Re: [PATCH] sha: prevent removal of memset as dead store in sha1_update()

From: Andi Kleen <andi@firstfloor.org>
To: Mikael Pettersson <mikpe@it.uu.se>
Cc: Andi Kleen <andi@firstfloor.org>,
	roel kluin <roel.kluin@gmail.com>,
	David Miller <davem@davemloft.net>,
	herbert@gondor.apana.org.au, linux-crypto@vger.kernel.org,
	akpm@linux-foundation.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] sha: prevent removal of memset as dead store in sha1_update()
Date: Fri, 26 Feb 2010 16:55:53 +0100	[thread overview]
Message-ID: <20100226155553.GD16335@basil.fritz.box> (raw)
In-Reply-To: <19335.60634.75667.480609@pilspetsen.it.uu.se>

> Instead of barrier(), this works with gcc-3.2.3 up to gcc-4.4.3
> for the purpose of making the memset() not disappear:
> 
> 	{
> 		struct s { char c[n]; };
> 		asm("" : : "m"(*(struct s *)p));
> 	}
> 
> Every byte in the [p,p+n[ range must be used. If you only use the
> first byte, via e.g. asm("" :: "m"(*(char*)p)), then the compiler
> _will_ skip scrubbing bytes beyond the first.
> 
> An explicit loop that uses each byte individually also works, but
> results in awful code with older compilers.

I suspect it would be worth asking for a __builtin for this.
iirc Visual C already has one.

BTW I remember finding a few more such cases in random user space
code in past code review.

-Andi
-- 
ak@linux.intel.com -- Speaking for myself only.