529669 - Holding down Cmd/Ctrl+Shift+P for a couple of seconds will stop Private Browsing mode transistions (Error: this._privateBrowsingService is null)

Status: NEW

(Firefox :: Private Browsing, defect)

Description

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.3a1pre) Gecko/20091118 Minefield/3.7a1pre (.NET CLR 3.5.30729)

Having two windows open and holding down the shortcut keys for a couple of seconds will cause the Private Browsing transistion to fail. It's not possible to start/stop the PB mode anymore:

Error: this._privateBrowsingService is null
Source File: chrome://browser/content/browser.js
Line: 9189

>  get privateBrowsingEnabled PBUI_get_privateBrowsingEnabled() {
>    return this._privateBrowsingService.privateBrowsingEnabled;
>  }


Error: browser.stop is not a function
Source File: file:///C:/mozilla/bin/minefield/components/nsSessionStore.js
Line: 1972

>      aTabData[t]._tabStillLoading = true;
>      if (!aTabData[t].entries || aTabData[t].entries.length == 0) {
>        // make sure to blank out this tab's content
>        // (just purging the tab's history won't be enough)
>        browser.contentDocument.location = "about:blank";
>        continue;
>      }
>      
>      browser.stop(); // in case about:blank isn't done yet

Steps:
1. Open two window
2. Hold down the Cmd/Ctrl+Shift+P key for a couple of seconds

After a while the above errors are thrown and no transistion happens anymore. You have to restart the browser.

Comment 1

[Henrik Skupin [:whimboo][⌚️UTC+2]]

When you try to change the state in the other window you will get the following exception:

Error: uncaught exception: [Exception... "'Failure' when calling method: [nsIPrivateBrowsingService::privateBrowsingEnabled]"  nsresult: "0x80004005 (NS_ERROR_FAILURE)"  location: "JS frame :: chrome://browser/content/browser.js :: PBUI_toggleMode :: line 9185"  data: no]

Comment 2

[(no longer active)]

I observed this on OS X with several windows open (~10), but for me, it was possible to transition into the PB mode.  I bet it's been because of bug 529667 (I tested using a build patched with the fix for that bug.)  If that is the case, then I don't think that this bug needs to block 3.6 (perhaps we can consider it for 3.6.1).

I have submitted a try server build with that fix so that Henrik can try it tomorrow, and I'll unmark the blocking request in case Henrik confirms that this doesn't prevent the user from further transitions in/out of the PB mode.

Comment 3

[Mike Beltzner [:beltzner, not reading bugmail]]

Ehsan: I agree with your assessment, but think the better approach is to renominate for blocking if this turns out to be more serious.

Comment 4

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Ehsan, which try server build I have to use? There are two in the builds folder on the FTP server.

Comment 5

[Dão Gottwald [:dao]]

https://build.mozilla.org/tryserver-builds/[e-mail]-try-2fbb1cc133fa/

Comment 6

[Henrik Skupin [:whimboo][⌚️UTC+2]]

The try server build doesn't fix the problem. See the attached screencast:
http://screencast.com/t/N2Y5Yzk3ZjMt

So we should handle it separately from bug 529667.

Comment 7

[Dão Gottwald [:dao]]

Still an issue despite bug 529667.

Comment 8

[Mike Beltzner [:beltzner, not reading bugmail]]

I don't think that this blocks, based on the steps required to reproduce. Would take a safe fix in a security branch, most likely.

Comment 9

[Dão Gottwald [:dao]]

The steps don't seem different from those in bug 526194.

Comment 10

[Henrik Skupin [:whimboo][⌚️UTC+2]]

With the difference that we do not run into an OOM and finally crash.

Comment 11

[(no longer active)]

(In reply to comment #10)
> With the difference that we do not run into an OOM and finally crash.

Yes, and that was the reason why I supported bug 526194 to be blocking.

I'm working on a fix here, but I'm kind of moving in the dark, since I haven't been able to reproduce this locally with the fix to bug 529667.

Henrik, can you please try a nightly with the fix to bug 529667 as well?  I know that you tested the try server build, but I'm suspecting whether it could have been a thing I missed to fix in the try server build, because the symptom that you're observing (the command being disabled in one of the windows) looks an awful lot like bug 529667.

Comment 12

[Dão Gottwald [:dao]]

(In reply to comment #10)
> With the difference that we do not run into an OOM and finally crash.

That's not a difference regarding the steps to reproduce. (Besides, I haven't seen said effects in bug 526194, I just got plenty of windows.)

Comment 13

[Henrik Skupin [:whimboo][⌚️UTC+2]]

(In reply to comment #11)
> Henrik, can you please try a nightly with the fix to bug 529667 as well?  I
> know that you tested the try server build, but I'm suspecting whether it could
> have been a thing I missed to fix in the try server build, because the symptom
> that you're observing (the command being disabled in one of the windows) looks
> an awful lot like bug 529667.

I have verified that bug 529667 is fixed but that one is still present. For me it needs 3-4 transistions and then the mentioned errors occur.

Comment 14

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Ehsan, after our conversation on IRC I was able to reproduce this bug on OS X now. Even with a fresh profile. In comment 0 I forgot to say that you have to disable the warning dialog when entering pb mode. On OS X it takes a bit longer - around 10-15s. No idea if that happens because I run all the other builds in VM's.

Comment 15

[(no longer active)]

Henrik, can you please try this build: <http://build.mozilla.org/tryserver-builds/[e-mail]-try-e9480c984599>?  I believe it should fix the problem.  I managed to finally see this on OS X, and I don't get this behavior with this build, but I need you to confirm it on multiple platforms.

Comment 16

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Tested that build in a XP VM and the issue is still present. Same errors appear.

Comment 17

[(no longer active)]

(In reply to comment #16)
> Tested that build in a XP VM and the issue is still present. Same errors
> appear.

Even the "this._privateBrowsingService is null" error?  Because I had a patch in that build specifically to fix that error.

Comment 18

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Oh, no sorry. I only see the following exception:

Error: uncaught exception: [Exception... "'Failure' when calling method: [nsIPrivateBrowsingService::privateBrowsingEnabled]"  nsresult: "0x80004005 (NS_ERROR_FAILURE)"  location: "JS frame :: chrome://browser/content/browser.js :: PBUI_toggleMode :: line 9188"  data: no]