Re: [PATCH] ext4: Don't reduce symlink i_mode by umask if no ACL support

From: Theodore Ts'o
Date: Thu May 09 2024 - 10:28:26 EST


On Thu, May 09, 2024 at 03:47:27PM +0200, Miklos Szeredi wrote:
> On Thu, 9 May 2024 at 15:41, David Howells <dhowells@xxxxxxxxxx> wrote:
>
> > diff --git a/fs/ext4/acl.h b/fs/ext4/acl.h
> > index ef4c19e5f570..566625286442 100644
> > --- a/fs/ext4/acl.h
> > +++ b/fs/ext4/acl.h
> > @@ -71,7 +71,8 @@ ext4_init_acl(handle_t *handle, struct inode *inode, struct inode *dir)
> > /* usually, the umask is applied by posix_acl_create(), but if
> > ext4 ACL support is disabled at compile time, we need to do
> > it here, because posix_acl_create() will never be called */
> > - inode->i_mode &= ~current_umask();
> > + if (!S_ISLNK(inode->i_mode))
> > + inode->i_mode &= ~current_umask();
>
> I think this should just be removed unconditionally, since the VFS now
> takes care of mode masking in vfs_prepare_mode().

The following is in the ext4 tree:

commit c77194965dd0dcc26f9c1671d2e74e4eb1248af5
Author: Max Kellermann <max.kellermann@xxxxxxxxx>
Date: Fri Mar 15 15:29:56 2024 +0100

Revert "ext4: apply umask if ACL support is disabled"

This reverts commit 484fd6c1de13b336806a967908a927cc0356e312. The
commit caused a regression because now the umask was applied to
symlinks and the fix is unnecessary because the umask/O_TMPFILE bug
has been fixed somewhere else already.

Fixes: https://lore.kernel.org/lkml/28DSITL9912E1.2LSZUVTGTO52Q@xxxxxxxxxxx/
Signed-off-by: Max Kellermann <max.kellermann@xxxxxxxxx>
Reviewed-by: Christian Brauner <brauner@xxxxxxxxxx>
Tested-by: Michael Forney <mforney@xxxxxxxxxxx>
Link: https://lore.kernel.org/r/20240315142956.2420360-1-max.kellermann@xxxxxxxxx
Signed-off-by: Theodore Ts'o <tytso@xxxxxxx>