1142752 - (NS_NOINTERFACE) [nsIWebProgress.DOMWindow] exception in_docShellsToWindows when opening customize mode

Status: RESOLVED FIXED

(DevTools :: Framework, defect)

Description

[Brian Grinstead [:bgrins]]

I believe this started after bug 1059308.

STR:
Open browser toolbox (with --jsdebugger flag or using menu item)
Open customize mode (you may need to do this twice if you opened BT via the menu item)

See this error in the Browser Console:

Handler function threw an exception: [Exception... "Component returned failure code: 0x80004002 (NS_NOINTERFACE) [nsIWebProgress.DOMWindow]"  nsresult: "0x80004002 (NS_NOINTERFACE)"  location: "JS frame :: resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js :: TabActor.prototype._docShellsToWindows/< :: line 1003"  data: no]
Stack: TabActor.prototype._docShellsToWindows/<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js:1003:10
TabActor.prototype._docShellsToWindows@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js:1000:0
TabActor.prototype._notifyDocShellsUpdate@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js:1026:18
TabActor.prototype._onDocShellCreated/<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js:978:6
makeInfallible/<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/DevToolsUtils.js:82:13

Comment 1

[Elbart]

Also happening when loading websites, e.g.
https://photographylife.com/what-is-exif-data
http://www.lawblog.de/
http://www.w3schools.com/tags/tryit.asp?filename=tryhtml_iframe

! Only with e10s off.

Affected: 
Nightly 40.0a1
Aurora 39.0a2 

Regression-range:

m-c:

Last good revision: 0189941a3fd5 (2015-03-06)
First bad revision: 43fb1f92e8d4 (2015-03-07)
Pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=0189941a3fd5&tochange=43fb1f92e8d4


m-i:

Last good revision: 1e4b76918021
First bad revision: 43fb1f92e8d4
Pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=1e4b76918021&tochange=43fb1f92e8d4


fx-team:

Last good revision: 28a727d25fa7
First bad revision: 7697ad4919e7
Pushlog:
https://hg.mozilla.org/integration/fx-team/pushloghtml?fromchange=28a727d25fa7&tochange=7697ad4919e7

Comment 2

[Hallvord R. M. Steen [:hallvors]]

I'm seeing this error in nightly when loading this page and opening the "Style editor" tab in devtools:

http://m.mlb.com/game/2015/08/20/415456/indians-vs-yankees#game=gid_2015_08_20_clemlb_nyamlb_1,game_state=Wrapup,tz=false,game_tab=wrap

This seems like a regression since the current release version because the style panel opens OK there. Is this still this bug? Should it be flagged as a high-pri regression, or should I report another one?

Comment 3

[Hallvord R. M. Steen [:hallvors]]

Right - comment #1 also confirms it happens in the wild. Probably happens with E10S off on any page with an IFRAME.

Handler function threw an exception: [Exception... "Component returned failure code: 0x80004002 (NS_NOINTERFACE) [nsIInterfaceRequestor.getInterface]"  nsresult: "0x80004002 (NS_NOINTERFACE)"  location: "JS frame :: resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js :: DebuggerProgressListener.prototype._getWindowsInDocShell/< :: line 2118"  data: no]
Stack: DebuggerProgressListener.prototype._getWindowsInDocShell/<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js:2118:1
DebuggerProgressListener.prototype._getWindowsInDocShell@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js:2117:12
DebuggerProgressListener.prototype.watch@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js:2092:21
TabActor.prototype._onDocShellCreated/<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/webbrowser.js:1149:9
makeInfallible/<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/DevToolsUtils.js:83:14
Line: 2118, column: 0

Comment 4

[Brian Grinstead [:bgrins]]

When I open this page [0] and then directly open the Style Editor (shift+F7) I am seeing an error, but it's different from the one in Comment 3.  Everything seems to generally function after the error happens though.  Are you seeing any breakage in the tools, or just the error in the console?

Forwarding request to Alex since the regression range in Comment 1 points to Bug 1059308.

Here's the error I see FWIW:

console.error: 
  Message: TypeError: this.transport is null
  Stack:
    DSC_send@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/main.js:1339:5
Actor<._sendEvent@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/protocol.js:888:5
Actor<.initialize/<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/protocol.js:866:11
emitOnObject@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/commonjs/sdk/event/core.js:112:9
emit@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/commonjs/sdk/event/core.js:89:38
MediaRuleActor<._matchesChange@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/stylesheets.js:327:5
BottomHost.prototype.create@resource://gre/modules/commonjs/toolkit/loader.js -> resource:///modules/devtools/framework/toolbox-hosts.js:81:5
Toolbox.prototype.open/<@resource://gre/modules/commonjs/toolkit/loader.js -> resource:///modules/devtools/framework/toolbox.js:338:26
TaskImpl_run@resource://gre/modules/Task.jsm:314:40
TaskImpl@resource://gre/modules/Task.jsm:275:3
createAsyncFunction/asyncFunction@resource://gre/modules/Task.jsm:249:14
Task_spawn@resource://gre/modules/Task.jsm:164:12
Toolbox.prototype.open@resource://gre/modules/commonjs/toolkit/loader.js -> resource:///modules/devtools/framework/toolbox.js:337:12
DevTools.prototype.showToolbox@resource:///modules/devtools/gDevTools.jsm:416:7
gDevToolsBrowser.selectToolCommand@resource:///modules/devtools/gDevTools.jsm:672:7
oncommand@chrome://browser/content/browser.xul:1:1

[0]: http://m.mlb.com/game/2015/08/20/415456/indians-vs-yankees#game=gid_2015_08_20_clemlb_nyamlb_1,game_state=Wrapup,tz=false,game_tab=wrap

Comment 5

[Hallvord R. M. Steen [:hallvors]]

The tool breaks completely: the CSS code never appears. Did you disable E10S when testing? I happened to use a profile with E10S off when I noticed.

Comment 6

[Alexandre Poirot [:ochameau]]

I'm able to reproduce.

Comment 7

[Alexandre Poirot [:ochameau]]

Note that we can see this exception while running this test with e10s turned on: browser/devtools/inspector/test/browser_inspector_remove-iframe-during-load.js

Comment 8

[Alexandre Poirot [:ochameau]]

Attachment: prevent the exception

This patch prevent this exception in various scenarios: browser toolbox, regular page with e10s turned off,
or in the test I talked about.

I think the platform behavior changed a bit. We receive webnavigation-create event
with docshell being immediately destroyed.
I verified and we still get notifications for the document iframes,
it's just that we get some notification for some immediately-zombie docshell.
I would guess it is related to about:blank document or something...

https://treeherder.mozilla.org/#/jobs?repo=try&revision=b4573a818f82

Note that this exception happens within an executeSoon callback,
and the related docshell was a zombie one anyway,
I don't expect this exception to actually break anything.

While trying to reproduce this, I've seen the style editor being broken
for various other reasons. Sometimes with exceptions (access to Dead wrapper exception),
 and some other times without any until you close the toolbox (getStylesheets pending request exception).
Both these exceptions, I think, are not related to iframe switching feature,
but results from internal races within style editor itself.
That would help if you could have a 100% reproducible STR for these two bugs.

Comment 10

[[SV Manager] Florin Mezei, QA (:FlorinMezei)]

For testing purposes - also seen when choosing to connect to google.

Comment 11

[Pulsebot]

https://hg.mozilla.org/integration/fx-team/rev/e232ece8e041

Comment 12

[Ryan VanderMeulen [:RyanVM]]

Backed out for OSX devtools failures.
https://treeherder.mozilla.org/logviewer.html#?job_id=4415483&repo=fx-team
https://treeherder.mozilla.org/logviewer.html#?job_id=4415502&repo=fx-team

https://hg.mozilla.org/integration/fx-team/rev/4e72db149697

Comment 13

[Alexandre Poirot [:ochameau]]

I imagine this keybinding failure is due to the other changeset I bundled into this landing.
Let's see:
  https://treeherder.mozilla.org/#/jobs?repo=try&revision=5f2d8a9e1c28

Comment 14

[Pulsebot]

https://hg.mozilla.org/integration/fx-team/rev/f5868487bb7f

Comment 15

[Phil Ringnalda (:philor)]

https://hg.mozilla.org/mozilla-central/rev/f5868487bb7f