NetBSD Problem Report #47322

From www@NetBSD.org  Thu Dec 13 16:30:55 2012
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [149.20.53.66])
	by www.NetBSD.org (Postfix) with ESMTP id A664563E33E
	for <gnats-bugs@gnats.NetBSD.org>; Thu, 13 Dec 2012 16:30:55 +0000 (UTC)
Message-Id: <20121213163054.63E2363E33E@www.NetBSD.org>
Date: Thu, 13 Dec 2012 16:30:54 +0000 (UTC)
From: richard.palo@baou.fr
Reply-To: richard.palo@baou.fr
To: gnats-bugs@NetBSD.org
Subject: pkgsrc/print/cups update for LIBS_SSP to include -libssp for solaris
X-Send-Pr-Version: www-1.0

>Number:         47322
>Category:       pkg
>Synopsis:       pkgsrc/print/cups update for LIBS_SSP to include -libssp for solaris
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    sbd
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Dec 13 16:35:00 +0000 2012
>Closed-Date:    Sat Nov 02 19:00:30 +0000 2013
>Last-Modified:  Sat Nov 02 19:00:30 +0000 2013
>Originator:     Richard PALO
>Release:        pkgsrc trunk
>Organization:
>Environment:
SunOS devzone 5.11 oi_151a7 i86pc i386 i86pc Solaris

>Description:
although cups will "build" and "install", other programs linking to any cups shared library using libssp_nonshared.a will fail with 'error symbol __stack_chk_guard not found' unless cups links at the same time with libssp.so.
>How-To-Repeat:
an example program that fails is gutenprint-lib:

richard@devzone:~/src/pkgsrc/print/gutenprint-lib/work.devzone/gutenprint-5.2.9$ ldd  -d /opt/pkg/lib/libcupsimage.so
        libtiff.so.5 =>  /opt/pkg/lib/libtiff.so.5
        liblzma.so.5 =>  /opt/pkg/lib/liblzma.so.5
        librt.so.1 =>    /lib/64/librt.so.1
        libpng15.so.15 =>        /opt/pkg/lib/libpng15.so.15
        libjpeg.so.8 =>  /opt/pkg/lib/libjpeg.so.8
        libcups.so.2 =>  /opt/pkg/lib/libcups.so.2
        libpthread.so.1 =>       /lib/64/libpthread.so.1
        libm.so.2 =>     /lib/64/libm.so.2
        libiconv.so.2 =>         /opt/pkg/lib/libiconv.so.2
        libgssapi.so.3 =>        /opt/pkg/lib/libgssapi.so.3
        libheimntlm.so.0 =>      /opt/pkg/lib/libheimntlm.so.0
        libkrb5.so.26 =>         /opt/pkg/lib/libkrb5.so.26
        libsqlite3.so.0 =>       /opt/pkg/lib/libsqlite3.so.0
        libdl.so.1 =>    /lib/64/libdl.so.1
        libhx509.so.5 =>         /opt/pkg/lib/libhx509.so.5
        libasn1.so.8 =>  /opt/pkg/lib/libasn1.so.8
        libwind.so.0 =>  /opt/pkg/lib/libwind.so.0
        libcom_err.so.1 =>       /opt/pkg/lib/libcom_err.so.1
        libheimbase.so.1 =>      /opt/pkg/lib/libheimbase.so.1
        libroken.so.18 =>        /opt/pkg/lib/libroken.so.18
        libresolv.so.2 =>        /lib/64/libresolv.so.2
        libnsl.so.1 =>   /lib/64/libnsl.so.1
        libsocket.so.1 =>        /lib/64/libsocket.so.1
        libssl.so.0.9.8 =>       /opt/pkg/lib/libssl.so.0.9.8
        libcrypto.so.0.9.8 =>    /opt/pkg/lib/libcrypto.so.0.9.8
        libdns_sd.so =>  /opt/pkg/lib/libdns_sd.so
        libz.so.1 =>     /opt/pkg/lib/libz.so.1
        libgcc_s.so.1 =>         /opt/pkg/gcc47/lib/amd64/libgcc_s.so.1
        libc.so.1 =>     /lib/64/libc.so.1
        libmd.so.1 =>    /lib/64/libmd.so.1
        libmp.so.2 =>    /lib/64/libmp.so.2
        libgcc_s.so.1 =>         /opt/pbulk/gcc47/lib/amd64/libgcc_s.so.1
        symbol not found: __stack_chk_guard             (/opt/pkg/lib/libcupsimage.so)
        symbol not found: __stack_chk_guard             (/opt/pkg/lib/libcups.so.2)



>Fix:
From what I can gather, normally the gcc compiler should do 'the right thing' when -fstack-protect is passed, but thanks to a hint gleaned here (http://gcc.gnu.org/ml/gcc-patches/2005-07/msg00066.html)
in addition to -lssp_nonshared, -lssp needs to be there as well.

Apparently a compiler problem, and a moving target, because I noticed that the pkgsrc Makefile for cups had this entry for SunOS: CONFIGURE_ENV+= LIBS_SSP=-lssp_nonshared

after applying the following, (and deinstalling fully cups), I was able to rebuild cups and get over this problem in gutenprint:

richard@devzone:~/src/pkgsrc/print/cups$ cvs diff
Index: Makefile
===================================================================
RCS file: /cvsroot/pkgsrc/print/cups/Makefile,v
retrieving revision 1.193
diff -u -p -r1.193 Makefile
--- Makefile    8 Oct 2012 13:45:36 -0000       1.193
+++ Makefile    13 Dec 2012 16:29:15 -0000
@@ -70,7 +70,8 @@ CONFIGURE_ENV+=               DSOFLAGS=${LDFLAGS:M*:Q
 .include "options.mk"

 .if ${OPSYS} == "SunOS"
-CONFIGURE_ENV+=                LIBS_SSP=-lssp_nonshared
+LIBS_SSP =             -lssp_nonshared -lssp
+CONFIGURE_ENV+=                LIBS_SSP=${LIBS_SSP:M*:Q}
 LIBS.SunOS+=           -liconv
 .endif

>Release-Note:

>Audit-Trail:

Responsible-Changed-From-To: pkg-manager->sbd
Responsible-Changed-By: hauke@NetBSD.org
Responsible-Changed-When: Sun, 16 Dec 2012 19:47:22 +0000
Responsible-Changed-Why:
Over to maintainer.


From: Richard PALO <richard.palo@baou.fr>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: pkg/47322 (pkgsrc/print/cups update for LIBS_SSP to include -libssp
 for solaris)
Date: Tue, 29 Jan 2013 11:10:16 +0100

 Le 16/12/12 20:47, hauke@NetBSD.org a écrit :
 > Synopsis: pkgsrc/print/cups update for LIBS_SSP to include -libssp for solaris
 >
 > Responsible-Changed-From-To: pkg-manager->sbd
 > Responsible-Changed-By: hauke@NetBSD.org
 > Responsible-Changed-When: Sun, 16 Dec 2012 19:47:22 +0000
 > Responsible-Changed-Why:
 > Over to maintainer.
 >
 >
 >
 >
 Any movement on this and the libtool patch (pkg/47359) for solaris?
 If there is a better suggestion, I'd be happy to try it.

From: Richard PALO <richard.palo@free.fr>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: pkg/47322 (pkgsrc/print/cups update for LIBS_SSP to include -libssp
 for solaris)
Date: Tue, 08 Oct 2013 17:04:31 +0200

 Le 16/12/12 20:47, hauke@NetBSD.org a écrit :
 > Synopsis: pkgsrc/print/cups update for LIBS_SSP to include -libssp for solaris
 >
 > Responsible-Changed-From-To: pkg-manager->sbd
 > Responsible-Changed-By: hauke@NetBSD.org
 > Responsible-Changed-When: Sun, 16 Dec 2012 19:47:22 +0000
 > Responsible-Changed-Why:
 > Over to maintainer.

 No response for quite awhile, so I'll reiterate with an extract of the 
 link provided:
 > But, if GCC decides to use TLS canary for -fstack-protector,
 > then it ought to avoid -lssp_nonshared -lssp, while if
 > it decides not to use them, then it should force -lssp_nonshared -lssp
 > (as glibc doesn't provide __stack_chk_guard on the architectures
 > that provide TLS canary).

 In absence of any pertinent objections, I'll push this with the builtin 
 mit-krb5 gssapi fixes affecting cups this weekend or so.

 BTW - I don't believe suppressing the 'LDFLAGS+= -lssp_nonshared' in 
 favour of 'CFLAGS+= -fno-stack-protector' is necessarily a great 
 alternative... That is, unless there are some great arguments for this 
 approach.

From: David Holland <dholland-pbugs@netbsd.org>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: pkg/47322 (pkgsrc/print/cups update for LIBS_SSP to include
 -libssp for solaris)
Date: Sat, 2 Nov 2013 18:36:14 +0000

 not sent to gnats

    --------

 From: Richard PALO <richard.palo@free.fr>
 To: sbd@NetBSD.org, pkgsrc-bugs@netbsd.org, richard.palo@baou.fr
 Subject: Re: pkg/47322 (pkgsrc/print/cups update for LIBS_SSP to include
 	-libssp for solaris)
 Date: Mon, 14 Oct 2013 14:10:41 +0200

 Le 08/10/13 17:05, Richard PALO a ?crit :
 > The following reply was made to PR pkg/47322; it has been noted by GNATS.
 > 
 >   No response for quite awhile, so I'll reiterate with an extract of the
 >   link provided:
 >   > But, if GCC decides to use TLS canary for -fstack-protector,
 >   > then it ought to avoid -lssp_nonshared -lssp, while if
 >   > it decides not to use them, then it should force -lssp_nonshared -lssp
 >   > (as glibc doesn't provide __stack_chk_guard on the architectures
 >   > that provide TLS canary).
 > 
 >   In absence of any pertinent objections, I'll push this with the builtin
 >   mit-krb5 gssapi fixes affecting cups this weekend or so.
 > 
 >   BTW - I don't believe suppressing the 'LDFLAGS+= -lssp_nonshared' in
 >   favour of 'CFLAGS+= -fno-stack-protector' is necessarily a great
 >   alternative... That is, unless there are some great arguments for this
 >   approach.
 > 
 > 
 As this was pushed this morning, the problem report can probably be closed.

State-Changed-From-To: open->closed
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Sat, 02 Nov 2013 19:00:30 +0000
State-Changed-Why:
submitter says it can be closed


>Unformatted:

Home	PR Database Search