TEST: ogr_geom_build_from_edges_4 ... Segmentation fault

[peifer]

# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x00007f62864a9c8e, pid=24083, tid=140061169284864
#
# JRE version: 6.0_23-b23
# Java VM: OpenJDK 64-Bit Server VM (20.0-b11 mixed mode linux-amd64 compressed oops)
# Derivative: IcedTea6 1.11pre
# Distribution: Debian GNU/Linux unstable (sid), package 6b23~pre11-1
# Problematic frame:
# C  [libgdal.so.1+0x859c8e]  CheckPoints(OGRLineString*, int, OGRLineString*, int, double*)+0xbe
#

[warmerdam]

I'm confused. The pasted report looks like it is being run from within a Java environment, but our test suite is normally run from a python script. Is this some Java specific autotest? Can you be more specific about what you ran?

I suspect this failure relates to Even's recent work in r23323.

[peifer]

What I ran: ./run_all.py

I've been running the autotests a couple of days ago and things were fine. Now I bumped into the segfault, after updating from svn.

[Even Rouault]

Hermann,

Did you also rebuild GDAL ? I highly suspect you didn't because it works fine on my Linux 32 et 64bit builds, on all Tamas Windows bots

It is expected that the update test segfault is you run it against not the latest version. If you are sure to run the latest build of GDAL, then please run from autotest/ogr,

valgrind --trace-children=yes python ogr_geom.py

Actually, the commit that would cause the crash is that one : r23327 : OGRBuildPolygonFromEdges(): fix crash on a LINESTRING EMPTY. It fixes the crash and adds a test case that would cause the crash on a non-updated GDAL version.

(The Java trackbace could come if you have the MDB driver configured with the supporting Java lib. The JVM installs its segfault handler and if the ogr_pgeo.py test is run before, it will catch the segfault.)

[peifer]

Silly me. I thought about rebuilding ... but then forgot to do so :-(
After rebuilding, the test result is: success. Sorry for the false alarm.