Re: [PATCH maint-1.6.5] block-sha1: avoid unaligned accesses on some big-endian systems

From: Jonathan Nieder <jrnieder@gmail.com>
Date: Jul 14, 2012, 19:50 UTC
Message-ID: <20120714195022.GB23242@burratino>
In-Reply-To: <CA+55aFy+y=TCoJUQarinaduibt4i-46TAuvpp7fsAmjDZj_+3w@mail.gmail.com>

Linus Torvalds wrote:

> On Sat, Jul 14, 2012 at 12:59 AM, Jonathan Nieder <jrnieder@gmail.com> wrote:

Show 7 quoted lines

>> Unfortunately, on big-endian architectures, if p is a pointer to
>> unsigned int then current gcc assumes it is properly aligned and
>> converts this construct to a 32-bit load.
>
> This patch seems to entirely depend on the location of the cast. And
> as far as I can tell, that workaround will in turn depend on just what
> gets inlined.

After the patch, what reason does gcc have to expect that 'block' is 32-bit aligned except when it is? The code (including the code I didn't touch) never casts from char * to int * except in get/put_be32 on arches that don't mind unaligned accesses.

[...]

> Anyway, the whole "noticed on alpha" makes no sense, since alpha isn't
> even big-endian. So the commit log is insane and misleading too.

Yes, true.  Thanks for catching it.

Jonathan