Closed Bug 586584 Opened 14 years ago Closed 14 years ago

run 64-bit Mac OS X test plugin out-of-process

Categories

(Core Graveyard :: Plug-ins, defect)

All
macOS
defect
Not set
normal

Tracking

(blocking2.0 beta8+)

RESOLVED FIXED
Tracking Status
blocking2.0 --- beta8+

People

(Reporter: jaas, Assigned: jaas)

References

Details

Attachments

(1 file)

We currently have a whitelist for out-of-process plugins on Mac OS X and the test plugin is not whitelisted. We do this on purpose for 32-bit builds because we want to test the Carbon event model but for 64-bit builds we should be running the test plugin out-of-process.
blocking2.0: --- → betaN+
Note that this is going to screw you in the unify step if you make the preference different and we switch to 32/64 universal builds.
The 32-bit pref will have to remain separate since we are still shipping a 32-bit build and we should still have a 32-bit tbox. So we'll either have separate prefs that match or some logic will be binary.
Assignee: joshmoz → nobody
Assignee: nobody → sgreenlay
Depends on: 596762
I think we need to be running OOPP test on Mac OS X for the feature freeze. Moving this to blocking beta 7. This should be resolved by bug 596762.
blocking2.0: betaN+ → beta7+
Fixed by patch in bug 596762.
Status: NEW → RESOLVED
Closed: 14 years ago
Resolution: --- → FIXED
The patch that enabled OOPP testing on Mac OS X was backed out because it introduced too much intermittent orange.

I have been trying to reproduce this orange on Mac OS X 10.6 for hours and I can't reproduce any of the failures we were seeing. I'm not sure what to do next, I think I need to see what is happening on an actual test slave. A quick list of tests that were know to have problems:

Bugs on tests that failed: bug 599391, bug 599378, bug 599267, bug 599087, bug 599086, bug 599076.
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Assignee: sgreenlay → joshmoz
Given the large amount of intermittent orange (that I can't reproduce) I don't see how this could make beta 7. Going to have to go back to beta 8.
blocking2.0: beta7+ → beta8+
Attached patch fix v1.0Splinter Review
I'm pretty sure the problem is the 1-second timeout for receiving the task_t mach IPC message after launching a child process. This would explain why it only happens on debug minis (they are slower) and I can never reproduce it on my fast machines. Also, the warning for that message shows up in most (all?) of the intermittent failures I've seen. This patch ups the timeout to 10 seconds and re-enables OOPP tests, confirming the fix on tryserver now.
Try server run passes with all green with fix v1.0. That seems to be it, I'll land it today.
Attachment #483076 - Flags: review?(benjamin)
Attachment #483076 - Flags: review?(benjamin) → review+
pushed to mozilla-central

http://hg.mozilla.org/mozilla-central/rev/12e53903b73f
Status: REOPENED → RESOLVED
Closed: 14 years ago14 years ago
Resolution: --- → FIXED
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.