please fix nthash FTBS in lunar

[Vladimir Petko]

nthash fails to build from source for armhf, riscv64 and s390x platform.

s390x platform is not available for its dependency libbtllib-dev, because libomp-dev provided by llvm-defaults is not available for s390x [1]

Build failed for riscv64 because it was built with libbtllib-dev earlier depended on libsdsl-dev. It was demoted to recommends and we can retry the build.

libbtllib-dev fails tests on armhf [2]

Objectives:
 - update arches to align with dependency (drop s390x)
 - fix armhf tests

[1] https://launchpadlibrarian.net/655100169/buildlog_ubuntu-lunar-s390x.btllib_1.4.10+dfsg-1ubuntu1_BUILDING.txt.gz
[2] https://launchpadlibrarian.net/655100633/buildlog_ubuntu-lunar-armhf.btllib_1.4.10+dfsg-1ubuntu1_BUILDING.txt.gz

[Vladimir Petko (vpa1977)]

btlib tests die on armhf due to the deadlock:
--------------------
(gdb) thread apply all bt

Thread 4 (Thread 0xf6a5c800 (LWP 3970) "seq_reader_fast"):
#0 __libc_do_syscall () at ../sysdeps/unix/sysv/linux/arm/libc-do-syscall.S:46
#1 0xf7cba47e in __futex_abstimed_wait_common32 (private=<optimized out>, cancel=<optimized out>, abstime=<optimized out>, op=<optimized out>, expected=<optimized out>, futex_word=<optimized out>) at futex-internal.c:40
#2 __futex_abstimed_wait_common (futex_word=futex_word@entry=0x423588, expected=expected@entry=0, clockid=clockid@entry=0, abstime=abstime@entry=0x0, private=private@entry=0, cancel=cancel@entry=true) at futex-internal.c:99
#3 0xf7cba55c in __GI___futex_abstimed_wait_cancelable64 (futex_word=futex_word@entry=0x423588, expected=expected@entry=0, clockid=clockid@entry=0, abstime=abstime@entry=0x0, private=private@entry=0) at futex-internal.c:139
#4 0xf7cbc876 in __pthread_cond_wait_common (abstime=0x0, clockid=0, mutex=0x423540, cond=0x423560) at pthread_cond_wait.c:503
#5 ___pthread_cond_wait (cond=0x423560, mutex=0x423540) at pthread_cond_wait.c:627
#6 0x0040d542 in std::condition_variable::wait<btllib::OrderQueueMPMC<btllib::SeqReader::Record>::write(btllib::OrderQueue<btllib::SeqReader::Record>::Block&)::{lambda()#1}>(std::unique_lock<std::mutex>&, btllib::OrderQueueMPMC<btllib::SeqReader::Record>::write(btllib::OrderQueue<btllib::SeqReader::Record>::Block&)::{lambda()#1}) (__p=..., __lock=..., this=0x423560) at /usr/include/c++/12/condition_variable:102
#7 btllib::OrderQueueMPMC<btllib::SeqReader::Record>::write (this=0xfffef910, block=...) at ../include/btllib/order_queue.hpp:243
#8 0x0040b1c8 in operator() (__closure=<optimized out>) at /usr/src/btllib-1.4.10+dfsg-1ubuntu1/src/btllib/seq_reader.cpp:420
#9 0xf7ebde16 in ?? () from /lib/arm-linux-gnueabihf/libstdc++.so.6
#10 0xf7cbd06e in start_thread (arg=0x417f6c05) at pthread_create.c:442
#11 0xf7d14140 in ?? () at ../sysdeps/unix/sysv/linux/arm/clone.S:74 from /lib/arm-linux-gnueabihf/libc.so.6
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

Thread 3 (Thread 0xf745d800 (LWP 3969) "seq_reader_fast"):
#0 __libc_do_syscall () at ../sysdeps/unix/sysv/linux/arm/libc-do-syscall.S:46
#1 0xf7cba47e in __futex_abstimed_wait_common32 (private=<optimized out>, cancel=<optimized out>, abstime=<optimized out>, op=<optimized out>, expected=<optimized out>, futex_word=<optimized out>) at futex-internal.c:40
#2 __futex_abstimed_wait_common (futex_word=futex_word@entry=0x423518, expected=expected@entry=0, clockid=clockid@entry=0, abstime=abstime@entry=0x0, private=private@entry=0, cancel=cancel@entry=true) at futex-internal.c:99
#3 0xf7cba55c in __GI___futex_abstimed_wait_cancelable64 (futex_word=futex_word@entry=0x423518, expected=expected@entry=0, clockid=clockid@entry=0, abstime=abstime@entry=0x0, private=private@entry=0) at futex-internal.c:139
#4 0xf7cbc876 in __pthread_cond_wait_common (abstime=0x0, clockid=0, mutex=0x4234d0, cond=0x4234f0) at pthread_cond_wait.c:503
#5 ___pthread_cond_wait (cond=0x4234f0, mutex=0x4234d0) at pthread_cond_wait.c:627
#6 0x0040d542 in std::condition_variable::wait<btllib::OrderQ...

[Vladimir Petko (vpa1977)]

Seems to be caused by absence of the integer overflow - there is addional condition for the queue `&&(num - target.last_tenant <= this->queue_size),` and target.lasttenant is coming up as `4294967295`

[Vladimir Petko (vpa1977)]

Filed bug and patch upstream: https://github.com/bcgsc/btllib/issues/80

[Launchpad Janitor (janitor)]

This bug was fixed in the package btllib - 1.4.10+dfsg-1ubuntu2

---------------
btllib (1.4.10+dfsg-1ubuntu2) lunar; urgency=medium

  [ Graham Inggs ]
  * Revert libbtllib-dev to Architecture: any

  [ Vladimir Petko ]
  * Avoid FTBFS due to deadlock on 32-bit architectures (LP: #2009757)

 -- Graham Inggs <email address hidden> Sun, 16 Apr 2023 12:12:42 +0000

[Launchpad Janitor (janitor)]

This bug was fixed in the package nthash - 2.3.0+dfsg-1ubuntu1

---------------
nthash (2.3.0+dfsg-1ubuntu1) lunar; urgency=medium

  * Link with -latomic on riscv64 to avoid FTBFS (LP: #2009757)

 -- Graham Inggs <email address hidden> Sun, 16 Apr 2023 12:59:10 +0000