Table of Contents
Late January work
2024-01-28
Here's this week's work.
lDebug
dpb.eld
- Add ADDRESS mode to specify address to read DPB from instead of a drive
- Add EXTENDED= mode with FREEDOS, MSDOS, and EDRDOS keywords to select format
- Add ONLY keyword to skip base DPB
- Add LIST SKIP number and LIST DRIVE drive modes to use DPB pointer chain from the DOS List of Lists
- Add LIST AT mode to specify address of the DOS List of Lists
This last point is interesting because it uses the debugger's getaddrX function with a 86 Mode segmented address, even in Protected Mode. This appears to have no ill effect. The default segment for the LIST AT address is read from the 86 Mode interrupt vector 31h's segment which will address the DOS data segment in FreeDOS and EDR-DOS. Therefore, a command like ext dpb.eld list at 26 drive C: is sufficient to access a DPB without using DOS at all, but not needing to find the exact address of the DPB either.
Other
- Disable reclaim.eld hint by default
- boot.asm: Allow trailing blanks after BOOT PROTOCOL= filenames even with CMDLINE=0
inicomp
Add the C mode test to display the number of dots that would be displayed during a full depack run. This is not used yet, but will allow to display a more useful progress indicator later.
bootimg
Display how many bytes are missing when the image becomes too full.
FreeDOS kernel
I submitted a bug report to the kernel on what happens when the fdconfig.sys SHELL= line has a leading digit at the beginning of the specified filename. This was reported to one of the mailing lists earlier. Todo: Note that inserting a backslash should work around the problem.
I also want to push the ecm-boot32-dl branch to merge it into the main kernel branch soon. Its Pull Request has been sitting for more than a week without any reactions.
Finally, I want to change the boot, boot32, and boot32lb loaders to move their buffer uses up to linear 22000h (or segmented 1FE0h:2200h) in order to allow loading with 128 KiB/cluster if the kernel file fits within 128 KiB, and with an /L parameter between 60 and 200 to SYS. With the FAT32 loaders this is probably not possible yet, and with the FAT12/FAT16 loaders it may depend on the trail of the file cluster data being padded with zeroes.
