From owner-freebsd-current@FreeBSD.ORG  Wed Nov 28 23:22:50 2007
Return-Path: <owner-freebsd-current@FreeBSD.ORG>
Delivered-To: current@hub.freebsd.org
Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34])
	by hub.freebsd.org (Postfix) with ESMTP id 85E3116A421;
	Wed, 28 Nov 2007 23:22:50 +0000 (UTC)
	(envelope-from luoqi@FreeBSD.org)
Received: from freefall.freebsd.org (freefall.freebsd.org
	[IPv6:2001:4f8:fff6::28])
	by mx1.freebsd.org (Postfix) with ESMTP id 8937413C43E;
	Wed, 28 Nov 2007 23:22:45 +0000 (UTC)
	(envelope-from luoqi@FreeBSD.org)
Received: from freefall.freebsd.org (luoqi@localhost [127.0.0.1])
	by freefall.freebsd.org (8.14.2/8.14.2) with ESMTP id lASNMjHv014323;
	Wed, 28 Nov 2007 23:22:45 GMT
	(envelope-from luoqi@freefall.freebsd.org)
Received: (from luoqi@localhost)
	by freefall.freebsd.org (8.14.2/8.14.1/Submit) id lASNMjNu014322;
	Wed, 28 Nov 2007 23:22:45 GMT (envelope-from luoqi)
Date: Wed, 28 Nov 2007 23:22:45 GMT
From: Luoqi Chen <luoqi@FreeBSD.org>
Message-Id: <200711282322.lASNMjNu014322@freefall.freebsd.org>
To: current@FreeBSD.org, ivoras@FreeBSD.org
Cc: 
Subject: RE: gprof's broken in 7-CURRENT
X-BeenThere: freebsd-current@freebsd.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Discussions about the use of FreeBSD-current
	<freebsd-current.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-current>, 
	<mailto:freebsd-current-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-current>
List-Post: <mailto:freebsd-current@freebsd.org>
List-Help: <mailto:freebsd-current-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-current>,
	<mailto:freebsd-current-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Wed, 28 Nov 2007 23:22:50 -0000

> Luoqi Chen wrote:
> > Has anyone else noticed that gprof no longer works with gcc 4.2
>
> It's a known bug (very irritating!), but I think noone is working on it
> so it won't be fixed in 7.0.

In that case, I'll fix it.

I've looked at glibc's code, it claims that mcount cannot clobber any register.
I don't know if there is a standard for mcount interface, but I guess the gcc
team would use the glibc implementation as the reference. I'll change our code
to do the same.

/* We need a special version of the `mcount' function since for ix86 it
   must not clobber any register.  This has several reasons:
     - there is a bug in gcc as of version 2.7.2.2 which prohibits the
       use of profiling together with nested functions
     - the ELF `fixup' function uses GCC's regparm feature
     - some (future) systems might want to pass parameters in registers.  */

	ASM_GLOBAL_DIRECTIVE C_SYMBOL_NAME(_mcount)
	ASM_TYPE_DIRECTIVE(C_SYMBOL_NAME(_mcount), @function)
	.align ALIGNARG(4)
C_LABEL(_mcount)
	/* Save the caller-clobbered registers.  */
	pushl %eax
	pushl %ecx
	pushl %edx

	movl 12(%esp), %edx
	movl 4(%ebp), %eax

	/* No need to access the PLT or GOT, __mcount_internal is an
	   internal function and we can make a relative call.  */
	call C_SYMBOL_NAME(__mcount_internal)

	/* Pop the saved registers.  Please note that `mcount' has no
	   return value.  */
	popl %edx
	popl %ecx
	popl %eax
	ret
	ASM_SIZE_DIRECTIVE(C_SYMBOL_NAME(_mcount))

Thanks
-lq