Le mardi 07 janvier 2014 à 09:31 -0500, Mike Miller a écrit : > On Tue, 7 Jan 2014 09:06:29 +0100, Sébastien Villemot wrote: > > On my side, I have fixed the FTBFS on s390x, which was due to some test > > crashing. This would need more investigation, but in the meantime I have > > just disabled that test. > > I think there was an upstream bug report about the dblquad test failing > at some point, may be related. > > > The archs on which we should disable the JIT are therefore: armel, > > armhf, mipsel, powerpc (for armhf I made a test on a porterbox, since > > the buildds have not taken it yet). > > > > There is also the case of mips where I can't understand the crash, I am > > going to replicate it on a porterbox. > > The mips build failure looks also related to JIT, doesn't it? Just not > in the unit test as with the other ports: > > | Operand is null > | %2 = call %octave_base_value* @octave_jit_binary_any_any(i32 0, > %octave_base_value* %0, %octave_base_value* %1) > | Operand is null > | ret %octave_base_value* %2 > | Broken module found, compilation aborted! > | Stack dump: > | 0. Running pass 'Module Verifier' on function > '@octave_jit_binary_any_any_0' > | /bin/sh: line 1: 16228 Aborted ../../run-octave -f -q > -H ./mk_doc_cache.m doc-cache ./macros.texi ../../scripts/DOCSTRINGS > ../../libinterp/DOCSTRINGS > | make[4]: *** [doc-cache] Error 1 I had not seen this error message in the build log (parallel builds make build logs harder to read since several processes are entangled). Indeed I agree with you that this must come from the JIT. -- .''`. Sébastien Villemot : :' : Debian Developer `. `' http://www.dynare.org/sebastien `- GPG Key: 4096R/381A7594
Attachment:
signature.asc
Description: This is a digitally signed message part