jnall%FSU.BITNET@cornellc.cit.cornell.edu (John Nall 904-644-5241) (05/20/89)
Formally, this is merely a message to the net stating that the Turbo C 2.0
files (.bat files, mods to Minix source files, etc.,) which were posted by
Deborah Mullen some time ago to allow one to compile Minix source code and
make a Minix boot disk under MS-DOS works equally well with Minix 1.4a.
Informally, by way of explanation of why it took me more than a month to
decide that it works OK: Carefully planned steps to check it out:
1. Get all the stuff ready, but first make a boot disk with the
*original* Minix versions of bootblok, kernel, mm, fs, init and
fsck, all carefully moved from Minix to MSDOS with doswrite.
2. Under MSDOS, use the new BUILD (Minix build + build.dif) to make
a boot disk. Use a new, formatted disk. It builds fine. Boot
it up. It works fine. Eureka! I can make a disk OK from MSDOS.
3. Now go through and compile/assemble/link all Minix source material
necessary to make boot disk, using Turbo C. Change the .out files
so that Build does not squawk, and build another Minix boot disk.
This time, using the Turbo C versions of bootblok, kernel, mm etc.
Use the same disk. But not formatted again, as I never do that.
4. Boot it up. It starts loading, disk light goes off, and the system
is hung tighter than a drum. CONCLUSION: Something is wrong with
the Turbo C compiled material.
5. Spend a considerable amount of time figuring out how to debug Minix
while it is loading. Conclusion: get access to a Zenith AT, which
has a built-in debug program in ROM. Learn to use it. Find out
that fsck is not where it is supposed to be. Put out cry for help.
Get told that MSDOS uses a different scheme if the disk has not been
formatted again.
MORAL: Always format the disk before using BUILD with the Turbo C material.
John Nall