Bug#920835: mlucas FTBFS on !amd64: test failures

2019-01-31 Thread Alex Vong
upstream to investigate. Cheers, Alex Juhani Numminen writes: > On Wed, 30 Jan 2019 05:27:48 +0800 Alex Vong wrote: >> Hello Adrian Bunk, >> >> The test log is not useful at all. I want to add a patch to enable >> verbose test log and let it build again to identify th

Bug#920834: mlucas: FTBFS on i386

2019-01-30 Thread Alex Vong
# I am changing the title since I think the i386 baseline is in fact not # violated. retitle 920834 mlucas: FTBFS on i386 thanks Alex Vong writes: > Hello Adrian Bunk, > > I figure out what's the problem. There's a typo in upstream src: > > s/SSE2_RADI16_CALC_TWID

Bug#920834: mlucas: FTBFS and baseline vioiation on i386

2019-01-30 Thread Alex Vong
Alex Vong writes: > Dear Adrian Bunk, > > Regarding FTBFS, I've commented it at <https://bugs.debian.org/920835>. > > Regarding baseline vioiation on i386, the autotool build system builds 2 > versions of the binary: generic-c and SSE2, and install it to > libexecd

Bug#920834: mlucas: FTBFS and baseline vioiation on i386

2019-01-29 Thread Alex Vong
Dear Adrian Bunk, Regarding FTBFS, I've commented it at . Regarding baseline vioiation on i386, the autotool build system builds 2 versions of the binary: generic-c and SSE2, and install it to libexecdir. Then the wrapper script selects the right binary at

Bug#920835: mlucas FTBFS on !amd64: test failures

2019-01-29 Thread Alex Vong
Hello Adrian Bunk, The test log is not useful at all. I want to add a patch to enable verbose test log and let it build again to identify the issues. Is it possible for me to get access to porter machine or should I upload to experimental and let buildd to build it? Thanks for your great work

Bug#860662: mlucas: FTBFS on i386: dh_auto_test: make -j64 check VERBOSE=1 returned exit code 2

2017-04-24 Thread Alex Vong
Hello, A RC bug was reported by Lucas Nussbaum last week. I tried to reproduce it in a QEMU VM with 64 cores but was unable to reproduce it precisely (pthread_create() dies with EAGAIN instead of ENOMEM). Recently, I find a similar bug[0], which will be fixed in the future by splitting big test

Bug#860662: mlucas: FTBFS on i386: dh_auto_test: make -j64 check VERBOSE=1 returned exit code 2

2017-04-23 Thread Alex Vong
Bug #860664: signature.asc Description: PGP signature

Bug#860662: mlucas: FTBFS on i386: dh_auto_test: make -j64 check VERBOSE=1 returned exit code 2

2017-04-23 Thread Alex Vong
Alex Vong <alexvong1...@gmail.com> writes: > Upon inspection, it seems the problem is caused by some serious memory > leaks: >>> executipthread_create:: Cannot allocate memory >>> FAIL self_test.test (exit status: 139) > > Building with '-fsanitize=leak' seems

Bug#860662: mlucas: FTBFS on i386: dh_auto_test: make -j64 check VERBOSE=1 returned exit code 2

2017-04-21 Thread Alex Vong
Thanks for the bug report. Upon inspection, it seems the problem is caused by some serious memory leaks: >> executipthread_create:: Cannot allocate memory >> FAIL self_test.test (exit status: 139) Building with '-fsanitize=leak' seems to confirm it: > make[1]: Leaving directory