Got a build and tested it against lucid firefox-3.5.
I didn't appear to get any problems (I'm reporting this bug using the Thumb-2 xulrunner library).
Attached is a gdb session transcript, where I set a breakpoint on the NS_InvokeByIndex implementation: the code is definitely Thumb-2, as can be observed by the instruction offsets not divisible by 4, and the mixture of instruction sizes (among other things).
I step some calls and returns through NS_InvokeByIndex; it seems to be doing the right thing.
Got a build and tested it against lucid firefox-3.5.
I didn't appear to get any problems (I'm reporting this bug using the Thumb-2 xulrunner library).
Attached is a gdb session transcript, where I set a breakpoint on the NS_InvokeByIndex implementation: the code is definitely Thumb-2, as can be observed by the instruction offsets not divisible by 4, and the mixture of instruction sizes (among other things).
I step some calls and returns through NS_InvokeByIndex; it seems to be doing the right thing.