1410072 - Permaorange test-verify browser/base/content/test/performance/browser_startup.js | Uncaught exception - [Exception... "Component returned failure code: 0x804b000a (NS_ERROR_MALFORMED_URI) [xpcIJSModuleLoader.getComponentLoadStack]"

Status: RESOLVED FIXED

(Firefox :: General, defect, P5)

Description

[Treeherder Bug Filer]

Filed by: nbeleuzu [at] mozilla.com

https://treeherder.mozilla.org/logviewer.html#?job_id=138210273&repo=autoland

https://queue.taskcluster.net/v1/task/KHarTODhQMSffzvRKuD-cg/runs/0/artifacts/public/logs/live_backing.log

Intermittent browser/base/content/test/performance/browser_startup.js | Uncaught exception - [Exception... "Component returned failure code: 0x804b000a (NS_ERROR_MALFORMED_URI) [xpcIJSModuleLoader.getComponentLoadStack]"  nsresult: "0x804b000a (NS_ERROR_MALFORMED_URI)"  location: "JS frame :: chrome://mochitests/content/browser/browser/base/content/test/performance/browser_startup.js :: data[phase].components< :: line 163"  data: no]

Comment 1

[Intermittent Failures Robot]

18 failures in 146 pushes (0.123 failures/push) were associated with this bug yesterday.    

Repository breakdown:
* mozilla-central: 12
* mozilla-inbound: 6

Platform breakdown:
* windows10-64: 5
* linux64: 4
* windows7-32: 3
* osx-10-10: 2
* windows7-32-nightly: 1
* windows10-64-nightly: 1
* macosx64-nightly: 1
* linux64-nightly: 1

For more details, see:
https://brasstacks.mozilla.com/orangefactor/?display=Bug&bugid=1410072&startday=2017-10-20&endday=2017-10-20&tree=all

Comment 2

[Intermittent Failures Robot]

23 failures in 864 pushes (0.027 failures/push) were associated with this bug in the last 7 days.    

Repository breakdown:
* mozilla-central: 15
* mozilla-inbound: 6
* autoland: 2

Platform breakdown:
* windows10-64: 6
* osx-10-10: 5
* linux64: 4
* windows7-32: 3
* windows7-32-nightly: 2
* windows10-64-nightly: 1
* macosx64-nightly: 1
* linux64-nightly: 1

For more details, see:
https://brasstacks.mozilla.com/orangefactor/?display=Bug&bugid=1410072&startday=2017-10-16&endday=2017-10-22&tree=all

Comment 3

[Geoff Brown [:gbrown]]

These failures were first seen with the test-verify run triggered by bug 1347532, but I do not know if it was a pre-existing problem. browser_startup.js has several other open bugs, but none that look like this.

Comment 4

[Firefox Bug Husbandry Bot]

https://wiki.mozilla.org/Bugmasters#Intermittent_Test_Failure_Cleanup

Comment 5

[Ryan VanderMeulen [:RyanVM]]

This is trivially-reproducible on Try across all platforms still. Florian, is this something you'd be interested in looking at?

Comment 6

[Florian Quèze [:florian]]

(In reply to Ryan VanderMeulen [:RyanVM] from comment #5)
> This is trivially-reproducible on Try across all platforms still. Florian,
> is this something you'd be interested in looking at?

Do you have a link to a try build where this failed?

Comment 7

[Ryan VanderMeulen [:RyanVM]]

https://treeherder.mozilla.org/logviewer.html#?job_id=153159640&repo=try

Comment 8

[Florian Quèze [:florian]]

I wonder if https://searchfox.org/mozilla-central/rev/e6fd22c37447579bd544254685c80fd0da221d93/js/xpconnect/loader/mozJSComponentLoader.cpp#1061 needs a change for devedition too. But that would give NS_ERROR_NOT_IMPLEMENTED, not NS_ERROR_MALFORMED_URI. I guess I'll need to push this to try with more debug logging to figure out what's actually going on.

https://hg.mozilla.org/try/rev/4fe83f30ea86#l1.12 in your try push (which doesn't match the patch we actually landed in bug 1426793 seems responsible for the "This test contains no passes, no fails and no todos." errors you are getting on some platforms.

Comment 9

[Ryan VanderMeulen [:RyanVM]]

Yes, the NS_ERROR_NOT_IMPLEMENTED condition was what I hit when I amended browser_startup.js to include DevEdition in that check as well. Rather than debug the component loader code, I decided to take the easy way out and skip the test on DevEdition instead (which in effect was no worse than what we were doing before, except we're skipping it at the manifest level instead of via an early return).

Anyway, that's orthogonal to this bug :). This bug is what I hit on everything that wasn't DevEdition in the TV job, and those other platforms/flavors wouldn't have been affected by the DevEdition-only changes I was making.

Comment 10

[Florian Quèze [:florian]]

(In reply to Florian Quèze [:florian] from comment #8)
> I guess
> I'll need to push this to try with more debug logging to figure out what's
> actually going on.

Actually, after sleeping on it the cause of the malformed URIs is obvious:
https://searchfox.org/mozilla-central/rev/e6fd22c37447579bd544254685c80fd0da221d93/browser/base/content/test/performance/browser_startup.js#160 modifies the stored data, so next run can't access the full URIs anymore.

So this issue really affects only test-verify.

Comment 11

[Intermittent Failures Robot]

15 failures in 590 pushes (0.025 failures/push) were associated with this bug in the last 7 days.    

Repository breakdown:
* try: 15

Platform breakdown:
* windows7-32-devedition: 2
* windows7-32: 2
* windows10-64-devedition: 2
* osx-10-10: 2
* macosx64-devedition: 2
* linux64-devedition: 2
* linux64: 2
* windows10-64: 1

For more details, see:
https://brasstacks.mozilla.com/orangefactor/?display=Bug&bugid=1410072&startday=2017-12-18&endday=2017-12-24&tree=all

Comment 12

[Florian Quèze [:florian]]

Attachment: Patch

To pass test-verify, the startup tests should never modify the startupRecorder data. I used Cu.cloneInto because it helped performance a lot for the startup flicker test, by removing all cross compartment wrappers.

browser_startup_images.js currently doesn't really need it, but I figured it's safer to do it everywhere, so that if someone copy/pastes the startup test skeleton in the future to add a new test, they don't reproduce the mistake I made in browser_startup.js

Try seems happy: https://treeherder.mozilla.org/#/jobs?repo=try&revision=4bb03db3ebea2b8419b43ae86a6175b6d15c66fd
The TV failures on this try push are from other changesets that landed on inbound recently but hadn't been pushed to try yet.

Comment 13

[Johann Hofmann [:johannh]]

Comment on attachment 8939885 [details] [diff] [review]
Patch

Review of attachment 8939885 [details] [diff] [review]:
-----------------------------------------------------------------

Thanks!

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/eaf8243e6e33
browser/base/content/test/performance tests should not modify startupRecorder data in order to pass test-verify, r=johannh.

Comment 15

[Tiberius Oros[:tiberius_oros]]

https://hg.mozilla.org/mozilla-central/rev/eaf8243e6e33