linux-kernel - Re: [PATCH] clk: vc5: Avoid divide by zero when rounding or setting rates

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: <9b8f68f6-dd88-cc1a-6f13-b4d2fedadead@gmail.com>
Date:   Wed, 18 Jul 2018 10:56:51 -0700
From:   Steve Longerbeam <slongerbeam@...il.com>
To:     Stephen Boyd <sboyd@...nel.org>,
        Marek Vasut <marek.vasut@...il.com>,
        Michael Turquette <mturquette@...libre.com>
Cc:     linux-clk@...r.kernel.org, linux-kernel@...r.kernel.org,
        Steve Longerbeam <steve_longerbeam@...tor.com>
Subject: Re: [PATCH] clk: vc5: Avoid divide by zero when rounding or setting
 rates

On 07/06/2018 11:23 AM, Stephen Boyd wrote:
> Quoting Steve Longerbeam (2018-05-31 18:59:17)
>> Add checks in the .round_rate and .set_rate ops for zero requested
>> rate or zero parent rate. If either are zero in .round_rate, just
>> return zero. If either are zero in .set_rate, return -EINVAL.
> Are you seeing problems when the clk is unparented and we're trying to
> recalculate the rate or change rates, and thus the parent frequency
> looks like 0?

The problem appeared when suspending the rcar-du driver.
The kernel tested is a Renesas BSP release (3.6.2), and in the
version of the rcar-du driver from that release, the driver calls
clk_set_rate() with a rate of zero in its suspend PM op. This is
fixed in mainline kernel. So the divide-by-zero in vc5 clock driver
probably will not show up in mainline.

>   Should this get a Fixes: tag so that it goes back to
> stable kernels?
>

Zero rates are not checked beginning with the initial commit
3e1aec4e2c ("clk: vc5: Add support for IDT VersaClock 5P49V5923 and 
5P49V5933").
so that would have to be the Fixes: tag.

Steve