NetBSD/sys/lkm/arch/mac68k/iwm/TODO

72 lines
1.9 KiB
Plaintext

$NetBSD: TODO,v 1.2 1999/03/27 05:48:53 scottr Exp $
981126
o Change file naming pattern: Other drivers (adb, if) use a prefix.
We could use
iwm.s (was: iwm.s)
iwm_regs.s (was: iwmRegs.s)
iwm_fd.c (was: fd.c)
iwm_fdvar.h (was: fdvar.h)
iwm_reg.h (was: iwmreg.h)
and "swim_*" when it comes to it...
The lkm sources would then be
iwm_mod.c (was: iwmModule.c)
iwm_mod.h (was: fdmod.h)
o location: Is obio fine, or should the driver files be placed in dev?
o LKM location: Is /sys/arch/mac68k/lkm appropriate or should the
lkm rather go to sys/lkm?
980607
o Do statistics in fdstart() about retries and errors.
o send-pr because disklabel(8) displays transition times in milliseconds
whereas <sys/disklabel.h> has them in microseconds (us).
980606
o The Performa 630, a CLASS_Q2 machine according to mac68k/mac68k/machdep.c,
fails to attach (initialize) the IWM. Apparently the driver tickles
the right spot, though: Booting with an inserted disk lets the driver
"find" two (unsupported) HD drives. What IWM chip does that box have?
[SWIM II]
Has Apple changed the IWM register layout with introducing
manual-insertion floppy drives?
[Yes; All 040 Macs need System 7 which does not fit on an 800k disk,
so the 800k code is legacy and need not be present at boot time.]
980602
o send-pr because static array fscknames[] cannot be disabled in
<sys/disklabel.h>
[FIXED as of 11/98 by J. Thorpe]
o send-pr because <bsd.kmod.mk> tries to unload a kernel module
under a different name than it was loaded and fails miserably.
[bin/4701]
o finish manpage
o Just how much KNF can I take (CAPITALIZATION, __P() noise) ?
Old stuff
o Consistent motor on/off management
o Order of responsibilities for iwm_softc_t <-> fd_softc_t
o Service SCC during floppy access
o Implement track buffer
o Support SWIM mode (720K/1440K MFM)
o Check for IWM/SWIM (original 800k Mac II's)
o Test driver with 2 drives