[Buildroot] [autobuild.buildroot.net] Build results for 2014-04-20

[Buildroot] [autobuild.buildroot.net] Build results for 2014-04-20

[Thomas Petazzoni]

Build statistics for 2014-04-20
===============================

        success : 60 
       failures : 24 
       timeouts : 1  
          TOTAL : 85 

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

           host-nodejs-0.10.12 | 5 
libstrophe-d408eaf2bbfe5ff5... | 3 
                  zeromq-4.0.4 | 2 
jack2-37976441044d69b91d61d... | 2 
                 cairo-1.12.10 | 1 
                    php-5.5.11 | 1 
               openswan-2.6.41 | 1 
                  cppcms-1.0.4 | 1 
               qt5webkit-5.2.1 | 1 
                 mplayer-1.1.1 | 1 
                   duma-2.5.15 | 1 
             lttng-tools-2.4.1 | 1 
                 qt5base-5.2.1 | 1 
             util-linux-2.24.1 | 1 
  Compiling ../librpc/ndr/ndr_ | 1 
                  python-2.7.6 | 1 
               vo-aacenc-0.1.3 | 1 

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

       arm |                  cairo-1.12.10 | NOK | http://autobuild.buildroot.net/results/2cb13a5bb92dabed219d49f49f0b9a2dfe65a0ca/
  mips64el |   Compiling ../librpc/ndr/ndr_ | TIM | http://autobuild.buildroot.net/results/2ca7cc97b9980a2066388aa5c6df41f67da1a60f/
      bfin |                   cppcms-1.0.4 | NOK | http://autobuild.buildroot.net/results/5ee69eb7492fbd462124d7a23b5b79003fc1dd64/
      bfin |                    duma-2.5.15 | NOK | http://autobuild.buildroot.net/results/c605022f2ff7168d28cc07e4176a3a3067aa7e36/
       arm |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/920f6a913673a08ffd45712b41daad59644b99f1/
       arm |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/8da7e6c8be07d6fcae3f18cdaf777ba16b636add/
       arm |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/480b55155facf61f69b49e124865c20d000270b4/
    x86_64 |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/3fcbf35ed720fa478adc4f6d6e8475ba8b68ca78/
       arm |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/c8d83c695bd97b7ce7f5907ca7b1d1d514f6cc07/
   powerpc | jack2-37976441044d69b91d61d... | NOK | http://autobuild.buildroot.net/results/4216666973fdb8038732c78a95b47c3faf1f6546/
microblazeel | jack2-37976441044d69b91d61d... | NOK | http://autobuild.buildroot.net/results/ede17f2fdb9b5cd4b974ef820d95b0eba863c420/
microblazeel | libstrophe-d408eaf2bbfe5ff5... | NOK | http://autobuild.buildroot.net/results/5bce2ff6b8cf6ec082b9ec0e5b7e4a80d7395225/
    xtensa | libstrophe-d408eaf2bbfe5ff5... | NOK | http://autobuild.buildroot.net/results/499ffd50ec58b54343d6606c0f1bf7cd04f14e1c/
      sh4a | libstrophe-d408eaf2bbfe5ff5... | NOK | http://autobuild.buildroot.net/results/b91a199ccdfe36d7f6d884bcf1642416ddf61264/
       arm |              lttng-tools-2.4.1 | NOK | http://autobuild.buildroot.net/results/03e897160964108d30390200d299025f6dadb31c/
       arm |                  mplayer-1.1.1 | NOK | http://autobuild.buildroot.net/results/257a10e9cb5022bb09e0c6a03844be5b5b3e0bd4/
       arc |                openswan-2.6.41 | NOK | http://autobuild.buildroot.net/results/71f92d80b1aa3ac98660fb88004efd0ea55e784e/
       sh4 |                     php-5.5.11 | NOK | http://autobuild.buildroot.net/results/f0592a6b04fd30b51796d69ca0b31cd8238a1dda/
     avr32 |                   python-2.7.6 | NOK | http://autobuild.buildroot.net/results/1aad2d677dbf7b0a3cb0120a35f45123124f51ab/
       arm |                  qt5base-5.2.1 | NOK | http://autobuild.buildroot.net/results/8eddd934bd80fdbcdf7a9dbf5d9f8b7ba69634d4/
    x86_64 |                qt5webkit-5.2.1 | NOK | http://autobuild.buildroot.net/results/527315bf9b506ec32a2fbd255526570c7486e5fb/
       arm |              util-linux-2.24.1 | NOK | http://autobuild.buildroot.net/results/27d22ff075e5e3e696d05ac6319043a5129abadc/
       arm |                vo-aacenc-0.1.3 | NOK | http://autobuild.buildroot.net/results/680b29cd824624eb8e4ec71187b9a6576444e72b/
      bfin |                   zeromq-4.0.4 | NOK | http://autobuild.buildroot.net/results/dc8dd7c1dd019dd06c4f806f2bd44d270eca4905/
      bfin |                   zeromq-4.0.4 | NOK | http://autobuild.buildroot.net/results/439df2ad4482d08f295dbc40b2e88bf9167a7794/


-- 
http://autobuild.buildroot.net

[Buildroot] Analysis of build results for 2014-04-20

[Thomas Petazzoni]

Hello,

As usual, analysis of the build results. Please help to fix them! Is
there some interest in me doing this on a regular basis? Or it's not
useful?

On Mon, 21 Apr 2014 08:30:08 +0200 (CEST), Thomas Petazzoni wrote:

>        arm |                  cairo-1.12.10 | NOK | http://autobuild.buildroot.net/results/2cb13a5bb92dabed219d49f49f0b9a2dfe65a0ca/

cairo-gl-private.h:287:5: error: unknown type name 'GLchar'

Could some OpenGL person look into this? Paul ? Yann ?

>   mips64el |   Compiling ../librpc/ndr/ndr_ | TIM | http://autobuild.buildroot.net/results/2ca7cc97b9980a2066388aa5c6df41f67da1a60f/

Another timeout to be investigated.

>       bfin |                   cppcms-1.0.4 | NOK | http://autobuild.buildroot.net/results/5ee69eb7492fbd462124d7a23b5b79003fc1dd64/

Fixed by
http://git.buildroot.net/buildroot/commit/package/cppcms?id=f9db7e6865e1805e8281136d5f2a6965f6a3b5c2.

>       bfin |                    duma-2.5.15 | NOK | http://autobuild.buildroot.net/results/c605022f2ff7168d28cc07e4176a3a3067aa7e36/

/home/test/test/1/output/host/usr/bfin-buildroot-uclinux-uclibc/sysroot/usr/lib/libc.a(strncpy.o): In function `*___GI_strncpy':
/usr/src/packages/BUILD/blackfin-toolchain-uclibc-full-2013R1/uClibc/libc/string/generic/strncpy.c:27: multiple definition of `_strncpy'
libduma.a(duma.o):duma.c:(.text+0x144): first defined here

>        arm |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/920f6a913673a08ffd45712b41daad59644b99f1/
>        arm |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/8da7e6c8be07d6fcae3f18cdaf777ba16b636add/
>        arm |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/480b55155facf61f69b49e124865c20d000270b4/
>     x86_64 |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/3fcbf35ed720fa478adc4f6d6e8475ba8b68ca78/
>        arm |            host-nodejs-0.10.12 | NOK | http://autobuild.buildroot.net/results/c8d83c695bd97b7ce7f5907ca7b1d1d514f6cc07/

All fixed by:

  http://git.buildroot.net/buildroot/commit/package/nodejs?id=fb80d283412b610fc0f5e4baf1c5bcdf093dadc2

Thanks Samuel!

>    powerpc | jack2-37976441044d69b91d61d... | NOK | http://autobuild.buildroot.net/results/4216666973fdb8038732c78a95b47c3faf1f6546/

Remains to be fixed.

> microblazeel | jack2-37976441044d69b91d61d... | NOK | http://autobuild.buildroot.net/results/ede17f2fdb9b5cd4b974ef820d95b0eba863c420/

Fixed by http://git.buildroot.net/buildroot/commit/package/jack2?id=5035af0cbb6a1dcb1e9b443fbf601be4fad737f9

> microblazeel | libstrophe-d408eaf2bbfe5ff5... | NOK | http://autobuild.buildroot.net/results/5bce2ff6b8cf6ec082b9ec0e5b7e4a80d7395225/
>     xtensa | libstrophe-d408eaf2bbfe5ff5... | NOK | http://autobuild.buildroot.net/results/499ffd50ec58b54343d6606c0f1bf7cd04f14e1c/
>       sh4a | libstrophe-d408eaf2bbfe5ff5... | NOK | http://autobuild.buildroot.net/results/b91a199ccdfe36d7f6d884bcf1642416ddf61264/

All fixed by
http://git.buildroot.net/buildroot/commit/package/libstrophe?id=c946a78996c25a65861672389a63ac59b3a5a4bc.

>        arm |              lttng-tools-2.4.1 | NOK | http://autobuild.buildroot.net/results/03e897160964108d30390200d299025f6dadb31c/

Compiler bug.

>        arm |                  mplayer-1.1.1 | NOK | http://autobuild.buildroot.net/results/257a10e9cb5022bb09e0c6a03844be5b5b3e0bd4/

Fixed by
http://git.buildroot.net/buildroot/commit/package/mplayer?id=e3de3ae58e830729bde1fcdc7da634567f4f3fa2.

>        arc |                openswan-2.6.41 | NOK | http://autobuild.buildroot.net/results/71f92d80b1aa3ac98660fb88004efd0ea55e784e/

/home/test/test/3/output/host/usr/lib/gcc/arc-buildroot-linux-uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/bin/ld: /home/test/test/3/output/host/usr/arc-buildroot-linux-uclibc/sysroot/usr/lib/crt1.o: relocation R_ARC_32_ME against `main' can not be used when making a shared object; recompile with -fPIC
/home/test/test/3/output/host/usr/arc-buildroot-linux-uclibc/sysroot/usr/lib/crt1.o: could not read symbols: Bad value
collect2: error: ld returned 1 exit status
make[4]: *** [pluto] Error 1

Could some ARC person look into this? Alexey, Anton?

>        sh4 |                     php-5.5.11 | NOK | http://autobuild.buildroot.net/results/f0592a6b04fd30b51796d69ca0b31cd8238a1dda/

This is for Gustavo :)

>      avr32 |                   python-2.7.6 | NOK | http://autobuild.buildroot.net/results/1aad2d677dbf7b0a3cb0120a35f45123124f51ab/

Fixed by http://git.buildroot.net/buildroot/commit/package/python?id=616dd6245be75798fa6d9fd057a0fb5d7020dc97.

>        arm |                  qt5base-5.2.1 | NOK | http://autobuild.buildroot.net/results/8eddd934bd80fdbcdf7a9dbf5d9f8b7ba69634d4/

Fatih, could you have a look at this issue?

>     x86_64 |                qt5webkit-5.2.1 | NOK | http://autobuild.buildroot.net/results/527315bf9b506ec32a2fbd255526570c7486e5fb/

Compiler bug.

>        arm |              util-linux-2.24.1 | NOK | http://autobuild.buildroot.net/results/27d22ff075e5e3e696d05ac6319043a5129abadc/

configure: error: chfn_chsh selected, but required PAM header file not available
make: *** [/scratch/peko/build/util-linux-2.24.1/.stamp_configured] Error 1

Fixed by
http://git.buildroot.net/buildroot/commit/package/util-linux?id=cb39b46f6007d02ba8db7611b5f11705b96e6cd1,
thanks Samuel!

>        arm |                vo-aacenc-0.1.3 | NOK | http://autobuild.buildroot.net/results/680b29cd824624eb8e4ec71187b9a6576444e72b/

Fixed by
http://git.buildroot.net/buildroot/commit/package/vo-aacenc?id=dc4d0e2f5cc783a826b3bfe4d31c00840407a8e2.

>       bfin |                   zeromq-4.0.4 | NOK | http://autobuild.buildroot.net/results/dc8dd7c1dd019dd06c4f806f2bd44d270eca4905/
>       bfin |                   zeromq-4.0.4 | NOK | http://autobuild.buildroot.net/results/439df2ad4482d08f295dbc40b2e88bf9167a7794/

I'll fix those ones today.

Thanks,

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

[Buildroot] Analysis of build results for 2014-04-20

[Anton Kolesov]

Hi Thomas,

> 
> >        arc |                openswan-2.6.41 | NOK |
> http://autobuild.buildroot.net/results/71f92d80b1aa3ac98660fb88004efd0ea
> 55e784e/
> 
> /home/test/test/3/output/host/usr/lib/gcc/arc-buildroot-linux-
> uclibc/4.8.0/../../../../arc-buildroot-linux-uclibc/bin/ld:
> /home/test/test/3/output/host/usr/arc-buildroot-linux-
> uclibc/sysroot/usr/lib/crt1.o: relocation R_ARC_32_ME against `main' can not
> be used when making a shared object; recompile with -fPIC
> /home/test/test/3/output/host/usr/arc-buildroot-linux-
> uclibc/sysroot/usr/lib/crt1.o: could not read symbols: Bad value
> collect2: error: ld returned 1 exit status
> make[4]: *** [pluto] Error 1
> 
> Could some ARC person look into this? Alexey, Anton?

[AK:] This is a limitation of ARC uClibc port: our crt1.o doesn't support PIE (though it seems simple hello world can be compiled). Our uClibc developer (added to CC) is going to fix this in near future, but he has a lot of other tasks to do right now, so this might not happen immediately.

Anton 

[Buildroot] Analysis of build results for 2014-04-20

[Thomas Petazzoni]

Dear Anton Kolesov,

On Mon, 21 Apr 2014 10:18:53 +0000, Anton Kolesov wrote:

> [AK:] This is a limitation of ARC uClibc port: our crt1.o doesn't support PIE (though it seems simple hello world can be compiled). Our uClibc developer (added to CC) is going to fix this in near future, but he has a lot of other tasks to do right now, so this might not happen immediately.

Why do we need a PIE executable here? As far as I understand, PIC is
mandatory for shared libraries, but PIE for executables is only needed
for security reasons, if you want to randomize the location of
executables in memory, at least in MMU-capable platforms.

So maybe we should just cook a patch to remove -fPIE when building
openswan ? I don't think we build any other package with -fPIE, so why
should openswan be different ?

Best regards,

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

[Buildroot] Analysis of build results for 2014-04-20

[Anton Kolesov]

Makefile.in of openswan contains -fPIE hardcoded . I suppose that is for the "security reasons", because this packages is IPsec implementation. I'm not a security expert, so I'm not sure how much PIE is important here. I would say that it is not much of the use to apply randomization to selected package, while leaving the rest of software on the same system without randomization.

Anton


> -----Original Message-----
> From: Thomas Petazzoni [mailto:thomas.petazzoni at free-electrons.com]
> Sent: 21 April 2014 15:58
> To: Anton Kolesov
> Cc: buildroot at uclibc.org; Alexey Brodkin; Vineet Gupta
> Subject: Re: Analysis of build results for 2014-04-20
> 
> Dear Anton Kolesov,
> 
> On Mon, 21 Apr 2014 10:18:53 +0000, Anton Kolesov wrote:
> 
> > [AK:] This is a limitation of ARC uClibc port: our crt1.o doesn't support PIE
> (though it seems simple hello world can be compiled). Our uClibc developer
> (added to CC) is going to fix this in near future, but he has a lot of other tasks
> to do right now, so this might not happen immediately.
> 
> Why do we need a PIE executable here? As far as I understand, PIC is
> mandatory for shared libraries, but PIE for executables is only needed
> for security reasons, if you want to randomize the location of
> executables in memory, at least in MMU-capable platforms.
> 
> So maybe we should just cook a patch to remove -fPIE when building
> openswan ? I don't think we build any other package with -fPIE, so why
> should openswan be different ?
> 
> Best regards,
> 
> Thomas
> --
> Thomas Petazzoni, CTO, Free Electrons
> Embedded Linux, Kernel and Android engineering
> http://free-electrons.com

[Buildroot] Analysis of build results for 2014-04-20

[Alexey Brodkin]

Hi Thomas,

On Mon, 2014-04-21 at 13:58 +0200, Thomas Petazzoni wrote:
> Dear Anton Kolesov,
> 
> On Mon, 21 Apr 2014 10:18:53 +0000, Anton Kolesov wrote:
> 
> > [AK:] This is a limitation of ARC uClibc port: our crt1.o doesn't support PIE (though it seems simple hello world can be compiled). Our uClibc developer (added to CC) is going to fix this in near future, but he has a lot of other tasks to do right now, so this might not happen immediately.
> 
> Why do we need a PIE executable here? As far as I understand, PIC is
> mandatory for shared libraries, but PIE for executables is only needed
> for security reasons, if you want to randomize the location of
> executables in memory, at least in MMU-capable platforms.

> So maybe we should just cook a patch to remove -fPIE when building
> openswan ? I don't think we build any other package with -fPIE, so why
> should openswan be different ?

As Anton has just answered -fPIE/-pie is enabled in top-level
Makefile.inc. But as I may see at least in master branch here
https://github.com/xelerance/Openswan/blob/master/Makefile.inc it's very
easy to override these settings with simple definitoin of USERCOMPILE
and USERLINK variables.

So definitely it is doable. But another good question would be if this
was done on purpose (which I cannot undersnad clearly from commit
message because too many items were dropped in at once here -
https://github.com/xelerance/Openswan/commit/6181d2b2999e112d47884dda48d1ca2916e2403e#diff-084b77e3e200296f6230945d5f0ea0ec) or not.

Another comment on PIE usage.
I know for sure that many architectures (and more to come) use PIE for
U-Boot (and here it is used on purpose for entire relocation of both
code and data), so definitely PIE is required feature of toolchains and
IMHO we don't want to escape it if it is used by packages developers.

Regards,
Alexey

[Buildroot] Analysis of build results for 2014-04-20

[Yann E. MORIN]

Thomas, All,

On 2014-04-21 11:49 +0200, Thomas Petazzoni spake thusly:
> >        arm |                  cairo-1.12.10 | NOK | http://autobuild.buildroot.net/results/2cb13a5bb92dabed219d49f49f0b9a2dfe65a0ca/
> 
> cairo-gl-private.h:287:5: error: unknown type name 'GLchar'
> 
> Could some OpenGL person look into this? Paul ? Yann ?

With only a cursory look and a bit of Google-fuu, it looks like we need
a full OpenGL provider in this case.

I'll look at it more in details, but probably not before this evening...

I was afraid my recent OpenGL virtual package as the reason for this
failure, but we already had such failures in the past. Pfew... :-)

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] Analysis of build results for 2014-04-20

[Eric Le Bihan]

Hi!
On Mon, Apr 21, 2014 at 04:24:44PM +0200, Yann E. MORIN wrote:
> Thomas, All,
>
> On 2014-04-21 11:49 +0200, Thomas Petazzoni spake thusly:
> > >        arm |                  cairo-1.12.10 | NOK | http://autobuild.buildroot.net/results/2cb13a5bb92dabed219d49f49f0b9a2dfe65a0ca/
> >
> > cairo-gl-private.h:287:5: error: unknown type name 'GLchar'
> >
> > Could some OpenGL person look into this? Paul ? Yann ?
>
> With only a cursory look and a bit of Google-fuu, it looks like we need
> a full OpenGL provider in this case.

In fact, looking at src/cairo-gl-private.h, we can see that GLchar is used in
functions related to shader loading. But the gl2.h file from sunxi-mali
defines prototypes using "char" where the official header from the Khronos
Group uses "GLchar". The gl2.h from Raspberry userland uses GLchar.

So the issue is with Sunxi Mali and has been fixed upstream:
https://github.com/linux-sunxi/sunxi-mali/commit/b36c94609c3a335c5bbe2019a8e8a1b261786b30

Best regards,
ELB

[Buildroot] Analysis of build results for 2014-04-20

[Yann E. MORIN]

Eric, All,

On 2014-04-21 19:03 +0200, Eric Le Bihan spake thusly:
> On Mon, Apr 21, 2014 at 04:24:44PM +0200, Yann E. MORIN wrote:
> > Thomas, All,
> >
> > On 2014-04-21 11:49 +0200, Thomas Petazzoni spake thusly:
> > > >        arm |                  cairo-1.12.10 | NOK | http://autobuild.buildroot.net/results/2cb13a5bb92dabed219d49f49f0b9a2dfe65a0ca/
> > >
> > > cairo-gl-private.h:287:5: error: unknown type name 'GLchar'
> > >
> > > Could some OpenGL person look into this? Paul ? Yann ?
> >
> > With only a cursory look and a bit of Google-fuu, it looks like we need
> > a full OpenGL provider in this case.
> 
> In fact, looking at src/cairo-gl-private.h, we can see that GLchar is used in
> functions related to shader loading. But the gl2.h file from sunxi-mali
> defines prototypes using "char" where the official header from the Khronos
> Group uses "GLchar". The gl2.h from Raspberry userland uses GLchar.
> 
> So the issue is with Sunxi Mali and has been fixed upstream:
> https://github.com/linux-sunxi/sunxi-mali/commit/b36c94609c3a335c5bbe2019a8e8a1b261786b30

Nice! Care to send a patch bumping sunxi-mali?

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] Analysis of build results for 2014-04-20

[Eric Le Bihan]

On Mon, Apr 21, 2014 at 07:17:03PM +0200, Yann E. MORIN wrote:
> Eric, All,
>
> On 2014-04-21 19:03 +0200, Eric Le Bihan spake thusly:
> > On Mon, Apr 21, 2014 at 04:24:44PM +0200, Yann E. MORIN wrote:
> > > Thomas, All,
> > >
> > > On 2014-04-21 11:49 +0200, Thomas Petazzoni spake thusly:
> > > > >        arm |                  cairo-1.12.10 | NOK | http://autobuild.buildroot.net/results/2cb13a5bb92dabed219d49f49f0b9a2dfe65a0ca/
> > > >
> > > > cairo-gl-private.h:287:5: error: unknown type name 'GLchar'
> > > >
> > > > Could some OpenGL person look into this? Paul ? Yann ?
> > >
> > > With only a cursory look and a bit of Google-fuu, it looks like we need
> > > a full OpenGL provider in this case.
> >
> > In fact, looking at src/cairo-gl-private.h, we can see that GLchar is used in
> > functions related to shader loading. But the gl2.h file from sunxi-mali
> > defines prototypes using "char" where the official header from the Khronos
> > Group uses "GLchar". The gl2.h from Raspberry userland uses GLchar.
> >
> > So the issue is with Sunxi Mali and has been fixed upstream:
> > https://github.com/linux-sunxi/sunxi-mali/commit/b36c94609c3a335c5bbe2019a8e8a1b261786b30
>
> Nice! Care to send a patch bumping sunxi-mali?

Too bad, the pull request fixing the issue has not been merged yet! So I'll
only post a small patch for sunxi-mali header files, as the pull request also
covers the *.pc files issue, which has not been acked by upstream yet.

Best regards,
ELB

[Buildroot] Analysis of build results for 2014-04-20

[Yann E. MORIN]

Eric, All,

On 2014-04-21 19:56 +0200, Eric Le Bihan spake thusly:
> On Mon, Apr 21, 2014 at 07:17:03PM +0200, Yann E. MORIN wrote:
> > Eric, All,
> >
> > On 2014-04-21 19:03 +0200, Eric Le Bihan spake thusly:
> > > On Mon, Apr 21, 2014 at 04:24:44PM +0200, Yann E. MORIN wrote:
> > > > Thomas, All,
> > > >
> > > > On 2014-04-21 11:49 +0200, Thomas Petazzoni spake thusly:
> > > > > >        arm |                  cairo-1.12.10 | NOK | http://autobuild.buildroot.net/results/2cb13a5bb92dabed219d49f49f0b9a2dfe65a0ca/
> > > > >
> > > > > cairo-gl-private.h:287:5: error: unknown type name 'GLchar'
> > > > >
> > > > > Could some OpenGL person look into this? Paul ? Yann ?
> > > >
> > > > With only a cursory look and a bit of Google-fuu, it looks like we need
> > > > a full OpenGL provider in this case.
> > >
> > > In fact, looking at src/cairo-gl-private.h, we can see that GLchar is used in
> > > functions related to shader loading. But the gl2.h file from sunxi-mali
> > > defines prototypes using "char" where the official header from the Khronos
> > > Group uses "GLchar". The gl2.h from Raspberry userland uses GLchar.
> > >
> > > So the issue is with Sunxi Mali and has been fixed upstream:
> > > https://github.com/linux-sunxi/sunxi-mali/commit/b36c94609c3a335c5bbe2019a8e8a1b261786b30
> >
> > Nice! Care to send a patch bumping sunxi-mali?
> 
> Too bad, the pull request fixing the issue has not been merged yet! So I'll
> only post a small patch for sunxi-mali header files, as the pull request also
> covers the *.pc files issue, which has not been acked by upstream yet.

Ah, OK. I'll have a look at your patch shortly. Thanks! :-)

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] Analysis of build results for 2014-04-20

[Thomas Petazzoni]

Dear Anton Kolesov,

On Mon, 21 Apr 2014 13:29:14 +0000, Anton Kolesov wrote:

> Makefile.in of openswan contains -fPIE hardcoded . I suppose that is
> for the "security reasons", because this packages is IPsec
> implementation. I'm not a security expert, so I'm not sure how much
> PIE is important here. I would say that it is not much of the use to
> apply randomization to selected package, while leaving the rest of
> software on the same system without randomization.

Since we don't build all our binaries -fPIE, I don't think it makes
much sense to build just openswan -fPIE. So probably a simple patch to
remove -fPIE from openswan makefiles should be OK.

Best regards,

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

[Buildroot] Analysis of build results for 2014-04-20

[Thomas Petazzoni]

Dear Alexey Brodkin,

On Mon, 21 Apr 2014 13:47:57 +0000, Alexey Brodkin wrote:

> As Anton has just answered -fPIE/-pie is enabled in top-level
> Makefile.inc. But as I may see at least in master branch here
> https://github.com/xelerance/Openswan/blob/master/Makefile.inc it's
> very easy to override these settings with simple definitoin of
> USERCOMPILE and USERLINK variables.
> 
> So definitely it is doable. But another good question would be if this
> was done on purpose (which I cannot undersnad clearly from commit
> message because too many items were dropped in at once here -
> https://github.com/xelerance/Openswan/commit/6181d2b2999e112d47884dda48d1ca2916e2403e#diff-084b77e3e200296f6230945d5f0ea0ec)
> or not.
> 
> Another comment on PIE usage.
> I know for sure that many architectures (and more to come) use PIE for
> U-Boot (and here it is used on purpose for entire relocation of both
> code and data), so definitely PIE is required feature of toolchains
> and IMHO we don't want to escape it if it is used by packages
> developers.

Ok. However in the mean time, I think we could disable it for openswan,
no?

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