[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZbnYitvLz7sWi727@casper.infradead.org>
Date: Wed, 31 Jan 2024 05:20:10 +0000
From: Matthew Wilcox <willy@...radead.org>
To: Theodore Ts'o <tytso@....edu>
Cc: Dave Chinner <david@...morbit.com>,
syzbot <syzbot+cdee56dbcdf0096ef605@...kaller.appspotmail.com>,
adilger.kernel@...ger.ca, chandan.babu@...cle.com, jack@...e.com,
linux-ext4@...r.kernel.org, linux-fsdevel@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-xfs@...r.kernel.org,
syzkaller-bugs@...glegroups.com
Subject: Re: current->journal_info got nested! (was Re: [syzbot] [xfs?]
[ext4?] general protection fault in jbd2__journal_start)
On Tue, Jan 30, 2024 at 11:58:22PM -0500, Theodore Ts'o wrote:
> Hmm, could XFS pre-fault target memory buffer for the bulkstat output
> before starting its transaction? Alternatively, ext4 could do a save
> of current->journal_info before starting to process the page fault,
> and restore it when it is done. Both of these seem a bit hacky, and
> the question is indeed, are there other avenues that might cause the
> transaction context nesting, such that a more general solution is
> called for?
I'd suggest that saving off current->journal_info is risky because
it might cover a real problem where you've taken a pagefault inside
a transaction (eg ext4 faulting while in the middle of a transaction on
the same filesystem that contains the faulting file).
Seems to me that we shouldn't be writing to userspace while in the
middle of a transaction. We could even assert that in copy_to_user()?
Powered by blists - more mailing lists