Disable some tests intermittently failing with sessionHistory and Fission
Categories
(Core :: DOM: Navigation, task, P3)
Tracking
()
Fission Milestone | M6c |
People
(Reporter: smaug, Assigned: smaug)
References
Details
(Whiteboard: [stockwell disabled])
Attachments
(2 files)
Assignee | ||
Comment 1•4 years ago
|
||
Comment 4•4 years ago
|
||
Olli, I disabled browser_privilegedmozilla_process_pref.js as well.
Updated•4 years ago
|
Updated•4 years ago
|
Updated•4 years ago
|
Updated•4 years ago
|
Comment 10•4 years ago
|
||
bugherder |
https://hg.mozilla.org/mozilla-central/rev/6b89ff1b983e
https://hg.mozilla.org/mozilla-central/rev/304e15dfa979
https://hg.mozilla.org/mozilla-central/rev/455ac1cc7e68
https://hg.mozilla.org/mozilla-central/rev/81146bda0b51
Comment 11•4 years ago
|
||
bugherder |
Updated•4 years ago
|
Updated•4 years ago
|
Updated•4 years ago
|
Comment 12•4 years ago
|
||
Updated•4 years ago
|
Updated•4 years ago
|
Updated•4 years ago
|
Comment 13•4 years ago
|
||
Comment 14•4 years ago
|
||
bugherder |
Updated•4 years ago
|
Updated•4 years ago
|
Comment 15•4 years ago
|
||
Tom, this browser/base/content/test/tabs/browser_new_tab_in_privilegedabout_process_pref.js test is currently skipped on Fission with a comment saying: "Pref and test only relevant for e10s".
Is this test relevant for Fission? I don't know if "e10s" in this comment means to differentiate between "pre-e10s" or "pre-Fission". The browser.tabs.remote.separatePrivilegedMozillaWebContentProcess
pref is currently enabled by default, with or without Fission.
[browser_new_tab_in_privilegedabout_process_pref.js]
skip-if = !e10s || (os == 'linux' && debug && bits == 64) || fission # Pref and test only relevant for e10s, Bug 1581500. Bug 1668809
Comment 16•4 years ago
|
||
So the intent of the Privileged Mozilla Content Process (PMCP) is to put FXA and AMO into a separate content process and then reject certain capabilities if a content process requests them and it's not the PMCP. Patch 1 Patch 2.
With Fission we can change that check to say "You have to be a content process for FXA/AMO" and eliminate the notion of the PMCP - the only difference between a Content Process and a PMCP is the checks in those two patches.
The real question is does Fission work on those sites using those capabilities when the PMCP is enabled? Or does Fission mess up the notion of the PMCP. If those sites were broken I'm sure someone'd have noticed, so maybe the question is - do those sites go into Fission Content Processes with Fission or do they go into one PMCP and share a process? If they do, I don't think it's urgent that we fix it, but we should put a bug on file to clean this stuff up.
In conclusion, I think it is probably relevant for Fission because if the test fails it could indicate a way that FXA and AMO may fail with Fission.
Comment 17•4 years ago
|
||
(In reply to Tom Ritter [:tjr] (ni? for response to sec-[advisories/bounties/ratings/cves]) from comment #16)
With Fission we can change that check to say "You have to be a content process for FXA/AMO" and eliminate the notion of the PMCP - the only difference between a Content Process and a PMCP is the checks in those two patches.
The real question is does Fission work on those sites using those capabilities when the PMCP is enabled? Or does Fission mess up the notion of the PMCP. If those sites were broken I'm sure someone'd have noticed, so maybe the question is - do those sites go into Fission Content Processes with Fission or do they go into one PMCP and share a process? If they do, I don't think it's urgent that we fix it, but we should put a bug on file to clean this stuff up.
Nika, can you answer Tom's questions about Fission and PMCP?
I don't know if the browser_new_tab_in_privilegedabout_process_pref.js test currently passes with Fission or not. It's currently skipped for !e10s || fission
.
Comment 18•4 years ago
|
||
This test is almost certainly failing for the same reason as some of the other tests like it, it is waiting for a process switch to happen using SSTabRestored
here: https://searchfox.org/mozilla-central/rev/31ddf859c57e812878a5f817e4097efb06de4d97/browser/base/content/test/tabs/browser_new_tab_in_privilegedabout_process_pref.js#169,185. This event never fires anymore, as we don't use sessionstore to perform process switches anymore. These can easily be fixed by just waiting for BrowserTestUtils.browserLoaded
instead.
(In reply to Tom Ritter [:tjr] (ni? for response to sec-[advisories/bounties/ratings/cves]) from comment #16)
The real question is does Fission work on those sites using those capabilities when the PMCP is enabled? Or does Fission mess up the notion of the PMCP. If those sites were broken I'm sure someone'd have noticed, so maybe the question is - do those sites go into Fission Content Processes with Fission or do they go into one PMCP and share a process? If they do, I don't think it's urgent that we fix it, but we should put a bug on file to clean this stuff up.
They go into a single PMCP and share a process with fission. All special process allocations are not impacted with Fission, only loads which would've ended up in the web
content process.
Comment 19•4 years ago
|
||
(In reply to Nika Layzell [:nika] (ni? for response) from comment #18)
This test is almost certainly failing for the same reason as some of the other tests like it, it is waiting for a process switch to happen using
SSTabRestored
here: https://searchfox.org/mozilla-central/rev/31ddf859c57e812878a5f817e4097efb06de4d97/browser/base/content/test/tabs/browser_new_tab_in_privilegedabout_process_pref.js#169,185. This event never fires anymore, as we don't use sessionstore to perform process switches anymore. These can easily be fixed by just waiting forBrowserTestUtils.browserLoaded
instead.
In that case, I filed new bug 1686133 to track fixing and re-enabling browser_new_tab_in_privilegedabout_process_pref.js test.
Description
•