linux-kernel - Re: [PATCH] spi: tegra: avoid unsigned->signed->unsigned promotion

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: <20131129214313.GA18036@mithrandir>
Date:	Fri, 29 Nov 2013 22:43:15 +0100
From:	Thierry Reding <thierry.reding@...il.com>
To:	Michal Nazarewicz <mpn@...gle.com>
Cc:	Laxman Dewangan <ldewangan@...dia.com>,
	Mark Brown <broonie@...nel.org>, linux-spi@...r.kernel.org,
	linux-tegra@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] spi: tegra: avoid unsigned->signed->unsigned promotion

On Fri, Nov 29, 2013 at 06:12:44PM +0100, Michal Nazarewicz wrote:
> From: Michal Nazarewicz <mina86@...a86.com>
> 
> u8 type, which is unsigned, is promoted to int, which is singned, when
> doing a binary shift.  Then, on 64-bit machines, it is further promoted
> to unsigned long which may lead to more significant half of the value
> to be all ones.  To avoid this, explicitly promote to an unsigned type.

So this would only become an problem on 64-bit Tegra SoCs, right? I
suspect that there are quite a few places where something similar is
done and which suffer from the same issue.

But I don't think there's a real issue here. Even on 64-bit platforms,
writel() will only write 32-bit registers, and therefore discard the
upper 32 bits.

I think in general perhaps a more proper fix would be to use only u32
instead of unsigned long for these cases, since then the problem goes
away. u32 is also the type of the value used by writel() so it makes
perfect sense to use it.

Thierry

Content of type "application/pgp-signature" skipped