swi-prolog 8.5.6 s390x thread test failure in Alpine

[brebs-gh]

Hi, 8.5.6 had one test failure on s390x in Alpine:

27/73 Test #27: swipl:thread .....................Subprocess aborted***Exception:  17.93 sec
Running scripts from thread ............
plunit_queue_gc:create: Required 2 retrys before all queues were GCed
.............................................[PROLOG SYSTEM ERROR:  Thread 23 () at Mon Jan 24 00:43:23 2022
    Unknown problem importing prolog_debug:assertion/1 into module thread
/builds/brebs/aports/testing/swi-prolog/src/swipl-8.5.6/src/os/pl-cstack.c:1006 C-stack dumps are not supported on this platform
PROLOG STACK:
     [11] import(prolog_debug:assertion/1)
     [10] do_autoload('/builds/brebs/aports/testing/swi-prolog/src/swipl-8.5.6/home/library/debug.pl', thread:assertion/1, prolog_debug)
      [8] setup_call_catcher_cleanup('$autoload':leave_sandbox(false), '$autoload':'$autoload3'(thread:assertion/1), _2816, '$autoload':restore_sandbox(false))
      [4] '$undefined_procedure'(thread, assertion, 1, _164)
      [3] '$c_call_prolog'
      [2] thread:assertion(all_done==all_done)
      [1] thread:ca_worker(<message_queue>(0x3ffb3804390), <message_queue>(0x3ffb38044a0), v(_26), plunit_concurrent_and:(_36 is _26*2), v(_26, _36))
      [0] '$c_call_prolog'

This had previously passed in 8.5.5

[brebs-gh]

It's passed after a no-change retry - presumably intermittent failure?

[JanWielemaker]

presumably intermittent failure?

Most likely. It does probably hint at an autoload concurrency issue, possibly one that is triggered on the s390x emulator due to its extreme timing? Possibly we can reproduce this on another platform by stressing this failed test. Bit too busy to work on that shortly though.