[Octave-bug-tracker] [bug #53733] Panic/segfault when calling Java method after forge:im[read|write]

[Adrian]

Follow-up Comment #34, bug #53733 (project octave):

Checking back, the single out of four GNU/Linux systems where the
originally reported bug did not seem to occur (see comment #1) was in
fact not as up-to-date as the others: its kernel was at 4.14 and
octave at 4.2.1. It executed the instructions cited at the start of
comment #1 without error. I did unfortunately not think of running it
from within gdb to see if that changed things. Once I selectively
upgraded octave to 4.2.2[*], still running the 4.14 kernel, this
system also exhibited the bug reported in comment #1.

So with this correction, the bug seems to be totally reproducible with
octave 4.2.2 on four Debian systems (1 stretch, 3 buster). When
running octave under gdb, the bug is a SIGSEGV signal upon execution
of the fourth and last line of the following code:


javaaddpath(".")
imwrite(zeros(1000), "foo.png");
graph = javaObject("Baz");
graph.addNodes(1:1000000);


When octave is run directly without attaching gdb (in GUI or CLI
mode), the bug also occurs. but on two systems leads to an application
crash with the signal 11 / SIGSEGV reported by the controlling
terminal, while on two other systems the interface freezes (the
octave-gui process ramps up to 100% CPU usage until killed).

So it seems that this bug could be (at least on Debian systems) a
regression in octave 4.2.2, not occuring on octave 4.0.3 and 4.2.1.

Let me know if some other test could help shed more light. I'll
upgrade gdb to the version used by Dmitri to see if that changes
things.


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?53733>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/