1599083 - privacy.firstparty.isolate;true: Crash in [@ IPCError-browser | invalid ClientSourceParent actor] when opening https://[::1]/

Status: VERIFIED FIXED

(Core :: DOM: Security, defect, P1)

Description

[Darkspirit]

STR: Open https://[::1]/ with privacy.firstparty.isolate enabled, get a tab crash.

mozregression --good 2019-11-01 --bad 2019-11-25 --pref privacy.firstparty.isolate:true -a https://[::1]/

8:52.19 INFO: Last good revision: f609d410b6621f3e82c65054c958e62e45d4ea13
8:52.19 INFO: First bad revision: 9f6892c03bd8dfcc604461ea90e71430c220fb74
8:52.19 INFO: Pushlog:
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=f609d410b6621f3e82c65054c958e62e45d4ea13&tochange=9f6892c03bd8dfcc604461ea90e71430c220fb74

9f6892c03bd8dfcc604461ea90e71430c220fb74 Mihai Alexandru Michis — Backed out changeset 14d6dd504f86 (bug 1534339) for causing Bug 1595179. CLOSED TREE
3a3dda1714d5fb4b818e0a78cc8291b6c98f056f Mihai Alexandru Michis — Backed out 2 changesets (bug 1585661) for causing lint failures.

Last good and first bad are correct. ?!

This bug is for crash report bp-08d13f23-06a6-4fac-ab01-e72bb0191125.

Top 10 frames of crashing thread:

0 libc-2.29.so __poll /build/glibc-suXNNi/glibc-2.29/io/../sysdeps/unix/sysv/linux/poll.c:29
1 libxul.so PollWrapper widget/gtk/nsAppShell.cpp:58
2 libglib-2.0.so.0.6200.2 g_main_context_iterate ../../../glib/gmain.c:3912
3 libglib-2.0.so.0.6200.2 <name omitted> ../../../glib/gmain.c:3978
4 libxul.so nsBaseAppShell::OnProcessNextEvent widget/nsBaseAppShell.cpp:259
5 libxul.so nsThread::ProcessNextEvent xpcom/threads/nsThread.cpp:1140
6 libxul.so <name omitted> xpcom/threads/nsThreadUtils.cpp:486
7 libxul.so mozilla::ipc::MessagePump::Run ipc/glue/MessagePump.cpp:110
8 libxul.so MessageLoop::Run ipc/chromium/src/base/message_loop.cc:290
9 libxul.so nsBaseAppShell::Run widget/nsBaseAppShell.cpp:137

Comment 1

[Gary Chen [:xeonchen]]

Seems correct, but it's unclear to me why this would happen for the backout.
Could be affected by another revision, but need further investigation.

Comment 2

[Gary Chen [:xeonchen]]

I just re-run the regression with wider range and try to find if the issue was there before.
And it shows it was caused by bug 1583700, which enabled DocumentChannel by default.

Comment 3

[Gary Chen [:xeonchen]]

Attachment: Bug 1599083 - remove old ClientSource from ClientManagerService before adding new one; r=mattwoodrow

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ce5c3283167c
remove old ClientSource from ClientManagerService before adding new one; r=mattwoodrow

Comment 5

[Raul Gurzau (:RaulG)]

https://hg.mozilla.org/mozilla-central/rev/ce5c3283167c

Comment 6

[Darkspirit]

Verified fixed. Thank you :)