mirror of
https://github.com/ctm/syn68k.git
synced 2024-12-03 14:49:38 +00:00
34 lines
1.5 KiB
Plaintext
34 lines
1.5 KiB
Plaintext
This is the output from running syngentest with various command line options
|
|
on two different platforms.
|
|
|
|
One platform was a hardware 68k (probably a 68040 running NEXTSTEP,
|
|
but it's been so long that I don't remember), done sometime in the mid
|
|
1990s. IIRC, the command line options were
|
|
|
|
../syngentest 1000 > 68k.1000
|
|
../syngentest -noncc 1000 > 68k.noncc.1000
|
|
../syngentest 100 > dump-68k-100-cc
|
|
../syngentest -noncc 100 > dump-68k-100-noncc
|
|
|
|
The other platform is an i686 Fedora 9 system and those command line switches
|
|
were:
|
|
|
|
../syngentest > 10000
|
|
../syngentest -noncc > 10000-noncc
|
|
../syngentest -noncc -notnative > 10000-noncc-notnative
|
|
../syngentest -notnative > 10000-notnative
|
|
|
|
Unfortunately, the memory location that we used to use to test on the 68k
|
|
is unavailable on modern Linux systems, so we can't actually compare the
|
|
old output from a real 680x0 to the new output, but the new output is
|
|
presumably correct, since we did a lot of testing with a real 68k a long time
|
|
ago and Executor runs with the libsyn68k used with this Fedora 9 syn68k.
|
|
|
|
The output when you include "-noncc" *should* be different from when
|
|
you don't include it (you can see the difference between 68k.1000 and
|
|
68k.noncc.1000. The output when you include "-notnative" "shouldn't"
|
|
be different, but it is, and that may well be a bug in Syn68k,
|
|
although it's a bug that has (IIRC) been there from day one. The
|
|
problem is that we get a different set of condition codes for divs_reg
|
|
in a particular case.
|