diff options
| author | eroman@chromium.org <eroman@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2010-05-05 23:25:29 +0000 |
|---|---|---|
| committer | eroman@chromium.org <eroman@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2010-05-05 23:25:29 +0000 |
| commit | 6627ef6ed1e99731c6c2f2e3a87179fc6448790e (patch) | |
| tree | 7bec24d1a585fb4666f79b9fcbb87bd7333cc202 /third_party/npapi/bindings | |
| parent | ea14138640443f60808ebb388d7dbdba630252d6 (diff) | |
| download | chromium_src-6627ef6ed1e99731c6c2f2e3a87179fc6448790e.zip chromium_src-6627ef6ed1e99731c6c2f2e3a87179fc6448790e.tar.gz chromium_src-6627ef6ed1e99731c6c2f2e3a87179fc6448790e.tar.bz2 | |
Fix a unit-test that was crashing on valgrind bot.
The problem is that ClientSocketPoolBaseHelper holds a pointer to a task, and tries to call a method on that pointer during destruction.
However, if destruction of ClientSocketPoolBaseHelper occurs during message loop destruction, then pending tasks are deleted, so this "backup_task" pointer that it accesses may be invalid.
I worked around the problem here by avoiding this destruction code-path in the unit-test. It would be good to make ClientSocketPoolBaseHelper able to handle this case though.
BUG=43291
Review URL: http://codereview.chromium.org/1920003
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@46521 0039d316-1c4b-4281-b951-d872f2087c98
Diffstat (limited to 'third_party/npapi/bindings')
0 files changed, 0 insertions, 0 deletions