Closed
Bug 1460610
Opened 7 years ago
Closed 7 years ago
Get rid of workerbootstrap-test@mozilla.org.xpi
Categories
(Testing :: Mochitest, defect)
Tracking
(firefox62 fixed)
RESOLVED
FIXED
mozilla62
Tracking | Status | |
---|---|---|
firefox62 | --- | fixed |
People
(Reporter: baku, Assigned: kmag)
References
Details
Attachments
(1 file)
I tried to remove this xpi in bug 1455956 but, somehow, SpecialPowers.jsm depends on it. This file is not used anymore and it can be removed.
Assignee | ||
Comment 1•7 years ago
|
||
I just ran into this too. It's not just specialpowers. Other parts of the mochitest harness fail, too. Apparently their frame scripts don't get loaded into new content process when the extension isn't there.
They don't depend on anything the extension does. Just replacing it with a skeleton extension with only an install.rdf fixes the problem.
Assignee | ||
Comment 2•7 years ago
|
||
OK. I see the problem now.
Without the workerbootstrap-test extension, we don't have any extensions installed in the profile during early startup, which means that the profile extensions directory doesn't exist. Since it doesn't exist, it doesn't get added to the content sandbox filesystem whitelist, which means that profile extension contents are not available in sandboxed content processes.
Assignee: nobody → kmaglione+bmo
Comment hidden (mozreview-request) |
Comment 4•7 years ago
|
||
mozreview-review |
Comment on attachment 8979063 [details]
Bug 1460610: Fix content process whitelist for profile extensions when none are present at startup.
https://reviewboard.mozilla.org/r/245324/#review252394
Can we just fix the sandboxing code to unconditionally whitelist the extensions directory?
Assignee | ||
Comment 5•7 years ago
|
||
mozreview-review-reply |
Comment on attachment 8979063 [details]
Bug 1460610: Fix content process whitelist for profile extensions when none are present at startup.
https://reviewboard.mozilla.org/r/245324/#review252394
No, I'm pretty sure we can't whitelist directories that don't already exist.
Comment 6•7 years ago
|
||
mozreview-review-reply |
Comment on attachment 8979063 [details]
Bug 1460610: Fix content process whitelist for profile extensions when none are present at startup.
https://reviewboard.mozilla.org/r/245324/#review252394
well that's a bummer...
Comment 7•7 years ago
|
||
mozreview-review |
Comment on attachment 8979063 [details]
Bug 1460610: Fix content process whitelist for profile extensions when none are present at startup.
https://reviewboard.mozilla.org/r/245324/#review252410
Attachment #8979063 -
Flags: review?(aswan) → review+
Assignee | ||
Comment 8•7 years ago
|
||
https://hg.mozilla.org/integration/mozilla-inbound/rev/120d17a176ffa033fe79632804755ae23940c336
Bug 1460610: Fix content process whitelist for profile extensions when none are present at startup. r=aswan
Assignee | ||
Comment 9•7 years ago
|
||
https://hg.mozilla.org/integration/mozilla-inbound/rev/7398284f1d7f62748dcc7eb556a24b5e0ee11bba
Bug 1460610: Follow-up: Fix test expectations. r=bustage DONTBUILD
Comment 10•7 years ago
|
||
bugherder |
https://hg.mozilla.org/mozilla-central/rev/120d17a176ff
https://hg.mozilla.org/mozilla-central/rev/7398284f1d7f
Status: NEW → RESOLVED
Closed: 7 years ago
status-firefox62:
--- → fixed
Resolution: --- → FIXED
Target Milestone: --- → mozilla62
Comment 11•7 years ago
|
||
Interestingly this causes a problem in Telemetry xpcshelltests. Specifically,
./mach test toolkit/components/telemetry/tests/unit
Now fails with the following message:
Error running mach:
['test', 'toolkit/components/telemetry/tests/unit']
The error occurred in code that was called by the mach command. This is either
a bug in the called code itself or in the way that mach is calling it.
You should consider filing a bug for this issue.
If filing a bug, please include the full output of mach, including this error
message.
The details of the failure are as follows:
ErrorMessage: Symlink target path does not exist: /home/chutten/mozilla-central/dom/workers/test/extensions/bootstrap/workerbootstrap-test@mozilla.org.xpi
File "/home/chutten/mozilla-central/testing/mach_commands.py", line 167, in test
argv=extra_args, test_objects=tests, **kwargs)
File "/home/chutten/mozilla-central/python/mach/mach/registrar.py", line 129, in dispatch
return self._run_command_handler(handler, context=context, **kwargs)
File "/home/chutten/mozilla-central/python/mach/mach/registrar.py", line 90, in _run_command_handler
result = fn(**kwargs)
File "/home/chutten/mozilla-central/testing/xpcshell/mach_commands.py", line 229, in run_xpcshell_test
driver.install_tests(test_objects)
File "/home/chutten/mozilla-central/python/mozbuild/mozbuild/controller/building.py", line 1341, in install_tests
'_tests', test_objs)
File "/home/chutten/mozilla-central/python/mozbuild/mozbuild/testing.py", line 267, in install_test_files
remove_unaccounted=False)
File "/home/chutten/mozilla-central/python/mozbuild/mozpack/copier.py", line 431, in copy
copy_results.append((destfile, f.copy(destfile, skip_if_older)))
File "/home/chutten/mozilla-central/python/mozbuild/mozpack/files.py", line 336, in copy
raise ErrorMessage('Symlink target path does not exist: %s' % self.path)
What makes this _extra_ interesting is that
./mach xpcshelltest toolkit/components/telemetry/tests/unit
runs just fine.
Flags: needinfo?(kmaglione+bmo)
Comment 12•7 years ago
|
||
:chutten, does a clobber fix the problem reported in comment 11?
Comment 13•7 years ago
|
||
Sorry for the delay.
Yes, it seems a clobber takes care of it.
Comment 14•7 years ago
|
||
I can't explain the "mach test" vs "mach xpcshell-test" difference but this is pretty typical when something is removed from the build. I'm not sure how we should handle this properly, touching CLOBBER would mean a bunch of recompilation for people not working with artifact builds who may not be affected by this...
Flags: needinfo?(kmaglione+bmo)
Comment 15•7 years ago
|
||
|mach test| does some extra querying into the build system that the individual test commands don't need to do, so it's not terribly surprising.
I guess it depends how widespread this issue is, but IMO it's usually better to touch CLOBBER when things like this arise. The time developers will spend trying to figure this issue out will likely exceed the time they would waste waiting for a CLOBBER build (especially since they can do other things while compiling).
You need to log in
before you can comment on or make changes to this bug.
Description
•