[Buildroot] [autobuild.buildroot.net] Your build results for 2017-03-18

[Marcin Nowakowski]

Hi Thomas,

On 19.03.2017 20:50, Thomas Petazzoni wrote:
> Hello,
>
> On Sun, 19 Mar 2017 19:06:19 +0000, Marcin Nowakowski wrote:
>
>> These failures are actually failures while building a host app in a
>> target package ... but this isn't happening for any arch so I'm
>> wondering if there are different host compiler variations across
>> build machines?
>
> I'm not sure what you mean by "this isn't happening for any arch". The
> failure is while building something with the host compiler, so the
> target architecture doesn't really matter.

Yes, of course. What I meant is that it didn't fail for all builds but 
only some, so does that mean that there are different build machines 
with different host toolchains?

>> Unfortunately the flag that is causing the failure is checked for the
>> target compiler only (and isn't really needed for the host anyway),
>> so I'll prepare a patch that removes it from the host build as
>> unnecessary, but the variations across build targets that don't
>> depend on the target itself have surprised me
>
> Indeed, the problem comes from the configure.ac script that tests if
> the target compiler supports -fstack-protector-strong and if it does,
> uses it in AM_CFLAGS. Unfortunately, this AM_CFLAGS also gets used when
> building the host tool for the documentation.
>
> Since anyway Buildroot will already be passing -fstack-protector in
> CFLAGS if enabled in the Buildroot configuration, I believe a patch to
> configure.ac to add a --enable/--disable option to enable/disable the
> addition of -fstack-protector-strong would be nice.
>
> Or alternatively: fix the real problem, and make sure it doesn't use
> AM_CFLAGS for the host build.

That's a patch (removal of AM_CFLAGS for host build) I will be sending 
to the library maintainer. I'll send a patch for BR if/when it is 
accepted upstream.

Marcin