On Tue, 1 Dec 1998, Colin Wood wrote:

> there are inline asm routines in /sys/arch/mac68k/dev/adb_direct.c, tho,
> and i have a feeling that perhaps the problem lies in that file?  routines
> in that file are most likely going to be the ones being run at the time
> you see the hang...

Kernels made with egcs-1.1.1 hang if and only if adb.o is made with
"-O2". Moreover, of all the functions defined in
/sys/arch/mac68k/dev/adb_direct.c, only _adbattach and _adbprint are
in any way different when compiled with -O2 vs -O. That's all I have
right now.