594487 - [SeaMonkey] leak test: "Exception / NS_ERROR_XPC_GS_RETURNED_FAILURE / nsIJSCID.getService / messageWakeupService.js :: anonymous :: line 56"

Status: VERIFIED FIXED

(Core :: IPC, defect)

Description

[Serge Gautherie (:sgautherie)]

{
* Call to xpconnect wrapped JSObject produced this error:  *
[Exception... "Component returned failure code: 0x80570016
(NS_ERROR_XPC_GS_RETURNED_FAILURE) [nsIJSCID.getService]"  nsresult:
"0x80570016 (NS_ERROR_XPC_GS_RETURNED_FAILURE)"  location: "JS frame ::
file:///builds/slave/comm-central-trunk-macosx-debug/build/objdir/mozilla/dist/SeaMonkeyDebug.app/Contents/MacOS/components/messageWakeupService.js
:: anonymous :: line 56"  data: no]
}

Fwiw, KaiRo seemed to wonder whether this should be active at all in current SeaMonkey setup...

Comment 1

[Serge Gautherie (:sgautherie)]

Code is
{
53   get messageManager() {
54     if (!this._messageManager)
55       this._messageManager = Cc["@mozilla.org/parentprocessmessagemanager;1"].
56                              getService(Ci.nsIFrameMessageManager);
57     return this._messageManager;
58   },
}

Comment 2

[Alon Zakai (:azakai)]

I'm not familiar with how SeaMonkey differs from stock Firefox, but must be some difference that is the issue. Is MOZ_IPC enabled there? Is the messagemanager stuff usable? And the parentprocessmessagemanager in particular?

Comment 3

[Serge Gautherie (:sgautherie)]

http://mxr.mozilla.org/comm-central/search?string=MOZ_IPC&case=1&find=%2Fsuite%2F
MOZ_IPC is disabled.

I don't know how to check/answer your two other questions.

Comment 4

[Alon Zakai (:azakai)]

Attachment: Patch? [Checked in: Comment 12]

(In reply to comment #3)
> MOZ_IPC is disabled.

That must be the problem then.

I guess the solution is to disable the wakeup service if MOZ_IPC is disabled, something like in this (untested) patch? Or is there a better way?

Comment 5

[Robert Kaiser]

I have no idea why Serge filed another bug for this, when bug 594430 will fix this by just packaging the message wakeup service so that it's available.

What I wondered about was why we even try to access it in non-IPC builds. Given that we want to enable OOPP ASAP (i.e. as soon as we can build with some kind of libxul), we should package it anyhow, though.

Comment 6

[Robert Kaiser]

Forget the part of my comment that talks about fixing, as the failure happens in builds that don't run from packages but from dist/bin.

The problem is really that we shouldn't include IPC-only code in non-IPC builds and Alon's patch is probably correct.

Comment 7

[Serge Gautherie (:sgautherie)]

Comment on attachment 473190 [details] [diff] [review]
Patch?
[Checked in: Comment 12]

I can't test due to bug 562820, but that's exactly what I was thinking about.

Comment 8

[Mark Finkle (:mfinkle) (use needinfo?)]

Comment on attachment 473190 [details] [diff] [review]
Patch?
[Checked in: Comment 12]

looks OK to me

Comment 9

[Serge Gautherie (:sgautherie)]

Comment on attachment 473190 [details] [diff] [review]
Patch?
[Checked in: Comment 12]

"approval2.0=?":
Trivial build fix for non-MOZ_IPC builds, like SeaMonkey ones.

Comment 10

[Serge Gautherie (:sgautherie)]

Ping for approval2.0.

Comment 11

[Justin Wood (:Callek)]

Comment on attachment 473190 [details] [diff] [review]
Patch?
[Checked in: Comment 12]

a=NPOTB ... I'll push in a few minutes.

Comment 12

[Drew Willcoxon :adw]

http://hg.mozilla.org/mozilla-central/rev/1d2729c14e46

Comment 13

[Serge Gautherie (:sgautherie)]

V.Fixed, per
http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1284865960.1284868738.29807.gz&fulltext=1
Linux comm-central-trunk leak test build on 2010/09/18 20:12:40

Comment 14

[Serge Gautherie (:sgautherie)]

Fwiw, should nsIMessageWakeupService.idl be somehow "ifdef'ed" too?