Date: Tue, 21 Jul 2009 22:34:36 +0100 (BST)
From: Troy Moure <twmoure@szypr.net>
To: Linus Torvalds <torvalds@linux-foundation.org>
cc: Krzysztof Oledzki <olel@ans.pl>, Greg KH <gregkh@suse.de>,
       linux-kernel@vger.kernel.org, Andrew Morton <akpm@linux-foundation.org>,
       stable@kernel.org, lwn@lwn.net
Subject: Re: Linux 2.6.27.27
In-Reply-To: <alpine.LFD.2.01.0907211133010.19335@localhost.localdomain>
Message-ID: <alpine.LFD.2.00.0907212211040.4630@troy-laptop>
References: <20090720040655.GA11940@kroah.com> <4A645A45.9060509@ans.pl> <20090720151008.GC10015@suse.de> <alpine.LFD.2.01.0907200848530.13838@localhost.localdomain> <alpine.LNX.1.10.0907202341580.29729@bizon.gios.gov.pl> <alpine.LFD.2.01.0907201504470.19335@localhost.localdomain>
 <alpine.LNX.1.10.0907210831210.29729@bizon.gios.gov.pl> <alpine.LNX.1.10.0907211154480.29729@bizon.gios.gov.pl> <alpine.LFD.2.01.0907210906400.19335@localhost.localdomain> <alpine.LFD.2.01.0907211133010.19335@localhost.localdomain>
User-Agent: Alpine 2.00 (LFD 1167 2008-08-23)
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3795
Lines: 77


On Tue, 21 Jul 2009, Linus Torvalds wrote:

> > Great. This is all about as perfect as could be asked for. Now it's just a 
> > question of trying to find the right code generation difference...
> 
> Ok, that "just" is turning out to be really painful.

I think I've found something interesting.  Look at the the code generated 
for edid_checksum() in driver/video/fbmon.c.  This is what I see for the 
-fno-strict-overflow kernel:

...
ffffffff803b37ed <edid_checksum>:
ffffffff803b37ed:       53                      push   %rbx
ffffffff803b37ee:       48 89 fb                mov    %rdi,%rbx
ffffffff803b37f1:       e8 8d fd ff ff          callq  ffffffff803b3583 <check_edid>
ffffffff803b37f6:       85 c0                   test   %eax,%eax
ffffffff803b37f8:       89 c6                   mov    %eax,%esi
ffffffff803b37fa:       74 08                   je     ffffffff803b3804 <edid_checksum+0x17>
ffffffff803b37fc:       48 89 df                mov    %rbx,%rdi
ffffffff803b37ff:       e8 c0 fe ff ff          callq  ffffffff803b36c4 <fix_edid>
ffffffff803b3804:       eb fe                   jmp    ffffffff803b3804 <edid_checksum+0x17>

ffffffff803b3806 <fb_parse_edid>:
ffffffff803b3806:       41 54                   push   %r12
ffffffff803b3808:       48 85 ff                test   %rdi,%rdi
...

That last insn in edid_checksum() doesn't look *quite* right to me...

The -fnone kernel has something a lot more sensible-looking:

ffffffff803b39dd <edid_checksum>:
ffffffff803b39dd:       53                      push   %rbx
ffffffff803b39de:       48 89 fb                mov    %rdi,%rbx
ffffffff803b39e1:       e8 8d fd ff ff          callq  ffffffff803b3773 <check_$
ffffffff803b39e6:       85 c0                   test   %eax,%eax
ffffffff803b39e8:       89 c6                   mov    %eax,%esi
ffffffff803b39ea:       74 08                   je     ffffffff803b39f4 <edid_c$
ffffffff803b39ec:       48 89 df                mov    %rbx,%rdi
ffffffff803b39ef:       e8 c0 fe ff ff          callq  ffffffff803b38b4 <fix_ed$
ffffffff803b39f4:       31 c9                   xor    %ecx,%ecx
ffffffff803b39f6:       31 f6                   xor    %esi,%esi
ffffffff803b39f8:       31 d2                   xor    %edx,%edx
ffffffff803b39fa:       eb 0a                   jmp    ffffffff803b3a06 <edid_c$
ffffffff803b39fc:       0f b6 c0                movzbl %al,%eax
ffffffff803b39ff:       8a 04 03                mov    (%rbx,%rax,1),%al
ffffffff803b3a02:       01 c1                   add    %eax,%ecx
ffffffff803b3a04:       09 c6                   or     %eax,%esi
ffffffff803b3a06:       88 d0                   mov    %dl,%al
ffffffff803b3a08:       ff c2                   inc    %edx
ffffffff803b3a0a:       81 fa 81 00 00 00       cmp    $0x81,%edx
ffffffff803b3a10:       75 ea                   jne    ffffffff803b39fc <edid_c$
ffffffff803b3a12:       84 c9                   test   %cl,%cl
ffffffff803b3a14:       0f 94 c0                sete   %al
ffffffff803b3a17:       40 84 f6                test   %sil,%sil
ffffffff803b3a1a:       5b                      pop    %rbx
ffffffff803b3a1b:       0f 95 c2                setne  %dl
ffffffff803b3a1e:       21 d0                   and    %edx,%eax
ffffffff803b3a20:       0f b6 c0                movzbl %al,%eax
ffffffff803b3a23:       c3                      retq

ffffffff803b3a24 <fb_parse_edid>:
ffffffff803b3a24:       41 54                   push   %r12
ffffffff803b3a26:       48 85 ff                test   %rdi,%rdi

Hope that helps narrow things down ;)

	Troy

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/