_get_write_access(handle, bh);
1148} 1148}
1149 1149
1150/*
1151 * The idea of this helper function is following:
1152 * if prepare_write has allocated some blocks, but not all of them, the
1153 * transaction must include the content of the newly allocated blocks.
1154 * This content is expected to be set to zeroes by block_prepare_write().
1155 * 2006/10/14 SAW
1156 */
1157static int ext4_prepare_failure(struct file *file, struct page *page,
1158 unsigned from, unsigned to)
1159{
1160 struct address_space *mapping;
1161 struct buffer_head *bh, *head, *next;
1162 unsigned block_start, block_end;
1163 unsigned blocksize;
1164 int ret;
1165 handle_t *handle = ext4_journal_current_handle();
1166
1167 mapping = page->mapping;
1168 if (ext4_should_writeback_data(mapping->host)) {
1169 /* optimization: no constraints about data */
1170skip:
1171 return ext4_journal_stop(handle);
1172 }
1173
1174 head = page_buffers(page);
1175 blocksize = head->b_size;
1176 for ( bh = head, block_start = 0;
1177 bh != head || !block_start;
1178 block_start = block_end, bh = next)
1179 {
1180 next = bh->b_this_page;
1181 block_end = block_start + blocksize;
1182 if (block_end <= from)
1183 continue;
1184 if (block_start >= to) {
1185 block_start = to;
1186 break;
1187 }
1188 if (!buffer_mapped(bh))
1189 /* prepare_write failed on this bh */
1190 break;
1191 if (ext4_should_journal_data(mapping->host)) {
1192 ret = do_journal_get_write_access(handle, bh);
1193 if (ret) {
1194 ext4_journal_stop(handle);
1195 return ret;
1196 }
1197 }
1198 /*
1199 * block_start here becomes the first block where the current iteration
1200 * of prepare_write failed.
1201 */
1202 }
1203 if (block_start <= from)
1204 goto skip;
1205
1206 /* commit allocated and zeroed buffers */
1207 return mapping->a_ops->commit_write(file, page, from, block_start);
1208}
1209
1210static int ext4_prepare_write(struct file *file, struct page *page, 1150static int ext4_prepare_write(struct file *file, struct page *page,
1211 unsigned from, unsigned to) 1151 unsigned from, unsigned to)
1212{ 1152{
1213 struct inode *inode = page->mapping->host; 1153 struct inode *inode = page->mapping->host;
1214 int ret, ret2; 1154 int ret, needed_blocks = ext4_writepage_trans_blocks(inode);
1215 int needed_blocks = ext4_writepage_trans_blocks(inode);
1216 handle_t *handle; 1155 handle_t *handle;
1217 int retries = 0; 1156 int retries = 0;
1218 1157
1219retry: 1158retry:
1220 handle = ext4_journal_start(inode, needed_blocks); 1159 handle = ext4_journal_start(inode, needed_blocks);
1221 if (IS_ERR(handle)) 1160 if (IS_ERR(handle)) {
1222 return PTR_ERR(handle); 1161 ret = PTR_ERR(handle);
1162 goto out;
1163 }
1223 if (test_opt(inode->i_sb, NOBH) && ext4_should_writeback_data(inode)) 1164 if (test_opt(inode->i_sb, NOBH) && ext4_should_writeback_data(inode))
1224 ret = nobh_prepare_write(page, from, to, ext4_get_block); 1165 ret = nobh_prepare_write(page, from, to, ext4_get_block);
1225 else 1166 else
1226 ret = block_prepare_write(page, from, to, ext4_get_block); 1167 ret = block_prepare_write(page, from, to, ext4_get_block);
1227 if (ret) 1168 if (ret)
1228 goto failure; 1169 goto prepare_write_failed;
1229 1170
1230 if (ext4_should_journal_data(inode)) { 1171 if (ext4_should_journal_data(inode)) {
1231 ret = walk_page_buffers(handle, page_buffers(page), 1172 ret = walk_page_buffers(handle, page_buffers(page),
1232 from, to, NULL, do_journal_get_write_access); 1173 from, to, NULL, do_journal_get_write_access);
1233 if (ret)
1234 /* fatal error, just put the handle and return */
1235 ext4_journal_stop(handle);
1236 } 1174 }
1237 return ret; 1175prepare_write_failed:
1238 1176 if (ret)
1239failure: 1177 ext4_journal_stop(handle);
1240 ret2 = ext4_prepare_failure(file, page, from, to);
1241 if (ret2 < 0)
1242 return ret2;
1243 if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries)) 1178 if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
1244 goto retry; 1179 goto retry;
1245 /* retry number exceeded, or other error like -EDQUOT */ 1180out:
1246 return ret; 1181 return ret;
1247} 1182}
1248 1183
diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c
index 406bf61ed510..8890eba1db52 100644
--- a/fs/fuse/dir.c
+++ b/fs/fuse/dir.c
@@ -195,7 +195,7 @@ static struct dentry_operations fuse_dentry_operations = {
195 .d_revalidate = fuse_dentry_revalidate, 195 .d_revalidate = fuse_dentry_revalidate,
196}; 196};
197 197
198static int valid_mode(int m) 198int fuse_valid_type(int m)
199{ 199{
200 return S_ISREG(m) || S_ISDIR(m) || S_ISLNK(m) || S_ISCHR(m) || 200 return S_ISREG(m) || S_ISDIR(m) || S_ISLNK(m) || S_ISCHR(m) ||
201 S_ISBLK(m) || S_ISFIFO(m) || S_ISSOCK(m); 201 S_ISBLK(m) || S_ISFIFO(m) || S_ISSOCK(m);
@@ -248,7 +248,8 @@ static struct dentry *fuse_lookup(struct inode *dir, struct dentry *entry,
248 fuse_put_request(fc, req); 248 fuse_put_request(fc, req);
249 /* Zero nodeid is same as -ENOENT, but with valid timeout */ 249 /* Zero nodeid is same as -ENOENT, but with valid timeout */
250 if (!err && outarg.nodeid && 250 if (!err && outarg.nodeid &&
251 (invalid_nodeid(outarg.nodeid) || !valid_mode(outarg.attr.mode))) 251 (invalid_nodeid(outarg.nodeid) ||
252 !fuse_valid_type(outarg.attr.mode)))
252 err = -EIO; 253 err = -EIO;
253 if (!err && outarg.nodeid) { 254 if (!err && outarg.nodeid) {
254 inode = fuse_iget(dir->i_sb, outarg.nodeid, outarg.generation, 255 inode = fuse_iget(dir->i_sb, outarg.nodeid, outarg.generation,
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index b98b20de7405..68ae87cbafab 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -552,3 +552,8 @@ int fuse_ctl_add_conn(struct fuse_conn *fc);
552 * Remove connection from control filesystem 552 * Remove connection from control filesystem
553 */ 553 */
554void fuse_ctl_remove_conn(struct fuse_conn *fc); 554void fuse_ctl_remove_conn(struct fuse_conn *fc);
555
556/**
557 * Is file type valid?
558 */
559int fuse_valid_type(int m);
diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index 5ab8e50e7808..608db81219a0 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -330,6 +330,8 @@ static int parse_fuse_opt(char *opt, struct fuse_mount_data *d, int is_bdev)
330 case OPT_ROOTMODE: 330 case OPT_ROOTMODE:
331 if (match_octal(&args[0], &value)) 331 if (match_octal(&args[0], &value))
332 return 0; 332 return 0;
333 if (!fuse_valid_type(value))
334 return 0;
333 d->rootmode = value; 335 d->rootmode = value;
334 d->rootmode_present = 1; 336 d->rootmode_present = 1;
335 break; 337 break;
diff --git a/fs/hostfs/hostfs_kern.c b/fs/hostfs/hostfs_kern.c
index 9baf69773ed1..fd301a910122 100644
--- a/fs/hostfs/hostfs_kern.c
+++ b/fs/hostfs/hostfs_kern.c
@@ -20,7 +20,6 @@
20#include "hostfs.h" 20#include "hostfs.h"
21#include "kern_util.h" 21#include "kern_util.h"
22#include "kern.h" 22#include "kern.h"
23#include "user_util.h"
24#include "init.h" 23#include "init.h"
25 24
26struct hostfs_inode_info { 25struct hostfs_inode_info {
@@ -939,7 +938,7 @@ static const struct address_space_operations hostfs_link_aops = {
939static int hostfs_fill_sb_common(struct super_block *sb, void *d, int silent) 938static int hostfs_fill_sb_common(struct super_block *sb, void *d, int silent)
940{ 939{
941 struct inode *root_inode; 940 struct inode *root_inode;
942 char *name, *data = d; 941 char *host_root_path, *req_root = d;
943 int err; 942 int err;
944 943
945 sb->s_blocksize = 1024; 944 sb->s_blocksize = 1024;
@@ -948,16 +947,16 @@ static int hostfs_fill_sb_common(struct super_block *sb, void *d, int silent)
948 sb->s_op = &hostfs_sbops; 947 sb->s_op = &hostfs_sbops;
949 948
950 /* NULL is printed as <NULL> by sprintf: avoid that. */ 949 /* NULL is printed as <NULL> by sprintf: avoid that. */
951 if (data == NULL) 950 if (req_root == NULL)
952 data = ""; 951 req_root = "";
953 952
954 err = -ENOMEM; 953 err = -ENOMEM;
955 name = kmalloc(strlen(root_ino) + 1 954 host_root_path = kmalloc(strlen(root_ino) + 1
956 + strlen(data) + 1, GFP_KERNEL); 955 + strlen(req_root) + 1, GFP_KERNEL);
957 if(name == NULL) 956 if(host_root_path == NULL)
958 goto out; 957 goto out;
959 958
960 sprintf(name, "%s/%s", root_ino, data); 959 sprintf(host_root_path, "%s/%s", root_ino, req_root);
961 960
962 root_inode = iget(sb, 0); 961 root_inode = iget(sb, 0);
963 if(root_inode == NULL) 962 if(root_inode == NULL)
@@ -967,10 +966,10 @@ static int hostfs_fill_sb_common(struct super_block *sb, void *d, int silent)
967 if(err) 966 if(err)
968 goto out_put; 967 goto out_put;
969 968
970 HOSTFS_I(root_inode)->host_filename = name; 969 HOSTFS_I(root_inode)->host_filename = host_root_path;
971 /* Avoid that in the error path, iput(root_inode) frees again name through 970 /* Avoid that in the error path, iput(root_inode) frees again
972 * hostfs_destroy_inode! */ 971 * host_root_path through hostfs_destroy_inode! */
973 name = NULL; 972 host_root_path = NULL;
974 973
975 err = -ENOMEM; 974 err = -ENOMEM;
976 sb->s_root = d_alloc_root(root_inode); 975 sb->s_root = d_alloc_root(root_inode);
@@ -990,7 +989,7 @@ static int hostfs_fill_sb_common(struct super_block *sb, void *d, int silent)
990 out_put: 989 out_put:
991 iput(root_inode); 990 iput(root_inode);
992 out_free: 991 out_free:
993 kfree(name); 992 kfree(host_root_path);
994 out: 993 out:
995 return(err); 994 return(err);
996} 995}
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 92d8ec859e22..cd3469720cbf 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1684,7 +1684,8 @@ go_ahead:
1684 * ... prune child dentries and writebacks if needed. 1684 * ... prune child dentries and writebacks if needed.
1685 */ 1685 */
1686 if (atomic_read(&old_dentry->d_count) > 1) { 1686 if (atomic_read(&old_dentry->d_count) > 1) {
1687 nfs_wb_all(old_inode); 1687 if (S_ISREG(old_inode->i_mode))
1688 nfs_wb_all(old_inode);
1688 shrink_dcache_parent(old_dentry); 1689 shrink_dcache_parent(old_dentry);
1689 } 1690 }
1690 nfs_inode_return_delegation(old_inode); 1691 nfs_inode_return_delegation(old_inode);
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index b1c98ea39b72..2877744cb606 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -432,10 +432,10 @@ static void nfs_direct_commit_result(struct rpc_task *task, void *calldata)
432 if (NFS_PROTO(data->inode)->commit_done(task, data) != 0) 432 if (NFS_PROTO(data->inode)->commit_done(task, data) != 0)
433 return; 433 return;
434 if (unlikely(task->tk_status < 0)) { 434 if (unlikely(task->tk_status < 0)) {
435 dreq->error = task->tk_status; 435 dprintk("NFS: %5u commit failed with error %d.\n",
436 task->tk_pid, task->tk_status);
436 dreq->flags = NFS_ODIRECT_RESCHED_WRITES; 437 dreq->flags = NFS_ODIRECT_RESCHED_WRITES;
437 } 438 } else if (memcmp(&dreq->verf, &data->verf, sizeof(data->verf))) {
438 if (memcmp(&dreq->verf, &data->verf, sizeof(data->verf))) {
439 dprintk("NFS: %5u commit verify failed\n", task->tk_pid); 439 dprintk("NFS: %5u commit verify failed\n", task->tk_pid);
440 dreq->flags = NFS_ODIRECT_RESCHED_WRITES; 440 dreq->flags = NFS_ODIRECT_RESCHED_WRITES;
441 } 441 }
@@ -531,9 +531,12 @@ static void nfs_direct_write_result(struct rpc_task *task, void *calldata)
531 531
532 spin_lock(&dreq->lock); 532 spin_lock(&dreq->lock);
533 533
534 if (unlikely(dreq->error != 0))
535 goto out_unlock;
534 if (unlikely(status < 0)) { 536 if (unlikely(status < 0)) {
537 /* An error has occured, so we should not commit */
538 dreq->flags = 0;
535 dreq->error = status; 539 dreq->error = status;
536 goto out_unlock;
537 } 540 }
538 541
539 dreq->count += data->res.count; 542 dreq->count += data->res.count;
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index af53c02f473b..44aa9b726573 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -341,8 +341,10 @@ nfs_setattr(struct dentry *dentry, struct iattr *attr)
341 lock_kernel(); 341 lock_kernel();
342 nfs_begin_data_update(inode); 342 nfs_begin_data_update(inode);
343 /* Write all dirty data */ 343 /* Write all dirty data */
344 filemap_write_and_wait(inode->i_mapping); 344 if (S_ISREG(inode->i_mode)) {
345 nfs_wb_all(inode); 345 filemap_write_and_wait(inode->i_mapping);
346 nfs_wb_all(inode);
347 }
346 /* 348 /*
347 * Return any delegations if we're going to change ACLs 349 * Return any delegations if we're going to change ACLs
348 */ 350 */
@@ -429,7 +431,8 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
429 int err; 431 int err;
430 432
431 /* Flush out writes to the server in order to update c/mtime */ 433 /* Flush out writes to the server in order to update c/mtime */
432 nfs_sync_mapping_range(inode->i_mapping, 0, 0, FLUSH_NOCOMMIT); 434 if (S_ISREG(inode->i_mode))
435 nfs_sync_mapping_range(inode->i_mapping, 0, 0, FLUSH_NOCOMMIT);
433 436
434 /* 437 /*
435 * We may force a getattr if the user cares about atime. 438 * We may force a getattr if the user cares about atime.
diff --git a/fs/nfs/super.c b/fs/nfs/super.c
index bb516a2cfbaf..f1eae44b9a1a 100644
--- a/fs/nfs/super.c
+++ b/fs/nfs/super.c
@@ -151,10 +151,10 @@ int __init register_nfs_fs(void)
151 if (ret < 0) 151 if (ret < 0)
152 goto error_0; 152 goto error_0;
153 153
154#ifdef CONFIG_NFS_V4
155 ret = nfs_register_sysctl(); 154 ret = nfs_register_sysctl();
156 if (ret < 0) 155 if (ret < 0)
157 goto error_1; 156 goto error_1;
157#ifdef CONFIG_NFS_V4
158 ret = register_filesystem(&nfs4_fs_type); 158 ret = register_filesystem(&nfs4_fs_type);
159 if (ret < 0) 159 if (ret < 0)
160 goto error_2; 160 goto error_2;
@@ -165,9 +165,9 @@ int __init register_nfs_fs(void)
165#ifdef CONFIG_NFS_V4 165#ifdef CONFIG_NFS_V4
166error_2: 166error_2:
167 nfs_unregister_sysctl(); 167 nfs_unregister_sysctl();
168#endif
168error_1: 169error_1:
169 unregister_filesystem(&nfs_fs_type); 170 unregister_filesystem(&nfs_fs_type);
170#endif
171error_0: 171error_0:
172 return ret; 172 return ret;
173} 173}
diff --git a/fs/nfs/sysctl.c b/fs/nfs/sysctl.c
index fcdcafbb3293..b62481dabae9 100644
--- a/fs/nfs/sysctl.c
+++ b/fs/nfs/sysctl.c
@@ -50,6 +50,14 @@ static ctl_table nfs_cb_sysctls[] = {
50 .proc_handler = &proc_dointvec_jiffies, 50 .proc_handler = &proc_dointvec_jiffies,
51 .strategy = &sysctl_jiffies, 51 .strategy = &sysctl_jiffies,
52 }, 52 },
53 {
54 .ctl_name = CTL_UNNUMBERED,
55 .procname = "nfs_congestion_kb",
56 .data = &nfs_congestion_kb,
57 .maxlen = sizeof(nfs_congestion_kb),
58 .mode = 0644,
59 .proc_handler = &proc_dointvec,
60 },
53 { .ctl_name = 0 } 61 { .ctl_name = 0 }
54}; 62};
55 63
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index febdade91670..797558941745 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -12,6 +12,7 @@
12#include <linux/pagemap.h> 12#include <linux/pagemap.h>
13#include <linux/file.h> 13#include <linux/file.h>
14#include <linux/writeback.h> 14#include <linux/writeback.h>
15#include <linux/swap.h>
15 16
16#include <linux/sunrpc/clnt.h> 17#include <linux/sunrpc/clnt.h>
17#include <linux/nfs_fs.h> 18#include <linux/nfs_fs.h>
@@ -37,8 +38,6 @@
37static struct nfs_page * nfs_update_request(struct nfs_open_context*, 38static struct nfs_page * nfs_update_request(struct nfs_open_context*,
38 struct page *, 39 struct page *,
39 unsigned int, unsigned int); 40 unsigned int, unsigned int);
40static void nfs_mark_request_dirty(struct nfs_page *req);
41static int nfs_wait_on_write_congestion(struct address_space *, int);
42static long nfs_flush_mapping(struct address_space *mapping, struct writeback_control *wbc, int how); 41static long nfs_flush_mapping(struct address_space *mapping, struct writeback_control *wbc, int how);
43static const struct rpc_call_ops nfs_write_partial_ops; 42static const struct rpc_call_ops nfs_write_partial_ops;
44static const struct rpc_call_ops nfs_write_full_ops; 43static const struct rpc_call_ops nfs_write_full_ops;
@@ -48,8 +47,6 @@ static struct kmem_cache *nfs_wdata_cachep;
48static mempool_t *nfs_wdata_mempool; 47static mempool_t *nfs_wdata_mempool;
49static mempool_t *nfs_commit_mempool; 48static mempool_t *nfs_commit_mempool;
50 49
51static DECLARE_WAIT_QUEUE_HEAD(nfs_write_congestion);
52
53struct nfs_write_data *nfs_commit_alloc(void) 50struct nfs_write_data *nfs_commit_alloc(void)
54{ 51{
55 struct nfs_write_data *p = mempool_alloc(nfs_commit_mempool, GFP_NOFS); 52 struct nfs_write_data *p = mempool_alloc(nfs_commit_mempool, GFP_NOFS);
@@ -211,6 +208,43 @@ static int wb_priority(struct writeback_control *wbc)
211} 208}
212 209
213/* 210/*
211 * NFS congestion control
212 */
213
214int nfs_congestion_kb;
215
216#define NFS_CONGESTION_ON_THRESH (nfs_congestion_kb >> (PAGE_SHIFT-10))
217#define NFS_CONGESTION_OFF_THRESH \
218 (NFS_CONGESTION_ON_THRESH - (NFS_CONGESTION_ON_THRESH >> 2))
219
220static int nfs_set_page_writeback(struct page *page)
221{
222 int ret = test_set_page_writeback(page);
223
224 if (!ret) {
225 struct inode *inode = page->mapping->host;
226 struct nfs_server *nfss = NFS_SERVER(inode);
227
228 if (atomic_inc_return(&nfss->writeback) >
229 NFS_CONGESTION_ON_THRESH)
230 set_bdi_congested(&nfss->backing_dev_info, WRITE);
231 }
232 return ret;
233}
234
235static void nfs_end_page_writeback(struct page *page)
236{
237 struct inode *inode = page->mapping->host;
238 struct nfs_server *nfss = NFS_SERVER(inode);
239
240 end_page_writeback(page);
241 if (atomic_dec_return(&nfss->writeback) < NFS_CONGESTION_OFF_THRESH) {
242 clear_bdi_congested(&nfss->backing_dev_info, WRITE);
243 congestion_end(WRITE);
244 }
245}
246
247/*
214 * Find an associated nfs write request, and prepare to flush it out 248 * Find an associated nfs write request, and prepare to flush it out
215 * Returns 1 if there was no write request, or if the request was 249 * Returns 1 if there was no write request, or if the request was
216 * already tagged by nfs_set_page_dirty.Returns 0 if the request 250 * already tagged by nfs_set_page_dirty.Returns 0 if the request
@@ -220,7 +254,8 @@ static int wb_priority(struct writeback_control *wbc)
220static int nfs_page_mark_flush(struct page *page) 254static int nfs_page_mark_flush(struct page *page)
221{ 255{
222 struct nfs_page *req; 256 struct nfs_page *req;
223 spinlock_t *req_lock = &NFS_I(page->mapping->host)->req_lock; 257 struct nfs_inode *nfsi = NFS_I(page->mapping->host);
258 spinlock_t *req_lock = &nfsi->req_lock;
224 int ret; 259 int ret;
225 260
226 spin_lock(req_lock); 261 spin_lock(req_lock);
@@ -244,11 +279,23 @@ static int nfs_page_mark_flush(struct page *page)
244 return ret; 279 return ret;
245 spin_lock(req_lock); 280 spin_lock(req_lock);
246 } 281 }
247 spin_unlock(req_lock); 282 if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) {
248 if (test_and_set_bit(PG_FLUSHING, &req->wb_flags) == 0) { 283 /* This request is marked for commit */
249 nfs_mark_request_dirty(req); 284 spin_unlock(req_lock);
250 set_page_writeback(page); 285 nfs_unlock_request(req);
286 return 1;
251 } 287 }
288 if (nfs_set_page_writeback(page) == 0) {
289 nfs_list_remove_request(req);
290 /* add the request to the inode's dirty list. */
291 radix_tree_tag_set(&nfsi->nfs_page_tree,
292 req->wb_index, NFS_PAGE_TAG_DIRTY);
293 nfs_list_add_request(req, &nfsi->dirty);
294 nfsi->ndirty++;
295 spin_unlock(req_lock);
296 __mark_inode_dirty(page->mapping->host, I_DIRTY_PAGES);
297 } else
298 spin_unlock(req_lock);
252 ret = test_bit(PG_NEED_FLUSH, &req->wb_flags); 299 ret = test_bit(PG_NEED_FLUSH, &req->wb_flags);
253 nfs_unlock_request(req); 300 nfs_unlock_request(req);
254 return ret; 301 return ret;
@@ -302,13 +349,8 @@ int nfs_writepage(struct page *page, struct writeback_control *wbc)
302 return err; 349 return err;
303} 350}
304 351
305/*
306 * Note: causes nfs_update_request() to block on the assumption
307 * that the writeback is generated due to memory pressure.
308 */
309int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc) 352int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
310{ 353{
311 struct backing_dev_info *bdi = mapping->backing_dev_info;
312 struct inode *inode = mapping->host; 354 struct inode *inode = mapping->host;
313 int err; 355 int err;
314 356
@@ -317,20 +359,12 @@ int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
317 err = generic_writepages(mapping, wbc); 359 err = generic_writepages(mapping, wbc);
318 if (err) 360 if (err)
319 return err; 361 return err;
320 while (test_and_set_bit(BDI_write_congested, &bdi->state) != 0) {
321 if (wbc->nonblocking)
322 return 0;
323 nfs_wait_on_write_congestion(mapping, 0);
324 }
325 err = nfs_flush_mapping(mapping, wbc, wb_priority(wbc)); 362 err = nfs_flush_mapping(mapping, wbc, wb_priority(wbc));
326 if (err < 0) 363 if (err < 0)
327 goto out; 364 goto out;
328 nfs_add_stats(inode, NFSIOS_WRITEPAGES, err); 365 nfs_add_stats(inode, NFSIOS_WRITEPAGES, err);
329 err = 0; 366 err = 0;
330out: 367out:
331 clear_bit(BDI_write_congested, &bdi->state);
332 wake_up_all(&nfs_write_congestion);
333 congestion_end(WRITE);
334 return err; 368 return err;
335} 369}
336 370
@@ -354,13 +388,15 @@ static int nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
354 } 388 }
355 SetPagePrivate(req->wb_page); 389 SetPagePrivate(req->wb_page);
356 set_page_private(req->wb_page, (unsigned long)req); 390 set_page_private(req->wb_page, (unsigned long)req);
391 if (PageDirty(req->wb_page))
392 set_bit(PG_NEED_FLUSH, &req->wb_flags);
357 nfsi->npages++; 393 nfsi->npages++;
358 atomic_inc(&req->wb_count); 394 atomic_inc(&req->wb_count);
359 return 0; 395 return 0;
360} 396}
361 397
362/* 398/*
363 * Insert a write request into an inode 399 * Remove a write request from an inode
364 */ 400 */
365static void nfs_inode_remove_request(struct nfs_page *req) 401static void nfs_inode_remove_request(struct nfs_page *req)
366{ 402{
@@ -373,6 +409,8 @@ static void nfs_inode_remove_request(struct nfs_page *req)
373 set_page_private(req->wb_page, 0); 409 set_page_private(req->wb_page, 0);
374 ClearPagePrivate(req->wb_page); 410 ClearPagePrivate(req->wb_page);
375 radix_tree_delete(&nfsi->nfs_page_tree, req->wb_index); 411 radix_tree_delete(&nfsi->nfs_page_tree, req->wb_index);
412 if (test_and_clear_bit(PG_NEED_FLUSH, &req->wb_flags))
413 __set_page_dirty_nobuffers(req->wb_page);
376 nfsi->npages--; 414 nfsi->npages--;
377 if (!nfsi->npages) { 415 if (!nfsi->npages) {
378 spin_unlock(&nfsi->req_lock); 416 spin_unlock(&nfsi->req_lock);
@@ -384,28 +422,9 @@ static void nfs_inode_remove_request(struct nfs_page *req)
384 nfs_release_request(req); 422 nfs_release_request(req);
385} 423}
386 424
387/*
388 * Add a request to the inode's dirty list.
389 */
390static void
391nfs_mark_request_dirty(struct nfs_page *req)
392{
393 struct inode *inode = req->wb_context->dentry->d_inode;
394 struct nfs_inode *nfsi = NFS_I(inode);
395
396 spin_lock(&nfsi->req_lock);
397 radix_tree_tag_set(&nfsi->nfs_page_tree,
398 req->wb_index, NFS_PAGE_TAG_DIRTY);
399 nfs_list_add_request(req, &nfsi->dirty);
400 nfsi->ndirty++;
401 spin_unlock(&nfsi->req_lock);
402 __mark_inode_dirty(inode, I_DIRTY_PAGES);
403}
404
405static void 425static void
406nfs_redirty_request(struct nfs_page *req) 426nfs_redirty_request(struct nfs_page *req)
407{ 427{
408 clear_bit(PG_FLUSHING, &req->wb_flags);
409 __set_page_dirty_nobuffers(req->wb_page); 428 __set_page_dirty_nobuffers(req->wb_page);
410} 429}
411 430
@@ -415,7 +434,11 @@ nfs_redirty_request(struct nfs_page *req)
415static inline int 434static inline int
416nfs_dirty_request(struct nfs_page *req) 435nfs_dirty_request(struct nfs_page *req)
417{ 436{
418 return test_bit(PG_FLUSHING, &req->wb_flags) == 0; 437 struct page *page = req->wb_page;
438
439 if (page == NULL || test_bit(PG_NEED_COMMIT, &req->wb_flags))
440 return 0;
441 return !PageWriteback(req->wb_page);
419} 442}
420 443
421#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4) 444#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
@@ -431,10 +454,48 @@ nfs_mark_request_commit(struct nfs_page *req)
431 spin_lock(&nfsi->req_lock); 454 spin_lock(&nfsi->req_lock);
432 nfs_list_add_request(req, &nfsi->commit); 455 nfs_list_add_request(req, &nfsi->commit);
433 nfsi->ncommit++; 456 nfsi->ncommit++;
457 set_bit(PG_NEED_COMMIT, &(req)->wb_flags);
434 spin_unlock(&nfsi->req_lock); 458 spin_unlock(&nfsi->req_lock);
435 inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS); 459 inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
436 __mark_inode_dirty(inode, I_DIRTY_DATASYNC); 460 __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
437} 461}
462
463static inline
464int nfs_write_need_commit(struct nfs_write_data *data)
465{
466 return data->verf.committed != NFS_FILE_SYNC;
467}
468
469static inline
470int nfs_reschedule_unstable_write(struct nfs_page *req)
471{
472 if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) {
473 nfs_mark_request_commit(req);
474 return 1;
475 }
476 if (test_and_clear_bit(PG_NEED_RESCHED, &req->wb_flags)) {
477 nfs_redirty_request(req);
478 return 1;
479 }
480 return 0;
481}
482#else
483static inline void
484nfs_mark_request_commit(struct nfs_page *req)
485{
486}
487
488static inline
489int nfs_write_need_commit(struct nfs_write_data *data)
490{
491 return 0;
492}
493
494static inline
495int nfs_reschedule_unstable_write(struct nfs_page *req)
496{
497 return 0;
498}
438#endif 499#endif
439 500
440/* 501/*
@@ -481,6 +542,7 @@ static void nfs_cancel_dirty_list(struct list_head *head)
481 while(!list_empty(head)) { 542 while(!list_empty(head)) {
482 req = nfs_list_entry(head->next); 543 req = nfs_list_entry(head->next);
483 nfs_list_remove_request(req); 544 nfs_list_remove_request(req);
545 nfs_end_page_writeback(req->wb_page);
484 nfs_inode_remove_request(req); 546 nfs_inode_remove_request(req);
485 nfs_clear_page_writeback(req); 547 nfs_clear_page_writeback(req);
486 } 548 }
@@ -494,6 +556,7 @@ static void nfs_cancel_commit_list(struct list_head *head)
494 req = nfs_list_entry(head->next); 556 req = nfs_list_entry(head->next);
495 dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS); 557 dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
496 nfs_list_remove_request(req); 558 nfs_list_remove_request(req);
559 clear_bit(PG_NEED_COMMIT, &(req)->wb_flags);
497 nfs_inode_remove_request(req); 560 nfs_inode_remove_request(req);
498 nfs_unlock_request(req); 561 nfs_unlock_request(req);
499 } 562 }
@@ -531,10 +594,10 @@ static inline int nfs_scan_commit(struct inode *inode, struct list_head *dst, un
531} 594}
532#endif 595#endif
533 596
534static int nfs_wait_on_write_congestion(struct address_space *mapping, int intr) 597static int nfs_wait_on_write_congestion(struct address_space *mapping)
535{ 598{
599 struct inode *inode = mapping->host;
536 struct backing_dev_info *bdi = mapping->backing_dev_info; 600 struct backing_dev_info *bdi = mapping->backing_dev_info;
537 DEFINE_WAIT(wait);
538 int ret = 0; 601 int ret = 0;
539 602
540 might_sleep(); 603 might_sleep();
@@ -542,31 +605,23 @@ static int nfs_wait_on_write_congestion(struct address_space *mapping, int intr)
542 if (!bdi_write_congested(bdi)) 605 if (!bdi_write_congested(bdi))
543 return 0; 606 return 0;
544 607
545 nfs_inc_stats(mapping->host, NFSIOS_CONGESTIONWAIT); 608 nfs_inc_stats(inode, NFSIOS_CONGESTIONWAIT);
546 609
547 if (intr) { 610 do {
548 struct rpc_clnt *clnt = NFS_CLIENT(mapping->host); 611 struct rpc_clnt *clnt = NFS_CLIENT(inode);
549 sigset_t oldset; 612 sigset_t oldset;
550 613
551 rpc_clnt_sigmask(clnt, &oldset); 614 rpc_clnt_sigmask(clnt, &oldset);
552 prepare_to_wait(&nfs_write_congestion, &wait, TASK_INTERRUPTIBLE); 615 ret = congestion_wait_interruptible(WRITE, HZ/10);
553 if (bdi_write_congested(bdi)) {
554 if (signalled())
555 ret = -ERESTARTSYS;
556 else
557 schedule();
558 }
559 rpc_clnt_sigunmask(clnt, &oldset); 616 rpc_clnt_sigunmask(clnt, &oldset);
560 } else { 617 if (ret == -ERESTARTSYS)
561 prepare_to_wait(&nfs_write_congestion, &wait, TASK_UNINTERRUPTIBLE); 618 break;
562 if (bdi_write_congested(bdi)) 619 ret = 0;
563 schedule(); 620 } while (bdi_write_congested(bdi));
564 } 621
565 finish_wait(&nfs_write_congestion, &wait);
566 return ret; 622 return ret;
567} 623}
568 624
569
570/* 625/*
571 * Try to update any existing write request, or create one if there is none. 626 * Try to update any existing write request, or create one if there is none.
572 * In order to match, the request's credentials must match those of 627 * In order to match, the request's credentials must match those of
@@ -577,14 +632,15 @@ static int nfs_wait_on_write_congestion(struct address_space *mapping, int intr)
577static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx, 632static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx,
578 struct page *page, unsigned int offset, unsigned int bytes) 633 struct page *page, unsigned int offset, unsigned int bytes)
579{ 634{
580 struct inode *inode = page->mapping->host; 635 struct address_space *mapping = page->mapping;
636 struct inode *inode = mapping->host;
581 struct nfs_inode *nfsi = NFS_I(inode); 637 struct nfs_inode *nfsi = NFS_I(inode);
582 struct nfs_page *req, *new = NULL; 638 struct nfs_page *req, *new = NULL;
583 unsigned long rqend, end; 639 unsigned long rqend, end;
584 640
585 end = offset + bytes; 641 end = offset + bytes;
586 642
587 if (nfs_wait_on_write_congestion(page->mapping, NFS_SERVER(inode)->flags & NFS_MOUNT_INTR)) 643 if (nfs_wait_on_write_congestion(mapping))
588 return ERR_PTR(-ERESTARTSYS); 644 return ERR_PTR(-ERESTARTSYS);
589 for (;;) { 645 for (;;) {
590 /* Loop over all inode entries and see if we find 646 /* Loop over all inode entries and see if we find
@@ -727,26 +783,12 @@ int nfs_updatepage(struct file *file, struct page *page,
727 783
728static void nfs_writepage_release(struct nfs_page *req) 784static void nfs_writepage_release(struct nfs_page *req)
729{ 785{
730 end_page_writeback(req->wb_page);
731 786
732#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4) 787 if (PageError(req->wb_page) || !nfs_reschedule_unstable_write(req)) {
733 if (!PageError(req->wb_page)) { 788 nfs_end_page_writeback(req->wb_page);
734 if (NFS_NEED_RESCHED(req)) { 789 nfs_inode_remove_request(req);
735 nfs_redirty_request(req); 790 } else
736 goto out; 791 nfs_end_page_writeback(req->wb_page);
737 } else if (NFS_NEED_COMMIT(req)) {
738 nfs_mark_request_commit(req);
739 goto out;
740 }
741 }
742 nfs_inode_remove_request(req);
743
744out:
745 nfs_clear_commit(req);
746 nfs_clear_reschedule(req);
747#else
748 nfs_inode_remove_request(req);
749#endif
750 nfs_clear_page_writeback(req); 792 nfs_clear_page_writeback(req);
751} 793}
752 794
@@ -879,6 +921,7 @@ out_bad:
879 nfs_writedata_release(data); 921 nfs_writedata_release(data);
880 } 922 }
881 nfs_redirty_request(req); 923 nfs_redirty_request(req);
924 nfs_end_page_writeback(req->wb_page);
882 nfs_clear_page_writeback(req); 925 nfs_clear_page_writeback(req);
883 return -ENOMEM; 926 return -ENOMEM;
884} 927}
@@ -924,6 +967,7 @@ static int nfs_flush_one(struct inode *inode, struct list_head *head, int how)
924 struct nfs_page *req = nfs_list_entry(head->next); 967 struct nfs_page *req = nfs_list_entry(head->next);
925 nfs_list_remove_request(req); 968 nfs_list_remove_request(req);
926 nfs_redirty_request(req); 969 nfs_redirty_request(req);
970 nfs_end_page_writeback(req->wb_page);
927 nfs_clear_page_writeback(req); 971 nfs_clear_page_writeback(req);
928 } 972 }
929 return -ENOMEM; 973 return -ENOMEM;
@@ -959,6 +1003,7 @@ out_err:
959 req = nfs_list_entry(head->next); 1003 req = nfs_list_entry(head->next);
960 nfs_list_remove_request(req); 1004 nfs_list_remove_request(req);
961 nfs_redirty_request(req); 1005 nfs_redirty_request(req);
1006 nfs_end_page_writeback(req->wb_page);
962 nfs_clear_page_writeback(req); 1007 nfs_clear_page_writeback(req);
963 } 1008 }
964 return error; 1009 return error;
@@ -986,22 +1031,28 @@ static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata)
986 nfs_set_pageerror(page); 1031 nfs_set_pageerror(page);
987 req->wb_context->error = task->tk_status; 1032 req->wb_context->error = task->tk_status;
988 dprintk(", error = %d\n", task->tk_status); 1033 dprintk(", error = %d\n", task->tk_status);
989 } else { 1034 goto out;
990#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
991 if (data->verf.committed < NFS_FILE_SYNC) {
992 if (!NFS_NEED_COMMIT(req)) {
993 nfs_defer_commit(req);
994 memcpy(&req->wb_verf, &data->verf, sizeof(req->wb_verf));
995 dprintk(" defer commit\n");
996 } else if (memcmp(&req->wb_verf, &data->verf, sizeof(req->wb_verf))) {
997 nfs_defer_reschedule(req);
998 dprintk(" server reboot detected\n");
999 }
1000 } else
1001#endif
1002 dprintk(" OK\n");
1003 } 1035 }
1004 1036
1037 if (nfs_write_need_commit(data)) {
1038 spinlock_t *req_lock = &NFS_I(page->mapping->host)->req_lock;
1039
1040 spin_lock(req_lock);
1041 if (test_bit(PG_NEED_RESCHED, &req->wb_flags)) {
1042 /* Do nothing we need to resend the writes */
1043 } else if (!test_and_set_bit(PG_NEED_COMMIT, &req->wb_flags)) {
1044 memcpy(&req->wb_verf, &data->verf, sizeof(req->wb_verf));
1045 dprintk(" defer commit\n");
1046 } else if (memcmp(&req->wb_verf, &data->verf, sizeof(req->wb_verf))) {
1047 set_bit(PG_NEED_RESCHED, &req->wb_flags);
1048 clear_bit(PG_NEED_COMMIT, &req->wb_flags);
1049 dprintk(" server reboot detected\n");
1050 }
1051 spin_unlock(req_lock);
1052 } else
1053 dprintk(" OK\n");
1054
1055out:
1005 if (atomic_dec_and_test(&req->wb_complete)) 1056 if (atomic_dec_and_test(&req->wb_complete))
1006 nfs_writepage_release(req); 1057 nfs_writepage_release(req);
1007} 1058}
@@ -1042,25 +1093,21 @@ static void nfs_writeback_done_full(struct rpc_task *task, void *calldata)
1042 if (task->tk_status < 0) { 1093 if (task->tk_status < 0) {
1043 nfs_set_pageerror(page); 1094 nfs_set_pageerror(page);
1044 req->wb_context->error = task->tk_status; 1095 req->wb_context->error = task->tk_status;
1045 end_page_writeback(page);
1046 nfs_inode_remove_request(req);
1047 dprintk(", error = %d\n", task->tk_status); 1096 dprintk(", error = %d\n", task->tk_status);
1048 goto next; 1097 goto remove_request;
1049 } 1098 }
1050 end_page_writeback(page);
1051 1099
1052#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4) 1100 if (nfs_write_need_commit(data)) {
1053 if (data->args.stable != NFS_UNSTABLE || data->verf.committed == NFS_FILE_SYNC) { 1101 memcpy(&req->wb_verf, &data->verf, sizeof(req->wb_verf));
1054 nfs_inode_remove_request(req); 1102 nfs_mark_request_commit(req);
1055 dprintk(" OK\n"); 1103 nfs_end_page_writeback(page);
1104 dprintk(" marked for commit\n");
1056 goto next; 1105 goto next;
1057 } 1106 }
1058 memcpy(&req->wb_verf, &data->verf, sizeof(req->wb_verf)); 1107 dprintk(" OK\n");
1059 nfs_mark_request_commit(req); 1108remove_request:
1060 dprintk(" marked for commit\n"); 1109 nfs_end_page_writeback(page);
1061#else
1062 nfs_inode_remove_request(req); 1110 nfs_inode_remove_request(req);
1063#endif
1064 next: 1111 next:
1065 nfs_clear_page_writeback(req); 1112 nfs_clear_page_writeback(req);
1066 } 1113 }
@@ -1248,6 +1295,7 @@ static void nfs_commit_done(struct rpc_task *task, void *calldata)
1248 while (!list_empty(&data->pages)) { 1295 while (!list_empty(&data->pages)) {
1249 req = nfs_list_entry(data->pages.next); 1296 req = nfs_list_entry(data->pages.next);
1250 nfs_list_remove_request(req); 1297 nfs_list_remove_request(req);
1298 clear_bit(PG_NEED_COMMIT, &(req)->wb_flags);
1251 dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS); 1299 dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
1252 1300
1253 dprintk("NFS: commit (%s/%Ld %d@%Ld)", 1301 dprintk("NFS: commit (%s/%Ld %d@%Ld)",
@@ -1483,15 +1531,22 @@ int nfs_wb_page(struct inode *inode, struct page* page)
1483 1531
1484int nfs_set_page_dirty(struct page *page) 1532int nfs_set_page_dirty(struct page *page)
1485{ 1533{
1534 spinlock_t *req_lock = &NFS_I(page->mapping->host)->req_lock;
1486 struct nfs_page *req; 1535 struct nfs_page *req;
1536 int ret;
1487 1537
1488 req = nfs_page_find_request(page); 1538 spin_lock(req_lock);
1539 req = nfs_page_find_request_locked(page);
1489 if (req != NULL) { 1540 if (req != NULL) {
1490 /* Mark any existing write requests for flushing */ 1541 /* Mark any existing write requests for flushing */
1491 set_bit(PG_NEED_FLUSH, &req->wb_flags); 1542 ret = !test_and_set_bit(PG_NEED_FLUSH, &req->wb_flags);
1543 spin_unlock(req_lock);
1492 nfs_release_request(req); 1544 nfs_release_request(req);
1545 return ret;
1493 } 1546 }
1494 return __set_page_dirty_nobuffers(page); 1547 ret = __set_page_dirty_nobuffers(page);
1548 spin_unlock(req_lock);
1549 return ret;
1495} 1550}
1496 1551
1497 1552
@@ -1514,6 +1569,26 @@ int __init nfs_init_writepagecache(void)
1514 if (nfs_commit_mempool == NULL) 1569 if (nfs_commit_mempool == NULL)
1515 return -ENOMEM; 1570 return -ENOMEM;
1516 1571
1572 /*
1573 * NFS congestion size, scale with available memory.
1574 *
1575 * 64MB: 8192k
1576 * 128MB: 11585k
1577 * 256MB: 16384k
1578 * 512MB: 23170k
1579 * 1GB: 32768k
1580 * 2GB: 46340k
1581 * 4GB: 65536k
1582 * 8GB: 92681k
1583 * 16GB: 131072k
1584 *
1585 * This allows larger machines to have larger/more transfers.
1586 * Limit the default to 256M
1587 */
1588 nfs_congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10);
1589 if (nfs_congestion_kb > 256*1024)
1590 nfs_congestion_kb = 256*1024;
1591
1517 return 0; 1592 return 0;
1518} 1593}
1519 1594
diff --git a/fs/nfsd/nfs3xdr.c b/fs/nfsd/nfs3xdr.c
index 6f677988c71d..7e4bb0af24d7 100644
--- a/fs/nfsd/nfs3xdr.c
+++ b/fs/nfsd/nfs3xdr.c
@@ -859,8 +859,8 @@ compose_entry_fh(struct nfsd3_readdirres *cd, struct svc_fh *fhp,
859#define NFS3_ENTRY_BAGGAGE (2 + 1 + 2 + 1) 859#define NFS3_ENTRY_BAGGAGE (2 + 1 + 2 + 1)
860#define NFS3_ENTRYPLUS_BAGGAGE (1 + 21 + 1 + (NFS3_FHSIZE >> 2)) 860#define NFS3_ENTRYPLUS_BAGGAGE (1 + 21 + 1 + (NFS3_FHSIZE >> 2))
861static int 861static int
862encode_entry(struct readdir_cd *ccd, const char *name, 862encode_entry(struct readdir_cd *ccd, const char *name, int namlen,
863 int namlen, off_t offset, ino_t ino, unsigned int d_type, int plus) 863 loff_t offset, ino_t ino, unsigned int d_type, int plus)
864{ 864{
865 struct nfsd3_readdirres *cd = container_of(ccd, struct nfsd3_readdirres, 865 struct nfsd3_readdirres *cd = container_of(ccd, struct nfsd3_readdirres,
866 common); 866 common);
@@ -880,7 +880,7 @@ encode_entry(struct readdir_cd *ccd, const char *name,
880 *cd->offset1 = htonl(offset64 & 0xffffffff); 880 *cd->offset1 = htonl(offset64 & 0xffffffff);
881 cd->offset1 = NULL; 881 cd->offset1 = NULL;
882 } else { 882 } else {
883 xdr_encode_hyper(cd->offset, (u64) offset); 883 xdr_encode_hyper(cd->offset, offset64);
884 } 884 }
885 } 885 }
886 886
diff --git a/fs/nfsd/nfs4acl.c b/fs/nfsd/nfs4acl.c
index 832673b14587..673a53c014a3 100644
--- a/fs/nfsd/nfs4acl.c
+++ b/fs/nfsd/nfs4acl.c
@@ -228,7 +228,7 @@ _posix_to_nfsv4_one(struct posix_acl *pacl, struct nfs4_acl *acl,
228 struct posix_acl_summary pas; 228 struct posix_acl_summary pas;
229 unsigned short deny; 229 unsigned short deny;
230 int eflag = ((flags & NFS4_ACL_TYPE_DEFAULT) ? 230 int eflag = ((flags & NFS4_ACL_TYPE_DEFAULT) ?
231 NFS4_INHERITANCE_FLAGS : 0); 231 NFS4_INHERITANCE_FLAGS | NFS4_ACE_INHERIT_ONLY_ACE : 0);
232 232
233 BUG_ON(pacl->a_count < 3); 233 BUG_ON(pacl->a_count < 3);
234 summarize_posix_acl(pacl, &pas); 234 summarize_posix_acl(pacl, &pas);
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 9e4067999209..af360705e551 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -750,9 +750,8 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
750 status = nfserr_clid_inuse; 750 status = nfserr_clid_inuse;
751 if (!cmp_creds(&conf->cl_cred, &rqstp->rq_cred) 751 if (!cmp_creds(&conf->cl_cred, &rqstp->rq_cred)
752 || conf->cl_addr != sin->sin_addr.s_addr) { 752 || conf->cl_addr != sin->sin_addr.s_addr) {
753 printk("NFSD: setclientid: string in use by client" 753 dprintk("NFSD: setclientid: string in use by client"
754 "(clientid %08x/%08x)\n", 754 "at %u.%u.%u.%u\n", NIPQUAD(conf->cl_addr));
755 conf->cl_clientid.cl_boot, conf->cl_clientid.cl_id);
756 goto out; 755 goto out;
757 } 756 }
758 } 757 }
@@ -3261,7 +3260,6 @@ __nfs4_state_shutdown(void)
3261 unhash_delegation(dp); 3260 unhash_delegation(dp);
3262 } 3261 }
3263 3262
3264 cancel_delayed_work(&laundromat_work);
3265 nfsd4_shutdown_recdir(); 3263 nfsd4_shutdown_recdir();
3266 nfs4_init = 0; 3264 nfs4_init = 0;
3267} 3265}
diff --git a/fs/nfsd/nfsfh.c b/fs/nfsd/nfsfh.c
index c2660cbfcd96..8d995bcef806 100644
--- a/fs/nfsd/nfsfh.c
+++ b/fs/nfsd/nfsfh.c
@@ -17,7 +17,6 @@
17#include <linux/stat.h> 17#include <linux/stat.h>
18#include <linux/dcache.h> 18#include <linux/dcache.h>
19#include <linux/mount.h> 19#include <linux/mount.h>
20#include <asm/pgtable.h>
21 20
22#include <linux/sunrpc/clnt.h> 21#include <linux/sunrpc/clnt.h>
23#include <linux/sunrpc/svc.h> 22#include <linux/sunrpc/svc.h>
diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 93628b02ef5d..875c11443817 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -614,6 +614,27 @@ static void ocfs2_dio_end_io(struct kiocb *iocb,
614 ocfs2_rw_unlock(inode, 0); 614 ocfs2_rw_unlock(inode, 0);
615} 615}
616 616
617/*
618 * ocfs2_invalidatepage() and ocfs2_releasepage() are shamelessly stolen
619 * from ext3. PageChecked() bits have been removed as OCFS2 does not
620 * do journalled data.
621 */
622static void ocfs2_invalidatepage(struct page *page, unsigned long offset)
623{
624 journal_t *journal = OCFS2_SB(page->mapping->host->i_sb)->journal->j_journal;
625
626 journal_invalidatepage(journal, page, offset);
627}
628
629static int ocfs2_releasepage(struct page *page, gfp_t wait)
630{
631 journal_t *journal = OCFS2_SB(page->mapping->host->i_sb)->journal->j_journal;
632
633 if (!page_has_buffers(page))
634 return 0;
635 return journal_try_to_free_buffers(journal, page, wait);
636}
637
617static ssize_t ocfs2_direct_IO(int rw, 638static ssize_t ocfs2_direct_IO(int rw,
618 struct kiocb *iocb, 639 struct kiocb *iocb,
619 const struct iovec *iov, 640 const struct iovec *iov,
@@ -661,5 +682,8 @@ const struct address_space_operations ocfs2_aops = {
661 .commit_write = ocfs2_commit_write, 682 .commit_write = ocfs2_commit_write,
662 .bmap = ocfs2_bmap, 683 .bmap = ocfs2_bmap,
663 .sync_page = block_sync_page, 684 .sync_page = block_sync_page,
664 .direct_IO = ocfs2_direct_IO 685 .direct_IO = ocfs2_direct_IO,
686 .invalidatepage = ocfs2_invalidatepage,
687 .releasepage = ocfs2_releasepage,
688 .migratepage = buffer_migrate_page,
665}; 689};
diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c
index 5a9779bb9236..eba282da500e 100644
--- a/fs/ocfs2/cluster/heartbeat.c
+++ b/fs/ocfs2/cluster/heartbeat.c
@@ -1234,6 +1234,7 @@ static ssize_t o2hb_region_dev_write(struct o2hb_region *reg,
1234 const char *page, 1234 const char *page,
1235 size_t count) 1235 size_t count)
1236{ 1236{
1237 struct task_struct *hb_task;
1237 long fd; 1238 long fd;
1238 int sectsize; 1239 int sectsize;
1239 char *p = (char *)page; 1240 char *p = (char *)page;
@@ -1319,20 +1320,28 @@ static ssize_t o2hb_region_dev_write(struct o2hb_region *reg,
1319 */ 1320 */
1320 atomic_set(&reg->hr_steady_iterations, O2HB_LIVE_THRESHOLD + 1); 1321 atomic_set(&reg->hr_steady_iterations, O2HB_LIVE_THRESHOLD + 1);
1321 1322
1322 reg->hr_task = kthread_run(o2hb_thread, reg, "o2hb-%s", 1323 hb_task = kthread_run(o2hb_thread, reg, "o2hb-%s",
1323 reg->hr_item.ci_name); 1324 reg->hr_item.ci_name);
1324 if (IS_ERR(reg->hr_task)) { 1325 if (IS_ERR(hb_task)) {
1325 ret = PTR_ERR(reg->hr_task); 1326 ret = PTR_ERR(hb_task);
1326 mlog_errno(ret); 1327 mlog_errno(ret);
1327 reg->hr_task = NULL;
1328 goto out; 1328 goto out;
1329 } 1329 }
1330 1330
1331 spin_lock(&o2hb_live_lock);
1332 reg->hr_task = hb_task;
1333 spin_unlock(&o2hb_live_lock);
1334
1331 ret = wait_event_interruptible(o2hb_steady_queue, 1335 ret = wait_event_interruptible(o2hb_steady_queue,
1332 atomic_read(&reg->hr_steady_iterations) == 0); 1336 atomic_read(&reg->hr_steady_iterations) == 0);
1333 if (ret) { 1337 if (ret) {
1334 kthread_stop(reg->hr_task); 1338 spin_lock(&o2hb_live_lock);
1339 hb_task = reg->hr_task;
1335 reg->hr_task = NULL; 1340 reg->hr_task = NULL;
1341 spin_unlock(&o2hb_live_lock);
1342
1343 if (hb_task)
1344 kthread_stop(hb_task);
1336 goto out; 1345 goto out;
1337 } 1346 }
1338 1347
@@ -1354,10 +1363,17 @@ out:
1354static ssize_t o2hb_region_pid_read(struct o2hb_region *reg, 1363static ssize_t o2hb_region_pid_read(struct o2hb_region *reg,
1355 char *page) 1364 char *page)
1356{ 1365{
1357 if (!reg->hr_task) 1366 pid_t pid = 0;
1367
1368 spin_lock(&o2hb_live_lock);
1369 if (reg->hr_task)
1370 pid = reg->hr_task->pid;
1371 spin_unlock(&o2hb_live_lock);
1372
1373 if (!pid)
1358 return 0; 1374 return 0;
1359 1375
1360 return sprintf(page, "%u\n", reg->hr_task->pid); 1376 return sprintf(page, "%u\n", pid);
1361} 1377}
1362 1378
1363struct o2hb_region_attribute { 1379struct o2hb_region_attribute {
@@ -1495,13 +1511,17 @@ out:
1495static void o2hb_heartbeat_group_drop_item(struct config_group *group, 1511static void o2hb_heartbeat_group_drop_item(struct config_group *group,
1496 struct config_item *item) 1512 struct config_item *item)
1497{ 1513{
1514 struct task_struct *hb_task;
1498 struct o2hb_region *reg = to_o2hb_region(item); 1515 struct o2hb_region *reg = to_o2hb_region(item);
1499 1516
1500 /* stop the thread when the user removes the region dir */ 1517 /* stop the thread when the user removes the region dir */
1501 if (reg->hr_task) { 1518 spin_lock(&o2hb_live_lock);
1502 kthread_stop(reg->hr_task); 1519 hb_task = reg->hr_task;
1503 reg->hr_task = NULL; 1520 reg->hr_task = NULL;
1504 } 1521 spin_unlock(&o2hb_live_lock);
1522
1523 if (hb_task)
1524 kthread_stop(hb_task);
1505 1525
1506 config_item_put(item); 1526 config_item_put(item);
1507} 1527}
@@ -1682,7 +1702,7 @@ out:
1682} 1702}
1683EXPORT_SYMBOL_GPL(o2hb_register_callback); 1703EXPORT_SYMBOL_GPL(o2hb_register_callback);
1684 1704
1685int o2hb_unregister_callback(struct o2hb_callback_func *hc) 1705void o2hb_unregister_callback(struct o2hb_callback_func *hc)
1686{ 1706{
1687 BUG_ON(hc->hc_magic != O2HB_CB_MAGIC); 1707 BUG_ON(hc->hc_magic != O2HB_CB_MAGIC);
1688 1708
@@ -1690,15 +1710,13 @@ int o2hb_unregister_callback(struct o2hb_callback_func *hc)
1690 __builtin_return_address(0), hc); 1710 __builtin_return_address(0), hc);
1691 1711
1692 if (list_empty(&hc->hc_item)) 1712 if (list_empty(&hc->hc_item))
1693 return 0; 1713 return;
1694 1714
1695 down_write(&o2hb_callback_sem); 1715 down_write(&o2hb_callback_sem);
1696 1716
1697 list_del_init(&hc->hc_item); 1717 list_del_init(&hc->hc_item);
1698 1718
1699 up_write(&o2hb_callback_sem); 1719 up_write(&o2hb_callback_sem);
1700
1701 return 0;
1702} 1720}
1703EXPORT_SYMBOL_GPL(o2hb_unregister_callback); 1721EXPORT_SYMBOL_GPL(o2hb_unregister_callback);
1704 1722
diff --git a/fs/ocfs2/cluster/heartbeat.h b/fs/ocfs2/cluster/heartbeat.h
index cac6223206a9..cc6d40b39771 100644
--- a/fs/ocfs2/cluster/heartbeat.h
+++ b/fs/ocfs2/cluster/heartbeat.h
@@ -70,7 +70,7 @@ void o2hb_setup_callback(struct o2hb_callback_func *hc,
70 void *data, 70 void *data,
71 int priority); 71 int priority);
72int o2hb_register_callback(struct o2hb_callback_func *hc); 72int o2hb_register_callback(struct o2hb_callback_func *hc);
73int o2hb_unregister_callback(struct o2hb_callback_func *hc); 73void o2hb_unregister_callback(struct o2hb_callback_func *hc);
74void o2hb_fill_node_map(unsigned long *map, 74void o2hb_fill_node_map(unsigned long *map,
75 unsigned bytes); 75 unsigned bytes);
76void o2hb_init(void); 76void o2hb_init(void);
diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
index 1718215fc018..69caf3e12fea 100644
--- a/fs/ocfs2/cluster/tcp.c
+++ b/fs/ocfs2/cluster/tcp.c
@@ -1638,17 +1638,8 @@ static void o2net_hb_node_up_cb(struct o2nm_node *node, int node_num,
1638 1638
1639void o2net_unregister_hb_callbacks(void) 1639void o2net_unregister_hb_callbacks(void)
1640{ 1640{
1641 int ret; 1641 o2hb_unregister_callback(&o2net_hb_up);
1642 1642 o2hb_unregister_callback(&o2net_hb_down);
1643 ret = o2hb_unregister_callback(&o2net_hb_up);
1644 if (ret < 0)
1645 mlog(ML_ERROR, "Status return %d unregistering heartbeat up "
1646 "callback!\n", ret);
1647
1648 ret = o2hb_unregister_callback(&o2net_hb_down);
1649 if (ret < 0)
1650 mlog(ML_ERROR, "Status return %d unregistering heartbeat down "
1651 "callback!\n", ret);
1652} 1643}
1653 1644
1654int o2net_register_hb_callbacks(void) 1645int o2net_register_hb_callbacks(void)
diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
index 6087c4749fee..c558442a0b44 100644
--- a/fs/ocfs2/dlm/dlmdomain.c
+++ b/fs/ocfs2/dlm/dlmdomain.c
@@ -138,8 +138,10 @@ static void dlm_unregister_domain_handlers(struct dlm_ctxt *dlm);
138 138
139void __dlm_unhash_lockres(struct dlm_lock_resource *lockres) 139void __dlm_unhash_lockres(struct dlm_lock_resource *lockres)
140{ 140{
141 hlist_del_init(&lockres->hash_node); 141 if (!hlist_unhashed(&lockres->hash_node)) {
142 dlm_lockres_put(lockres); 142 hlist_del_init(&lockres->hash_node);
143 dlm_lockres_put(lockres);
144 }
143} 145}
144 146
145void __dlm_insert_lockres(struct dlm_ctxt *dlm, 147void __dlm_insert_lockres(struct dlm_ctxt *dlm,
@@ -655,6 +657,8 @@ void dlm_unregister_domain(struct dlm_ctxt *dlm)
655 dlm_kick_thread(dlm, NULL); 657 dlm_kick_thread(dlm, NULL);
656 658
657 while (dlm_migrate_all_locks(dlm)) { 659 while (dlm_migrate_all_locks(dlm)) {
660 /* Give dlm_thread time to purge the lockres' */
661 msleep(500);
658 mlog(0, "%s: more migration to do\n", dlm->name); 662 mlog(0, "%s: more migration to do\n", dlm->name);
659 } 663 }
660 dlm_mark_domain_leaving(dlm); 664 dlm_mark_domain_leaving(dlm);
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index 77e4e6169a0d..6edffca99d98 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -2424,6 +2424,57 @@ static void dlm_deref_lockres_worker(struct dlm_work_item *item, void *data)
2424 dlm_lockres_put(res); 2424 dlm_lockres_put(res);
2425} 2425}
2426 2426
2427/* Checks whether the lockres can be migrated. Returns 0 if yes, < 0
2428 * if not. If 0, numlocks is set to the number of locks in the lockres.
2429 */
2430static int dlm_is_lockres_migrateable(struct dlm_ctxt *dlm,
2431 struct dlm_lock_resource *res,
2432 int *numlocks)
2433{
2434 int ret;
2435 int i;
2436 int count = 0;
2437 struct list_head *queue, *iter;
2438 struct dlm_lock *lock;
2439
2440 assert_spin_locked(&res->spinlock);
2441
2442 ret = -EINVAL;
2443 if (res->owner == DLM_LOCK_RES_OWNER_UNKNOWN) {
2444 mlog(0, "cannot migrate lockres with unknown owner!\n");
2445 goto leave;
2446 }
2447
2448 if (res->owner != dlm->node_num) {
2449 mlog(0, "cannot migrate lockres this node doesn't own!\n");
2450 goto leave;
2451 }
2452
2453 ret = 0;
2454 queue = &res->granted;
2455 for (i = 0; i < 3; i++) {
2456 list_for_each(iter, queue) {
2457 lock = list_entry(iter, struct dlm_lock, list);
2458 ++count;
2459 if (lock->ml.node == dlm->node_num) {
2460 mlog(0, "found a lock owned by this node still "
2461 "on the %s queue! will not migrate this "
2462 "lockres\n", (i == 0 ? "granted" :
2463 (i == 1 ? "converting" :
2464 "blocked")));
2465 ret = -ENOTEMPTY;
2466 goto leave;
2467 }
2468 }
2469 queue++;
2470 }
2471
2472 *numlocks = count;
2473 mlog(0, "migrateable lockres having %d locks\n", *numlocks);
2474
2475leave:
2476 return ret;
2477}
2427 2478
2428/* 2479/*
2429 * DLM_MIGRATE_LOCKRES 2480 * DLM_MIGRATE_LOCKRES
@@ -2437,14 +2488,12 @@ static int dlm_migrate_lockres(struct dlm_ctxt *dlm,
2437 struct dlm_master_list_entry *mle = NULL; 2488 struct dlm_master_list_entry *mle = NULL;
2438 struct dlm_master_list_entry *oldmle = NULL; 2489 struct dlm_master_list_entry *oldmle = NULL;
2439 struct dlm_migratable_lockres *mres = NULL; 2490 struct dlm_migratable_lockres *mres = NULL;
2440 int ret = -EINVAL; 2491 int ret = 0;
2441 const char *name; 2492 const char *name;
2442 unsigned int namelen; 2493 unsigned int namelen;
2443 int mle_added = 0; 2494 int mle_added = 0;
2444 struct list_head *queue, *iter; 2495 int numlocks;
2445 int i; 2496 int wake = 0;
2446 struct dlm_lock *lock;
2447 int empty = 1, wake = 0;
2448 2497
2449 if (!dlm_grab(dlm)) 2498 if (!dlm_grab(dlm))
2450 return -EINVAL; 2499 return -EINVAL;
@@ -2458,42 +2507,16 @@ static int dlm_migrate_lockres(struct dlm_ctxt *dlm,
2458 * ensure this lockres is a proper candidate for migration 2507 * ensure this lockres is a proper candidate for migration
2459 */ 2508 */
2460 spin_lock(&res->spinlock); 2509 spin_lock(&res->spinlock);
2461 if (res->owner == DLM_LOCK_RES_OWNER_UNKNOWN) { 2510 ret = dlm_is_lockres_migrateable(dlm, res, &numlocks);
2462 mlog(0, "cannot migrate lockres with unknown owner!\n"); 2511 if (ret < 0) {
2463 spin_unlock(&res->spinlock);
2464 goto leave;
2465 }
2466 if (res->owner != dlm->node_num) {
2467 mlog(0, "cannot migrate lockres this node doesn't own!\n");
2468 spin_unlock(&res->spinlock); 2512 spin_unlock(&res->spinlock);
2469 goto leave; 2513 goto leave;
2470 } 2514 }
2471 mlog(0, "checking queues...\n");
2472 queue = &res->granted;
2473 for (i=0; i<3; i++) {
2474 list_for_each(iter, queue) {
2475 lock = list_entry (iter, struct dlm_lock, list);
2476 empty = 0;
2477 if (lock->ml.node == dlm->node_num) {
2478 mlog(0, "found a lock owned by this node "
2479 "still on the %s queue! will not "
2480 "migrate this lockres\n",
2481 i==0 ? "granted" :
2482 (i==1 ? "converting" : "blocked"));
2483 spin_unlock(&res->spinlock);
2484 ret = -ENOTEMPTY;
2485 goto leave;
2486 }
2487 }
2488 queue++;
2489 }
2490 mlog(0, "all locks on this lockres are nonlocal. continuing\n");
2491 spin_unlock(&res->spinlock); 2515 spin_unlock(&res->spinlock);
2492 2516
2493 /* no work to do */ 2517 /* no work to do */
2494 if (empty) { 2518 if (numlocks == 0) {
2495 mlog(0, "no locks were found on this lockres! done!\n"); 2519 mlog(0, "no locks were found on this lockres! done!\n");
2496 ret = 0;
2497 goto leave; 2520 goto leave;
2498 } 2521 }
2499 2522
@@ -2729,15 +2752,26 @@ int dlm_empty_lockres(struct dlm_ctxt *dlm, struct dlm_lock_resource *res)
2729{ 2752{
2730 int ret; 2753 int ret;
2731 int lock_dropped = 0; 2754 int lock_dropped = 0;
2755 int numlocks;
2732 2756
2757 spin_lock(&res->spinlock);
2733 if (res->owner != dlm->node_num) { 2758 if (res->owner != dlm->node_num) {
2734 if (!__dlm_lockres_unused(res)) { 2759 if (!__dlm_lockres_unused(res)) {
2735 mlog(ML_ERROR, "%s:%.*s: this node is not master, " 2760 mlog(ML_ERROR, "%s:%.*s: this node is not master, "
2736 "trying to free this but locks remain\n", 2761 "trying to free this but locks remain\n",
2737 dlm->name, res->lockname.len, res->lockname.name); 2762 dlm->name, res->lockname.len, res->lockname.name);
2738 } 2763 }
2764 spin_unlock(&res->spinlock);
2765 goto leave;
2766 }
2767
2768 /* No need to migrate a lockres having no locks */
2769 ret = dlm_is_lockres_migrateable(dlm, res, &numlocks);
2770 if (ret >= 0 && numlocks == 0) {
2771 spin_unlock(&res->spinlock);
2739 goto leave; 2772 goto leave;
2740 } 2773 }
2774 spin_unlock(&res->spinlock);
2741 2775
2742 /* Wheee! Migrate lockres here! Will sleep so drop spinlock. */ 2776 /* Wheee! Migrate lockres here! Will sleep so drop spinlock. */
2743 spin_unlock(&dlm->spinlock); 2777 spin_unlock(&dlm->spinlock);
diff --git a/fs/ocfs2/dlm/dlmthread.c b/fs/ocfs2/dlm/dlmthread.c
index 8ffa0916eb86..2b264c6ba039 100644
--- a/fs/ocfs2/dlm/dlmthread.c
+++ b/fs/ocfs2/dlm/dlmthread.c
@@ -256,18 +256,14 @@ static void dlm_run_purge_list(struct dlm_ctxt *dlm,
256 break; 256 break;
257 } 257 }
258 258
259 mlog(0, "removing lockres %.*s:%p from purgelist\n", 259 dlm_lockres_get(lockres);
260 lockres->lockname.len, lockres->lockname.name, lockres);
261 list_del_init(&lockres->purge);
262 dlm_lockres_put(lockres);
263 dlm->purge_count--;
264 260
265 /* This may drop and reacquire the dlm spinlock if it 261 /* This may drop and reacquire the dlm spinlock if it
266 * has to do migration. */ 262 * has to do migration. */
267 mlog(0, "calling dlm_purge_lockres!\n");
268 if (dlm_purge_lockres(dlm, lockres)) 263 if (dlm_purge_lockres(dlm, lockres))
269 BUG(); 264 BUG();
270 mlog(0, "DONE calling dlm_purge_lockres!\n"); 265
266 dlm_lockres_put(lockres);
271 267
272 /* Avoid adding any scheduling latencies */ 268 /* Avoid adding any scheduling latencies */
273 cond_resched_lock(&dlm->spinlock); 269 cond_resched_lock(&dlm->spinlock);
diff --git a/fs/ocfs2/heartbeat.c b/fs/ocfs2/heartbeat.c
index 8fc52d6d0ce7..b25ef63781ba 100644
--- a/fs/ocfs2/heartbeat.c
+++ b/fs/ocfs2/heartbeat.c
@@ -164,8 +164,10 @@ int ocfs2_register_hb_callbacks(struct ocfs2_super *osb)
164 } 164 }
165 165
166 status = o2hb_register_callback(&osb->osb_hb_up); 166 status = o2hb_register_callback(&osb->osb_hb_up);
167 if (status < 0) 167 if (status < 0) {
168 mlog_errno(status); 168 mlog_errno(status);
169 o2hb_unregister_callback(&osb->osb_hb_down);
170 }
169 171
170bail: 172bail:
171 return status; 173 return status;
@@ -173,18 +175,11 @@ bail:
173 175
174void ocfs2_clear_hb_callbacks(struct ocfs2_super *osb) 176void ocfs2_clear_hb_callbacks(struct ocfs2_super *osb)
175{ 177{
176 int status;
177
178 if (ocfs2_mount_local(osb)) 178 if (ocfs2_mount_local(osb))
179 return; 179 return;
180 180
181 status = o2hb_unregister_callback(&osb->osb_hb_down); 181 o2hb_unregister_callback(&osb->osb_hb_down);
182 if (status < 0) 182 o2hb_unregister_callback(&osb->osb_hb_up);
183 mlog_errno(status);
184
185 status = o2hb_unregister_callback(&osb->osb_hb_up);
186 if (status < 0)
187 mlog_errno(status);
188} 183}
189 184
190void ocfs2_stop_heartbeat(struct ocfs2_super *osb) 185void ocfs2_stop_heartbeat(struct ocfs2_super *osb)
diff --git a/fs/partitions/Kconfig b/fs/partitions/Kconfig
index 74552c60b671..6e8bb66fe619 100644
--- a/fs/partitions/Kconfig
+++ b/fs/partitions/Kconfig
@@ -235,5 +235,4 @@ config EFI_PARTITION
235 select CRC32 235 select CRC32
236 help 236 help
237 Say Y here if you would like to use hard disks under Linux which 237 Say Y here if you would like to use hard disks under Linux which
238 were partitioned using EFI GPT. Presently only useful on the 238 were partitioned using EFI GPT.
239 IA-64 platform.
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index e46d237b10f9..8a7d0035ad7a 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -541,7 +541,7 @@ int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
541 if (!get_capacity(disk) || !(state = check_partition(disk, bdev))) 541 if (!get_capacity(disk) || !(state = check_partition(disk, bdev)))
542 return 0; 542 return 0;
543 if (IS_ERR(state)) /* I/O error reading the partition table */ 543 if (IS_ERR(state)) /* I/O error reading the partition table */
544 return PTR_ERR(state); 544 return -EIO;
545 for (p = 1; p < state->limit; p++) { 545 for (p = 1; p < state->limit; p++) {
546 sector_t size = state->parts[p].size; 546 sector_t size = state->parts[p].size;
547 sector_t from = state->parts[p].from; 547 sector_t from = state->parts[p].from;
diff --git a/fs/proc/Makefile b/fs/proc/Makefile
index a6b3a8f878f0..bce38e3f06cb 100644
--- a/fs/proc/Makefile
+++ b/fs/proc/Makefile
@@ -8,8 +8,9 @@ proc-y := nommu.o task_nommu.o
8proc-$(CONFIG_MMU) := mmu.o task_mmu.o 8proc-$(CONFIG_MMU) := mmu.o task_mmu.o
9 9
10proc-y += inode.o root.o base.o generic.o array.o \ 10proc-y += inode.o root.o base.o generic.o array.o \
11 proc_tty.o proc_misc.o proc_sysctl.o 11 proc_tty.o proc_misc.o
12 12
13proc-$(CONFIG_PROC_SYSCTL) += proc_sysctl.o
13proc-$(CONFIG_PROC_KCORE) += kcore.o 14proc-$(CONFIG_PROC_KCORE) += kcore.o
14proc-$(CONFIG_PROC_VMCORE) += vmcore.o 15proc-$(CONFIG_PROC_VMCORE) += vmcore.o
15proc-$(CONFIG_PROC_DEVICETREE) += proc_devtree.o 16proc-$(CONFIG_PROC_DEVICETREE) += proc_devtree.o
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 01f7769da8e6..989af5e55d1b 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -1558,29 +1558,20 @@ static ssize_t proc_pid_attr_read(struct file * file, char __user * buf,
1558 size_t count, loff_t *ppos) 1558 size_t count, loff_t *ppos)
1559{ 1559{
1560 struct inode * inode = file->f_path.dentry->d_inode; 1560 struct inode * inode = file->f_path.dentry->d_inode;
1561 unsigned long page; 1561 char *p = NULL;
1562 ssize_t length; 1562 ssize_t length;
1563 struct task_struct *task = get_proc_task(inode); 1563 struct task_struct *task = get_proc_task(inode);
1564 1564
1565 length = -ESRCH;
1566 if (!task) 1565 if (!task)
1567 goto out_no_task; 1566 return -ESRCH;
1568
1569 if (count > PAGE_SIZE)
1570 count = PAGE_SIZE;
1571 length = -ENOMEM;
1572 if (!(page = __get_free_page(GFP_KERNEL)))
1573 goto out;
1574 1567
1575 length = security_getprocattr(task, 1568 length = security_getprocattr(task,
1576 (char*)file->f_path.dentry->d_name.name, 1569 (char*)file->f_path.dentry->d_name.name,
1577 (void*)page, count); 1570 &p);
1578 if (length >= 0)
1579 length = simple_read_from_buffer(buf, count, ppos, (char *)page, length);
1580 free_page(page);
1581out:
1582 put_task_struct(task); 1571 put_task_struct(task);
1583out_no_task: 1572 if (length > 0)
1573 length = simple_read_from_buffer(buf, count, ppos, p, length);
1574 kfree(p);
1584 return length; 1575 return length;
1585} 1576}
1586 1577
diff --git a/fs/proc/internal.h b/fs/proc/internal.h
index c932aa65e198..f771889183c3 100644
--- a/fs/proc/internal.h
+++ b/fs/proc/internal.h
@@ -11,7 +11,11 @@
11 11
12#include <linux/proc_fs.h> 12#include <linux/proc_fs.h>
13 13
14#ifdef CONFIG_PROC_SYSCTL
14extern int proc_sys_init(void); 15extern int proc_sys_init(void);
16#else
17static inline void proc_sys_init(void) { }
18#endif
15 19
16struct vmalloc_info { 20struct vmalloc_info {
17 unsigned long used; 21 unsigned long used;
diff --git a/fs/proc/root.c b/fs/proc/root.c
index 5834a744c2a9..41f17037f738 100644
--- a/fs/proc/root.c
+++ b/fs/proc/root.c
@@ -79,9 +79,7 @@ void __init proc_root_init(void)
79 proc_device_tree_init(); 79 proc_device_tree_init();
80#endif 80#endif
81 proc_bus = proc_mkdir("bus", NULL); 81 proc_bus = proc_mkdir("bus", NULL);
82#ifdef CONFIG_SYSCTL
83 proc_sys_init(); 82 proc_sys_init();
84#endif
85} 83}
86 84
87static int proc_root_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat 85static int proc_root_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat
diff --git a/fs/reiserfs/item_ops.c b/fs/reiserfs/item_ops.c
index b9b423b22a8b..9475557ab499 100644
--- a/fs/reiserfs/item_ops.c
+++ b/fs/reiserfs/item_ops.c
@@ -23,7 +23,7 @@ static void sd_decrement_key(struct cpu_key *key)
23{ 23{
24 key->on_disk_key.k_objectid--; 24 key->on_disk_key.k_objectid--;
25 set_cpu_key_k_type(key, TYPE_ANY); 25 set_cpu_key_k_type(key, TYPE_ANY);
26 set_cpu_key_k_offset(key, (loff_t) (-1)); 26 set_cpu_key_k_offset(key, (loff_t)(~0ULL >> 1));
27} 27}
28 28
29static int sd_is_left_mergeable(struct reiserfs_key *key, unsigned long bsize) 29static int sd_is_left_mergeable(struct reiserfs_key *key, unsigned long bsize)
diff --git a/fs/reiserfs/xattr.c b/fs/reiserfs/xattr.c
index f01389fd162e..c8178b7b9212 100644
--- a/fs/reiserfs/xattr.c
+++ b/fs/reiserfs/xattr.c
@@ -54,82 +54,48 @@
54static struct reiserfs_xattr_handler *find_xattr_handler_prefix(const char 54static struct reiserfs_xattr_handler *find_xattr_handler_prefix(const char
55 *prefix); 55 *prefix);
56 56
57static struct dentry *create_xa_root(struct super_block *sb) 57/* Returns the dentry referring to the root of the extended attribute
58 * directory tree. If it has already been retrieved, it is used. If it
59 * hasn't been created and the flags indicate creation is allowed, we
60 * attempt to create it. On error, we return a pointer-encoded error.
61 */
62static struct dentry *get_xa_root(struct super_block *sb, int flags)
58{ 63{
59 struct dentry *privroot = dget(REISERFS_SB(sb)->priv_root); 64 struct dentry *privroot = dget(REISERFS_SB(sb)->priv_root);
60 struct dentry *xaroot; 65 struct dentry *xaroot;
61 66
62 /* This needs to be created at mount-time */ 67 /* This needs to be created at mount-time */
63 if (!privroot) 68 if (!privroot)
64 return ERR_PTR(-EOPNOTSUPP); 69 return ERR_PTR(-ENODATA);
65 70
66 xaroot = lookup_one_len(XAROOT_NAME, privroot, strlen(XAROOT_NAME)); 71 mutex_lock(&privroot->d_inode->i_mutex);
67 if (IS_ERR(xaroot)) { 72 if (REISERFS_SB(sb)->xattr_root) {
73 xaroot = dget(REISERFS_SB(sb)->xattr_root);
68 goto out; 74 goto out;
69 } else if (!xaroot->d_inode) {
70 int err;
71 mutex_lock(&privroot->d_inode->i_mutex);
72 err =
73 privroot->d_inode->i_op->mkdir(privroot->d_inode, xaroot,
74 0700);
75 mutex_unlock(&privroot->d_inode->i_mutex);
76
77 if (err) {
78 dput(xaroot);
79 dput(privroot);
80 return ERR_PTR(err);
81 }
82 REISERFS_SB(sb)->xattr_root = dget(xaroot);
83 } 75 }
84 76
85 out:
86 dput(privroot);
87 return xaroot;
88}
89
90/* This will return a dentry, or error, refering to the xa root directory.
91 * If the xa root doesn't exist yet, the dentry will be returned without
92 * an associated inode. This dentry can be used with ->mkdir to create
93 * the xa directory. */
94static struct dentry *__get_xa_root(struct super_block *s)
95{
96 struct dentry *privroot = dget(REISERFS_SB(s)->priv_root);
97 struct dentry *xaroot = NULL;
98
99 if (IS_ERR(privroot) || !privroot)
100 return privroot;
101
102 xaroot = lookup_one_len(XAROOT_NAME, privroot, strlen(XAROOT_NAME)); 77 xaroot = lookup_one_len(XAROOT_NAME, privroot, strlen(XAROOT_NAME));
103 if (IS_ERR(xaroot)) { 78 if (IS_ERR(xaroot)) {
104 goto out; 79 goto out;
105 } else if (!xaroot->d_inode) { 80 } else if (!xaroot->d_inode) {
106 dput(xaroot); 81 int err = -ENODATA;
107 xaroot = NULL; 82 if (flags == 0 || flags & XATTR_CREATE)
108 goto out; 83 err = privroot->d_inode->i_op->mkdir(privroot->d_inode,
84 xaroot, 0700);
85 if (err) {
86 dput(xaroot);
87 xaroot = ERR_PTR(err);
88 goto out;
89 }
109 } 90 }
110 91 REISERFS_SB(sb)->xattr_root = dget(xaroot);
111 REISERFS_SB(s)->xattr_root = dget(xaroot);
112 92
113 out: 93 out:
94 mutex_unlock(&privroot->d_inode->i_mutex);
114 dput(privroot); 95 dput(privroot);
115 return xaroot; 96 return xaroot;
116} 97}
117 98
118/* Returns the dentry (or NULL) referring to the root of the extended
119 * attribute directory tree. If it has already been retrieved, it is used.
120 * Otherwise, we attempt to retrieve it from disk. It may also return
121 * a pointer-encoded error.
122 */
123static inline struct dentry *get_xa_root(struct super_block *s)
124{
125 struct dentry *dentry = dget(REISERFS_SB(s)->xattr_root);
126
127 if (!dentry)
128 dentry = __get_xa_root(s);
129
130 return dentry;
131}
132
133/* Opens the directory corresponding to the inode's extended attribute store. 99/* Opens the directory corresponding to the inode's extended attribute store.
134 * If flags allow, the tree to the directory may be created. If creation is 100 * If flags allow, the tree to the directory may be created. If creation is
135 * prohibited, -ENODATA is returned. */ 101 * prohibited, -ENODATA is returned. */
@@ -138,21 +104,11 @@ static struct dentry *open_xa_dir(const struct inode *inode, int flags)
138 struct dentry *xaroot, *xadir; 104 struct dentry *xaroot, *xadir;
139 char namebuf[17]; 105 char namebuf[17];
140 106
141 xaroot = get_xa_root(inode->i_sb); 107 xaroot = get_xa_root(inode->i_sb, flags);
142 if (IS_ERR(xaroot)) { 108 if (IS_ERR(xaroot))
143 return xaroot; 109 return xaroot;
144 } else if (!xaroot) {
145 if (flags == 0 || flags & XATTR_CREATE) {
146 xaroot = create_xa_root(inode->i_sb);
147 if (IS_ERR(xaroot))
148 return xaroot;
149 }
150 if (!xaroot)
151 return ERR_PTR(-ENODATA);
152 }
153 110
154 /* ok, we have xaroot open */ 111 /* ok, we have xaroot open */
155
156 snprintf(namebuf, sizeof(namebuf), "%X.%X", 112 snprintf(namebuf, sizeof(namebuf), "%X.%X",
157 le32_to_cpu(INODE_PKEY(inode)->k_objectid), 113 le32_to_cpu(INODE_PKEY(inode)->k_objectid),
158 inode->i_generation); 114 inode->i_generation);
@@ -821,7 +777,7 @@ int reiserfs_delete_xattrs(struct inode *inode)
821 777
822 /* Leftovers besides . and .. -- that's not good. */ 778 /* Leftovers besides . and .. -- that's not good. */
823 if (dir->d_inode->i_nlink <= 2) { 779 if (dir->d_inode->i_nlink <= 2) {
824 root = get_xa_root(inode->i_sb); 780 root = get_xa_root(inode->i_sb, XATTR_REPLACE);
825 reiserfs_write_lock_xattrs(inode->i_sb); 781 reiserfs_write_lock_xattrs(inode->i_sb);
826 err = vfs_rmdir(root->d_inode, dir); 782 err = vfs_rmdir(root->d_inode, dir);
827 reiserfs_write_unlock_xattrs(inode->i_sb); 783 reiserfs_write_unlock_xattrs(inode->i_sb);
diff --git a/fs/smbfs/request.c b/fs/smbfs/request.c
index 42261dbdf60f..723f7c667661 100644
--- a/fs/smbfs/request.c
+++ b/fs/smbfs/request.c
@@ -181,6 +181,7 @@ static int smb_setup_request(struct smb_request *req)
181 req->rq_errno = 0; 181 req->rq_errno = 0;
182 req->rq_fragment = 0; 182 req->rq_fragment = 0;
183 kfree(req->rq_trans2buffer); 183 kfree(req->rq_trans2buffer);
184 req->rq_trans2buffer = NULL;
184 185
185 return 0; 186 return 0;
186} 187}
diff --git a/fs/splice.c b/fs/splice.c
index 2fca6ebf4cc2..5428b0ff3b6f 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -576,76 +576,21 @@ static int pipe_to_file(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
576 if (this_len + offset > PAGE_CACHE_SIZE) 576 if (this_len + offset > PAGE_CACHE_SIZE)
577 this_len = PAGE_CACHE_SIZE - offset; 577 this_len = PAGE_CACHE_SIZE - offset;
578 578
579 /*
580 * Reuse buf page, if SPLICE_F_MOVE is set and we are doing a full
581 * page.
582 */
583 if ((sd->flags & SPLICE_F_MOVE) && this_len == PAGE_CACHE_SIZE) {
584 /*
585 * If steal succeeds, buf->page is now pruned from the
586 * pagecache and we can reuse it. The page will also be
587 * locked on successful return.
588 */
589 if (buf->ops->steal(pipe, buf))
590 goto find_page;
591
592 page = buf->page;
593 if (add_to_page_cache(page, mapping, index, GFP_KERNEL)) {
594 unlock_page(page);
595 goto find_page;
596 }
597
598 page_cache_get(page);
599
600 if (!(buf->flags & PIPE_BUF_FLAG_LRU))
601 lru_cache_add(page);
602 } else {
603find_page: 579find_page:
604 page = find_lock_page(mapping, index); 580 page = find_lock_page(mapping, index);
605 if (!page) { 581 if (!page) {
606 ret = -ENOMEM; 582 ret = -ENOMEM;
607 page = page_cache_alloc_cold(mapping); 583 page = page_cache_alloc_cold(mapping);
608 if (unlikely(!page)) 584 if (unlikely(!page))
609 goto out_ret; 585 goto out_ret;
610
611 /*
612 * This will also lock the page
613 */
614 ret = add_to_page_cache_lru(page, mapping, index,
615 GFP_KERNEL);
616 if (unlikely(ret))
617 goto out;
618 }
619 586
620 /* 587 /*
621 * We get here with the page locked. If the page is also 588 * This will also lock the page
622 * uptodate, we don't need to do more. If it isn't, we
623 * may need to bring it in if we are not going to overwrite
624 * the full page.
625 */ 589 */
626 if (!PageUptodate(page)) { 590 ret = add_to_page_cache_lru(page, mapping, index,
627 if (this_len < PAGE_CACHE_SIZE) { 591 GFP_KERNEL);
628 ret = mapping->a_ops->readpage(file, page); 592 if (unlikely(ret))
629 if (unlikely(ret)) 593 goto out;
630 goto out;
631
632 lock_page(page);
633
634 if (!PageUptodate(page)) {
635 /*
636 * Page got invalidated, repeat.
637 */
638 if (!page->mapping) {
639 unlock_page(page);
640 page_cache_release(page);
641 goto find_page;
642 }
643 ret = -EIO;
644 goto out;
645 }
646 } else
647 SetPageUptodate(page);
648 }
649 } 594 }
650 595
651 ret = mapping->a_ops->prepare_write(file, page, offset, offset+this_len); 596 ret = mapping->a_ops->prepare_write(file, page, offset, offset+this_len);
@@ -682,18 +627,25 @@ find_page:
682 } 627 }
683 628
684 ret = mapping->a_ops->commit_write(file, page, offset, offset+this_len); 629 ret = mapping->a_ops->commit_write(file, page, offset, offset+this_len);
685 if (!ret) { 630 if (ret) {
631 if (ret == AOP_TRUNCATED_PAGE) {
632 page_cache_release(page);
633 goto find_page;
634 }
635 if (ret < 0)
636 goto out;
686 /* 637 /*
687 * Return the number of bytes written and mark page as 638 * Partial write has happened, so 'ret' already initialized by
688 * accessed, we are now done! 639 * number of bytes written, Where is nothing we have to do here.
689 */ 640 */
641 } else
690 ret = this_len; 642 ret = this_len;
691 mark_page_accessed(page); 643 /*
692 balance_dirty_pages_ratelimited(mapping); 644 * Return the number of bytes written and mark page as
693 } else if (ret == AOP_TRUNCATED_PAGE) { 645 * accessed, we are now done!
694 page_cache_release(page); 646 */
695 goto find_page; 647 mark_page_accessed(page);
696 } 648 balance_dirty_pages_ratelimited(mapping);
697out: 649out:
698 page_cache_release(page); 650 page_cache_release(page);
699 unlock_page(page); 651 unlock_page(page);
@@ -706,9 +658,9 @@ out_ret:
706 * key here is the 'actor' worker passed in that actually moves the data 658 * key here is the 'actor' worker passed in that actually moves the data
707 * to the wanted destination. See pipe_to_file/pipe_to_sendpage above. 659 * to the wanted destination. See pipe_to_file/pipe_to_sendpage above.
708 */ 660 */
709static ssize_t __splice_from_pipe(struct pipe_inode_info *pipe, 661ssize_t __splice_from_pipe(struct pipe_inode_info *pipe,
710 struct file *out, loff_t *ppos, size_t len, 662 struct file *out, loff_t *ppos, size_t len,
711 unsigned int flags, splice_actor *actor) 663 unsigned int flags, splice_actor *actor)
712{ 664{
713 int ret, do_wakeup, err; 665 int ret, do_wakeup, err;
714 struct splice_desc sd; 666 struct splice_desc sd;
@@ -802,6 +754,7 @@ static ssize_t __splice_from_pipe(struct pipe_inode_info *pipe,
802 754
803 return ret; 755 return ret;
804} 756}
757EXPORT_SYMBOL(__splice_from_pipe);
805 758
806ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out, 759ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out,
807 loff_t *ppos, size_t len, unsigned int flags, 760 loff_t *ppos, size_t len, unsigned int flags,
diff --git a/fs/sysfs/file.c b/fs/sysfs/file.c
index 8d4d839a9d88..fc4633378dc0 100644
--- a/fs/sysfs/file.c
+++ b/fs/sysfs/file.c
@@ -168,12 +168,12 @@ sysfs_read_file(struct file *file, char __user *buf, size_t count, loff_t *ppos)
168 ssize_t retval = 0; 168 ssize_t retval = 0;
169 169
170 down(&buffer->sem); 170 down(&buffer->sem);
171 if (buffer->orphaned) {
172 retval = -ENODEV;
173 goto out;
174 }
175 if (buffer->needs_read_fill) { 171 if (buffer->needs_read_fill) {
176 if ((retval = fill_read_buffer(file->f_path.dentry,buffer))) 172 if (buffer->orphaned)
173 retval = -ENODEV;
174 else
175 retval = fill_read_buffer(file->f_path.dentry,buffer);
176 if (retval)
177 goto out; 177 goto out;
178 } 178 }
179 pr_debug("%s: count = %zd, ppos = %lld, buf = %s\n", 179 pr_debug("%s: count = %zd, ppos = %lld, buf = %s\n",
@@ -629,6 +629,60 @@ void sysfs_remove_file_from_group(struct kobject *kobj,
629} 629}
630EXPORT_SYMBOL_GPL(sysfs_remove_file_from_group); 630EXPORT_SYMBOL_GPL(sysfs_remove_file_from_group);
631 631
632struct sysfs_schedule_callback_struct {
633 struct kobject *kobj;
634 void (*func)(void *);
635 void *data;
636 struct work_struct work;
637};
638
639static void sysfs_schedule_callback_work(struct work_struct *work)
640{
641 struct sysfs_schedule_callback_struct *ss = container_of(work,
642 struct sysfs_schedule_callback_struct, work);
643
644 (ss->func)(ss->data);
645 kobject_put(ss->kobj);
646 kfree(ss);
647}
648
649/**
650 * sysfs_schedule_callback - helper to schedule a callback for a kobject
651 * @kobj: object we're acting for.
652 * @func: callback function to invoke later.
653 * @data: argument to pass to @func.
654 *
655 * sysfs attribute methods must not unregister themselves or their parent
656 * kobject (which would amount to the same thing). Attempts to do so will
657 * deadlock, since unregistration is mutually exclusive with driver
658 * callbacks.
659 *
660 * Instead methods can call this routine, which will attempt to allocate
661 * and schedule a workqueue request to call back @func with @data as its
662 * argument in the workqueue's process context. @kobj will be pinned
663 * until @func returns.
664 *
665 * Returns 0 if the request was submitted, -ENOMEM if storage could not
666 * be allocated.
667 */
668int sysfs_schedule_callback(struct kobject *kobj, void (*func)(void *),
669 void *data)
670{
671 struct sysfs_schedule_callback_struct *ss;
672
673 ss = kmalloc(sizeof(*ss), GFP_KERNEL);
674 if (!ss)
675 return -ENOMEM;
676 kobject_get(kobj);
677 ss->kobj = kobj;
678 ss->func = func;
679 ss->data = data;
680 INIT_WORK(&ss->work, sysfs_schedule_callback_work);
681 schedule_work(&ss->work);
682 return 0;
683}
684EXPORT_SYMBOL_GPL(sysfs_schedule_callback);
685
632 686
633EXPORT_SYMBOL_GPL(sysfs_create_file); 687EXPORT_SYMBOL_GPL(sysfs_create_file);
634EXPORT_SYMBOL_GPL(sysfs_remove_file); 688EXPORT_SYMBOL_GPL(sysfs_remove_file);
diff --git a/fs/sysfs/inode.c b/fs/sysfs/inode.c
index ccb7d722c558..4de5c6b89918 100644
--- a/fs/sysfs/inode.c
+++ b/fs/sysfs/inode.c
@@ -222,13 +222,17 @@ const unsigned char * sysfs_get_name(struct sysfs_dirent *sd)
222 222
223static inline void orphan_all_buffers(struct inode *node) 223static inline void orphan_all_buffers(struct inode *node)
224{ 224{
225 struct sysfs_buffer_collection *set = node->i_private; 225 struct sysfs_buffer_collection *set;
226 struct sysfs_buffer *buf; 226 struct sysfs_buffer *buf;
227 227
228 mutex_lock_nested(&node->i_mutex, I_MUTEX_CHILD); 228 mutex_lock_nested(&node->i_mutex, I_MUTEX_CHILD);
229 if (node->i_private) { 229 set = node->i_private;
230 list_for_each_entry(buf, &set->associates, associates) 230 if (set) {
231 list_for_each_entry(buf, &set->associates, associates) {
232 down(&buf->sem);
231 buf->orphaned = 1; 233 buf->orphaned = 1;
234 up(&buf->sem);
235 }
232 } 236 }
233 mutex_unlock(&node->i_mutex); 237 mutex_unlock(&node->i_mutex);
234} 238}
diff --git a/fs/ufs/balloc.c b/fs/ufs/balloc.c
index bcc44084e004..841ac25fd950 100644
--- a/fs/ufs/balloc.c
+++ b/fs/ufs/balloc.c
@@ -244,62 +244,87 @@ failed:
244 * We can come here from ufs_writepage or ufs_prepare_write, 244 * We can come here from ufs_writepage or ufs_prepare_write,
245 * locked_page is argument of these functions, so we already lock it. 245 * locked_page is argument of these functions, so we already lock it.
246 */ 246 */
247static void ufs_change_blocknr(struct inode *inode, unsigned int beg, 247static void ufs_change_blocknr(struct inode *inode, sector_t beg,
248 unsigned int count, unsigned int oldb, 248 unsigned int count, sector_t oldb,
249 unsigned int newb, struct page *locked_page) 249 sector_t newb, struct page *locked_page)
250{ 250{
251 const unsigned mask = (1 << (PAGE_CACHE_SHIFT - inode->i_blkbits)) - 1; 251 const unsigned blks_per_page =
252 1 << (PAGE_CACHE_SHIFT - inode->i_blkbits);
253 const unsigned mask = blks_per_page - 1;
252 struct address_space * const mapping = inode->i_mapping; 254 struct address_space * const mapping = inode->i_mapping;
253 pgoff_t index, cur_index; 255 pgoff_t index, cur_index, last_index;
254 unsigned end, pos, j; 256 unsigned pos, j, lblock;
257 sector_t end, i;
255 struct page *page; 258 struct page *page;
256 struct buffer_head *head, *bh; 259 struct buffer_head *head, *bh;
257 260
258 UFSD("ENTER, ino %lu, count %u, oldb %u, newb %u\n", 261 UFSD("ENTER, ino %lu, count %u, oldb %llu, newb %llu\n",
259 inode->i_ino, count, oldb, newb); 262 inode->i_ino, count,
263 (unsigned long long)oldb, (unsigned long long)newb);
260 264
261 BUG_ON(!locked_page); 265 BUG_ON(!locked_page);
262 BUG_ON(!PageLocked(locked_page)); 266 BUG_ON(!PageLocked(locked_page));
263 267
264 cur_index = locked_page->index; 268 cur_index = locked_page->index;
265 269 end = count + beg;
266 for (end = count + beg; beg < end; beg = (beg | mask) + 1) { 270 last_index = end >> (PAGE_CACHE_SHIFT - inode->i_blkbits);
267 index = beg >> (PAGE_CACHE_SHIFT - inode->i_blkbits); 271 for (i = beg; i < end; i = (i | mask) + 1) {
272 index = i >> (PAGE_CACHE_SHIFT - inode->i_blkbits);
268 273
269 if (likely(cur_index != index)) { 274 if (likely(cur_index != index)) {
270 page = ufs_get_locked_page(mapping, index); 275 page = ufs_get_locked_page(mapping, index);
271 if (!page || IS_ERR(page)) /* it was truncated or EIO */ 276 if (!page)/* it was truncated */
277 continue;
278 if (IS_ERR(page)) {/* or EIO */
279 ufs_error(inode->i_sb, __FUNCTION__,
280 "read of page %llu failed\n",
281 (unsigned long long)index);
272 continue; 282 continue;
283 }
273 } else 284 } else
274 page = locked_page; 285 page = locked_page;
275 286
276 head = page_buffers(page); 287 head = page_buffers(page);
277 bh = head; 288 bh = head;
278 pos = beg & mask; 289 pos = i & mask;
279 for (j = 0; j < pos; ++j) 290 for (j = 0; j < pos; ++j)
280 bh = bh->b_this_page; 291 bh = bh->b_this_page;
281 j = 0; 292
293
294 if (unlikely(index == last_index))
295 lblock = end & mask;
296 else
297 lblock = blks_per_page;
298
282 do { 299 do {
283 if (buffer_mapped(bh)) { 300 if (j >= lblock)
284 pos = bh->b_blocknr - oldb; 301 break;
285 if (pos < count) { 302 pos = (i - beg) + j;
286 UFSD(" change from %llu to %llu\n", 303
287 (unsigned long long)pos + oldb, 304 if (!buffer_mapped(bh))
288 (unsigned long long)pos + newb); 305 map_bh(bh, inode->i_sb, oldb + pos);
289 bh->b_blocknr = newb + pos; 306 if (!buffer_uptodate(bh)) {
290 unmap_underlying_metadata(bh->b_bdev, 307 ll_rw_block(READ, 1, &bh);
291 bh->b_blocknr); 308 wait_on_buffer(bh);
292 mark_buffer_dirty(bh); 309 if (!buffer_uptodate(bh)) {
293 ++j; 310 ufs_error(inode->i_sb, __FUNCTION__,
311 "read of block failed\n");
312 break;
294 } 313 }
295 } 314 }
296 315
316 UFSD(" change from %llu to %llu, pos %u\n",
317 (unsigned long long)pos + oldb,
318 (unsigned long long)pos + newb, pos);
319
320 bh->b_blocknr = newb + pos;
321 unmap_underlying_metadata(bh->b_bdev,
322 bh->b_blocknr);
323 mark_buffer_dirty(bh);
324 ++j;
297 bh = bh->b_this_page; 325 bh = bh->b_this_page;
298 } while (bh != head); 326 } while (bh != head);
299 327
300 if (j)
301 set_page_dirty(page);
302
303 if (likely(cur_index != index)) 328 if (likely(cur_index != index))
304 ufs_put_locked_page(page); 329 ufs_put_locked_page(page);
305 } 330 }
@@ -457,8 +482,9 @@ u64 ufs_new_fragments(struct inode *inode, void *p, u64 fragment,
457 if (result) { 482 if (result) {
458 ufs_clear_frags(inode, result + oldcount, newcount - oldcount, 483 ufs_clear_frags(inode, result + oldcount, newcount - oldcount,
459 locked_page != NULL); 484 locked_page != NULL);
460 ufs_change_blocknr(inode, fragment - oldcount, oldcount, tmp, 485 ufs_change_blocknr(inode, fragment - oldcount, oldcount,
461 result, locked_page); 486 uspi->s_sbbase + tmp,
487 uspi->s_sbbase + result, locked_page);
462 ufs_cpu_to_data_ptr(sb, p, result); 488 ufs_cpu_to_data_ptr(sb, p, result);
463 *err = 0; 489 *err = 0;
464 UFS_I(inode)->i_lastfrag = max_t(u32, UFS_I(inode)->i_lastfrag, fragment + count); 490 UFS_I(inode)->i_lastfrag = max_t(u32, UFS_I(inode)->i_lastfrag, fragment + count);
diff --git a/fs/ufs/ialloc.c b/fs/ufs/ialloc.c
index b868878009b6..c28a8b6f2feb 100644
--- a/fs/ufs/ialloc.c
+++ b/fs/ufs/ialloc.c
@@ -343,9 +343,8 @@ cg_found:
343 lock_buffer(bh); 343 lock_buffer(bh);
344 ufs2_inode = (struct ufs2_inode *)bh->b_data; 344 ufs2_inode = (struct ufs2_inode *)bh->b_data;
345 ufs2_inode += ufs_inotofsbo(inode->i_ino); 345 ufs2_inode += ufs_inotofsbo(inode->i_ino);
346 ufs2_inode->ui_birthtime.tv_sec = 346 ufs2_inode->ui_birthtime = cpu_to_fs64(sb, CURRENT_TIME.tv_sec);
347 cpu_to_fs32(sb, CURRENT_TIME_SEC.tv_sec); 347 ufs2_inode->ui_birthnsec = cpu_to_fs32(sb, CURRENT_TIME.tv_nsec);
348 ufs2_inode->ui_birthtime.tv_usec = 0;
349 mark_buffer_dirty(bh); 348 mark_buffer_dirty(bh);
350 unlock_buffer(bh); 349 unlock_buffer(bh);
351 if (sb->s_flags & MS_SYNCHRONOUS) 350 if (sb->s_flags & MS_SYNCHRONOUS)
diff --git a/fs/ufs/inode.c b/fs/ufs/inode.c
index fb34ad03e224..f18b79122fa3 100644
--- a/fs/ufs/inode.c
+++ b/fs/ufs/inode.c
@@ -212,7 +212,7 @@ repeat:
212 brelse (result); 212 brelse (result);
213 goto repeat; 213 goto repeat;
214 } else { 214 } else {
215 *phys = tmp + blockoff; 215 *phys = uspi->s_sbbase + tmp + blockoff;
216 return NULL; 216 return NULL;
217 } 217 }
218 } 218 }
@@ -282,9 +282,9 @@ repeat:
282 } 282 }
283 283
284 if (!phys) { 284 if (!phys) {
285 result = sb_getblk(sb, tmp + blockoff); 285 result = sb_getblk(sb, uspi->s_sbbase + tmp + blockoff);
286 } else { 286 } else {
287 *phys = tmp + blockoff; 287 *phys = uspi->s_sbbase + tmp + blockoff;
288 result = NULL; 288 result = NULL;
289 *err = 0; 289 *err = 0;
290 *new = 1; 290 *new = 1;
@@ -368,7 +368,7 @@ repeat:
368 brelse (result); 368 brelse (result);
369 goto repeat; 369 goto repeat;
370 } else { 370 } else {
371 *phys = tmp + blockoff; 371 *phys = uspi->s_sbbase + tmp + blockoff;
372 goto out; 372 goto out;
373 } 373 }
374 } 374 }
@@ -389,9 +389,9 @@ repeat:
389 389
390 390
391 if (!phys) { 391 if (!phys) {
392 result = sb_getblk(sb, tmp + blockoff); 392 result = sb_getblk(sb, uspi->s_sbbase + tmp + blockoff);
393 } else { 393 } else {
394 *phys = tmp + blockoff; 394 *phys = uspi->s_sbbase + tmp + blockoff;
395 *new = 1; 395 *new = 1;
396 } 396 }
397 397
@@ -601,7 +601,7 @@ static void ufs_set_inode_ops(struct inode *inode)
601 ufs_get_inode_dev(inode->i_sb, UFS_I(inode))); 601 ufs_get_inode_dev(inode->i_sb, UFS_I(inode)));
602} 602}
603 603
604static void ufs1_read_inode(struct inode *inode, struct ufs_inode *ufs_inode) 604static int ufs1_read_inode(struct inode *inode, struct ufs_inode *ufs_inode)
605{ 605{
606 struct ufs_inode_info *ufsi = UFS_I(inode); 606 struct ufs_inode_info *ufsi = UFS_I(inode);
607 struct super_block *sb = inode->i_sb; 607 struct super_block *sb = inode->i_sb;
@@ -613,8 +613,10 @@ static void ufs1_read_inode(struct inode *inode, struct ufs_inode *ufs_inode)
613 */ 613 */
614 inode->i_mode = mode = fs16_to_cpu(sb, ufs_inode->ui_mode); 614 inode->i_mode = mode = fs16_to_cpu(sb, ufs_inode->ui_mode);
615 inode->i_nlink = fs16_to_cpu(sb, ufs_inode->ui_nlink); 615 inode->i_nlink = fs16_to_cpu(sb, ufs_inode->ui_nlink);
616 if (inode->i_nlink == 0) 616 if (inode->i_nlink == 0) {
617 ufs_error (sb, "ufs_read_inode", "inode %lu has zero nlink\n", inode->i_ino); 617 ufs_error (sb, "ufs_read_inode", "inode %lu has zero nlink\n", inode->i_ino);
618 return -1;
619 }
618 620
619 /* 621 /*
620 * Linux now has 32-bit uid and gid, so we can support EFT. 622 * Linux now has 32-bit uid and gid, so we can support EFT.
@@ -643,9 +645,10 @@ static void ufs1_read_inode(struct inode *inode, struct ufs_inode *ufs_inode)
643 for (i = 0; i < (UFS_NDADDR + UFS_NINDIR) * 4; i++) 645 for (i = 0; i < (UFS_NDADDR + UFS_NINDIR) * 4; i++)
644 ufsi->i_u1.i_symlink[i] = ufs_inode->ui_u2.ui_symlink[i]; 646 ufsi->i_u1.i_symlink[i] = ufs_inode->ui_u2.ui_symlink[i];
645 } 647 }
648 return 0;
646} 649}
647 650
648static void ufs2_read_inode(struct inode *inode, struct ufs2_inode *ufs2_inode) 651static int ufs2_read_inode(struct inode *inode, struct ufs2_inode *ufs2_inode)
649{ 652{
650 struct ufs_inode_info *ufsi = UFS_I(inode); 653 struct ufs_inode_info *ufsi = UFS_I(inode);
651 struct super_block *sb = inode->i_sb; 654 struct super_block *sb = inode->i_sb;
@@ -658,8 +661,10 @@ static void ufs2_read_inode(struct inode *inode, struct ufs2_inode *ufs2_inode)
658 */ 661 */
659 inode->i_mode = mode = fs16_to_cpu(sb, ufs2_inode->ui_mode); 662 inode->i_mode = mode = fs16_to_cpu(sb, ufs2_inode->ui_mode);
660 inode->i_nlink = fs16_to_cpu(sb, ufs2_inode->ui_nlink); 663 inode->i_nlink = fs16_to_cpu(sb, ufs2_inode->ui_nlink);
661 if (inode->i_nlink == 0) 664 if (inode->i_nlink == 0) {
662 ufs_error (sb, "ufs_read_inode", "inode %lu has zero nlink\n", inode->i_ino); 665 ufs_error (sb, "ufs_read_inode", "inode %lu has zero nlink\n", inode->i_ino);
666 return -1;
667 }
663 668
664 /* 669 /*
665 * Linux now has 32-bit uid and gid, so we can support EFT. 670 * Linux now has 32-bit uid and gid, so we can support EFT.
@@ -668,12 +673,12 @@ static void ufs2_read_inode(struct inode *inode, struct ufs2_inode *ufs2_inode)
668 inode->i_gid = fs32_to_cpu(sb, ufs2_inode->ui_gid); 673 inode->i_gid = fs32_to_cpu(sb, ufs2_inode->ui_gid);
669 674
670 inode->i_size = fs64_to_cpu(sb, ufs2_inode->ui_size); 675 inode->i_size = fs64_to_cpu(sb, ufs2_inode->ui_size);
671 inode->i_atime.tv_sec = fs32_to_cpu(sb, ufs2_inode->ui_atime.tv_sec); 676 inode->i_atime.tv_sec = fs64_to_cpu(sb, ufs2_inode->ui_atime);
672 inode->i_ctime.tv_sec = fs32_to_cpu(sb, ufs2_inode->ui_ctime.tv_sec); 677 inode->i_ctime.tv_sec = fs64_to_cpu(sb, ufs2_inode->ui_ctime);
673 inode->i_mtime.tv_sec = fs32_to_cpu(sb, ufs2_inode->ui_mtime.tv_sec); 678 inode->i_mtime.tv_sec = fs64_to_cpu(sb, ufs2_inode->ui_mtime);
674 inode->i_mtime.tv_nsec = 0; 679 inode->i_atime.tv_nsec = fs32_to_cpu(sb, ufs2_inode->ui_atimensec);
675 inode->i_atime.tv_nsec = 0; 680 inode->i_ctime.tv_nsec = fs32_to_cpu(sb, ufs2_inode->ui_ctimensec);
676 inode->i_ctime.tv_nsec = 0; 681 inode->i_mtime.tv_nsec = fs32_to_cpu(sb, ufs2_inode->ui_mtimensec);
677 inode->i_blocks = fs64_to_cpu(sb, ufs2_inode->ui_blocks); 682 inode->i_blocks = fs64_to_cpu(sb, ufs2_inode->ui_blocks);
678 inode->i_generation = fs32_to_cpu(sb, ufs2_inode->ui_gen); 683 inode->i_generation = fs32_to_cpu(sb, ufs2_inode->ui_gen);
679 ufsi->i_flags = fs32_to_cpu(sb, ufs2_inode->ui_flags); 684 ufsi->i_flags = fs32_to_cpu(sb, ufs2_inode->ui_flags);
@@ -690,6 +695,7 @@ static void ufs2_read_inode(struct inode *inode, struct ufs2_inode *ufs2_inode)
690 for (i = 0; i < (UFS_NDADDR + UFS_NINDIR) * 4; i++) 695 for (i = 0; i < (UFS_NDADDR + UFS_NINDIR) * 4; i++)
691 ufsi->i_u1.i_symlink[i] = ufs2_inode->ui_u2.ui_symlink[i]; 696 ufsi->i_u1.i_symlink[i] = ufs2_inode->ui_u2.ui_symlink[i];
692 } 697 }
698 return 0;
693} 699}
694 700
695void ufs_read_inode(struct inode * inode) 701void ufs_read_inode(struct inode * inode)
@@ -698,6 +704,7 @@ void ufs_read_inode(struct inode * inode)
698 struct super_block * sb; 704 struct super_block * sb;
699 struct ufs_sb_private_info * uspi; 705 struct ufs_sb_private_info * uspi;
700 struct buffer_head * bh; 706 struct buffer_head * bh;
707 int err;
701 708
702 UFSD("ENTER, ino %lu\n", inode->i_ino); 709 UFSD("ENTER, ino %lu\n", inode->i_ino);
703 710
@@ -720,14 +727,17 @@ void ufs_read_inode(struct inode * inode)
720 if ((UFS_SB(sb)->s_flags & UFS_TYPE_MASK) == UFS_TYPE_UFS2) { 727 if ((UFS_SB(sb)->s_flags & UFS_TYPE_MASK) == UFS_TYPE_UFS2) {
721 struct ufs2_inode *ufs2_inode = (struct ufs2_inode *)bh->b_data; 728 struct ufs2_inode *ufs2_inode = (struct ufs2_inode *)bh->b_data;
722 729
723 ufs2_read_inode(inode, 730 err = ufs2_read_inode(inode,
724 ufs2_inode + ufs_inotofsbo(inode->i_ino)); 731 ufs2_inode + ufs_inotofsbo(inode->i_ino));
725 } else { 732 } else {
726 struct ufs_inode *ufs_inode = (struct ufs_inode *)bh->b_data; 733 struct ufs_inode *ufs_inode = (struct ufs_inode *)bh->b_data;
727 734
728 ufs1_read_inode(inode, ufs_inode + ufs_inotofsbo(inode->i_ino)); 735 err = ufs1_read_inode(inode,
736 ufs_inode + ufs_inotofsbo(inode->i_ino));
729 } 737 }
730 738
739 if (err)
740 goto bad_inode;
731 inode->i_version++; 741 inode->i_version++;
732 ufsi->i_lastfrag = 742 ufsi->i_lastfrag =
733 (inode->i_size + uspi->s_fsize - 1) >> uspi->s_fshift; 743 (inode->i_size + uspi->s_fsize - 1) >> uspi->s_fshift;
@@ -803,12 +813,12 @@ static void ufs2_update_inode(struct inode *inode, struct ufs2_inode *ufs_inode)
803 ufs_inode->ui_gid = cpu_to_fs32(sb, inode->i_gid); 813 ufs_inode->ui_gid = cpu_to_fs32(sb, inode->i_gid);
804 814
805 ufs_inode->ui_size = cpu_to_fs64(sb, inode->i_size); 815 ufs_inode->ui_size = cpu_to_fs64(sb, inode->i_size);
806 ufs_inode->ui_atime.tv_sec = cpu_to_fs32(sb, inode->i_atime.tv_sec); 816 ufs_inode->ui_atime = cpu_to_fs64(sb, inode->i_atime.tv_sec);
807 ufs_inode->ui_atime.tv_usec = 0; 817 ufs_inode->ui_atimensec = cpu_to_fs32(sb, inode->i_atime.tv_nsec);
808 ufs_inode->ui_ctime.tv_sec = cpu_to_fs32(sb, inode->i_ctime.tv_sec); 818 ufs_inode->ui_ctime = cpu_to_fs64(sb, inode->i_ctime.tv_sec);
809 ufs_inode->ui_ctime.tv_usec = 0; 819 ufs_inode->ui_ctimensec = cpu_to_fs32(sb, inode->i_ctime.tv_nsec);
810 ufs_inode->ui_mtime.tv_sec = cpu_to_fs32(sb, inode->i_mtime.tv_sec); 820 ufs_inode->ui_mtime = cpu_to_fs64(sb, inode->i_mtime.tv_sec);
811 ufs_inode->ui_mtime.tv_usec = 0; 821 ufs_inode->ui_mtimensec = cpu_to_fs32(sb, inode->i_mtime.tv_nsec);
812 822
813 ufs_inode->ui_blocks = cpu_to_fs64(sb, inode->i_blocks); 823 ufs_inode->ui_blocks = cpu_to_fs64(sb, inode->i_blocks);
814 ufs_inode->ui_flags = cpu_to_fs32(sb, ufsi->i_flags); 824 ufs_inode->ui_flags = cpu_to_fs32(sb, ufsi->i_flags);
@@ -888,6 +898,8 @@ void ufs_delete_inode (struct inode * inode)
888 loff_t old_i_size; 898 loff_t old_i_size;
889 899
890 truncate_inode_pages(&inode->i_data, 0); 900 truncate_inode_pages(&inode->i_data, 0);
901 if (is_bad_inode(inode))
902 goto no_delete;
891 /*UFS_I(inode)->i_dtime = CURRENT_TIME;*/ 903 /*UFS_I(inode)->i_dtime = CURRENT_TIME;*/
892 lock_kernel(); 904 lock_kernel();
893 mark_inode_dirty(inode); 905 mark_inode_dirty(inode);
@@ -898,4 +910,7 @@ void ufs_delete_inode (struct inode * inode)
898 ufs_warning(inode->i_sb, __FUNCTION__, "ufs_truncate failed\n"); 910 ufs_warning(inode->i_sb, __FUNCTION__, "ufs_truncate failed\n");
899 ufs_free_inode (inode); 911 ufs_free_inode (inode);
900 unlock_kernel(); 912 unlock_kernel();
913 return;
914no_delete:
915 clear_inode(inode); /* We must guarantee clearing of inode... */
901} 916}
diff --git a/fs/ufs/truncate.c b/fs/ufs/truncate.c
index 749581fa7729..79c54c85fb58 100644
--- a/fs/ufs/truncate.c
+++ b/fs/ufs/truncate.c
@@ -74,7 +74,7 @@ static int ufs_trunc_direct(struct inode *inode)
74 unsigned i, tmp; 74 unsigned i, tmp;
75 int retry; 75 int retry;
76 76
77 UFSD("ENTER\n"); 77 UFSD("ENTER: ino %lu\n", inode->i_ino);
78 78
79 sb = inode->i_sb; 79 sb = inode->i_sb;
80 uspi = UFS_SB(sb)->s_uspi; 80 uspi = UFS_SB(sb)->s_uspi;
@@ -96,8 +96,8 @@ static int ufs_trunc_direct(struct inode *inode)
96 block2 = ufs_fragstoblks (frag3); 96 block2 = ufs_fragstoblks (frag3);
97 } 97 }
98 98
99 UFSD("frag1 %llu, frag2 %llu, block1 %llu, block2 %llu, frag3 %llu," 99 UFSD("ino %lu, frag1 %llu, frag2 %llu, block1 %llu, block2 %llu,"
100 " frag4 %llu\n", 100 " frag3 %llu, frag4 %llu\n", inode->i_ino,
101 (unsigned long long)frag1, (unsigned long long)frag2, 101 (unsigned long long)frag1, (unsigned long long)frag2,
102 (unsigned long long)block1, (unsigned long long)block2, 102 (unsigned long long)block1, (unsigned long long)block2,
103 (unsigned long long)frag3, (unsigned long long)frag4); 103 (unsigned long long)frag3, (unsigned long long)frag4);
@@ -163,7 +163,7 @@ next1:
163 mark_inode_dirty(inode); 163 mark_inode_dirty(inode);
164 next3: 164 next3:
165 165
166 UFSD("EXIT\n"); 166 UFSD("EXIT: ino %lu\n", inode->i_ino);
167 return retry; 167 return retry;
168} 168}
169 169
@@ -248,7 +248,7 @@ static int ufs_trunc_indirect(struct inode *inode, u64 offset, void *p)
248 } 248 }
249 ubh_brelse (ind_ubh); 249 ubh_brelse (ind_ubh);
250 250
251 UFSD("EXIT\n"); 251 UFSD("EXIT: ino %lu\n", inode->i_ino);
252 252
253 return retry; 253 return retry;
254} 254}
@@ -262,7 +262,7 @@ static int ufs_trunc_dindirect(struct inode *inode, u64 offset, void *p)
262 void *dind; 262 void *dind;
263 int retry = 0; 263 int retry = 0;
264 264
265 UFSD("ENTER\n"); 265 UFSD("ENTER: ino %lu\n", inode->i_ino);
266 266
267 sb = inode->i_sb; 267 sb = inode->i_sb;
268 uspi = UFS_SB(sb)->s_uspi; 268 uspi = UFS_SB(sb)->s_uspi;
@@ -312,7 +312,7 @@ static int ufs_trunc_dindirect(struct inode *inode, u64 offset, void *p)
312 } 312 }
313 ubh_brelse (dind_bh); 313 ubh_brelse (dind_bh);
314 314
315 UFSD("EXIT\n"); 315 UFSD("EXIT: ino %lu\n", inode->i_ino);
316 316
317 return retry; 317 return retry;
318} 318}
@@ -327,7 +327,7 @@ static int ufs_trunc_tindirect(struct inode *inode)
327 void *tind, *p; 327 void *tind, *p;
328 int retry; 328 int retry;
329 329
330 UFSD("ENTER\n"); 330 UFSD("ENTER: ino %lu\n", inode->i_ino);
331 331
332 retry = 0; 332 retry = 0;
333 333
@@ -348,7 +348,7 @@ static int ufs_trunc_tindirect(struct inode *inode)
348 } 348 }
349 349
350 for (i = tindirect_block ; i < uspi->s_apb ; i++) { 350 for (i = tindirect_block ; i < uspi->s_apb ; i++) {
351 tind = ubh_get_addr32 (tind_bh, i); 351 tind = ubh_get_data_ptr(uspi, tind_bh, i);
352 retry |= ufs_trunc_dindirect(inode, UFS_NDADDR + 352 retry |= ufs_trunc_dindirect(inode, UFS_NDADDR +
353 uspi->s_apb + ((i + 1) << uspi->s_2apbshift), tind); 353 uspi->s_apb + ((i + 1) << uspi->s_2apbshift), tind);
354 ubh_mark_buffer_dirty(tind_bh); 354 ubh_mark_buffer_dirty(tind_bh);
@@ -372,19 +372,21 @@ static int ufs_trunc_tindirect(struct inode *inode)
372 } 372 }
373 ubh_brelse (tind_bh); 373 ubh_brelse (tind_bh);
374 374
375 UFSD("EXIT\n"); 375 UFSD("EXIT: ino %lu\n", inode->i_ino);
376 return retry; 376 return retry;
377} 377}
378 378
379static int ufs_alloc_lastblock(struct inode *inode) 379static int ufs_alloc_lastblock(struct inode *inode)
380{ 380{
381 int err = 0; 381 int err = 0;
382 struct super_block *sb = inode->i_sb;
382 struct address_space *mapping = inode->i_mapping; 383 struct address_space *mapping = inode->i_mapping;
383 struct ufs_sb_private_info *uspi = UFS_SB(inode->i_sb)->s_uspi; 384 struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
384 unsigned i, end; 385 unsigned i, end;
385 sector_t lastfrag; 386 sector_t lastfrag;
386 struct page *lastpage; 387 struct page *lastpage;
387 struct buffer_head *bh; 388 struct buffer_head *bh;
389 u64 phys64;
388 390
389 lastfrag = (i_size_read(inode) + uspi->s_fsize - 1) >> uspi->s_fshift; 391 lastfrag = (i_size_read(inode) + uspi->s_fsize - 1) >> uspi->s_fshift;
390 392
@@ -424,6 +426,20 @@ static int ufs_alloc_lastblock(struct inode *inode)
424 set_page_dirty(lastpage); 426 set_page_dirty(lastpage);
425 } 427 }
426 428
429 if (lastfrag >= UFS_IND_FRAGMENT) {
430 end = uspi->s_fpb - ufs_fragnum(lastfrag) - 1;
431 phys64 = bh->b_blocknr + 1;
432 for (i = 0; i < end; ++i) {
433 bh = sb_getblk(sb, i + phys64);
434 lock_buffer(bh);
435 memset(bh->b_data, 0, sb->s_blocksize);
436 set_buffer_uptodate(bh);
437 mark_buffer_dirty(bh);
438 unlock_buffer(bh);
439 sync_dirty_buffer(bh);
440 brelse(bh);
441 }
442 }
427out_unlock: 443out_unlock:
428 ufs_put_locked_page(lastpage); 444 ufs_put_locked_page(lastpage);
429out: 445out:
diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index e2bea6a661f0..69e9e80735d2 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -1829,11 +1829,11 @@ xfs_buf_init(void)
1829 if (!xfs_buf_zone) 1829 if (!xfs_buf_zone)
1830 goto out_free_trace_buf; 1830 goto out_free_trace_buf;
1831 1831
1832 xfslogd_workqueue = create_freezeable_workqueue("xfslogd"); 1832 xfslogd_workqueue = create_workqueue("xfslogd");
1833 if (!xfslogd_workqueue) 1833 if (!xfslogd_workqueue)
1834 goto out_free_buf_zone; 1834 goto out_free_buf_zone;
1835 1835
1836 xfsdatad_workqueue = create_freezeable_workqueue("xfsdatad"); 1836 xfsdatad_workqueue = create_workqueue("xfsdatad");
1837 if (!xfsdatad_workqueue) 1837 if (!xfsdatad_workqueue)
1838 goto out_destroy_xfslogd_workqueue; 1838 goto out_destroy_xfslogd_workqueue;
1839 1839