Buildroot Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
To: buildroot@busybox.net
Subject: [Buildroot] Analysis of build results for 2016-02-21
Date: Mon, 22 Feb 2016 23:21:44 +0100	[thread overview]
Message-ID: <20160222232144.61b99b32@free-electrons.com> (raw)
In-Reply-To: <CALdGskKUp-GtxeOBWq-5Tzp7b2zBad3=RoRxeN1eaJHPcg5Xww@mail.gmail.com>

Hello Olivier,

On Mon, 22 Feb 2016 21:05:00 +0200, Olivier Schonken wrote:

> arm |                     cups-2.1.2 | NOK |
> http://autobuild.buildroot.net/results/07ccca8f80c56ab0ff48bfee876885b1c8c3916e/
> arc |                     cups-2.1.2 | NOK |
> http://autobuild.buildroot.net/results/b169b939e38cb761319a92d2871f4f59ddb5c4ac/
> 
> Both error cases looks like it has to do with uclibc not behaving as
> expected at link time for the ippserver build (Both builds link statically
> with position independent execution also set).  For the arm platform it
> can't find the Scrt1.o file, and for the ARC compiler, __init_array_end is
> undefined.
> 
> I haven't considered all the options yet, but two that come to mind are
> disabling cups support for uclibc type compilers, or else to disable the
> cups test-framework.

Disabling for uClibc is completely overkill: this problem only occurs
on ARM with static linking and on ARC. CUPS builds perfectly fine on
ARM/uClibc when dynamic linking is used.

The problem is that CUPS forcefully uses PIE support, and:

 1/ Our static-only toolchain for ARM that uses uClibc-ng does not
    contain Scrt1.o, while a dynamic-linking enabled uClibc-ng ARM
    toolchain does have it. Due to this, trying to link a minimal C
    program with -fPIE -pie fails. I'm Cc'ing Waldemar, the uClibc-ng
    maintainer, to get his input on this.

 2/ PIE is not supported by the ARC architecture, and should not be
    used. Unfortunately, the configure script of CUPS only does an
    AC_TRY_COMPILE() test to verify the support of PIE in the compiler,
    but on ARC, it only fails at link time, so the test should be an
    AC_TRY_LINK(). Olivier, could you try to fix the cups stuff so that
    we can autoreconf it properly, and therefore fix such issues in the
    cups configure.ac ? I'm also Cc'ing Alexey from Synopsys, who can
    give more details about the PIE issue on ARC.

Thanks for your investigation!

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

  parent reply	other threads:[~2016-02-22 22:21 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-02-22 19:05 [Buildroot] Analysis of build results for 2016-02-21 Olivier Schonken
2016-02-22 21:55 ` Romain Naour
2016-02-22 22:21 ` Thomas Petazzoni [this message]
2016-02-23  7:34   ` Olivier Schonken
2016-02-23  8:11     ` Thomas Petazzoni
2016-02-23 11:13       ` Olivier Schonken
2016-02-24 21:30       ` Waldemar Brodkorb
  -- strict thread matches above, loose matches on Subject: below --
2016-02-22  7:30 [Buildroot] [autobuild.buildroot.net] Build " Thomas Petazzoni
2016-02-22 14:05 ` [Buildroot] Analysis of build " Thomas Petazzoni
2016-02-23 10:12   ` John Keeping
2016-02-23 10:40     ` Jeroen Roovers
2016-02-23 11:03       ` John Keeping
2016-02-23 12:30       ` Thomas Petazzoni
2016-02-24 11:08   ` Eric Limpens
2016-02-24 20:45     ` Arnout Vandecappelle
2016-02-24 20:59       ` Eric Limpens
2016-02-24 21:42   ` Waldemar Brodkorb
2016-02-24 22:09     ` Arnout Vandecappelle
2016-02-27 15:19   ` Romain Naour
2016-02-27 22:05   ` Bernd Kuhls
2016-03-01  9:38     ` Thomas Petazzoni
2016-02-28  0:17   ` Bernd Kuhls
2016-03-01  9:39     ` Thomas Petazzoni
2016-03-03 13:56   ` Gustavo Zacarias

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20160222232144.61b99b32@free-electrons.com \
    --to=thomas.petazzoni@free-electrons.com \
    --cc=buildroot@busybox.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox