Hello all,
I am still getting some very weird observations when look into why
FreeCOM weirds out:
When I recompiled FreeCOM on my end with GCC, and tried to run "fetch
mem" inside the METADOS setup, the command.com grew the heap area very
quickly when handling the complex batch files, and the stack pointer
(%sp), which was patched to start at a rather low address (0x39a0),
quickly collided with the heap.
On closer inspection, I found that this is not actually the case: the
end of the heap was not being advanced (sbrk(0) == 0x1724 < 0x39a0), but
malloc( ) was already returning large values for some strange reason.
More interestingly, when I recompiled FreeCOM again with both `xms-swap'
_and_ `gcc' options, FreeCOM could not even successfully start
fdauto.bat, giving this error, followed by a crash:
The path to COMMAND.COM must be fully qualified!
That means including drive letter and beginning with a backslash.
For example: C:\FDOS
COMMAND.COM assumes this path now:
A:\5
Unable to open file 'A:\5'
Invalid Opcode at F77B 05C8 0246 F135 FFB0 00DD 0003 FFA8 FFB0 FFAF
1068 1384 5CEA
Running fdauto.bat with tracing (the F8 key) seems to suggest that the
crash happened right after the `xmssize.com 5 >NUL' command --- I have
uploaded a screen shot at
https://user-images.githubusercontent.com/5507843/44596095-a44eeb00-a7fd-11e8-9aea-f2ffe5712d53.png
.
(In case it helps, I have also uploaded my recompiled FreeCOM --- with
additional debugging output --- and the linker map file, at
https://github.com/tkchia/gcc-ia16/files/2319135/command.com.gz and
https://github.com/tkchia/gcc-ia16/files/2319136/command.map.txt .)
Thank you!
--
https://github.com/tkchia
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Freedos-devel mailing list
Freedos-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freedos-devel
