[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <Pine.LNX.4.64.1011091352460.29956@wotan.suse.de>
Date: Tue, 9 Nov 2010 14:00:12 +0100 (CET)
From: Michael Matz <matz@...e.de>
To: Dave Korn <dave.korn.cygwin@...il.com>
Cc: Andi Kleen <andi@...stfloor.org>,
Richard Guenther <richard.guenther@...il.com>,
Andreas Schwab <schwab@...ux-m68k.org>, Jim <jim876@...all.nl>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
gcc@....gnu.org
Subject: Re: gcc 4.5.1 / as 2.20.51.0.11 miscompiling drivers/char/i8k.c ?
Hi,
On Mon, 8 Nov 2010, Dave Korn wrote:
> void foo (void)
> {
> int x, y, z;
> x = 23;
> asm ("do something" : "=r" (y) : "r" (x) );
> z = y + 1;
> }
The case in i8k.c really is different. It does use the value by
influencing the return value and the callers use the returned value in
conditionals and the like. It really, really _is_ used :-) and if GCC
removes the asm (which up to now is only speculation) then it's a GCC bug.
The code outlines like so:
int i8k_smm (regs) {
int rc;
asm (... : "=r"(rc) ...);
if (rc != 0 || ...)
return -EINVAL;
return 0;
}
...
struct regs regs = {.eax = ...}
return i8k_smm(regs) ?: regs.eax;
...
My speculation is, that the asm is not removed but rather that regs.eax
isn't reloaded after the asm because the memory clobber doesn't clobber
automatic variables.
Ciao,
Michael.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists