Bug #943661 “apport-gtk crashed with SIGSEGV in get_gsubgpos_tab...” : Bugs : apport package : Ubuntu

Revision history for this message

Jean-Louis Dupond (dupondje) wrote on 2012-02-29:

#1

ApportLog.txt Edit (1.4 KiB, text/plain; charset="utf-8")
Dependencies.txt Edit (4.5 KiB, text/plain; charset="utf-8")
Disassembly.txt Edit (606 bytes, text/plain; charset="utf-8")
ProcMaps.txt Edit (48.9 KiB, text/plain; charset="utf-8")
ProcStatus.txt Edit (830 bytes, text/plain; charset="utf-8")
Registers.txt Edit (763 bytes, text/plain; charset="utf-8")
Stacktrace.txt Edit (8.2 KiB, text/plain; charset="utf-8")
ThreadStacktrace.txt Edit (15.3 KiB, text/plain; charset="utf-8")

visibility:

private → public

Revision history for this message

Apport retracing service (apport) wrote on 2012-02-29:

#2

StacktraceTop:
get_gsubgpos_table (face=0x0, table_tag=1196643650) at /build/buildd/pango1.0-1.29.5/./pango/opentype/hb-ot-layout.cc:559
hb_ot_layout_table_find_script (face=<optimized out>, table_tag=<optimized out>, script_tag=1818326126, script_index=0x7f9699e67ce4) at /build/buildd/pango1.0-1.29.5/./pango/opentype/hb-ot-layout.cc:361
pango_ot_ruleset_new_for (info=<optimized out>, script=<optimized out>, language=<optimized out>) at /build/buildd/pango1.0-1.29.5/./pango/pango-ot-ruleset.c:213
pango_ot_ruleset_new_from_description (info=<optimized out>, desc=0x7f9699e67da0) at /build/buildd/pango1.0-1.29.5/./pango/pango-ot-ruleset.c:259
pango_ot_ruleset_get_for_description (info=0x17c88a0, desc=0x7f9699e67da0) at /build/buildd/pango1.0-1.29.5/./pango/pango-ot-ruleset.c:109

Revision history for this message

Apport retracing service (apport) wrote on 2012-02-29: Stacktrace.txt

#3

Stacktrace.txt Edit (29.6 KiB, text/plain)

Revision history for this message

Apport retracing service (apport) wrote on 2012-02-29: ThreadStacktrace.txt

#4

ThreadStacktrace.txt Edit (58.2 KiB, text/plain)

Changed in apport (Ubuntu):
importance:	Undecided → Medium
summary:	- apport-gtk crashed with SIGSEGV in pango_ot_ruleset_new_for() + apport-gtk crashed with SIGSEGV in get_gsubgpos_table()
tags:	removed: need-amd64-retrace

Revision history for this message

Martin Pitt (pitti) wrote on 2012-03-01:

#5

I can reproduce with "apport-bug libreoffice".

Changed in apport (Ubuntu):
status:	New → Triaged

Martin Pitt (pitti) on 2012-03-01

tags:

added: whoopsie-merge

Revision history for this message

Martin Pitt (pitti) wrote on 2012-03-02:

#6

Fixed in trunk r2198.

Changed in apport (Ubuntu):
assignee:	nobody → Martin Pitt (pitti)
status:	Triaged → Fix Committed

Revision history for this message

Launchpad Janitor (janitor) wrote on 2012-03-02:

#7

Download full text (5.5 KiB)

This bug was fixed in the package apport - 1.94-0ubuntu1

---------------
apport (1.94-0ubuntu1) precise; urgency=low

  [ Martin Pitt ]
  * New upstream release:
    - apport: Set the group of written reports to "whoopsie" if that group
      exists.
    - Fix tests to run properly against the system-installed modules and
      binaries.
    - test/run: Run under LC_MESSAGES=C to avoid test failures due to
      translated strings.
    - general-hooks/generic.py: Also attach xsession-errors for programs that
      link to libgtk-3.
    - launchpad.py: Properly handle "Expired" status, to avoid marking new
      bugs as duplicates of expired ones. (LP: #941854)
    - apport: Fix crash if the "whoopsie" group does not exist. (LP: #942326)
    - report.py, crash_signature(): Do not put "<module>" frames into Python
      crash signatures that happen outside of function/method calls. Fall back
      to the file/line number as a frame description instead. This will do a
      much better job at disambiguating e. g. different ImportError crashes.
      (LP: #920403)
    - Make "binary changed since the time of the crash" error message more
      comprehensible, thanks Paolo Rotolo. (LP: #942830)
    - crashdb.py, check_duplicate(): It can happen that a bug gets identified
      as being a duplicate of bug S by symbolic signatures and a duplicate of
      bug A by address signatures. Empirical evidence shows that this is due
      to the unavoidable jitter in stack traces (A and S not being identified
      as duplicates as their signatures differ slightly) and not a logic
      error. So instead of erroring out, duplicate all three bugs and keep the
      lowest number as the master ID. (LP: #943117)
    - Revert the usage of multiple nested threads during data collection, and
      switch back to only using one UI thread. The UI implementations can, and
      now do, decide between showing a spinner and showing a progress dialog
      in the ui_*_info_collection_progress() methods. This fixes libX11
      crashes when multiple UI threads do changes concurrently (LP: #901675),
      and also avoids multi-thread induced crashes in Pango (LP: #943661). The
      removal of the collect() method also fixes the new crashes in it.
      (LP: #942098, #939803)
    - ui.py, get_desktop_entry(): Fix crash on uninstalled package.
      (LP: #940984)
    - data/unkillable_shutdown: Fix crash on race condition when PID goes away
      while the report is created. (LP: #546369)
    - apport/hookutils.py, pci_devices(): Fix crash on unexpected lines from
      lspci. (LP: #904489)
    - Drop hardcoded "Ubuntu" words again which crept in with the whoopsie
      support merge. Use the DistroRelease: field.
    - apport-kde: Fix Home page URL in KApplication metadata.
    - apport-gtk: Fix resizability and size after hiding details.
      (LP: #405418)
    - test/run: Drop "local" argument. This now tests against the source tree
      when run in the source tree root, and against the system
      libraries/programs when run from anywhere else.
    - test/run: Consider command line arguments as test names and only run
      those when given. Also support just runni...

This bug was fixed in the package apport - 1.94-0ubuntu1

---------------
apport (1.94-0ubuntu1) precise; urgency=low

[ Martin Pitt ]
  * New upstream release:
    - apport: Set the group of written reports to "whoopsie" if that group
      exists.
    - Fix tests to run properly against the system-installed modules and
      binaries.
    - test/run: Run under LC_MESSAGES=C to avoid test failures due to
      translated strings.
    - general-hooks/generic.py: Also attach xsession-errors for programs that
      link to libgtk-3.
    - launchpad.py: Properly handle "Expired" status, to avoid marking new
      bugs as duplicates of expired ones. (LP: #941854)
    - apport: Fix crash if the "whoopsie" group does not exist. (LP: #942326)
    - report.py, crash_signature(): Do not put "<module>" frames into Python
      crash signatures that happen outside of function/method calls. Fall back
      to the file/line number as a frame description instead. This will do a
      much better job at disambiguating e. g. different ImportError crashes.
      (LP: #920403)
    - Make "binary changed since the time of the crash" error message more
      comprehensible, thanks Paolo Rotolo. (LP: #942830)
    - crashdb.py, check_duplicate(): It can happen that a bug gets identified
      as being a duplicate of bug S by symbolic signatures and a duplicate of
      bug A by address signatures. Empirical evidence shows that this is due
      to the unavoidable jitter in stack traces (A and S not being identified
      as duplicates as their signatures differ slightly) and not a logic
      error. So instead of erroring out, duplicate all three bugs and keep the
      lowest number as the master ID. (LP: #943117)
    - Revert the usage of multiple nested threads during data collection, and
      switch back to only using one UI thread. The UI implementations can, and
      now do, decide between showing a spinner and showing a progress dialog
      in the ui_*_info_collection_progress() methods. This fixes libX11
      crashes when multiple UI threads do changes concurrently (LP: #901675),
      and also avoids multi-thread induced crashes in Pango (LP: #943661). The
      removal of the collect() method also fixes the new crashes in it.
      (LP: #942098, #939803)
    - ui.py, get_desktop_entry(): Fix crash on uninstalled package.
      (LP: #940984)
    - data/unkillable_shutdown: Fix crash on race condition when PID goes away
      while the report is created. (LP: #546369)
    - apport/hookutils.py, pci_devices(): Fix crash on unexpected lines from
      lspci. (LP: #904489)
    - Drop hardcoded "Ubuntu" words again which crept in with the whoopsie
      support merge. Use the DistroRelease: field.
    - apport-kde: Fix Home page URL in KApplication metadata.
    - apport-gtk: Fix resizability and size after hiding details.
      (LP: #405418)
    - test/run: Drop "local" argument. This now tests against the source tree
      when run in the source tree root, and against the system
      libraries/programs when run from anywhere else.
    - test/run: Consider command line arguments as test names and only run
      those when given. Also support just running a single test.
    - testsuite: Force the skipping of online tests when $SKIP_ONLINE_TESTS is
      set.
    - hookutils.py, xsession_errors(): Add a reasonable default pattern which
      matches glib-style warnings, errors, criticals etc. and X window errors.
      In data/general-hooks/generic.py, call it with that default instead of
      the rather incomplete custom pattern. (LP: #932660)
    - packaging.py: Add get_package_origin() method, and implement it for
      apt-dpkg.
    - report.py, add_package_info(): Add "[origin: ...]" tag to "Package" and
      "Dependencies" fields for any package which is not native to the
      distribution. If any such package is present, tag the report with
      "third-party-packages" in data/general-hooks/generic.py. (LP: #927912)
    - apport/packaging.py: Add get_uninstalled_package() method as a helper
      method for the test suite. Use it instead of a hardcoded Debian/Ubuntu
      specific name in test/test_hooks.py.
    - test/test_ui_{gtk,kde}.py: Add test cases for complete UI workflow runs
      for reporting a bug against an installed/uninstalled package, and
      reporting a crash with and without showing details. This reproduces the
      recent crashes like LP #901675 or LP #943661.
    - test_ui.py: Add a test case for reporting a complete report on
      uninstalled package. This happens when reporting a problem from a
      different machine through copying a .crash file.
    - test/run: Add a test that there are no hardcoded "Ubuntu" words in the
      source. The code should use the DistroRelease: field or lsb_release.
  * debian/apport-retrace.install: Ship crash-digger. There is no reason any
    more not to, as it's now very easy to set up a retracer bot environment.
  * debian/apport.install: Install crash.{class,jar} into test suite
    directory, so that the test_java_crashes.py test can run.
  * debian/control: Tighten dependencies to ensure that we don't run a newer
    UI package against an older python-apport, which would cause crashes due
    to API mismatches. (LP: #939702)
  * Drop test/test_backend_rpm.py, as we also drop the RPM backend in the
    Ubuntu branch.
  * debian/rules: Update test suite invocation, the "local" argument is
    obsolete.

[ Brian Murray ]
  * data/package-hooks/source_linux.py: add in ProcFB
 -- Martin Pitt <martin.pitt@ubuntu.com>   Fri, 02 Mar 2012 15:32:07 +0100

Changed in apport (Ubuntu):
status:	Fix Committed → Fix Released

Ubuntu
apport package

apport-gtk crashed with SIGSEGV in get_gsubgpos_table()

Bug Description

Related branches

Other bug subscribers

Bug attachments

Remote bug watches

Ubuntuapport package

apport-gtk crashed with SIGSEGV in get_gsubgpos_table()

Bug Description

Related branches

Other bug subscribers

Bug attachments

Remote bug watches

Ubuntu
apport package