781704 - browser shutdown prints this warning 11 times: WARNING: NS_ENSURE_SUCCESS(rv, false) failed with result 0x8000FFFF: file /moz/mi/content/base/src/nsContentUtils.cpp, line 2929

Status: REOPENED

(Core :: DOM: Core & HTML, defect, P5)

Description

[Luke Wagner [:luke]]

It seems like this warning should either be removed or fixed if it prints 11 times on every browser startup.

Comment 1

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

I don't see this ...

Comment 2

[Luke Wagner [:luke]]

For reference, this is m-i tip on a debug Linux 64-bit build.

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

I don't see the warning either when starting browser.
Also, the warning is *very* useful for debugging.

Comment 4

[Luke Wagner [:luke]]

If it's useful, great, don't remove it.  I was wrong earlier, though, these error messages are printed on *shutdown*.  I can reproduce on both OSX and Linux with clean profile.

Comment 5

[Luke Wagner [:luke]]

Attachment: backtrace

Here is the backtrace.  Seems like we are trying to run code when tearing down a window.

Comment 6

[Nathan Froyd [:froydnj]]

...and this then triggers bug 781702 as well.

This warning also gets spewed into debug mochitest-other logs, maybe other tests as well; the two warnings combine for several percent of the total size of mochitest-other logs.

Comment 7

[Luke Wagner [:luke]]

Noisy Firefox debug spew taxes all devs, just like compiler warning spew: important things get lost in the noise and it takes extra time to find what you need (except worse, since warnings tend to only bug people hacking on the same code).  I know personally that I would really appreciate it if we had a general goal of "no debug spew during normal FF operation".  This is much more modest than requesting that all assertions be fatal and actually *increases* the utility of non-fatal assertions by increasing their visibility.  To wit:

  /moz/mi/d/dist/bin $ MOZ_QUIET=1 ./firefox -no-remote -P test 2>&1 | wc -l
  42

cc'ing people who might've had this discussion before.

Comment 8

[David Mandelin [:dmandelin]]

(In reply to Luke Wagner [:luke] from comment #7)
> Noisy Firefox debug spew taxes all devs, just like compiler warning spew:
> important things get lost in the noise and it takes extra time to find what
> you need (except worse, since warnings tend to only bug people hacking on
> the same code). 

Agreed. Want to post about this on m.d.platform?

Comment 9

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

We have this argument every 6 months and have continually failed to come to consensus.

If there is a specific warning that is firing during normal operation, we should fix it.

Comment 10

[Olli Pettay [:smaug][[e-mail]]]

If we had easy way to check whether we're shutting down...

Comment 11

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Are we necessarily shutting down?  It just looks like we're tearing down the window in that backtrace.

It should be pretty easy for XBL to figure out if we're in nsDocument::Destroy, I think.

Comment 12

[Olli Pettay [:smaug][[e-mail]]]

Comment 4 talks about shutdown.

Again, this particular warning has helped debugging many bugs, so we should not remove it.
It might be ok to make it fire less likely in case we know for sure the error case doesn't
indicate anything problematic.

Comment 13

[Nathan Froyd [:froydnj]]

It definitely can happen at other times than at shutdown.  And it's generally somewhere below nsDocument::Destroy.  So khuey's suggestion in comment 11 would not be amiss.  Only issue is that it's not happening in a document method, but inside nsContentUtils...

Comment 14

[Firefox Bug Husbandry Bot]

https://bugzilla.mozilla.org/show_bug.cgi?id=1472046

Move all DOM bugs that haven’t been updated in more than 3 years and has no one currently assigned to P5.

If you have questions, please contact :mdaly.