NetBSD Problem Report #51089

From martin@aprisoft.de  Sun Apr 24 07:02:55 2016
Return-Path: <martin@aprisoft.de>
Received: from mail.netbsd.org (mail.netbsd.org [199.233.217.200])
	(using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
	(Client CN "mail.netbsd.org", Issuer "Postmaster NetBSD.org" (verified OK))
	by mollari.NetBSD.org (Postfix) with ESMTPS id EC10D7A467
	for <gnats-bugs@gnats.NetBSD.org>; Sun, 24 Apr 2016 07:02:55 +0000 (UTC)
Message-Id: <20160424070245.D9B10ED0E4F@emmas.aprisoft.de>
Date: Sun, 24 Apr 2016 09:02:45 +0200 (CEST)
From: martin@NetBSD.org
Reply-To: martin@NetBSD.org
To: gnats-bugs@NetBSD.org
Subject: random crashes building arm on amd64 -current
X-Send-Pr-Version: 3.95

>Number:         51089
>Category:       toolchain
>Synopsis:       random crashes building arm on amd64 -current
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    toolchain-manager
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Apr 24 07:05:00 +0000 2016
>Closed-Date:    Fri Dec 21 07:55:48 +0000 2018
>Last-Modified:  Fri Dec 21 07:55:48 +0000 2018
>Originator:     Martin Husemann
>Release:        NetBSD 7.99.28
>Organization:
The NetBSD Foundation, Inc.
>Environment:
System: NetBSD martins.aprisoft.de 7.99.28 NetBSD 7.99.28 (GENERIC) #63: Sat Apr 23 15:19:11 CEST 2016 martin@martins.aprisoft.de:/ssd/src/sys/arch/amd64/compile/GENERIC amd64
Architecture: x86_64
Machine: amd64
>Description:

When using a -current amd64 machine to build releases for arm (eabi) I get
random crashes:

/evbarm/ssd/src/compat/arm/oabi/external/bsd/elftoolchain/lib/libelf/arm--netbsdelf-e.core
./evbarm/ssd/src/compat/arm/oabi/external/bsd/file/lib/arm--netbsdelf-e.core

file says about them:

./evbarm/ssd/src/compat/arm/oabi/external/bsd/elftoolchain/lib/libelf/arm--netbsdelf-e.core: ELF 64-bit LSB core file x86-64, version 1 (SYSV)
./evbarm/ssd/src/compat/arm/oabi/external/bsd/file/lib/arm--netbsdelf-e.core: ELF 64-bit LSB core file x86-64, version 1 (SYSV)

but gdb is unhappy:

Reading symbols from /ssd/tools/bin/arm--netbsdelf-eabi-elfedit...(no debugging symbols found)...done.
"/ssd/obj/./evbarm/ssd/src/compat/arm/oabi/external/bsd/elftoolchain/lib/libelf/arm--netbsdelf-e.core" is not a core dump: File truncated


The volume this happens on has plenty of space left.

Repeating the build (with -u) eventually gets me through.

>How-To-Repeat:
see above. Maybe aslr related?

>Fix:
n/a

>Release-Note:

>Audit-Trail:
From: Thomas Klausner <wiz@NetBSD.org>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: toolchain/51089: random crashes building arm on amd64 -current
Date: Sun, 24 Apr 2016 09:09:48 +0200

 On Sun, Apr 24, 2016 at 07:05:00AM +0000, martin@NetBSD.org wrote:
 > >Number:         51089
 > >Category:       toolchain
 > >Synopsis:       random crashes building arm on amd64 -current

 I see random crashes too building amd64 on amd64.
  Thomas

From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: toolchain/51089: random crashes building arm on amd64 -current
Date: Sun, 24 Apr 2016 09:17:33 +0200

 On Sun, Apr 24, 2016 at 07:10:01AM +0000, Thomas Klausner wrote:
 >  I see random crashes too building amd64 on amd64.

 Try after

   sysctl -w security.pax.aslr.global=0

 and see if it still happens.

 Are the core files usable in your case?

 Martin

From: Thomas Klausner <wiz@NetBSD.org>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: toolchain/51089: random crashes building arm on amd64 -current
Date: Sun, 24 Apr 2016 09:47:26 +0200

 On Sun, Apr 24, 2016 at 07:20:01AM +0000, Martin Husemann wrote:
 >  Try after
 >  
 >    sysctl -w security.pax.aslr.global=0
 >  
 >  and see if it still happens.

 Ok.

 >  Are the core files usable in your case?

 I had one build that produced a core dump but the path was cut, so I'm
 not sure which program it was, and I don't have the log any longer,
 and I can't find the core.

 The next one failed with
 --- subr_callback.d ---
 -I/archive/foreign/src/lib/librump/../../sys/rump/../lib/libkern/../../../common/lib/libc/arch/i386/string -I/archive/foreign/src/lib/librump/../../sys/rump/../lib/libkern/../../../common/include -Wp,-iremap,/archive/build/amd64.gcc.20160423: -Wp,-iremap,/archive/foreign/src:/usr/src -Wp,-iremap,/archive/foreign/xsrc:/usr/xsrc  -D_FORTIFY_SOURCE=2     /archive/foreign/src/lib/librump/../../sys/rump/../kern/subr_callback.c &&  mv subr_callback.d.tmp subr_callback.d
 ...
 --- subr_callback.d ---
 nbmkdep: compile failed.
 *** [subr_callback.d] Error code 1
 nbmake[8]: stopped in /archive/foreign/src/lib/librump
 ...

 I'll update and try again.
  Thomas

From: Alexander Nasonov <alnsn@yandex.ru>
To: gnats-bugs@NetBSD.org
Cc: toolchain-manager@netbsd.org, gnats-admin@netbsd.org,
	netbsd-bugs@netbsd.org
Subject: Re: toolchain/51089: random crashes building arm on amd64 -current
Date: Sun, 24 Apr 2016 09:36:55 +0100

 martin@NetBSD.org wrote:
 > >How-To-Repeat:
 > see above. Maybe aslr related?

 I turned on aslr a while ago and I built releases several times without
 any problem. I needed to take two additional steps to prevent aslr
 related crashes:

  1. Set ulimit -s `ulimit -s`. See bin/49821.
  2. if pax.mprotect is globally on, it must be turned off using
     paxctl for cc1 and cc1plus.

 Alex

From: Thomas Klausner <tk@giga.or.at>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: toolchain/51089: random crashes building arm on amd64 -current
Date: Sun, 24 Apr 2016 12:02:46 +0200

 On Sun, Apr 24, 2016 at 07:20:01AM +0000, Martin Husemann wrote:
 >  Try after
 >  
 >    sysctl -w security.pax.aslr.global=0
 >  
 >  and see if it still happens.

 My latest build was successful with this setting.
  Thomas

State-Changed-From-To: open->feedback
State-Changed-By: mrg@NetBSD.org
State-Changed-When: Thu, 20 Dec 2018 22:22:50 +0000
State-Changed-Why:
is this now fixed?  i haven't seen any random crashes for ages.


State-Changed-From-To: feedback->closed
State-Changed-By: martin@NetBSD.org
State-Changed-When: Fri, 21 Dec 2018 07:55:48 +0000
State-Changed-Why:
Right, haven't seen this in ages either


>Unformatted:

Home	PR Database Search