fs: introduce inode operation ->update_time
Btrfs has to make sure we have space to allocate new blocks in order to modify
the inode, so updating time can fail. We've gotten around this by having our
own file_update_time but this is kind of a pain, and Christoph has indicated he
would like to make xfs do something different with atime updates. So introduce
->update_time, where we will deal with i_version an a/m/c time updates and
indicate which changes need to be made. The normal version just does what it
has always done, updates the time and marks the inode dirty, and then
filesystems can choose to do something different.
I've gone through all of the users of file_update_time and made them check for
errors with the exception of the fault code since it's complicated and I wasn't
quite sure what to do there, also Jan is going to be pushing the file time
updates into page_mkwrite for those who have it so that should satisfy btrfs and
make it not a big deal to check the file_update_time() return code in the
generic fault path. Thanks,
Signed-off-by: Josef Bacik <josef@redhat.com>
diff --git a/Documentation/filesystems/Locking b/Documentation/filesystems/Locking
index 4fca82e..d5a269a 100644
--- a/Documentation/filesystems/Locking
+++ b/Documentation/filesystems/Locking
@@ -62,6 +62,7 @@
int (*removexattr) (struct dentry *, const char *);
void (*truncate_range)(struct inode *, loff_t, loff_t);
int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len);
+ void (*update_time)(struct inode *, struct timespec *, int);
locking rules:
all may block
@@ -89,6 +90,8 @@
removexattr: yes
truncate_range: yes
fiemap: no
+update_time: no
+
Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on
victim.
cross-directory ->rename() has (per-superblock) ->s_vfs_rename_sem.
diff --git a/Documentation/filesystems/vfs.txt b/Documentation/filesystems/vfs.txt
index 0d04920..b2aa722 100644
--- a/Documentation/filesystems/vfs.txt
+++ b/Documentation/filesystems/vfs.txt
@@ -364,6 +364,7 @@
ssize_t (*listxattr) (struct dentry *, char *, size_t);
int (*removexattr) (struct dentry *, const char *);
void (*truncate_range)(struct inode *, loff_t, loff_t);
+ void (*update_time)(struct inode *, struct timespec *, int);
};
Again, all methods are called without any locks being held, unless
@@ -475,6 +476,9 @@
truncate_range: a method provided by the underlying filesystem to truncate a
range of blocks , i.e. punch a hole somewhere in a file.
+ update_time: called by the VFS to update a specific time or the i_version of
+ an inode. If this is not defined the VFS will update the inode itself
+ and call mark_inode_dirty_sync.
The Address Space Object
========================
diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index 504e61b..9562109 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -962,7 +962,9 @@
if (err)
goto out;
- file_update_time(file);
+ err = file_update_time(file);
+ if (err)
+ goto out;
if (file->f_flags & O_DIRECT) {
written = generic_file_direct_write(iocb, iov, &nr_segs,
diff --git a/fs/inode.c b/fs/inode.c
index a79555e..f0335fc 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -1487,6 +1487,27 @@
return 0;
}
+/*
+ * This does the actual work of updating an inodes time or version. Must have
+ * had called mnt_want_write() before calling this.
+ */
+static int update_time(struct inode *inode, struct timespec *time, int flags)
+{
+ if (inode->i_op->update_time)
+ return inode->i_op->update_time(inode, time, flags);
+
+ if (flags & S_ATIME)
+ inode->i_atime = *time;
+ if (flags & S_VERSION)
+ inode_inc_iversion(inode);
+ if (flags & S_CTIME)
+ inode->i_ctime = *time;
+ if (flags & S_MTIME)
+ inode->i_mtime = *time;
+ mark_inode_dirty_sync(inode);
+ return 0;
+}
+
/**
* touch_atime - update the access time
* @path: the &struct path to update
@@ -1524,8 +1545,14 @@
if (mnt_want_write(mnt))
return;
- inode->i_atime = now;
- mark_inode_dirty_sync(inode);
+ /*
+ * File systems can error out when updating inodes if they need to
+ * allocate new space to modify an inode (such is the case for
+ * Btrfs), but since we touch atime while walking down the path we
+ * really don't care if we failed to update the atime of the file,
+ * so just ignore the return value.
+ */
+ update_time(inode, &now, S_ATIME);
mnt_drop_write(mnt);
}
EXPORT_SYMBOL(touch_atime);
@@ -1604,18 +1631,20 @@
* usage in the file write path of filesystems, and filesystems may
* choose to explicitly ignore update via this function with the
* S_NOCMTIME inode flag, e.g. for network filesystem where these
- * timestamps are handled by the server.
+ * timestamps are handled by the server. This can return an error for
+ * file systems who need to allocate space in order to update an inode.
*/
-void file_update_time(struct file *file)
+int file_update_time(struct file *file)
{
struct inode *inode = file->f_path.dentry->d_inode;
struct timespec now;
- enum { S_MTIME = 1, S_CTIME = 2, S_VERSION = 4 } sync_it = 0;
+ int sync_it = 0;
+ int ret;
/* First try to exhaust all avenues to not sync */
if (IS_NOCMTIME(inode))
- return;
+ return 0;
now = current_fs_time(inode->i_sb);
if (!timespec_equal(&inode->i_mtime, &now))
@@ -1628,21 +1657,16 @@
sync_it |= S_VERSION;
if (!sync_it)
- return;
+ return 0;
/* Finally allowed to write? Takes lock. */
if (mnt_want_write_file(file))
- return;
+ return 0;
- /* Only change inode inside the lock region */
- if (sync_it & S_VERSION)
- inode_inc_iversion(inode);
- if (sync_it & S_CTIME)
- inode->i_ctime = now;
- if (sync_it & S_MTIME)
- inode->i_mtime = now;
- mark_inode_dirty_sync(inode);
+ ret = update_time(inode, &now, sync_it);
mnt_drop_write_file(file);
+
+ return ret;
}
EXPORT_SYMBOL(file_update_time);
diff --git a/fs/ncpfs/file.c b/fs/ncpfs/file.c
index 3ff5fcc..122e260 100644
--- a/fs/ncpfs/file.c
+++ b/fs/ncpfs/file.c
@@ -221,6 +221,10 @@
already_written = 0;
+ errno = file_update_time(file);
+ if (errno)
+ goto outrel;
+
bouncebuffer = vmalloc(bufsize);
if (!bouncebuffer) {
errno = -EIO; /* -ENOMEM */
@@ -252,8 +256,6 @@
}
vfree(bouncebuffer);
- file_update_time(file);
-
*ppos = pos;
if (pos > i_size_read(inode)) {
diff --git a/fs/ntfs/file.c b/fs/ntfs/file.c
index 8639169..7389d2d 100644
--- a/fs/ntfs/file.c
+++ b/fs/ntfs/file.c
@@ -2096,7 +2096,9 @@
err = file_remove_suid(file);
if (err)
goto out;
- file_update_time(file);
+ err = file_update_time(file);
+ if (err)
+ goto out;
written = ntfs_file_buffered_write(iocb, iov, nr_segs, pos, ppos,
count);
out:
diff --git a/fs/pipe.c b/fs/pipe.c
index 95ebb56..49c1065 100644
--- a/fs/pipe.c
+++ b/fs/pipe.c
@@ -654,8 +654,11 @@
wake_up_interruptible_sync_poll(&pipe->wait, POLLIN | POLLRDNORM);
kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
}
- if (ret > 0)
- file_update_time(filp);
+ if (ret > 0) {
+ int err = file_update_time(filp);
+ if (err)
+ ret = err;
+ }
return ret;
}
diff --git a/fs/splice.c b/fs/splice.c
index f847684..47c4c1a 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -1003,8 +1003,10 @@
mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD);
ret = file_remove_suid(out);
if (!ret) {
- file_update_time(out);
- ret = splice_from_pipe_feed(pipe, &sd, pipe_to_file);
+ ret = file_update_time(out);
+ if (!ret)
+ ret = splice_from_pipe_feed(pipe, &sd,
+ pipe_to_file);
}
mutex_unlock(&inode->i_mutex);
} while (ret > 0);
diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index 8d214b8..9f7ec15 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -586,8 +586,11 @@
* lock above. Eventually we should look into a way to avoid
* the pointless lock roundtrip.
*/
- if (likely(!(file->f_mode & FMODE_NOCMTIME)))
- file_update_time(file);
+ if (likely(!(file->f_mode & FMODE_NOCMTIME))) {
+ error = file_update_time(file);
+ if (error)
+ return error;
+ }
/*
* If we're writing the file then make sure to clear the setuid and
diff --git a/include/linux/fs.h b/include/linux/fs.h
index cdc1a96..57fc705 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1684,6 +1684,7 @@
void (*truncate_range)(struct inode *, loff_t, loff_t);
int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
u64 len);
+ int (*update_time)(struct inode *, struct timespec *, int);
} ____cacheline_aligned;
struct seq_file;
@@ -1843,6 +1844,13 @@
spin_unlock(&inode->i_lock);
}
+enum file_time_flags {
+ S_ATIME = 1,
+ S_MTIME = 2,
+ S_CTIME = 4,
+ S_VERSION = 8,
+};
+
extern void touch_atime(struct path *);
static inline void file_accessed(struct file *file)
{
@@ -2579,7 +2587,7 @@
extern int inode_newsize_ok(const struct inode *, loff_t offset);
extern void setattr_copy(struct inode *inode, const struct iattr *attr);
-extern void file_update_time(struct file *file);
+extern int file_update_time(struct file *file);
extern int generic_show_options(struct seq_file *m, struct dentry *root);
extern void save_mount_options(struct super_block *sb, char *options);
diff --git a/mm/filemap.c b/mm/filemap.c
index 21e5abf..51070f1 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2463,7 +2463,9 @@
if (err)
goto out;
- file_update_time(file);
+ err = file_update_time(file);
+ if (err)
+ goto out;
/* coalesce the iovecs and go direct-to-BIO for O_DIRECT */
if (unlikely(file->f_flags & O_DIRECT)) {
diff --git a/mm/filemap_xip.c b/mm/filemap_xip.c
index a4eb311..213ca1f 100644
--- a/mm/filemap_xip.c
+++ b/mm/filemap_xip.c
@@ -426,7 +426,9 @@
if (ret)
goto out_backing;
- file_update_time(filp);
+ ret = file_update_time(filp);
+ if (ret)
+ goto out_backing;
ret = __xip_file_write (filp, buf, count, pos, ppos);
