Hello everyone, The test below creates a sparse file and then fills a hole with O_DIRECT. As far as I can tell from reading generic_osync_inode, the filesystem metadata is only forced to disk if i_size changes during the file write. I've tested ext3, xfs and reiserfs and they all skip the commit when filling holes. I would argue that filling holes via O_DIRECT is supposed to commit the metadata required to find those file blocks later. At least on ext3, O_SYNC does force a commit on fill hole...

Hi Linus, Here's a set of patches that remove all calls to iget() and all read_inode() functions. They should be removed for two reasons: firstly they don't lend themselves to good error handling, and secondly their presence is a temptation for code outside a filesystem to call iget() to access inodes within that filesystem. There are a few benefits to this: (1) Error handling gets simpler as you can return an error code rather than having to call is_bad_inode(). (2) You can now...

Remove the old iget() call and the read_inode() superblock operation it uses as these are really obsolete, and the use of read_inode() does not produce proper error handling (no distinction between ENOMEM and EIO when marking an inode bad). Furthermore, this removes the temptation to use iget() to find an inode by number in a filesystem from code outside that filesystem. iget_locked() should be used instead. A new function is added in an earlier patch (iget_failed) that is to be called to mark a...

Stop the OPENPROMFS filesystem from using iget() and read_inode(). Replace openpromfs_read_inode() with openpromfs_iget(), and call that instead of iget(). openpromfs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. openpromfs_fill_super() returns any error incurred when getting the root inode instead of ENOMEM (not that it currently incurs any other error). Signed-off-by: David Howells <dhowells@redhat.com> --- fs/o...

Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. Signed-off-by: David Howells <dhowells@redhat.com> --- fs/ufs/inode.c | 34 ++++++++++++++++++++-------------- fs/ufs/namei.c ...

Stop the SYSV filesystem from using iget() and read_inode(). Replace sysv_read_inode() with sysv_iget(), and call that instead of iget(). sysv_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. Signed-off-by: David Howells <dhowells@redhat.com> --- fs/sysv/inode.c | 25 ++++++++++++++++--------- fs/sysv/namei.c | 6 +++--- fs/sysv/super.c | 4 ++-- fs/sysv/sysv.h | 1 + 4 files changed, 22 insertions(+), 14 ...

Stop the PROCFS filesystem from using iget() and read_inode(). Merge procfs_read_inode() into procfs_get_inode(), and have that call iget_locked() instead of iget(). Signed-off-by: David Howells <dhowells@redhat.com> --- fs/proc/inode.c | 60 ++++++++++++++++++++++++++----------------------------- 1 files changed, 28 insertions(+), 32 deletions(-) diff --git a/fs/proc/inode.c b/fs/proc/inode.c index 0e4d37c..7a563c5 100644 --- a/fs/proc/inode.c +++ b/fs/proc/inode.c @@ -78,11 +78,...

Stop the JFFS2 filesystem from using iget() and read_inode(). Replace jffs2_read_inode() with jffs2_iget(), and call that instead of iget(). jffs2_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. jffs2_do_fill_super() returns any error incurred when getting the root inode instead of EINVAL. Signed-off-by: David Howells <dhowells@redhat.com> --- fs/jffs2/dir.c | 6 +++-- fs/jffs2/fs.c | 56 ++++++++++++...

Stop the ISOFS filesystem from using read_inode(). Make isofs_read_inode() return an error code, and make isofs_iget() pass it on. Signed-off-by: David Howells <dhowells@redhat.com> --- fs/isofs/inode.c | 25 +++++++++++++++++-------- 1 files changed, 17 insertions(+), 8 deletions(-) diff --git a/fs/isofs/inode.c b/fs/isofs/inode.c index 043b470..28d990b 100644 --- a/fs/isofs/inode.c +++ b/fs/isofs/inode.c @@ -54,7 +54,7 @@ static void isofs_put_super(struct super_block *sb) ret...

Stop the FAT filesystem from using iget() and read_inode(). Replace the call to iget() with a call to ilookup(). Signed-off-by: David Howells <dhowells@redhat.com> --- fs/fat/inode.c | 6 ++---- 1 files changed, 2 insertions(+), 4 deletions(-) diff --git a/fs/fat/inode.c b/fs/fat/inode.c index 960ed3d..9ae2e9f 100644 --- a/fs/fat/inode.c +++ b/fs/fat/inode.c @@ -629,8 +629,6 @@ static const struct super_operations fat_sops = { .clear_inode = fat_clear_inode, .remount_fs = fa...

Stop the EXT4 filesystem from using iget() and read_inode(). Replace ext4_read_inode() with ext4_iget(), and call that instead of iget(). ext4_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ext4_fill_super() returns any error incurred when getting the root inode instead of EINVAL. Signed-off-by: David Howells <dhowells@redhat.com> Acked-by: "Theodore Ts'o" <tytso@mit.edu> Acked-by: Jan Kara <jack@suse.cz> ...

Stop the EXT2 filesystem from using iget() and read_inode(). Replace ext2_read_inode() with ext2_iget(), and call that instead of iget(). ext2_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ext2_fill_super() returns any error incurred when getting the root inode instead of EINVAL. Signed-off-by: David Howells <dhowells@redhat.com> Acked-by: "Theodore Ts'o" <tytso@mit.edu> --- fs/ext2/ext2.h | 2 +- fs/ex...

Stop the EFS filesystem from using iget() and read_inode(). Replace efs_read_inode() with efs_iget(), and call that instead of iget(). efs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. efs_fill_super() returns any error incurred when getting the root inode instead of EACCES. Signed-off-by: David Howells <dhowells@redhat.com> --- fs/efs/inode.c | 25 +++++++++++++++++-------- fs/efs/namei.c | 23...

Stop the autofs filesystem from using iget() and read_inode(). Replace autofs_read_inode() with autofs_iget(), and call that instead of iget(). autofs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. Signed-off-by: David Howells <dhowells@redhat.com> --- fs/autofs/autofs_i.h | 1 + fs/autofs/inode.c | 27 ++++++++++++++++++--------- fs/autofs/root.c | 22 ++++++++++++++++++---- 3 files changed, 37 inserti...

Stop the BEFS filesystem from using iget() and read_inode(). Replace befs_read_inode() with befs_iget(), and call that instead of iget(). befs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. befs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. Signed-off-by: David Howells <dhowells@redhat.com> Acked-by: Will Dyson <will_dyson@pobox.com> --- fs/befs/linuxvfs.c | 39 +++++...

Use iget_failed() in GFS2 to kill a failed inode. Signed-off-by: David Howells <dhowells@redhat.com> --- fs/gfs2/inode.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c index 34f7bcd..498844f 100644 --- a/fs/gfs2/inode.c +++ b/fs/gfs2/inode.c @@ -185,7 +185,7 @@ fail_put: ip->i_gl->gl_object = NULL; gfs2_glock_put(ip->i_gl); fail: - iput(inode); + iget_failed(inode); return ERR_PTR(error); } -

Introduce a function to register failure in an inode construction path. This includes marking the inode under construction as bad, unlocking it and releasing it. Signed-off-by: David Howells <dhowells@redhat.com> --- Documentation/filesystems/porting | 18 +++++++++++++----- fs/bad_inode.c | 14 ++++++++++++++ include/linux/fs.h | 1 + 3 files changed, 28 insertions(+), 5 deletions(-) diff --git a/Documentation/filesystems/porting b/Documentati...

Add an ERR_CAST() macro to complement ERR_PTR and co. for the purposes of casting an error entyped as one pointer type to an error of another pointer type whilst making it explicit as to what is going on. This provides a replacement for the ERR_PTR(PTR_ERR(p)) construct. Signed-off-by: David Howells <dhowells@redhat.com> --- include/linux/err.h | 12 ++++++++++++ 1 files changed, 12 insertions(+), 0 deletions(-) diff --git a/include/linux/err.h b/include/linux/err.h index 1ab1d44.....

On Fri, 12 Oct 2007 12:36:01 +0100, Andrew Clayton wrote: Ignore the numbers below. under normal conditions the raid array and single drive show basically the same numbers. The raid array numbers While we're talking number's... raid array, strace over NFS open("test", O_WRONLY|O_CREAT|O_EXCL|O_TRUNC, 0600) = 3 <0.018330> open("test", O_WRONLY|O_CREAT|O_EXCL|O_TRUNC, 0600) = 3 <0.026398> open("test", O_WRONLY|O_CREAT|O_EXCL|O_TRUNC, 0600) = 3 <8.927449> open("test", O_WRON...