Re: nfblock vs udev
Due to that type of work, hobbes is currently unavaliable as buildd, BTW.
Could you please test to see if debian's atari-bootstrap works on your
Tested this - when running the bootstrap from floppy as GEM program
(calling it as bootstrap.prg), the system reboots spontaneously (and will
run through a memtest after that, so it's really rather like a cold boot).
Changing the name of the bootstrap to bootstrap.tos succeeds ib booting
The atari-bootstrap binary hasn't been recompiled since sarge, but I
have an open bug report that it doesn't work with CT-60's.
Probably time to look into rebuilding that toolchain.
The 6.0snapshot version that is in the archive probably is even identical
to the one I used all along. I would suspect an interaction with NVDI or
GEM rather than a bug in the bootstrap.
One thing that struck me as odd: the decompression of the kernel image
went a hell of a lot faster when run as GEM program. The kernel should be
loaded to ST-RAm (-s flag) but it looked as though it wasn't.
What details does the bug report give? Kernel hang, or reboot? The kernel
hang was quite a common thing for me with Debian kernels due to the
botched keyboard init. If the bug reporter has old hardware extensions
that hang off the ACIAs, and reports a hang it may well be that my latest
patch to linux-m68k will fix that symptom.
(Apologies if I seem unresponsive at this time - the mail system at my old
lab is migrated to postfix by a new admin, and that often results in no
mail going out for a few days ...)