NetBSD Problem Report #38969

From simonb@thistledown.com.au  Tue Jun 17 06:24:13 2008
Return-Path: <simonb@thistledown.com.au>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
	by narn.NetBSD.org (Postfix) with ESMTP id ACEAD63B842
	for <gnats-bugs@gnats.NetBSD.org>; Tue, 17 Jun 2008 06:24:13 +0000 (UTC)
Message-Id: <20080617062412.5FE9DAFD04@thoreau.thistledown.com.au>
Date: Tue, 17 Jun 2008 16:24:12 +1000 (EST)
From: Simon Burge <simonb@NetBSD.org>
Reply-To: Simon Burge <simonb@NetBSD.org>
To: gnats-bugs@gnats.NetBSD.org
Subject: ddb backtrace on amd64 doesn't show function arguments
X-Send-Pr-Version: 3.95

>Number:         38969
>Category:       port-amd64
>Synopsis:       ddb backtrace on amd64 doesn't show function arguments
>Confidential:   no
>Severity:       serious
>Priority:       low
>Responsible:    port-amd64-maintainer
>State:          closed
>Class:          change-request
>Submitter-Id:   net
>Arrival-Date:   Tue Jun 17 06:25:00 +0000 2008
>Closed-Date:    Fri Nov 06 22:47:15 +0000 2009
>Last-Modified:  Fri Nov 06 22:47:15 +0000 2009
>Originator:     Simon Burge
>Release:        Any?
>Organization:
>Environment:
	Architecture: amd64
	Machine: amd64
>Description:
	ddb's "backtrace" command doesn't show any function arguments
	on an amd64 machine.  One example is:

		db{5}> bt
		breakpoint() at netbsd:breakpoint+0x5
		panic() at netbsd:panic+0x260
		__kernassert() at netbsd:__kernassert+0x2d
		uipc_usrreq() at netbsd:uipc_usrreq+0x850
		sosend() at netbsd:sosend+0x506
		soo_write() at netbsd:soo_write+0x2d
		dofilewrite() at netbsd:dofilewrite+0x81
		sys_write() at netbsd:sys_write+0x72
		syscall() at netbsd:syscall+0x9a

>How-To-Repeat:
	Type "bt" at a ddb prompt on an amd64 machine.

>Fix:
	Not sure.  Just changing amd64/md64/db_trace.c:db_numargs()
	to return "6" instead of "0" didn't look to provide realistic
	function args at a glance - I think this is similar to
	sparc/mips/<other-register-args-archs> do?  Maybe tempt the
	compiler somehow to save function arguments on the stack?

>Release-Note:

>Audit-Trail:
From: David Laight <david@l8s.co.uk>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-amd64/38969: ddb backtrace on amd64 doesn't show function arguments
Date: Wed, 18 Jun 2008 22:32:37 +0100

 On Tue, Jun 17, 2008 at 06:25:00AM +0000, Simon Burge wrote:
 > >Number:         38969
 > >Category:       port-amd64
 > >Synopsis:       ddb backtrace on amd64 doesn't show function arguments
 > 
 > >Fix:
 > 	Not sure.  Just changing amd64/md64/db_trace.c:db_numargs()
 > 	to return "6" instead of "0" didn't look to provide realistic
 > 	function args at a glance - I think this is similar to
 > 	sparc/mips/<other-register-args-archs> do?  Maybe tempt the
 > 	compiler somehow to save function arguments on the stack?

 The problem is that the first 6? arguments are passed in registers
 so are very hard to locate.

 	David

 -- 
 David Laight: david@l8s.co.uk

From: Simon Burge <simonb@NetBSD.org>
To: David Laight <david@l8s.co.uk>
Cc: gnats-bugs@NetBSD.org, port-amd64-maintainer@netbsd.org,
    gnats-admin@netbsd.org, netbsd-bugs@netbsd.org
Subject: Re: port-amd64/38969: ddb backtrace on amd64 doesn't show function arguments 
Date: Thu, 19 Jun 2008 15:09:54 +1000

 David Laight wrote:

 >  On Tue, Jun 17, 2008 at 06:25:00AM +0000, Simon Burge wrote:
 >  > >Number:         38969
 >  > >Category:       port-amd64
 >  > >Synopsis:       ddb backtrace on amd64 doesn't show function arguments
 >  > 
 >  > >Fix:
 >  > 	Not sure.  Just changing amd64/md64/db_trace.c:db_numargs()
 >  > 	to return "6" instead of "0" didn't look to provide realistic
 >  > 	function args at a glance - I think this is similar to
 >  > 	sparc/mips/<other-register-args-archs> do?  Maybe tempt the
 >  > 	compiler somehow to save function arguments on the stack?
 >  
 >  The problem is that the first 6? arguments are passed in registers
 >  so are very hard to locate.

 That's what I meant by "tempt the compiler ...".  Sure it'd be a
 performance hit, but if there was some flag to say "save every arg on
 the stack on each function entry" that'd make backtraces much more
 useful.

 Cheers,
 Simon.

State-Changed-From-To: open->closed
State-Changed-By: dsl@NetBSD.org
State-Changed-When: Fri, 06 Nov 2009 22:47:15 +0000
State-Changed-Why:
This isn't going to change....


>Unformatted:

Home	PR Database Search