459790 - Following Error console link causes uncaught exception ( 0x80004002 (NS_NOINTERFACE) [nsISupports.QueryInterface] )

Status: VERIFIED FIXED

(Core :: XPConnect, defect)

Description

[Ronald Killmer]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.8.1.17) Gecko/20080829 Firefox/2.0.0.17
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1b2pre) Gecko/20081013 Shredder/3.0b1pre

Error: uncaught exception: [Exception... "Component returned failure code: 0x80004002 (NS_NOINTERFACE) [nsISupports.QueryInterface]"  nsresult: "0x80004002 (NS_NOINTERFACE)"  location: "native frame :: <unknown filename> :: <TOP_LEVEL> :: line 0"  data: no]

Reproducible: Always

Steps to Reproduce:
1. Opened a message with CSS errors 
2. Clicked the link in Error Console Warning msg. that opened view source window
3. This error flashed in Console just before view source window opened



Messages with the bad CSS were posted to news.mozilla.org/mozilla.support.firefox in thread "My FF303 cannot display this webpage correctly. Why ?" by Andrew D.

Comment 1

[IU]

Change (1) Product to Core (2) Platform to All (3) Version to Trunk (4) Keywords add "regression"

This bug is also reproducible in the following manner:

1. Create a new profile
2. Launch Firefox and open Error Console
3. Using your mouse, alternately switch to each of the two tabs by clicking it.
4. Notice the exception in Error Console

Regression range:

Works: http://hg.mozilla.org/mozilla-central/rev/ee5135f3f773
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1b2pre) Gecko/20081012 Minefield/3.1b2pre

Fails: http://hg.mozilla.org/mozilla-central/rev/163e9f2fa5c1
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1b2pre) Gecko/20081013 Minefield/3.1b2pre

Comment 2

[Nikolay Shopik]

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1b2pre) Gecko/20081014 Shredder/3.0b1pre
confirm

Comment 3

[IU]

Could someone please make the following changes to this bug; otherwise, I'll dupe it:

Change: (1) Product to Core (2) Platform to All (3) Version to Trunk (4)
Keywords add "regression"

Thanks

Comment 4

[[e-mail]]

I can reproduce this simply by tabbing through the Error Console.

The problem seems to be that the XPConnect and/or the JavaScript engine is reporting QueryInterface failures on XUL elements as JavaScript errors.

Comment 5

[Ben Turner (not reading bugmail, use the needinfo flag!)]

This is probably fallout from bug 455436... Sigh.

Comment 6

[asmpgmr]

The following messages appear in the error console using SeaMonkey 2.0a2 since 10/13 which I suspect may be a result of the patch for bug 455436:


Error: uncaught exception: [Exception... "Component returned failure code: 0x80004002 (NS_NOINTERFACE) [nsISupports.QueryInterface]" nsresult: "0x80004002 (NS_NOINTERFACE)" location: "native frame :: <unknown filename> :: <TOP_LEVEL> :: line 0" data: no]

Error: uncaught exception: 2147500034

Comment 7

[Magnus Melin [:mkmelin]]

"Error: uncaught exception: 2147500034" was bug 459784 for thunderbird... If you get it in seamonkey, file a bug for it. (It's not this bug.)

Comment 8

[asmpgmr]

Opened bug 460325 against SeaMonkey for the second message (though it's likely a Core issue). The first part should still stand as it is exactly the same message as reported.

Comment 9

[Fred Jendrzejewski]

Using hg bisect the first version with the regression is:
 20343:3aec0eec68a1 from Ben Turner

Comment 10

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Thanks folks, I'm aware of the problem (see comment 5).

Comment 11

[Danial Horton]

Its not a bug in the console itself rather the interface, if you click in the firefox search box, history search, bookmark search, or open any menu's this also causes the error.

Comment 12

[Johnny Stenback (:jst)]

Ben, we should probably see if we can do something about this for 1.9.1...

Comment 13

[WADA:World Anti-bad-Duping Agency]

Adding "nsISupports.QueryInterface" in bug summary for ease of search.

Comment 14

[Ronald Killmer]

To add to Comment #11, This error is also reported by Shredder when selecting View Message Source to use the View Source window.

Comment 15

[Nomis101]

I also get this error if I click on a link from a warning/error in the error console (I also get another error, but this is bug 463176). E.g. open the addressbook to trigger the "timed textboxes" warning in the error console and click on that link.

Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.1b2pre) Gecko/20081112 Shredder/3.0b1pre

Comment 17

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Attachment: Patch, v1

Here's one option. We simply ignore the NS_NOINTERFACE result of QI.

The alternative is to eat *all* exceptions that come from QI, but I really don't like that because it hides bugs.

Comment 18

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Attachment: Patch, v1.1

Use 'GetWrappedNativeOfJSObject' rather than 'WrapJS' for testing the nsIException. Thanks!

Comment 19

[Johnny Stenback (:jst)]

Comment on attachment 349077 [details] [diff] [review]
Patch, v1.1

Looks good!

Comment 20

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Comment on attachment 349077 [details] [diff] [review]
Patch, v1.1

Drivers, this patch will significantly reduce noise in the JS error console due to a regression in our exception handling. I think we might want this for b2 to avoid a ton of useless "I see this error in the console when I..." bug reports.

Comment 21

[Mike Connor [:mconnor]]

Comment on attachment 349077 [details] [diff] [review]
Patch, v1.1

a=mconnor, so we don't get a whole bunch of reports of this from the beta...

Comment 22

[Ilja Sekler]

I've built Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1b2pre) Gecko/20081118 Thunderbird/3.0b1pre with the patch <https://bugzilla.mozilla.org/attachment.cgi?id=349077>, it segfaults reproducibly when clicking on the link in the warning

Warning: Timed textboxes are deprecated. Consider using type="search" instead.
Source File: chrome://messenger/content/messenger.xul
Line: 0

A build from the same source but without the patch doesn't crash.

hg identify
9a933d62a4ec tip

Comment 23

[Ilja Sekler]

(In reply to comment #22)
> Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1b2pre)
> Gecko/20081118 Thunderbird/3.0b1pre

Oops, it should have been

Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1b2pre) Gecko/20081120 Thunderbird/3.0b1pre

of course. Sorry for bugspam.

Comment 24

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Attachment: Patch I checked in

This is the patch I checked in. Reviewing last night I realized that do_QueryWrappedNative is not null-safe as are the other do_Query* functions. I'm assuming this was the cause of the crash reported above.

Comment 25

[Ilja Sekler]

(In reply to comment #24)
> Created an attachment (id=349249) [...]
> 
> This is the patch I checked in. [...]

Confirming that this has eliminated the crash.

hg identify
da9024b330aa tip

Comment 26

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Pushed changeset 13ebc220fad3 to m-c for b2.

Comment 27

[asmpgmr]

Will this also fix bug 460325 ?

Comment 29

[Marcia Knous [:marcia]]

verified fixed using Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.1b2pre) Gecko/20081121 Minefield/3.1b2pre and Intel Mac 10.4,  Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1b2pre) Gecko/20081121 Minefield/3.1b2pre and Win Vista equivalent.

many thanks to bent as this error message was driving me crazy.

Comment 30

[Wladimir Palant]

For some reason, this had the opposite effect for me. In build 20081120 everything was fine, in build 20081123 I get "uncaught exception: 2147500034" pointing to this code in Adblock Plus 1.0 release candidate (executes during browser startup):

  Components.classes["@mozilla.org/privatebrowsing;1"]
            .getService(Components.interfaces.nsIPrivateBrowsingService);

The private browsing service is asked for nsIClassInfo and that's where the error comes from. I was pretty surprised to see an exception from code inside a try..catch block - but seeing JS_ReportPendingException in the patch here this makes sense.

Comment 31

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Looks like I missed some edge cases. Followup patch in a sec.

Comment 32

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Attachment: Followup patch, fixes throwing numbers

This applies on top of attachment 349249 [details] [diff] [review] (which was already checked in). Fixes the case where JS throws an nsresult ('throw Cr.NS_NOINTERFACE') and also the case where the context options are already set to not report exceptions.

Comment 33

[Ben Turner (not reading bugmail, use the needinfo flag!)]

See bug 466538, bug 460325, and comment 30 for examples that should be fixed with the new patch (I hope). I'd like to dupe those over if so.

Comment 34

[Shawn Wilsher :sdwilsh]

http://hg.mozilla.org/mozilla-central/rev/d50fb8c0d589

Comment 37

[Nochum Sossonko [:Natch]]

removing fixed1.9.1 b/c these bugs are verified1.9.1, sorry for the spam.

Comment 38

[Nochum Sossonko [:Natch]]

http://hg.mozilla.org/releases/mozilla-1.9.1/rev/d50fb8c0d589 (because it wasn't noted before)

Comment 39

[Tony Chung [:tchung]]

sorry for the inconvenience.  i meant to detect any bugs before the branch merge that were still RESO fixed, but had a verified1.9.1 keyword next to them.

Comment 40

[Tony Chung [:tchung]]

Verified fix on Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1b3pre) Gecko/20090115 Minefield/3.1b3pre  and

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2a1pre) Gecko/20090115 Minefield/3.2a1pre