Crash on complex example program while printing goals.

[matthewhammer]

I have a single .pl file that causes a segmentation fault after I try to suspend the program and print the goals.

The file is called ssa.pl, and is here: matthewhammer/prolog_playground@ea2f9f2

Here's what I see when I try to suspend run_example. and print the goals:

?- [ssa].
Warning: /home/matthew/prolog_playground/examples/ssa.pl:311:
Warning:    Singleton variables: [Nodes]
Warning: /home/matthew/prolog_playground/examples/ssa.pl:345:
Warning:    Singleton variables: [PhiMapOut,VCountersAfterKids,VStacksOut]
Warning: /home/matthew/prolog_playground/examples/ssa.pl:422:
Warning:    Singleton variables: [PhisHere]
Warning: /home/matthew/prolog_playground/examples/ssa.pl:480:
Warning:    Local definition of ssa:pairs_values/2 overrides weak import from pairs
true.

?- run_example.
^CAction (h for help) ? goals
[130,578,533] ssa:dict_pairs(defs{x:[entry, then1], y:[else1], z:[join]}, _15748, _15434)
     [12] ssa:insert_phis([entry, then1, else1, join], df{}, defs{x:[entry, then1], y:[else1], z:[join]}, _15314)

SWI-Prolog [thread 1 (main) at Sun Aug 17 19:26:26 2025]: received fatal signal 11 (segv)
C-stack trace labeled "crash":
  [0] PL_advance_hash_table_enum() at ??:? [0x7e2ac9559b6a]
  [1] Sdprintf() at ??:? [0x7e2ac955ea4d]
  [2] PL_changed_cwd() at ??:? [0x7e2ac9564cd7]
  [3] __restore_rt() at libc_sigaction.c:? [0x7e2ac9242520]
  [4] PL_describe_context() at ??:? [0x7e2ac952c529]
  [5] PL_interrupt() at ??:? [0x7e2ac952c93c]
  [6] Sflush() at ??:? [0x7e2ac95b40f9]
  [7] PL_get_blob() at ??:? [0x7e2ac9585943]
  [8] PL_malloc_atomic() at ??:? [0x7e2ac9571711]
  [9] PL_unify_blob() at ??:? [0x7e2ac95a46c8]
  [10] PL_unify_blob() at ??:? [0x7e2ac95a456c]
  [11] PL_toplevel() at ??:? [0x7e2ac95bbe6f]
  [12] swipl(+0x11a5) [0x5e4d9a1681a5]
  [13] __libc_start_call_main() at ./csu/../sysdeps/nptl/libc_start_call_main.h:58 [0x7e2ac9229d90]
  [14] call_init() at ./csu/../csu/libc-start.c:128 [0x7e2ac9229e40]
  [15] swipl(+0x10c5) [0x5e4d9a1680c5]
Prolog stack:
  [130578533] ssa:dict_pairs/3 [PC=0 in supervisor]
  [12] ssa:insert_phis/4 [PC=12 in clause 1]
  [11] ssa:ssa_convert/2 [PC=95 in clause 1]
  [10] ssa:run_example/0 [PC=10 in clause 1]
  [9] $toplevel:toplevel_call/1 [PC=3 in clause 1]
  [8] $toplevel:stop_backtrace/1 [PC=4 in clause 1]
  [7] $tabling:$wfs_call/2 [PC=17 in clause 1]
  [5] $toplevel:$execute_goal2/3 [PC=29 in clause 1]
  [3] $toplevel:$query_loop/0 [PC=39 in clause 2]
  [2] $toplevel:$runtoplevel/0 [PC=19 in clause 1]
  [1] $toplevel:$toplevel/0 [PC=3 in clause 1]
  [0] system:$c_call_prolog/0 [PC=0 in top query clause]
Running on_halt hooks with status 139
Killing 291220 with default signal handlers
Segmentation fault (core dumped)

[JanWielemaker]

Thanks. Without.a detailed instruction to reproduce this is of little value. In particular as the stack trace is mostly bogus, probably due to lacking symbols (release version on MacOS?)

[matthewhammer]

Thanks.

$ swipl --version
SWI-Prolog version 8.4.2 for x86_64-linux

Missing symbols in stack trace -- I'm on Ubuntu; I can list what I get from apt-cache show swi-prolog if that's more valuable. Please just let me know how to provide something valuable, and I can try to do so.

Reproduction --

• load that file that I linked (ssa.pl),
• do that predicate run_example,
• wait a second or two, or several, ...
• suspend it, and try to list the goals

[JanWielemaker]

I missed there was the source. Sorry. It doesn't reproduce (tried 5 times), which is not so strange as we are on 9.3.28 these days. You could try the PPA or build from source to get a more recent version.

As for the crash results, the output is generally useful, but only if the program is not stripped. I do not have control over how the binaries are generated on the many platforms. It might be useful to add printing the version when printing a crash report.

[kamahen]

@matthewhammer wrote:

despite my initial best efforts, it was "of little value". I feel disappointed in how this community responds if this response is representative of it.

My experience is that this community responds far better than almost any software group I have encountered, whether free or paid. Most of the discussion (including some bug reports) is in https://swi-prolog.discourse.group/

However, many of the community members are not native English speakers, and responses can seem to be rude that weren't intended that way. In this case, "of little value" simply refers to the stack trace that unfortunately contains little information of value because it lacks symbols.

I no longer use the PPA because I primarily use Debian (on a Chromebook) and the Debian maintainers are quite conservative with updates; instead, I build from source with symbols on (and install it locally using the flag -DCMAKE_INSTALL_PREFIX=$HOME/.local ... SWi-Prolog and emacs are the only open software whose quality and response to bug reports are such that I use the latest "experimental" release.