[Buildroot] [autobuild.buildroot.net] Build results for 2013-09-13

[Michael Rommel]

Hi Thomas,

thanks for the info. I already had a look at the http://git.buildroot.net/buildroot-test/plain/utils/br-reproduce-build
script and the make without any additional arguments is still running as we speak.

But it is true, the machine has many more packages installed. I originally thought it might have to do that
the .mk file checks for 
  ifeq ($(BR2_PACKAGE_LIBGCRYPT),y)

whereas the Config file does a
select BR2_PACKAGE_LIBGCRYPT if !BR2_PACKAGE_OPENSSL

but in the end it shouldn't matter and during the three builds I did, openssl was built first everytime and
the lib detection worked...

Cheers,

  Michael.

-- 
Michael Rommel, Erlangen, Germany

On Sep 14, 2013, at 17:03 , Thomas Petazzoni <thomas.petazzoni at free-electrons.com> wrote:

> Dear Michael Rommel,
> 
> On Sat, 14 Sep 2013 14:36:26 +0200, Michael Rommel wrote:
> 
>> I thought I take care of one of the build failures from yesterday and picked the line:
>> 
>>      arm |                  libssh2-1.4.3 | NOK | http://autobuild.buildroot.net/results/e93e24d210dc7a2a51758536b7b3964b8a3251b4/
>> 
>> I tried to reproduce the error, but was unable to do so in three attempts.
>> 1. used a clean i568 toolchain without configuration and gave only a make toolchain libssh2
>> 2. used a clean arm toolchain with a default .config file and made a make toolchain libssh2
>> 3. used the config file from the autobuilder and did a make toolchain libssh2
> 
> If you do 'make toolchain libssh2', you're not reproducing what the
> autobuilder did, because you're telling Buildroot to just build the
> toolchain, and then libssh2 (and its dependencies).
> 
> What the autobuilder is just a simple 'make'. So many more things might
> have been built before libssh2 and its dependencies, and it is
> sometimes what is affecting the build and causing the failure.
> 
> Another possibility is that your host system isn't necessarily
> comparable to the one of the autobuilders. All the build tests executed
> on the Free Electrons server take place inside a minimal Debian chroot,
> that has only the minimal set of packages needed to get Buildroot
> running. This is most likely different from your machine, which
> probably has a lot more packages installed, including a number of
> development packages for some libraries.
> 
> I'd recommend to first try to reproduce with just a plain 'make'. If
> you still can't reproduce the failure, then I'll have a look myself
> using the autobuilder environment.
> 
> Maybe I should be providing a script that downloads the complete
> chroot, and reproduce the build failure within the chroot, so that
> people can get the exact same behavior as the one in the autobuilder
> machine.
> 
> Anyway, thanks for having taken the time to look at this build failure!
> 
> Thomas
> -- 
> Thomas Petazzoni, Free Electrons
> Kernel, drivers, real-time and embedded Linux
> development, consulting, training and support.
> http://free-electrons.com