777739 - gsignal.c:3393: signal name `load_complete' is invalid for instance `0x7f396804abf0'

Status: NEW

(Core :: Disability Access APIs, defect)

Description

[Hubert Figuiere [:hub]]

When running Aurora 16 on Fedora Linux 17, I get the following messages

(firefox:23368): GLib-GObject-WARNING **: gsignal.c:3393: signal name `load_complete' is invalid for instance `0x7f396804abf0'

I didn't have them in Aurora 15.

It is my own debug build from source.

Comment 1

[Trevor Saunders (:tbsaunde)]

(In reply to Hub Figuiere [:hub] from comment #0)
> When running Aurora 16 on Fedora Linux 17, I get the following messages
> 
> (firefox:23368): GLib-GObject-WARNING **: gsignal.c:3393: signal name
> `load_complete' is invalid for instance `0x7f396804abf0'
> 
> I didn't have them in Aurora 15.
> 
> It is my own debug build from source.

so, then I guess you can a stack trace for it?

Comment 2

[Hubert Figuiere [:hub]]

yeah I haven't gotten it again. gdb is currently attached :-)

Comment 3

[Karl Tomlinson (:karlt)]

I'm guessing http://hg.mozilla.org/mozilla-central/annotate/98c2a42a3aef/accessible/src/atk/AccessibleWrap.cpp#l1234 but a break point at g_log will find the stack
(if you can reproduce).

Something like
"p ((char*(*)())g_type_name)(**(long unsigned int**) atkObj )"
should give the object type.

Comment 4

[Karl Tomlinson (:karlt)]

Looks like AtkDocument should be the interface, but I don't know the implementation.
http://developer.gnome.org/atk/1.32/AtkDocument.html#AtkDocument-load-complete

Comment 5

[Ginn Chen]

FWIW: I didn't see this message with Firefox 17.0a1 on Solaris x86.

Comment 6

[Hubert Figuiere [:hub]]

Indeed, it is the stack trace.

It happened when click on a button and some javascript put up an alert telling me a mandatory field was mandatory. Breakpoint reached BEFORE the alert was displayed.

#0  g_log (log_domain=log_domain@entry=0x3b59a37e44 "GLib-GObject", log_level=log_level@entry=G_LOG_LEVEL_WARNING, format=format@entry=
    0x3b59a3cf68 "%s: signal name `%s' is invalid for instance `%p'") at gmessages.c:788
#1  0x0000003b59a288f5 in g_signal_emit_by_name (instance=0x7f3d28f34650, detailed_signal=0x7f3de0c30cc8 "load_complete") at gsignal.c:3393
#2  0x00007f3de045cc51 in AccessibleWrap::FirePlatformEvent (this=0x7f3d5a569080, aEvent=<optimized out>)
    at /home/hub/source/mozilla/aurora-src/accessible/src/atk/AccessibleWrap.cpp:1245
#3  0x00007f3de043c8af in nsEventShell::FireEvent (aEvent=0x7f3d2ca66a00)
    at /home/hub/source/mozilla/aurora-src/accessible/src/base/nsEventShell.cpp:29
#4  0x00007f3de044a2bb in DocAccessible::ProcessPendingEvent (this=0x7f3dc06046e0, aEvent=0x7f3d2ca66a00)
    at /home/hub/source/mozilla/aurora-src/accessible/src/generic/DocAccessible.cpp:1797
#5  0x00007f3de0433648 in NotificationController::WillRefresh (this=0x7f3dbead8ba0, aTime=...)
    at /home/hub/source/mozilla/aurora-src/accessible/src/base/NotificationController.cpp:323
#6  0x00007f3ddfcd04bf in nsRefreshDriver::Notify (this=0x7f3dc7e7d6c0, aTimer=<optimized out>)
    at /home/hub/source/mozilla/aurora-src/layout/base/nsRefreshDriver.cpp:336
#7  0x00007f3ddfcd08e0 in nsRefreshDriver::Notify (this=<optimized out>, aTimer=<optimized out>)
    at /home/hub/source/mozilla/aurora-src/layout/base/nsRefreshDriver.cpp:306
#8  0x00007f3de057876c in nsTimerImpl::Fire (this=this@entry=0x7f3d2977d9c0)
    at /home/hub/source/mozilla/aurora-src/xpcom/threads/nsTimerImpl.cpp:476
#9  0x00007f3de0578841 in nsTimerEvent::Run (this=<optimized out>) at /home/hub/source/mozilla/aurora-src/xpcom/threads/nsTimerImpl.cpp:556
#10 0x00007f3de0575a1b in nsThread::ProcessNextEvent (this=0x7f3de2c23ac0, mayWait=<optimized out>, result=0x7fffaf3780d0)
    at /home/hub/source/mozilla/aurora-src/xpcom/threads/nsThread.cpp:624
#11 0x00007f3de0583415 in NS_InvokeByIndex_P (that=<optimized out>, methodIndex=8, paramCount=<optimized out>, params=<optimized out>)
    at /home/hub/source/mozilla/aurora-src/xpcom/reflect/xptcall/src/md/unix/xptcinvoke_x86_64_unix.cpp:163
#12 0x00007f3de0210e2d in Invoke (this=0x7fffaf378078) at /home/hub/source/mozilla/aurora-src/js/xpconnect/src/XPCWrappedNative.cpp:3082
#13 Call (this=0x7fffaf378078) at /home/hub/source/mozilla/aurora-src/js/xpconnect/src/XPCWrappedNative.cpp:2416
#14 XPCWrappedNative::CallMethod (ccx=..., mode=<optimized out>)
    at /home/hub/source/mozilla/aurora-src/js/xpconnect/src/XPCWrappedNative.cpp:2382
#15 0x00007f3de02142ae in XPC_WN_CallMethod (cx=0x7f3d293163a0, argc=1, vp=0x7f3dd4388508)
    at /home/hub/source/mozilla/aurora-src/js/xpconnect/src/XPCWrappedNativeJSOps.cpp:1474
#16 0x00007f3de086a2ac in CallJSNative (args=..., native=<optimized out>, cx=0x7f3d293163a0)
    at /home/hub/source/mozilla/aurora-src/js/src/jscntxtinlines.h:382
#17 js::InvokeKernel (cx=cx@entry=0x7f3d293163a0, args=..., construct=construct@entry=js::NO_CONSTRUCT)
    at /home/hub/source/mozilla/aurora-src/js/src/jsinterp.cpp:344
#18 0x00007f3de085b91d in js::Interpret (cx=cx@entry=0x7f3d293163a0, entryFrame=entryFrame@entry=0x7f3dd4388300, interpMode=interpMode@entry=
    js::JSINTERP_NORMAL) at /home/hub/source/mozilla/aurora-src/js/src/jsinterp.cpp:2442

Comment 7

[Hubert Figuiere [:hub]]

This is Aurora rebuilt from this morning after hg pull -u

Comment 8

[Trevor Saunders (:tbsaunde)]

(In reply to Hub Figuiere [:hub] from comment #6)
> Indeed, it is the stack trace.
> 
> It happened when click on a button and some javascript put up an alert
> telling me a mandatory field was mandatory. Breakpoint reached BEFORE the
> alert was displayed.

I'd want to see what mFlags for the accessible the event is being fired on is, and what Role() returns, but my gues is this is a result of fireing load events for aria documents.  Since we don't give accessibles for those the atk document interface.

> #1  0x0000003b59a288f5 in g_signal_emit_by_name (instance=0x7f3d28f34650,

Comment 9

[Hubert Figuiere [:hub]]

In frame 2 (AccessibleWrap::FirePlatformEvent())

this->mFlags == 514
this->Role() == mozilla::a11y::roles::DIALOG

Comment 10

[Trevor Saunders (:tbsaunde)]

(In reply to Hub Figuiere [:hub] from comment #9)
> In frame 2 (AccessibleWrap::FirePlatformEvent())
> 
> this->mFlags == 514
> this->Role() == mozilla::a11y::roles::DIALOG

if I'm converting mFlags to hex correctly and looking up the set bits correctly that's a hyper text accessible with a children flag set.  Which seems to confirm the document event fired on aira doc theory.  That would mean this is more or less harmless since iirc the aria documents event people wanted it on windows, but we should probably fix this someday probably by making the atkobject for the aria doc a atkdocument.

Comment 11

[Leho Kraav (:macmaN @lkraav)]

I'm getting a bunch of these invalid signal name "load_complete" errors in a session. Currently on 18.0.2, stock Mozilla binary.

Comment 12

[Armin K.]

I am still seeing bunch of these with Firefox 20.0

Comment 13

[Wellington Torrejais da Silva]

Hi,

I still seeing this in 55.
Is this: https://github.com/ueno/libfep/commit/9521c75a706f1ef48a638bf399554a9e361b1e00 the same problem?

Thanks...

Comment 14

[Frank McCormick]

I am still seeing this in 2022. Firefox 95.0.2

I am running 95.0.2 and my .xsession-errors log is still flooded with:

signal name 'load_complete' is invalid for instance '0x7fcb46e0d230' of type 'MaiAtkType139'

Running Fedora 35 . Same thing on Debian Sid.