Bug#769864: vcmi: FTBFS on armhf, hurd-any, kfreebsd-any, mipsel and ppc64el

[James Cowgill]

On Thu, 27 Nov 2014 14:16:05 +0100 Johannes Schauer <j.schauer at email.de> wrote:
> The armhf problem might be due to heavy swapping bringing the build to a
> standstill. At the point where the build stops, each compilation unit will
> require about 2.3 GB of memory. Since armhf boxes build with
> DEB_BUILD_OPTIONS=parallel=5 this means that 11 GB of memory might be consumed
> at a time which doesn't go well on a machine with only 4 GB of physical memory.
> The problem might get solved with the next upload which removes the --parallel
> option from the dh invocation in debian/rules.
> 
> I still have no idea about the mipsel ICE but applied for access to the mipsel
> porterbox to investigate this issue. I tried setting up a qemu based
> environment but failed because it seems that it is not possible to just use
> Debian installer but instead a manual debootstrap has to be done. I didn't find
> the time to fiddle around with that.

Hi,

I'm willing to bet that the mipsel failure was an out of memory issue as
well. If it was an actual ICE, then I would expect it to be terminated
by SIGABRT rather than SIGKILL. Also the buildd it ran on has a massive
1GB of RAM (which is unlucky, I think the others have 4 or 8 GB).

Thanks,
James