Re: [PATCH/RFC] Enable java on sh64-linux (Take 2)

[Kaz Kojima <kkojima at rr dot iij4u dot or dot jp>]

Andrew Haley <aph@redhat.com> wrote:
>> FAIL: ArrayStore execution - gij test
>> FAIL: ArrayStore execution - gij test
>> FAIL: Divide_1 execution - gij test
>> FAIL: Divide_1 execution - gij test
>
> This points to a bad divide overflow handler.

I'd like to see the divide overflow handler.

>> FAIL: PR3096 -O3 execution - source compiled test
>> FAIL: Process_2 output - source compiled test
>> FAIL: Process_2 output - gij test
>> FAIL: Process_2 output - bytecode->native test
>> FAIL: Process_2 -O3 output - source compiled test
>> FAIL: Process_2 output - gij test
>> FAIL: Process_2 -O3 output - bytecode->native test
[snip]
>> FAIL: Thread_Wait_2 -O3 execution - source compiled test
>> FAIL: Thread_Wait_2 -O3 execution - bytecode->native test
>> FAIL: Thread_Wait_Interrupt execution - gij test
>> FAIL: Thread_Wait_Interrupt -O3 execution - source compiled test
>> FAIL: Thread_Wait_Interrupt execution - gij test
>> FAIL: Thread_Wait_Interrupt -O3 execution - bytecode->native test
>
> These are perhaps the thread bugs you mentioned.

Some of them never exited because the control thread of linuxthread
become a zombie and the main thread suspended forever.  Weird.

>> FAIL: negzero execution - gij test
>> FAIL: negzero execution - gij test
>> FAIL: pr6388 output - gij test
>> FAIL: pr6388 output - gij test
>
> These look like weird interpreter bugs.  Libffi problems, maybe?

I agree with you.  I'll take a closer look at how the ffi closure
function does in these case.  Thanks for your suggestions.

Regards,
	kaz