Received-SPF: pass (google.com: domain of linux-ext4+bounces-1749-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Date: Mon, 25 Mar 2024 20:03:36 +0100
From: Jan Kara <jack@suse.cz>
To: Andreas Dilger <adilger@dilger.ca>
Cc: Jan Kara <jack@suse.cz>, Ted Tso <tytso@mit.edu>,
	linux-ext4@vger.kernel.org,
	syzbot+a43d4f48b8397d0e41a9@syzkaller.appspotmail.com
Subject: Re: [PATCH 2/2] ext4: Do not create EA inode under buffer lock
Message-ID: <20240325190336.lgnoze44yrdjjb7o@quack3>
References: <20240209111418.22308-1-jack@suse.cz>
 <20240321162657.27420-2-jack@suse.cz>
 <79FE751F-CFA8-4C46-B3D7-507C6A3BCFD3@dilger.ca>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <79FE751F-CFA8-4C46-B3D7-507C6A3BCFD3@dilger.ca>

On Fri 22-03-24 12:06:16, Andreas Dilger wrote:
> On Mar 21, 2024, at 10:26 AM, Jan Kara <jack@suse.cz> wrote:
> > 
> > ext4_xattr_set_entry() creates new EA inodes while holding buffer lock
> > on the external xattr block. This is problematic as it nests all the
> > allocation locking (which acquires locks on other buffers) under the
> > buffer lock. This can even deadlock when the filesystem is corrupted and
> > e.g. quota file is setup to contain xattr block as data block. Move the
> > allocation of EA inode out of ext4_xattr_set_entry() into the callers.
> 
> This looks like it will allocate a new inode for every setxattr called,
> even if the xattr is small and will likely fit inside the inode itself?
> This would seem to add a lot of extra overhead for the 99% of cases when
> an external inode is not needed.

This is not the case AFAICT. We call ext4_xattr_inode_lookup_create() only
in:

       if (i->value && i->in_inode) {

so that means we've already decided we need to put the xattr value in the
EA inode. Note that ext4_xattr_set_handle() for smaller xattr value first
calls ext4_xattr_block_set() with i.in_inode == 0 and if that fails due to
ENOSPC, it sets i.in_inode = 1 and tries again.

So I think everything is fine.

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR