The patch titled
ext2: don't inherit inappropriate inode flags from parent
has been added to the -mm tree. Its filename is
ext2-dont-inherit-inappropriate-inode-flags-from-parent.patch
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/SubmitChecklist when testing your code ***
See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
out what to do about this
The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/
------------------------------------------------------
Subject: ext2: don't inherit inappropriate inode flags from parent
From: "Duane Griffin" <[email protected]>
Fix the TOPDIR flag inheritance bug reported at
http://bugzilla.kernel.org/show_bug.cgi?id=9866.
Some file attributes have sticky behaviour in ext2/3 whereas I believe
it should not. Namely, a directory marked with chattr +T serves "to be
the top of directory hierarchies for the purposes of the Orlov block
allocator", as the manpage says.
It seems wrong that subdirectories stickily inherit the +T flag: they
are obviously not top-level! Even files get the T attribute this way,
even though the chattr manpages only documents it for directories.
Also prevent inheritance of other flags that Andreas Dilger identified as
inappropriate to inherit
At present BTREE is the only flag that new ext2 inodes do NOT inherit from
their parent. In addition prevent the flags DIRTY, ECOMPR, INDEX, IMAGIC
and TOPDIR from being inherited.
Signed-off-by: Duane Griffin <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
---
fs/ext2/ialloc.c | 2 +-
include/linux/ext2_fs.h | 4 ++++
2 files changed, 5 insertions(+), 1 deletion(-)
diff -puN fs/ext2/ialloc.c~ext2-dont-inherit-inappropriate-inode-flags-from-parent fs/ext2/ialloc.c
--- a/fs/ext2/ialloc.c~ext2-dont-inherit-inappropriate-inode-flags-from-parent
+++ a/fs/ext2/ialloc.c
@@ -565,7 +565,7 @@ got:
inode->i_blocks = 0;
inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME_SEC;
memset(ei->i_data, 0, sizeof(ei->i_data));
- ei->i_flags = EXT2_I(dir)->i_flags & ~EXT2_BTREE_FL;
+ ei->i_flags = EXT2_I(dir)->i_flags & EXT2_FL_INHERITED;
if (S_ISLNK(mode))
ei->i_flags &= ~(EXT2_IMMUTABLE_FL|EXT2_APPEND_FL);
/* dirsync is only applied to directories */
diff -puN include/linux/ext2_fs.h~ext2-dont-inherit-inappropriate-inode-flags-from-parent include/linux/ext2_fs.h
--- a/include/linux/ext2_fs.h~ext2-dont-inherit-inappropriate-inode-flags-from-parent
+++ a/include/linux/ext2_fs.h
@@ -194,6 +194,10 @@ struct ext2_group_desc
#define EXT2_FL_USER_VISIBLE FS_FL_USER_VISIBLE /* User visible flags */
#define EXT2_FL_USER_MODIFIABLE FS_FL_USER_MODIFIABLE /* User modifiable flags */
+/* Flags that should be inherited by new inodes from their parent. */
+#define EXT2_FL_INHERITED ~(EXT2_DIRTY_FL | EXT2_ECOMPR_FL | EXT2_BTREE_FL |\
+ EXT2_INDEX_FL | EXT2_IMAGIC_FL | EXT2_TOPDIR_FL)
+
/*
* ioctl commands
*/
_
Patches currently in -mm which might be from [email protected] are
jbd-replace-potentially-false-assertion-with-if-block.patch
jbd-eliminate-duplicated-code-in-revocation-table-init-destroy-functions.patch
jbd-tidy-up-revoke-cache-initialisation-and-destruction.patch
ext2-dont-inherit-inappropriate-inode-flags-from-parent.patch
ext3-dont-inherit-inappropriate-inode-flags-from-parent.patch
ext4-dont-inherit-inappropriate-inode-flags-from-parent.patch
On Jun 04, 2008 21:37 -0700, Andrew Morton wrote:
> +/* Flags that should be inherited by new inodes from their parent. */
> +#define EXT2_FL_INHERITED ~(EXT2_DIRTY_FL | EXT2_ECOMPR_FL | EXT2_BTREE_FL |\
> + EXT2_INDEX_FL | EXT2_IMAGIC_FL | EXT2_TOPDIR_FL)
This doesn't make things much better, unless it is just as a stop-gap
measure to get into the kernel while there is more discussion about which
flags need to be inherited... Duane did mention follow-up patches, so
I have no real objection to this patch as-is, so long as we don't consider
the matter closed.
Minor nit - "EXT2_BTREE_FL" is the long-deprecated name for "EXT2_INDEX_FL",
so we may as well just avoid it entirely here.
Yes, this patch avoids a few more of the cases that are bad, but will
again be incorrect as soon as some new flag enters use.
Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.
On Wed, 04 Jun 2008 23:30:43 -0600 Andreas Dilger <[email protected]> wrote:
> On Jun 04, 2008 21:37 -0700, Andrew Morton wrote:
> > +/* Flags that should be inherited by new inodes from their parent. */
> > +#define EXT2_FL_INHERITED ~(EXT2_DIRTY_FL | EXT2_ECOMPR_FL | EXT2_BTREE_FL |\
> > + EXT2_INDEX_FL | EXT2_IMAGIC_FL | EXT2_TOPDIR_FL)
>
> This doesn't make things much better, unless it is just as a stop-gap
> measure to get into the kernel while there is more discussion about which
> flags need to be inherited... Duane did mention follow-up patches, so
> I have no real objection to this patch as-is, so long as we don't consider
> the matter closed.
>
> Minor nit - "EXT2_BTREE_FL" is the long-deprecated name for "EXT2_INDEX_FL",
> so we may as well just avoid it entirely here.
>
> Yes, this patch avoids a few more of the cases that are bad, but will
> again be incorrect as soon as some new flag enters use.
>
ah, OK, thanks. I'll drop them again pending finalisation of all this.