Bug#740232: resiprocate ftbfs on amd64 (1 test failure)

[Daniel Pocock]

On 27/02/14 15:37, Daniel Pocock wrote:
> On 27/02/14 15:18, Matthias Klose wrote:
>> Am 27.02.2014 14:02, schrieb Daniel Pocock:
>>> On 27/02/14 11:12, Matthias Klose wrote:
>>>> Package: resiprocate
>>>> Version: 1.9.1-1
>>>> Severity: serious
>>>> Tags: sid jessie
>>>>
>>>> [...]
>>>> STACK | 20140227-095642.674 | testXMLCursor | RESIP:CONTENTS | 47431674565824 |
>>>> XMLCursor.cxx:260 | XMLCursor::nextSibling0x2b238c249d00[<resource
>>>> uri="sip:adam-friends at exa] 0x2b238c24a010[<list xmlns="urn:ietf:params:xml:ns]
>>>> All OK
>>>> PASS: testXMLCursor
>>>> ====================
>>>> 1 of 18 tests failed
>>>> ====================
>>> This is just the summary line
>>>
>>> Could you look earlier in the log for the "FAIL" and provide the context
>>> around that?  e.g.
>>>
>>>        grep -C 20 "FAIL:" resiprocate-build.log
>> PASS: testMD5Stream
>> lt-testParseBuffer: testParseBuffer.cxx:41: int main(int, char**): Assertion
>> `result == "test"' failed.
>> /bin/bash: line 5: 12077 Aborted                 ${dir}$tst
>> FAIL: testParseBuffer
>>
>>
> 
> 
> Great, that is the same failure as the Ubuntu build log link
> 
> I notice Ubuntu builds with gcc 4.8 now.  The other builds we've done on
> amd64 have been with gcc 4.6 and gcc 4.7 and the tests have always run
> successfully there.
> 
> Are you aware of any gcc 4.8 issues that could be exposing a fault in
> this code or causing it to fail?  Have other packages had similar issues
> on the latest Ubuntu?
> 
> I'll test the latest trunk with this gcc
> 
> 


I just did the following:

- on a virtual machine with Ubuntu saucy, amd64, gcc-4.8, 1.9.1-1 builds OK

- upgraded the VM to latest trusty with apt-get dist-upgrade

- build failed with a compiler error (see below)

- starting the build again (by calling make in the source tree without
cleaning it), it continues without repeating the error

- the build machine is a very stable HP server with ECC RAM, unlikely to
be a random hardware issue, nothing else is showing signs of trouble

I'm suspicious about this gcc in trusty now, not sure this should be an
RC bug against reSIProcate but I'm happy to investigate some more on a
machine with sid as well.


/usr/include/srtp/config.h:134:0: note: this is the location of the
previous definition
 #define PACKAGE_VERSION ""
 ^
g++: internal compiler error: Killed (program cc1plus)
Please submit a full bug report,
with preprocessed source if appropriate.
See <file:///usr/share/doc/gcc-4.8/README.Bugs> for instructions.
make[5]: *** [Conversation.lo] Error 1
make[5]: *** Waiting for unfinished jobs....