Joanmarie Diggs Reporter
	Description • 9 years ago

Steps to reproduce:
1. Load "data:text/html,<math><mtable><mtr><mtd><mi>x</mi></mtd></mtr></mtable></math>" in Nightly
2. In a python console do the following:

$ python3
>>> import pyatspi
>>> firefox = [app for app in pyatspi.Registry.getDesktop(0) if app.name == 'Firefox'][0]
>>> rows = pyatspi.utils.findAllDescendants(firefox, lambda x: x and x.getRole() == pyatspi.ROLE_TABLE_ROW)
>>>>>> for row in rows: print(row.getAttributes())

Expected result: Nightly would not crash.

Actual result: Nightly crashes reliably. See below.

I'm pretty sure this is a very recent regression. I'll try to pin that down next. Also, it doesn't seem to happen for regular tables; just MathML tables.

(gdb) bt
#0  0x00007ffff2cff8c2 in mozilla::a11y::HTMLTableRowAccessible::GroupPosition() () at /home/jd/Downloads/foxes/firefox/libxul.so
#1  0x00007ffff2cf2470 in mozilla::a11y::Accessible::NativeAttributes() () at /home/jd/Downloads/foxes/firefox/libxul.so
#2  0x00007ffff2cee63a in mozilla::a11y::Accessible::Attributes() () at /home/jd/Downloads/foxes/firefox/libxul.so
#3  0x00007ffff2cca9a7 in GetAttributeSet(mozilla::a11y::Accessible*) () at /home/jd/Downloads/foxes/firefox/libxul.so
#4  0x00007ffff2ccc280 in getAttributesCB () at /home/jd/Downloads/foxes/firefox/libxul.so
#5  0x00007fffdf56eb10 in impl_GetAttributes () at /lib64/libatk-bridge-2.0.so.0
#6  0x00007fffdf56d07a in handle_message () at /lib64/libatk-bridge-2.0.so.0
#7  0x00007fffef49a663 in _dbus_object_tree_dispatch_and_unlock () at /lib64/libdbus-1.so.3
#8  0x00007fffef48c104 in dbus_connection_dispatch () at /lib64/libdbus-1.so.3
#9  0x00007fffdf33b085 in message_queue_dispatch () at /lib64/libatspi.so.0
#10 0x00007fffeef39a8a in g_main_context_dispatch () at /lib64/libglib-2.0.so.0
#11 0x00007fffeef39e20 in g_main_context_iterate.isra () at /lib64/libglib-2.0.so.0
#12 0x00007fffeef39ecc in g_main_context_iteration () at /lib64/libglib-2.0.so.0
#13 0x00007ffff14af47f in nsAppShell::ProcessNextNativeEvent(bool) () at /home/jd/Downloads/foxes/firefox/libxul.so
#14 0x00007ffff287183d in nsBaseAppShell::DoProcessNextNativeEvent(bool, unsigned int) ()
    at /home/jd/Downloads/foxes/firefox/libxul.so
#15 0x00007ffff14aeb05 in nsBaseAppShell::OnProcessNextEvent(nsIThreadInternal*, bool, unsigned int) ()
    at /home/jd/Downloads/foxes/firefox/libxul.so
#16 0x00007ffff143a2a6 in nsThread::ProcessNextEvent(bool, bool*) () at /home/jd/Downloads/foxes/firefox/libxul.so
#17 0x00007ffff143fe9e in NS_ProcessNextEvent(nsIThread*, bool) () at /home/jd/Downloads/foxes/firefox/libxul.so
#18 0x00007ffff14511af in mozilla::ipc::MessagePump::Run(base::MessagePump::Delegate*) ()
    at /home/jd/Downloads/foxes/firefox/libxul.so
#19 0x00007ffff18bb8e3 in MessageLoop::Run() () at /home/jd/Downloads/foxes/firefox/libxul.so
#20 0x00007ffff286be78 in nsBaseAppShell::Run() () at /home/jd/Downloads/foxes/firefox/libxul.so
#21 0x00007ffff2e3b39c in nsAppStartup::Run() () at /home/jd/Downloads/foxes/firefox/libxul.so
#22 0x00007ffff2e78be7 in XREMain::XRE_mainRun() () at /home/jd/Downloads/foxes/firefox/libxul.so
#23 0x00007ffff2e7b26d in XREMain::XRE_main(int, char**, nsXREAppData const*) () at /home/jd/Downloads/foxes/firefox/libxul.so
#24 0x00007ffff2e7b51c in XRE_main () at /home/jd/Downloads/foxes/firefox/libxul.so
#25 0x0000000000408ce6 in do_main(int, char**, nsIFile*) ()
#26 0x00000000004052cb in main ()

	Joanmarie Diggs Reporter
	Comment 1 • 9 years ago

41.0a1 (2015-06-29) - Doesn't crash
42.0a1 (2015-06-30) - Crashes

	Frédéric Wang (:fredw) Assignee
	Comment 2 • 9 years ago

It seems that this code was added in bug 1177268 and merge to mozilla-central on 2015-06-30. Probably some special checks should be done for MathML tables.

	Marco Zehe (:MarcoZ)
	Comment 4 • 9 years ago

Comment on attachment 8628224 [details] [diff] [review]
Patch

r=me. Sorry I overlooked this when reviewing the original patch!

	Frédéric Wang (:fredw) Assignee
	Comment 5 • 9 years ago

https://treeherder.mozilla.org/#/jobs?repo=try&revision=af72943d393f

	Pulsebot
	Comment 6 • 9 years ago

https://hg.mozilla.org/integration/mozilla-inbound/rev/dcf40cb0dc96

	Ryan VanderMeulen [:RyanVM]
	Comment 7 • 9 years ago

https://hg.mozilla.org/mozilla-central/rev/dcf40cb0dc96

	Marco Zehe (:MarcoZ)
	Comment 8 • 9 years ago

Comment on attachment 8628224 [details] [diff] [review]
Patch

Approval Request Comment
[Feature/regressing bug #]: bug 1177268
[User impact if declined]: Reproducible crash.
[Describe test coverage new/current, TreeHerder]: TreeHerder, manual.
[Risks and why]: None, fixes crash.
[String/UUID change made/needed]: None.

	Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)
	Comment 9 • 9 years ago

Comment on attachment 8628224 [details] [diff] [review]
Patch

Approving for uplift to Aurora. Patch reviewed, seems safe and stable as it has been in m-c for a few days already.

	Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)
	Comment 10 • 9 years ago

Joanmarie, could you please verify that the fix works for you? If it does, please change the status to "Verified". Thanks!

	Ryan VanderMeulen [:RyanVM]
	Comment 11 • 9 years ago

https://hg.mozilla.org/releases/mozilla-aurora/rev/e796c66eb475