Re: [hatari-devel] Possible bug in 060 exception stack frames

[Toni Wilen <twilen@xxxxxxxxxx>]

>> But have you ever run the test program there? I think using -(a7) or
(a7)+
>> in conjunction with mulu.l is rather unusual.

I modified it to work with Amiga and managed to compile it.

Results are different:

Testing 68060 ISP started:
	64-bit multiply...passed
	64-bit divide...passed
	cmp2,chk2...passed
	movep...passed
	Effective addresses...passed
	cas...

It crashed at this point. (with trace exception? Perhaps CAS isn't
emulated in Amiga because it isn't supposed to be used in Amiga anyway)

Log shows these instructions being tested:

32-666 [1039 109-507]: 68060 unimplemented opcode 4C01, PC=083f01ac
32-666 [1039 107-516]: 68060 unimplemented opcode 4C01, PC=083f01fc
32-674 [1040 109-020]: 68060 unimplemented opcode 4C01, PC=083f0246
32-674 [1040 113-030]: 68060 unimplemented opcode 4C01, PC=083f0292
32-677 [1040 109-040]: 68060 unimplemented opcode 4C01, PC=083f02e6
32-677 [1040 105-050]: 68060 unimplemented opcode 4C01, PC=083f0336
32-678 [1040 113-060]: 68060 unimplemented opcode 4C01, PC=083f038a
32-678 [1040 113-070]: 68060 unimplemented opcode 4C01, PC=083f03de
32-681 [1040 113-080]: 68060 unimplemented opcode 4C01, PC=083f0432
32-682 [1040 113-300]: 68060 unimplemented opcode 4C41, PC=083f0bd0
32-682 [1040 101-310]: 68060 unimplemented opcode 4C41, PC=083f0c18
32-683 [1040 113-320]: 68060 unimplemented opcode 4C41, PC=083f0c70
32-683 [1040 101-330]: 68060 unimplemented opcode 4C41, PC=083f0ccc
32-683 [1040 109-340]: 68060 unimplemented opcode 4C41, PC=083f0d0c
32-684 [1040 113-350]: 68060 unimplemented opcode 4C41, PC=083f0d4c
32-684 [1040 109-360]: 68060 unimplemented opcode 4C41, PC=083f0d9c
32-685 [1040 113-370]: 68060 unimplemented opcode 4C41, PC=083f0ddc
32-686 [1040 109-380]: 68060 unimplemented opcode 4C7C, PC=083f0e1c
32-686 [1040 113-390]: 68060 unimplemented opcode 4C41, PC=083f0e70
32-690 [1041 110-081]: 68060 unimplemented opcode 00EE, PC=083f1896

FPU test crashes almost immediately:

effadd_0_pc:
	fmul.x		#0xc00000008000000000000000,%fp0
	move.w		%ccr,SCCR(%a6)

For some reason FPSP returns to middle of following move to CCR.

So I guess either emulation is different or 68060 FPSP is different. I
didn't test with included FPSP yet.