Re: Porter assistance for BSD unique test failures
Steven Chamberlain <email@example.com> writes:
> On 07/02/13 14:05, Petr Salinger wrote:
>> It seems that it builds fine under 8.3 and 9.0 kernels from wheezy, while
>> it fails under 8.1 kernel from squeeze in otherwise same (kvm, almost
>> wheezy) kfreebsd-i386 environment.
>> The "export MALLOC_CHECK_=7" generates failure under 8.1 kernel reliable.
> Interesting. I wonder if this could be related to the FTBFS of
> openjdk-7 seen on 8.1 but not 9.0 kfreebsd, which seemed to be some
> problem affecting the dl-minimal implementation of realloc()
>> Inconsistency detected by ld.so: dl-minimal.c: 138: realloc: Assertion `ptr == alloc_last_block' failed!
> If so, this would be a much easier testcase than spending nearly a day
> building openjdk-7...
The openjdk build failures where happening independent from kernels but
based on sbuild/schroot versions. That was how far I was able to track
it down back then. It was *not* related to env filtering (failure was
independant of putting the building step in a env -u)