1361230 - Problems with Windows Hooks when JAWS is running

Status: RESOLVED INCOMPLETE

(Core :: Disability Access APIs, defect)

Description

[(No longer employed by Mozilla) Aaron Klotz]

I'm filing this as a separate bug but I suspect that a bunch of other dependencies of bug 1350984 are just dupes of this one.

Under JAWS I noticed that sometimes Nightly just stops responding to input. In addition to characters typed into input fields, there are also more serious hangs where Window activation isn't working.

The thing is, these don't look like hangs to Windows. Furthermore, when I break into the debugger and look at the chrome main thread, it isn't stuck on anything.

OTOH, weird stuff happens when I run Spy++. One time a hung Nightly suddenly started working again when I started Spy++. The other time, Nightly was hung, Spy++ crashed, and then Nightly started working again.

This sounds to me like something isn't working right with JAWS's Window hooks.

This is going to be a real pain in the arse to investigate. I'm going to start by adding some WH_DEBUG hooks to enable logging of hook procedures.

Comment 1

[(No longer employed by Mozilla) Aaron Klotz]

Blocking jhook.dll in child processes is... interesting. Looks like JAWS sets a timer and retries its injections in a loop. Oh, and the entire GUI starts to lock up.

Comment 2

[(No longer employed by Mozilla) Aaron Klotz]

(In reply to Aaron Klotz [:aklotz] (a11y work receiving priority right now) from comment #1)
> Oh, and the entire GUI starts to lock up.

Actually I think this was due to the debugger. When no debugger is attached, things look like they run a lot better, even despite the constant re-hooking attempts in the child processes!