Bug#895467: libsearpc FTBFS on 64bit big endian: test failure
Hi Adrian and thanks for your response, On 16.04.2018 15:13, Adrian Bunk wrote: > Third solution: > Fix the bugs. Well that was actually the first solution I meant, but I just didn't have much hope finding a solution, so thank you very much for your help. > After fixing the two above bugs, libsearpc builds on s390x. I've added patches to the package and will send them upstream. > An en passant finding (not required to fix the FTBFS) is the following > that is documented to be unsupported, and might break on any > architecture at any time: > https://sources.debian.org/src/libsearpc/3.0.8-3/lib/searpc-named-pipe-transport.c/#L236 I will report that to upstream, too. Thx, Moritz signature.asc Description: OpenPGP digital signature
Bug#895467: libsearpc FTBFS on 64bit big endian: test failure
On Mon, Apr 16, 2018 at 12:09:36PM +0200, Moritz Schlarb wrote: >... > Two other "solutions", I can think of would be: > - Using a switch on the current architecture in debian/rules to just > don't test on these arches > - Exclude the arches from Architecture: in debian/control >... Third solution: Fix the bugs. This happens to work on 64bit little endian, but passing a pointer to a 64bit variable for a 32bit value is wrong: https://sources.debian.org/src/libsearpc/3.0.8-3/lib/searpc-named-pipe-transport.c/#L255 Same problem as above, I'd bet the "len = 0" was added as "fix" when this bug hit on 64bit little endian: https://sources.debian.org/src/libsearpc/3.0.8-3/lib/searpc-named-pipe-transport.c/#L224 After fixing the two above bugs, libsearpc builds on s390x. An en passant finding (not required to fix the FTBFS) is the following that is documented to be unsupported, and might break on any architecture at any time: https://sources.debian.org/src/libsearpc/3.0.8-3/lib/searpc-named-pipe-transport.c/#L236 > Best regards, > Moritz cu Adrian -- "Is there not promise of rain?" Ling Tan asked suddenly out of the darkness. There had been need of rain for many days. "Only a promise," Lao Er said. Pearl S. Buck - Dragon Seed
Bug#895467: libsearpc FTBFS on 64bit big endian: test failure
Hi there, I'm unsure how to handle this. The current breakage of the package since 3.0.8-2 comes from me re-enabling the upstream tests, which were previously disabled with a comment that they were broken upstream (so, at https://buildd.debian.org/status/logs.php?pkg=libsearpc=3.0.8-1, all builds ran fine but right now, they fail on some architectures - s390x, ppc64 and sparc64: https://buildd.debian.org/status/package.php?p=libsearpc). Since the package still builds and tests fine on *most* architectures, I'd beg to differ that the test suite is broken in general. Now of course myself and upstream can look into the problem, but since it maybe is specific to an exotic architecture that we might not have access to, I would not have high hopes in that approach. Two other "solutions", I can think of would be: - Using a switch on the current architecture in debian/rules to just don't test on these arches - Exclude the arches from Architecture: in debian/control Would any of those options be appropriate? Best regards, Moritz On 11.04.2018 22:24, Adrian Bunk wrote: > Source: libsearpc > Version: 3.0.8-2 > Severity: serious > > https://buildd.debian.org/status/package.php?p=libsearpc=sid > > ... > FAIL: test-searpc > = > > Loaded 1 suites: > Started > > (process:10121): Searpc-WARNING **: 17:59:21.601: failed to read rpc request: > Bad address > > (process:10121): Searpc-WARNING **: 17:59:21.601: failed to read rpc > response: Connection reset by peer > F > (process:10121): Searpc-WARNING **: 17:59:21.923: failed to read rpc request: > Bad address > > (process:10121): Searpc-WARNING **: 17:59:21.923: failed to send rpc call: > Broken pipe > F > (process:10121): Searpc-WARNING **: 17:59:22.122: failed to read rpc request: > Bad address > > (process:10121): Searpc-WARNING **: 17:59:22.125: failed to read rpc > response: Connection reset by peer > F > > 1) Failure: > searpc::pipe_simple_call [searpc.c:427] > Expression is not true: error == NULL > Transport Error > > 2) Failure: > searpc::pipe_large_request [searpc.c:454] > Expression is not true: error == NULL > Transport Error > > 3) Failure: > searpc::pipe_concurrent_clients [searpc.c:483] > Expression is not true: error == NULL > Transport Error > > System error: `fork()` call failed (12) - Cannot allocate memory > FAIL test-searpc (exit status: 255) > > > Testsuite summary for libsearpc 3.0.8 > > # TOTAL: 1 > # PASS: 0 > # SKIP: 0 > # XFAIL: 0 > # FAIL: 1 > # XPASS: 0 > # ERROR: 0 > > See tests/test-suite.log > Please report to i...@seafile.com > > make[5]: *** [Makefile:687: test-suite.log] Error 1 >
Bug#895467: libsearpc FTBFS on 64bit big endian: test failure
Source: libsearpc Version: 3.0.8-2 Severity: serious https://buildd.debian.org/status/package.php?p=libsearpc=sid ... FAIL: test-searpc = Loaded 1 suites: Started (process:10121): Searpc-WARNING **: 17:59:21.601: failed to read rpc request: Bad address (process:10121): Searpc-WARNING **: 17:59:21.601: failed to read rpc response: Connection reset by peer F (process:10121): Searpc-WARNING **: 17:59:21.923: failed to read rpc request: Bad address (process:10121): Searpc-WARNING **: 17:59:21.923: failed to send rpc call: Broken pipe F (process:10121): Searpc-WARNING **: 17:59:22.122: failed to read rpc request: Bad address (process:10121): Searpc-WARNING **: 17:59:22.125: failed to read rpc response: Connection reset by peer F 1) Failure: searpc::pipe_simple_call [searpc.c:427] Expression is not true: error == NULL Transport Error 2) Failure: searpc::pipe_large_request [searpc.c:454] Expression is not true: error == NULL Transport Error 3) Failure: searpc::pipe_concurrent_clients [searpc.c:483] Expression is not true: error == NULL Transport Error System error: `fork()` call failed (12) - Cannot allocate memory FAIL test-searpc (exit status: 255) Testsuite summary for libsearpc 3.0.8 # TOTAL: 1 # PASS: 0 # SKIP: 0 # XFAIL: 0 # FAIL: 1 # XPASS: 0 # ERROR: 0 See tests/test-suite.log Please report to i...@seafile.com make[5]: *** [Makefile:687: test-suite.log] Error 1
