Re: bisected boot regression post 2.6.25-rc3.. please revert

From: Ingo Molnar <mingo@elte.hu>
To: Segher Boessenkool <segher@kernel.crashing.org>
Cc: Arjan van de Ven <arjan@linux.intel.com>,
	hans.rosenfeld@amd.com, torvalds@linux-foundation.org,
	linux-kernel@vger.kernel.org, "H. Peter Anvin" <hpa@zytor.com>,
	Thomas Gleixner <tglx@linutronix.de>
Subject: Re: bisected boot regression post 2.6.25-rc3.. please revert
Date: Tue, 4 Mar 2008 07:58:53 +0100	[thread overview]
Message-ID: <20080304065853.GB32287@elte.hu> (raw)
In-Reply-To: <962471e5d2b8c88858dd5e0f50269751@kernel.crashing.org>

* Segher Boessenkool <segher@kernel.crashing.org> wrote:

> so (on 32-bit) ~_PAGE_PSE is ~0x80L is 0xffffff7f, which when cast to 
> 64-bit is 0x00000000ffffff7f, so in
>
> 	(~PAGE_MASK & ~_PAGE_USER & ~_PAGE_PSE & ~_PAGE_NX)
>
> all the high bits are lost, while the original
>
> 	~(PAGE_MASK | _PAGE_USER | _PAGE_PSE | _PAGE_NX)
>
> works as intended, since the bit inversion is done on a 64-bit number.

but we really are interested in the low bits here (lets ignore the NX 
bit for now) and the patch has been in the queue for a long time (more 
than a month), so if there was a trivial mask mixup problem it would 
have shown on the first day. So i suspect some gcc bug instead - and 
certainly the colorful mixture of types and signs in this expression 
might have surprised a new version of gcc somewhere.

	Ingo