linux-kernel - Re: [PATCH] module: correctly exit module_kallsyms_on_each

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>] [day] [month] [year] [list]

Message-ID: <YKZwzbkh2/qhLk+d@alley>
Date:   Thu, 20 May 2021 16:23:09 +0200
From:   Petr Mladek <pmladek@...e.com>
To:     Jon Mediero <jmdr@...root.org>
Cc:     Jessica Yu <jeyu@...nel.org>, Miroslav Benes <mbenes@...e.cz>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] module: correctly exit module_kallsyms_on_each_symbol
 when fn() != 0

On Thu 2021-05-20 14:23:26, Jon Mediero wrote:
> Commit 013c1667cf78 ("kallsyms: refactor
> {,module_}kallsyms_on_each_symbol") replaced the return inside the
> nested loop with a break, changing the semantics of the function: the
> break only exits the innermost loop, so the code continues iterating the
> symbols of the next module instead of exiting.
> 
> Fixes: 013c1667cf78 ("kallsyms: refactor {,module_}kallsyms_on_each_symbol")
> Signed-off-by: Jon Mediero <jmdr@...root.org>

Great catch!

Reviewed-by: Petr Mladek <pmladek@...e.com>

> ---
> As far as I can tell the incorrect behaviour is not causing problems at
> the moment: module_kallsyms_on_each_symbol is no longer exported and the
> only existing caller (klp_find_object_symbol) does not care about the
> return value nor the extra iterations.

It would actually break livepatching a very subtle way.

klp_find_callback() must not longer get called when there are more
symbols of the same name and args->count matches the requested
args->pos. The buggy version will always find only the last
occurence of such a symbol name.

Best Regards,
Petr