From mboxrd@z Thu Jan  1 00:00:00 1970
From: Alexey Brodkin <Alexey.Brodkin@synopsys.com>
Date: Fri, 17 Jul 2015 15:31:57 +0000
Subject: [Buildroot] [arc-buildroot] [autobuild.buildroot.net] arc build
 results for 2015-07-14
In-Reply-To: <1437142401.1824.6.camel@embedded.rocks>
References: <20150715063016.118A21016B5@stock.ovh.net>
 <1436982736.3682.16.camel@synopsys.com>
 <1437142401.1824.6.camel@embedded.rocks>
Message-ID: <1437147117.3374.4.camel@synopsys.com>
List-Id: <buildroot.busybox.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: buildroot@busybox.net

Hi J?rg,

On Fri, 2015-07-17 at 16:13 +0200, J?rg Krause wrote:
> Dear Alexey, Thomas,
> 
> On Mi, 2015-07-15 at 17:52 +0000, Alexey Brodkin wrote:
> > Hi Thomas,
> > 
> > On Wed, 2015-07-15 at 08:30 +0200, Thomas Petazzoni wrote:
> > > Those results are limited to the arc architecture.
> > > 
> > > Build statistics for 2015-07-14
> > > ===============================
> > > 
> > >         success : 6  
> > >        failures : 23 
> > >        timeouts : 0  
> > >           TOTAL : 29 
> > > 
> > > Classification of failures by reason
> > > ====================================
> > > 
> > >                   zeromq-4.0.5 | 9 
> > >              util-linux-2.26.2 | 5 
> > >                   boost-1.58.0 | 3 
> > >                alsa-lib-1.0.29 | 3 
> > >              btrfs-progs-4.1.1 | 1 
> > >              libselinux-2.1.13 | 1 
> > >                    tinc-1.0.24 | 1 
> > > 
> > > Detail of failures
> > > ===================
> > > 
> > >          arc |                   boost-1.58.0 | NOK | 
> > > http://autobuild.buildroot.net/results/596c2eb1378946b8c0fea6b3ebcf
> > > 80ee7e14995a/
> > >          arc |                   boost-1.58.0 | NOK | 
> > > http://autobuild.buildroot.net/results/0585f6c926042d7c859150dbf7b3
> > > 3928c4687273/
> > >          arc |                   boost-1.58.0 | NOK | 
> > > http://autobuild.buildroot.net/results/16fd658176b4e4927f7c7dbe5598
> > > b7a38b819af9/
> > 
> > Looks like an issue with boost that for some reason
> > uses syscall (non-existent on some arches including ARC)
> > directly instead of libc's eventfd().
> > ----------------->8-------------------
> > "syscall(__NR_eventfd, 0);"
> > ----------------->8-------------------
> 
> __NR_eventfd needs __ARCH_WANT_SYSCALL_NO_FLAGS to be defined in
> asm/unistd.h. I am not a toolchain expert so I do not know how to
> handle this.

Mentioned syscall and define go from kernel headers.
For old platforms like x86, ARM etc that system call was implemented.

Even though it is now obsolete for compatibility reasons it was left
as it is on those platforms.

Newer platforms like ARC and probably others strictly follow so-called
UAPI and so use "include/uapi/asm-generic/unistd.h" and so don't support 
__NR_eventfd syscall relying instead on newer implementation __NR_eventfd2.

IMHO it's an issue in boost that somebody (probably me) has to fix.

-Alexey