linux-kernel - Re: [PATCH 1/2] xfs: ensure tmp_logflags is initialized in xfs_bmap_del_extent

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20231128161905.GT2766956@frogsfrogsfrogs>
Date:   Tue, 28 Nov 2023 08:19:05 -0800
From:   "Darrick J. Wong" <djwong@...nel.org>
To:     Christoph Hellwig <hch@...radead.org>
Cc:     Jiachen Zhang <zhangjiachen.jaycee@...edance.com>,
        Chandan Babu R <chandan.babu@...cle.com>,
        Dave Chinner <dchinner@...hat.com>,
        Allison Henderson <allison.henderson@...cle.com>,
        Zhang Tianci <zhangtianci.1997@...edance.com>,
        Brian Foster <bfoster@...hat.com>, Ben Myers <bpm@....com>,
        linux-xfs@...r.kernel.org, linux-kernel@...r.kernel.org,
        xieyongji@...edance.com, me@...x.top
Subject: Re: [PATCH 1/2] xfs: ensure tmp_logflags is initialized in
 xfs_bmap_del_extent_real

On Tue, Nov 28, 2023 at 12:19:23AM -0800, Christoph Hellwig wrote:
> On Tue, Nov 28, 2023 at 01:32:01PM +0800, Jiachen Zhang wrote:
> > In the case of returning -ENOSPC, ensure tmp_logflags is initialized by 0.
> > Otherwise the caller __xfs_bunmapi will set uninitialized illegal
> > tmp_logflags value into xfs log, which might cause unpredictable error
> > in the log recovery procedure.
> 
> This looks good:
> 
> Reviewed-by: Christoph Hellwig <hch@....de>
> 
> But I wonder if removing the local flags variable and always directly
> assigning to *logflagsp might be more robust in the long run.

Yes, I think it's better to eliminate opportunities for subtle logic
bombs by not open-coding variable aliasing.  Perhaps this function
should set *logflagsp = 0 at the start of the function so that we don't
have to deal with uninitialized outparams, especially since the caller
uses it even on an error return.

--D