From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from hetzner.pbcl.net (mail.pbcl.net [88.198.119.4]) by mail.openembedded.org (Postfix) with ESMTP id 58D8971AC8 for ; Wed, 11 Jan 2017 15:57:04 +0000 (UTC) Received: from blundell.swaffham-prior.co.uk ([91.216.112.25] helo=tp13) by hetzner.pbcl.net with esmtpsa (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.87) (envelope-from ) id 1cRLGj-0005AX-25; Wed, 11 Jan 2017 16:57:02 +0100 Message-ID: <1484150216.3966.19.camel@pbcl.net> From: Phil Blundell To: "Burton, Ross" , Khem Raj Date: Wed, 11 Jan 2017 15:56:56 +0000 In-Reply-To: References: <20170111035102.18447-1-raj.khem@gmail.com> <20170111035102.18447-2-raj.khem@gmail.com> X-Mailer: Evolution 3.20.5-1 Mime-Version: 1.0 Cc: OE-core Subject: Re: [PATCH 2/2] glibc: Upgrade to 2.25 X-BeenThere: openembedded-core@lists.openembedded.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Patches and discussions about the oe-core layer List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 11 Jan 2017 15:57:05 -0000 Content-Type: multipart/alternative; boundary="=-w8ldDBn0iVH0ZIBGTws7" --=-w8ldDBn0iVH0ZIBGTws7 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 8bit On Wed, 2017-01-11 at 12:19 +0000, Burton, Ross wrote: > > > On 11 January 2017 at 03:51, Khem Raj wrote: > > > > Signed-off-by: Khem Raj > > > This fails when building gdk-pixbuf for x86-64 with gobject- introspection enabled: > > > > > > > > > > > > > | /data/poky-master/tmp-glibc/work/corei7-64-poky-linux/gdk- pixbuf/2.36.1-r0/build/gdk-pixbuf/tmp- introspectylz98kc7/.libs/GdkPixbuf-2.0: Relink `/data/poky- master/tmp-glibc/sysroots/intel-corei7-64//usr/lib/libpng16.so.16' with `/data/poky-master/tmp-glibc/sysroots/intel-corei7- 64//lib/libpthread.so.0' for IFUNC symbol `longjmp' > > > The gdk-pixbuf link does use -lpthread, is this saying that libpng should be linked against pthread too?  I can replicate on demand if you have any suggestions. > > That's what it's saying, and this was introduced by the fix for: https://sourceware.org/bugzilla/show_bug.cgi?id=20019 But I think this new behaviour is in itself a bug.  It isn't reasonable to expect any library that calls longjmp() to link -lpthread just in case it might end up coexisting with another library that does.  (The fact that libpthread defines longjmp is arguably a bug in itself, but one that we are stuck with.) p. --=-w8ldDBn0iVH0ZIBGTws7 Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: quoted-printable
On Wed, 2017-01-11 at 12:19 +0000, Burton, Ro= ss wrote:

On 11 January 2017 at 03:51, Khem = Raj <raj.khem@gmail.com> wrote:
Signed-off-by: Khem Raj <raj= .khem@gmail.com>


This fails when= building gdk-pixbuf for x86-64 with gobject-introspection enabled:

| /data/po= ky-master/tmp-glibc/work/corei7-64-poky-linux/gdk-pixbuf/2.36.1-r0/build/gd= k-pixbuf/tmp-introspectylz98kc7/.libs/GdkPixbuf-2.0: Relink `/data/poky-mas= ter/tmp-glibc/sysroots/intel-corei7-64//usr/lib/libpng16.so.16' with `/data= /poky-master/tmp-glibc/sysroots/intel-corei7-64//lib/libpthread.so.0' for I= FUNC symbol `longjmp'

=
The gdk-pixbuf link does use -lpthread, is this saying that libpng sho= uld be linked against pthread too?  I can replicate on demand if you h= ave any suggestions.

That's what it's saying, and this was intr= oduced by the fix for:


But I think this new beh= aviour is in itself a bug.  It isn't reasonable to expect any library = that calls longjmp() to link -lpthread just in case it might end up coexist= ing with another library that does.  (The fact that libpthread defines= longjmp is arguably a bug in itself, but one that we are stuck with.)

p.

--=-w8ldDBn0iVH0ZIBGTws7--