segfault in libwebkitgtk-3.0.so.0 on amd64

Bug #800741 reported by Curtis Hovey
4
This bug affects 1 person
Affects Status Importance Assigned to Milestone
HTML5-browser
Fix Released
High
Curtis Hovey
Launchpad itself
Fix Released
High
Curtis Hovey
webkit (Ubuntu)
New
Undecided
Unassigned

Bug Description

natty amd64 users report that running that Lp test runner will cause a segault
.bin/test -1cvvv --layer=YUI

The segfault is in the JIT code.

Revision history for this message
Curtis Hovey (sinzui) wrote :
Revision history for this message
Curtis Hovey (sinzui) wrote :

This test runner passes on Lucid amd64, but it is using the oldest combination of the import matrix: python2.6, pygtk, webkit 1.0. The natty users are python2.6, gir, webkit 3.0

description: updated
Curtis Hovey (sinzui)
Changed in launchpad:
status: New → In Progress
importance: Undecided → High
Revision history for this message
Curtis Hovey (sinzui) wrote :

natty amd64, python2.6, pygtk, webkit 1.0 also segfaults. the natty amd64 has a JIT issue in both 1.0 and 3.0 libs.

Revision history for this message
Curtis Hovey (sinzui) wrote :

The segefault is caused by repeated calls to start and stop the gtk main_loop in a single python process. WebKit does not like the repeated start and stops in main_loop, maybe because some state is left behind.

The test browser is stable called repeatedly in a subprocess. This is slower than working the browser in proc.

Changed in html5-browser:
status: In Progress → Fix Committed
Curtis Hovey (sinzui)
Changed in html5-browser:
status: Fix Committed → Fix Released
Changed in launchpad:
assignee: nobody → Curtis Hovey (sinzui)
status: In Progress → Fix Committed
Curtis Hovey (sinzui)
Changed in launchpad:
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.