[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Thomas Petazzoni]

Build statistics for 2014-11-22
===============================

        success : 181
       failures : 11 
       timeouts : 2  
          TOTAL : 194

Classification of failures by reason
====================================

                       unknown | 2 
               qt5webkit-5.3.2 | 2 
                   erlang-17.3 | 1 
                 webkit-1.11.5 | 1 
                   nginx-1.6.2 | 1 
                     radvd-2.8 | 1 
              uemacs-4.0.15-lt | 1 
                  thrift-0.9.1 | 1 
                    ruby-2.1.5 | 1 
                  cppcms-1.0.4 | 1 
                      qt-4.8.6 | 1 

Detail of failures
===================

         arm |                   cppcms-1.0.4 | NOK | http://autobuild.buildroot.net/results/22512cfee28885a0733159a49696d8cb39f67ccc/
         arm |                    erlang-17.3 | NOK | http://autobuild.buildroot.net/results/1ceaa6925d1ded236d01ae9feb31f5f915f6b51a/
         arc |                    nginx-1.6.2 | NOK | http://autobuild.buildroot.net/results/730105fc0a89b381b3b29192d07f28ef1f13cbb3/
    mips64el |                       qt-4.8.6 | NOK | http://autobuild.buildroot.net/results/7c11b91939030c18108ef4528453406faa47cbce/
         arm |                qt5webkit-5.3.2 | NOK | http://autobuild.buildroot.net/results/22b6c928e8d795f950888cb3e1cdd4ad97d98e32/
         arm |                qt5webkit-5.3.2 | NOK | http://autobuild.buildroot.net/results/5aff88b8c3fcbb21be5c96a0b0c358b1d71c3e96/
         arc |                      radvd-2.8 | NOK | http://autobuild.buildroot.net/results/bba1ebd9bbab02adc226242c65ae46cef7240563/
     powerpc |                     ruby-2.1.5 | NOK | http://autobuild.buildroot.net/results/53fea641f66cfbb1fb5992cdbb9a03567ff7f442/
        bfin |                   thrift-0.9.1 | NOK | http://autobuild.buildroot.net/results/8ac073a4e9c3a228dd22d80ea02b9dc3fc739164/
        bfin |               uemacs-4.0.15-lt | NOK | http://autobuild.buildroot.net/results/df9b693d7f53daba0d25f52132ca1594dd1273fa/
      mipsel |                        unknown | TIM | http://autobuild.buildroot.net/results/c26e0934bb45d99871a963d74f6c65ba44d7ffb7/
         arm |                        unknown | TIM | http://autobuild.buildroot.net/results/724e4197595ce1458c3a6644cb1b136c6e7e1ab5/
         arm |                  webkit-1.11.5 | NOK | http://autobuild.buildroot.net/results/e259cefcb0a2258cd55cfa8250d8a2bd4011ca5b/


-- 
http://autobuild.buildroot.net

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Samuel Martin]

On Sun, Nov 23, 2014 at 8:30 AM, Thomas Petazzoni
<thomas.petazzoni@free-electrons.com> wrote:
> Build statistics for 2014-11-22
> ===============================
>
>         success : 181
>        failures : 11
>        timeouts : 2
>           TOTAL : 194
>
> Classification of failures by reason
> ====================================
>
>                        unknown | 2
>                qt5webkit-5.3.2 | 2
>                    erlang-17.3 | 1
>                  webkit-1.11.5 | 1
>                    nginx-1.6.2 | 1
>                      radvd-2.8 | 1
>               uemacs-4.0.15-lt | 1
>                   thrift-0.9.1 | 1
>                     ruby-2.1.5 | 1
>                   cppcms-1.0.4 | 1
>                       qt-4.8.6 | 1
>
> Detail of failures
> ===================

[...]

>          arc |                    nginx-1.6.2 | NOK | http://autobuild.buildroot.net/results/730105fc0a89b381b3b29192d07f28ef1f13cbb3/

Nginx failure root cause seems to be a kernel with no sysctl syscall.
This seems to happen only on arc.

Anton, Alexey, is it intended that CONFIG_SYSCTL_SYSCALL is disabled
in the arc kernel config?

Anyway, I posted a fix for this:
http://patchwork.ozlabs.org/patch/413440/

Regards,

-- 
Samuel

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Thomas Petazzoni]

Dear Samuel Martin,

On Sun, 23 Nov 2014 18:24:02 +0100, Samuel Martin wrote:

> Nginx failure root cause seems to be a kernel with no sysctl syscall.
> This seems to happen only on arc.
> 
> Anton, Alexey, is it intended that CONFIG_SYSCTL_SYSCALL is disabled
> in the arc kernel config?

Yes, see
http://lists.busybox.net/pipermail/buildroot/2014-November/112629.html.

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Samuel Martin]

On Sun, Nov 23, 2014 at 6:26 PM, Thomas Petazzoni
<thomas.petazzoni@free-electrons.com> wrote:
> Dear Samuel Martin,
>
> On Sun, 23 Nov 2014 18:24:02 +0100, Samuel Martin wrote:
>
>> Nginx failure root cause seems to be a kernel with no sysctl syscall.
>> This seems to happen only on arc.
>>
>> Anton, Alexey, is it intended that CONFIG_SYSCTL_SYSCALL is disabled
>> in the arc kernel config?
>
> Yes, see
> http://lists.busybox.net/pipermail/buildroot/2014-November/112629.html.

Oh! I missed it! :-S
Thanks.

>
> Thomas
> --
> Thomas Petazzoni, CTO, Free Electrons
> Embedded Linux, Kernel and Android engineering
> http://free-electrons.com

-- 
Samuel

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Alexey Brodkin]

Hi Samuel,

On Sun, 2014-11-23 at 18:29 +0100, Samuel Martin wrote:
> On Sun, Nov 23, 2014 at 6:26 PM, Thomas Petazzoni
> <thomas.petazzoni@free-electrons.com> wrote:
> > Dear Samuel Martin,
> >
> > On Sun, 23 Nov 2014 18:24:02 +0100, Samuel Martin wrote:
> >
> >> Nginx failure root cause seems to be a kernel with no sysctl syscall.
> >> This seems to happen only on arc.
> >>
> >> Anton, Alexey, is it intended that CONFIG_SYSCTL_SYSCALL is disabled
> >> in the arc kernel config?
> >
> > Yes, see
> > http://lists.busybox.net/pipermail/buildroot/2014-November/112629.html.
> 
> Oh! I missed it! :-S
> Thanks.

Actually I'm wondering why "BR2_PACKAGE_NGINX_RTSIG_MODULE=y" is
implemented in "package/nginx/Config.in"?

As I may see from http://nginx.org/en/CHANGES sysctl was substituted
with procfs in nginx:
--->---
Changes with nginx 0.3.58                                 14 Aug 2006
...
    *) Change: now the procfs interface instead of sysctl is used on
Linux.
--->---

Also from here
http://hg.nginx.org/nginx/file/2c10db908b8c/auto/os/linux#l29 I may
conclude that rt signals were used for Linux kernels between 2.2.19 and
2.6.17.

In other words it looks like rt signals should not be used in modern
nginx running on top of modern Linux kernel.

Probably there's a real need for rt signals to be used still - I'm not
an expert in nginx of any kind.

If there's no such requirement/need for rt signals I would propose to
remove BR2_PACKAGE_NGINX_RTSIG_MODULE completely so users don't select
deprecated options.

-Alexey

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Samuel Martin]

On Wed, Nov 26, 2014 at 9:18 PM, Alexey Brodkin
<Alexey.Brodkin@synopsys.com> wrote:
> Hi Samuel,
>
> On Sun, 2014-11-23 at 18:29 +0100, Samuel Martin wrote:
>> On Sun, Nov 23, 2014 at 6:26 PM, Thomas Petazzoni
>> <thomas.petazzoni@free-electrons.com> wrote:
>> > Dear Samuel Martin,
>> >
>> > On Sun, 23 Nov 2014 18:24:02 +0100, Samuel Martin wrote:
>> >
>> >> Nginx failure root cause seems to be a kernel with no sysctl syscall.
>> >> This seems to happen only on arc.
>> >>
>> >> Anton, Alexey, is it intended that CONFIG_SYSCTL_SYSCALL is disabled
>> >> in the arc kernel config?
>> >
>> > Yes, see
>> > http://lists.busybox.net/pipermail/buildroot/2014-November/112629.html.
>>
>> Oh! I missed it! :-S
>> Thanks.
>
> Actually I'm wondering why "BR2_PACKAGE_NGINX_RTSIG_MODULE=y" is
> implemented in "package/nginx/Config.in"?
>
> As I may see from http://nginx.org/en/CHANGES sysctl was substituted
> with procfs in nginx:
> --->---
> Changes with nginx 0.3.58                                 14 Aug 2006
> ...
>     *) Change: now the procfs interface instead of sysctl is used on
> Linux.
> --->---
>
> Also from here
> http://hg.nginx.org/nginx/file/2c10db908b8c/auto/os/linux#l29 I may
> conclude that rt signals were used for Linux kernels between 2.2.19 and
> 2.6.17.
>
> In other words it looks like rt signals should not be used in modern
> nginx running on top of modern Linux kernel.
>
> Probably there's a real need for rt signals to be used still - I'm not
> an expert in nginx of any kind.
>
> If there's no such requirement/need for rt signals I would propose to
> remove BR2_PACKAGE_NGINX_RTSIG_MODULE completely so users don't select
> deprecated options.

Well, after ringing nginx people, I got this answer [1] (wondering
about deprecating rtsig module).

So, I'll cook a patch disabling rtsig module, at least for arc architecture.


[1] http://mailman.nginx.org/pipermail/nginx-devel/2014-November/006295.html

Regards,


-- 
Samuel

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Alexey Brodkin]

Hi Samuel,

On Wed, 2014-11-26 at 21:48 +0100, Samuel Martin wrote:

> Well, after ringing nginx people, I got this answer [1] (wondering
> about deprecating rtsig module).
> 
> So, I'll cook a patch disabling rtsig module, at least for arc architecture.

Well I'd prefer to have less arch-specific stuff in Buildroot so it's
easier to maintain then.

And probably better solution is to implement proper check in nginx
itself instead of relying on user's preference or educated guess.

For example in "radvd" I'm going to do it with simple inclusion of:
--->---
AC_CHECK_FUNCS(sysctl)
--->---
in "configure.ac".

-Alexey

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Thomas Petazzoni]

Dear Alexey Brodkin,

On Wed, 26 Nov 2014 20:57:34 +0000, Alexey Brodkin wrote:

> For example in "radvd" I'm going to do it with simple inclusion of:
> --->---
> AC_CHECK_FUNCS(sysctl)
> --->---
> in "configure.ac".

Did you notice that this is already done?

See
http://git.buildroot.net/buildroot/commit/?id=cb54da1b03f2e6d684416c614021735dfe276606.

Best regards,

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Alexey Brodkin]

Hi Thomas,

On Wed, 2014-11-26 at 22:00 +0100, Thomas Petazzoni wrote:
> Dear Alexey Brodkin,
> 
> On Wed, 26 Nov 2014 20:57:34 +0000, Alexey Brodkin wrote:
> 
> > For example in "radvd" I'm going to do it with simple inclusion of:
> > --->---
> > AC_CHECK_FUNCS(sysctl)
> > --->---
> > in "configure.ac".
> 
> Did you notice that this is already done?
> 
> See
> http://git.buildroot.net/buildroot/commit/?id=cb54da1b03f2e6d684416c614021735dfe276606.

No unfortunately I didn't noticed this change.
I don't follow mailing list regularly so it would be helpful if you add
me explicitly in Cc if I might be interested in stuff :)

And frankly I'm not sure if we need to check for a header still.
So in your patch I would simply replace HAVE_SYS_SYSCTL_H with
HAVE_SYSCTL in "device-linux.c".

So I would implement it like this:
--->---
diff --git a/configure.ac b/configure.ac
index 0556805..349e4c1 100644
--- a/configure.ac
+++ b/configure.ac
@@ -160,7 +160,6 @@ AC_CHECK_HEADERS( \
 	net/if_types.h \
 	sys/param.h \
 	sys/sockio.h \
-	sys/sysctl.h \
 	sys/time.h \
 	time.h \
 )
@@ -183,6 +182,7 @@ AC_MSG_RESULT(no))
 dnl Checks for library functions.
 AC_CHECK_FUNCS(getopt_long)
 AC_CHECK_FUNCS(ppoll)
+AC_CHECK_FUNCS(sysctl)
 
 CONDITIONAL_SOURCES="device-${arch}.${OBJEXT} ${CONDITIONAL_SOURCES}"
 if test x${arch} = xlinux ; then
diff --git a/device-linux.c b/device-linux.c
index 1ecaa6b..fa7d978 100644
--- a/device-linux.c
+++ b/device-linux.c
@@ -183,7 +183,7 @@ int check_ip6_forwarding(void)
 		value = -1;
 	}
 
-#ifdef HAVE_SYS_SYSCTL_H
+#ifdef HAVE_SYSCTL
 	int forw_sysctl[] = { SYSCTL_IP6_FORWARDING };
 	size_t size = sizeof(value);
 	if (!fp && sysctl(forw_sysctl, sizeof(forw_sysctl) /
sizeof(forw_sysctl[0]), &value, &size, NULL, 0) < 0) {
--
--->---

What do you think?

Also I was going to push my change upstream in radvd and already made a
pull-request https://github.com/reubenhwk/radvd/pull/37 but with your
help I realized that I forgot to modify "device-linux.c".

-Alexey

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Thomas Petazzoni]

Dear Alexey Brodkin,

On Wed, 26 Nov 2014 21:19:07 +0000, Alexey Brodkin wrote:

> No unfortunately I didn't noticed this change.
> I don't follow mailing list regularly so it would be helpful if you add
> me explicitly in Cc if I might be interested in stuff :)

I think I had the intention of doing it, but forgot it when actually
sending the patches. Sorry about that.

> And frankly I'm not sure if we need to check for a header still.
> So in your patch I would simply replace HAVE_SYS_SYSCTL_H with
> HAVE_SYSCTL in "device-linux.c".

Yes, I agree this is arguable.

> Also I was going to push my change upstream in radvd and already made a
> pull-request https://github.com/reubenhwk/radvd/pull/37 but with your
> help I realized that I forgot to modify "device-linux.c".

Hehe :-)

In any case, if you submit a fix upstream, next time we upgrade, we'll
get your version.

Best regards,

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Alexey Brodkin]

Hi Thomas,

On Wed, 2014-11-26 at 22:30 +0100, Thomas Petazzoni wrote:
> Dear Alexey Brodkin,
> 
> On Wed, 26 Nov 2014 21:19:07 +0000, Alexey Brodkin wrote:
> 
> > No unfortunately I didn't noticed this change.
> > I don't follow mailing list regularly so it would be helpful if you add
> > me explicitly in Cc if I might be interested in stuff :)
> 
> I think I had the intention of doing it, but forgot it when actually
> sending the patches. Sorry about that.

Just to let you know my patch was applied upstream -
https://github.com/reubenhwk/radvd/commit/2b09fdf64a5f4eb1db940c21fa65c016176a1f79

So on the next version bump of radvd your patch should be removed.

-Alexey

[Buildroot] [autobuild.buildroot.net] Build results for 2014-11-22

[Thomas Petazzoni]

Dear Alexey Brodkin,

On Thu, 27 Nov 2014 09:08:08 +0000, Alexey Brodkin wrote:

> > I think I had the intention of doing it, but forgot it when actually
> > sending the patches. Sorry about that.
> 
> Just to let you know my patch was applied upstream -
> https://github.com/reubenhwk/radvd/commit/2b09fdf64a5f4eb1db940c21fa65c016176a1f79
> 
> So on the next version bump of radvd your patch should be removed.

Excellent, thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com