> The reason I brought this all up is that XX0 access would not work for me.
>
> The disk had a dangerously dedicated label, but I wanted to overwrite the
> front of the disk. Impossible. I've noticed this also in the case where
> you have slices but want to go to a dangerously dedicated label- no can
> do.
The second case should work even on alphas unless the first slice already
begins at offset 0. The labels and their write protection will go away
when the slice table for the dangerously dedicated disk is written (and
synced by closing all minors on the disk or by calling the ioctl for this
as in sysinstall). At worst you resurrect an old label if you didn't
clear it before changing the slice table.
> So, what's the answer about what to do? I sure wouldn't want to leap
> in and 'fix' it because I don't have a good feel for the ins and outs
> of this stuff here (odd- I usually have a strong sense of knowing
> what's right, but this house of cards gives me the creeps).
The correct fix for the immediate problem of write protection on XX0 is
to remove the alpha ifdefs in subr_diskslice.c and then properly fix the
bug in the alpha sysinstall which caused them to be hacked in. sysinstall
and libdisk have an alarming number of __alpha__ ifdefs. The dangerously
dedicated case seems to be required, so it should be possible to use XX0c
instad of XX0 for almost everything. The problem is probably that XX0c
doesn't exist until you write suitable magic to XX0 (XX0c is the 'c'
partition on the first BSD slice, so it doesn't exist if there are no BSD
slices). For the i386 the boot sector contains suitable magic.
Bruce
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
