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

[ Thread Index | Date Index | More Archives ]

>> NaN test does look correct in softfloat emulation, I mistook invalid
>> encoding test (that does not check for SNaN) as NaN test.
>> (I'll try to compile the test program later)
> Any news on this? I would like to get it finished. Or should we just
> ignore the failing sNaN test?

I tried to compile fpsp040 but I this time I get stupid errors.. Is
pre-compiled (".sa" like with 060) possible with 68040, at least if they
are PC-relative. Or just include FPSP object files, if they are
compatible. Or even blob that is absolute address compiled to $00080000
or something would be good enough for testing :)

Errors like:

ssin.o: in function `ssincos':
ssin.o:(*ABS*+0xffffffe0): multiple definition of `X'
ssin.o:(*ABS*+0xffffffe2): multiple definition of `XDCARE'
ssin.o:(*ABS*+0xffffffe4): multiple definition of `XFRAC'

Why does it think those are some global variables?
(and even if it would have compiled without errors, I'd still wouldn't
be sure if result is 100% correct)

btw, there is not much point in Amiga FPSP binaries. There is already
"official" 68040 and 68060 support libraries that also set up MMU
correctly and patch few system functions to work more optimally with
68040+ CPUs. These also include FPSP.

Plain 68040/68060 FPU support library would be mostly useless, at least
when used with real hardware.

Mail converted by MHonArc 2.6.19+