294740 - Startup always does chrome reg twice

Status: RESOLVED DUPLICATE of bug 294150

(Toolkit :: Add-ons Manager, defect)

Description

[Boris Zbarsky [:bzbarsky]]

Marking this critical because this makes the fix for bug 281988 not work in my
build, which is Quite Bad.

Every time I start Firefox, it goes through xpcom/chrome registration twice. 
Specifically, this is the startup output:


----------------------------------------------------------------
No Persistent Registry Found.
Type Manifest File:
/home/bzbarsky/moztest/.mozilla/firefox/nl14hnzt.default/xpti.dat
nsNativeComponentLoader: autoregistering begins.
*** Registering nsSystemPrefModule components (all right -- a generic module!)

[ .... bunch more modules .... ]

++WEBSHELL == 1
WARNING: NS_ENSURE_TRUE(NS_SUCCEEDED(rv)) failed, file
../../../mozilla/extensions/cookie/nsPermissionManager.cpp, line 624
++DOMWINDOW == 1
*** Item Installed via directory addition to Install Location: app-global Item
ID: {972ce4c6-7e08-4474-a285-3208198ce6fd}, attempting to register...
*** Item Installed/Upgraded at Install Location: app-global Item ID:
{972ce4c6-7e08-4474-a285-3208198ce6fd}, attempting to register...
*** loading the extensions datasource
*** ... success, item is compatible
*** After installation: 0
WARNING: nsExceptionService ignoring thread destruction after shutdown, file
../../../mozilla/xpcom/base/nsExceptionService.cpp, line 191
+++ JavaScript debugging hooks removed.
--WEBSHELL == 0

[ this is us shutting down... then asserting about main thread being held past  
  XPCOM shutdown, then: ]

No Persistent Registry Found.
Type Manifest File:
/home/bzbarsky/moztest/.mozilla/firefox/nl14hnzt.default/xpti.dat
nsNativeComponentLoader: autoregistering begins.
*** Registering nsSystemPrefModule components (all right -- a generic module!)

[ and away we go again ]

*** Item Installed via directory addition to Install Location: app-global Item
ID: {972ce4c6-7e08-4474-a285-3208198ce6fd}, attempting to register...
*** Item Installed/Upgraded at Install Location: app-global Item ID:
{972ce4c6-7e08-4474-a285-3208198ce6fd}, attempting to register...
*** loading the extensions datasource
*** ... success, item is compatible
*** After installation: 0
WARNING: EM tried to force us to restart twice! Forcefully preventing that.,
file ../../../mozilla/toolkit/xre/nsAppRunner.cpp, line 1966

----------------------------------------------------------------

And then Firefox actually brings up a window, and I get to browse.  Next
startup, I get this all over again.

972ce4c6-7e08-4474-a285-3208198ce6fd seems to be the "classic theme" extension.
 The theme's certainly there.

The .mozconfig for this build is:

ac_add_options --enable-svg
ac_add_options --enable-mathml
ac_add_options --enable-crypto
ac_add_options --enable-jsd
ac_add_options --enable-xtf
ac_add_options --disable-freetype2
ac_add_options --srcdir=../mozilla
ac_add_options --enable-application=browser
ac_add_options --enable-debug
ac_add_options --disable-optimize
ac_add_options --enable-default-toolkit=gtk2
ac_add_options --enable-xft
ac_add_options --enable-extensions=all,-typeaheadfind

Comment 1

[Brendan Eich [:brendan]]

Benjamin, can you take a look?  This is messing up ability to debug on Linux,
and on that account alone is delaying alpha (cuz it's delaying the people
working on bug 281988).  It may be causing general startup badness.

/be

Comment 2

[Darin Fisher]

There is surely something odd with your build or your profile (or mine!) since
try as I might, I cannot reproduce this problem.

Try setting the environment variable NO_EM_RESTART=1, then nsAppRunner.cpp
should refuse to restart no matter what.  Maybe that will help.

Comment 3

[Darin Fisher]

If setting, NO_EM_RESTART=1 doesn't help, then also try XRE_IMPORT_PROFILES=1
just in case we are somehow restarting due to wierd interaction with the profile
migration code.  If neither of those environment variables does the trick for
you, then something is definitely wacky.

Comment 4

[Benjamin Smedberg]

Your debug output indicates that you haven't updated from bug 294150 - this is
most likely a dup of that bug.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Hmm.  I was pretty sure I was using a build pulled yesterday afternoon, but I
just updated to this morning and the problem did go away.... <sigh>.

*** This bug has been marked as a duplicate of 294150 ***

Comment 6

[Brendan Eich [:brendan]]

Hmm, I am up to date in toolkit/mozapps yet still see at least one extra restart:

$ NO_EM_RESTART=1 MOZ_NO_REMOTE=1 ./firefox -P test
https://bugzilla.mozilla.org/show_bug.cgi?id=289074
No Persistent Registry Found.
Type Manifest File: /home/brendan/.mozilla/firefox/x2kp0qpw.test/xpti.dat
nsNativeComponentLoader: autoregistering begins.
. . .
nsNativeComponentLoader: autoregistering succeeded
*** Deferring registration of sample JS components
nsNativeComponentLoader: registering deferred (0)
*** Registering sample JS components
nsNativeComponentLoader: registering deferred (0)
nsNativeComponentLoader: registering deferred (0)
nsNativeComponentLoader: registering deferred (0)
nsNativeComponentLoader: autoregistering begins.
nsNativeComponentLoader: SelfRegisterDll(libwidget_gtk.so) Load FAILED with
error: /home/brendan/src/firefox/mozilla/dist/bin/components/libwidget_gtk.so:
cannot open shared object file: No such file or directory
nsNativeComponentLoader: autoregistering succeeded
nsNativeComponentLoader: registering deferred (0)
pldhash: for the table at address 0x82bc300, the given entrySize of 44 probably
favors chaining over double hashing.
GFX: dpi=96 t2p=0.0666667 p2t=15 depth=24
++WEBSHELL == 1
++DOMWINDOW == 1
*** loading the extensions datasource
WARNING: EM tried to force us to restart twice! Forcefully preventing that.,
file nsAppRunner.cpp, line 1989
WARNING: NS_ENSURE_TRUE(NS_SUCCEEDED(rv)) failed, file nsStringBundle.cpp, line 273
++WEBSHELL == 2
. . .

What am I missing?

/be