linux-kernel - Re: [PATCH] hwmon: (nzxt-kraken2) mark and order concurrent accesses

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

Date:   Tue, 30 Mar 2021 14:53:02 -0300
From:   Jonas Malaco <jonas@...tocubo.io>
To:     Guenter Roeck <linux@...ck-us.net>
Cc:     Jean Delvare <jdelvare@...e.com>, linux-hwmon@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] hwmon: (nzxt-kraken2) mark and order concurrent accesses

On Tue, Mar 30, 2021 at 03:51:21AM -0700, Guenter Roeck wrote:
> [ ... ]
> 
> Then please explain why _this_ use of time_after() is wrong but all
> others in the kernel are not. Also, please note that we are not
> concerned with code generation by the compiler as long as the
> generated code is correct (and I don't see any indication that
> it isn't).

The accesses to priv->temp_input[], ->fan_input[] and ->updated (where
this relates to your question about time_after()) are not wrong, but
undefined.

But if we are not concerned with code that is currently generated
correctly, which indeed is the case in the five arch x compiler
combinations I tested, then there simply is no point to this patch.

Thanks for going through this with me,
Jonas

P.S. Admittedly from a sample way too small, but in the kernel
time_after(jiffies, x) calls do not generally appear to involve an
expression x with a data race.  And there are a few calls where
READ_ONCE() is indeed used in x.