DEFAULTSETS, SANESETS and THESETS are set in the install.md scripts
according to the same logic for all architectures but sgi.

Consolidate setting them in one place in install.sub and incorporate
the sgi specific MDSETS and MDSANESETS.

Introduce MDSANESETS complementing MDSETS which makes the construction
of DEFAULTSETS, THESETS and SANESETS clearer in install.sub.

While here, adjust the comment about how MDSETS used now.

OK deraadt@

Move files created by root during install/upgrade in a subdir of
/tmp with proper permissions so that unprivileged programs can not
tamper with them.

positive feedback from deraadt
OK halex

If manual disklabel editing is requested during install, don't emit
verbiage explaining what disklabels are.

If you don't know, you should be using (A)uto!

ok rpe@ halex@

Remove the $FSTABFLAG variable and use -F directly. The installer
uses disklabel UIDs unconditionally for a while already.

OK krw@
"looks good" deraadt@

Replace identical code in the MD scripts of the installer with a new
MI function disklabel_autolayout() which now handles all aspects of
the disklabel auto-layout and autopartitioning case for the root disk.
Remove get_disklabel_template() and merge it with the new function.

"move forward" deraadt@

Add the new template file based autopartitioning feature of disklabel(8)
to the OpenBSD installer. It is available during unattended installation.
The template file is fetched from an url, provided as answer to a new
question in the response file:

URL to autopartitioning template for disklabel = url

Original diff from and OK henning@
'no objection' krw@

last rev fixed upgrades, but broke installs, sigh. use on-bsd.rd sdboot
for LIF/header creation, pre-disklabel. post-install/upgrade, this
gets redone, this time with -r /mnt to pick up the new sdboot file
ok krw jsing miod

Revert the first chunk of 1.25.

md_prep_disklabel should run disklabel (for fresh disks intended to be
bootable, which do contain a LIF yet). It should also be run in
md_installboot for the upgrade case.
Diagnosis with jsing and kettenis, buglet spotted by rpe
Testing... waiting for that.

Align install.md files with installer style.

- { foo ; bar ; } -> { foo; bar; }
- if foo ; then -> if foo; then

OK halex@

Use the new installboot for hppa installs.

Remove AUTOROOT variable which is a leftover of r1.183 of install.sub

diff from Philipp e1c1bac6253dc54a1e89ddc046585792 at osteo dot net

ok krw@ halex@

since disklabel -W is no more, zap it from all the install.md's

ok krw@

As non-mp install.md's neither set NCPU, nor will have a bsd.mp,
we could as well pull the 'mv bsd.mp bsd' and the associated checks
out of there.

ok deraadt@ "makes sense" todd@

First crack at enabling the installation of a DUID version of
/etc/fstab, after asking user. Current default is existing behaviour.

Feedback & suggestions deraadt@, halex@, jsing@, todd@.

ok deraadt@

swap kernels without changing directory; other install.md should be modified like this

add support for bsd.mp

We want to install a LIF label on the disk before we run disklabel -A
or disklabel -E. Otherwise that disk will indicate the wrong bounds.
(nothing at all -> 0; MBR -> probably 63; LIF -> 256).

Nuke now superfluous ARCH=ARCH lines in install.md and the sed processing of
those lines in list2sh.awk.

ok deraadt@

Add '-h' flag, and '*' as a unit specifier for 'p' Editor command.
Both cause partition sizes to be displayed using a human readable
format with the units automatically chosen by looking at the smallest
partition in the disk label. Remove forceable humanization in 'A'
code and use '-h' in install scripts.

Prodded & ok deraadt@, verbiage tweaks from jmc@

Make bootblock installation as silent as possible by default. While
there, remove a lot of cruft from the various md_installboot functions
ok halex

Remind people that mountpoints must now be entered using the disklabel
command; the script does not ask afterwards. Note I am not adjusting
the macppc or sgi install.md scripts. I ask their maintainers to clean
them before I will maintain them further. Please

Tweak pattern used to select partition lines for display so the
'# /dev/r...' line is not selected.

Noticed by deraadt@

No longer need '-f' or '-p' when displaying layout. A few patterns
missed the '# size offset ...' line at the top of the partition
list.

Where appropriate display the auto-allocation layout of the root
disk and ask if the installee wants to use it, edit it, or create
their own custom disklabel. Most one-disk installs will not need
to see fdisk or disklabel.

i386 and macppc by me, adapted for others by deraadt@.

ok deraadt@

No point in checking disklabel output for 'disk label corrupted' message
since it isn't emitted anymore.

ok deraadt@

Redo serial console configuration logic. Smaller, easier to
understand.

Add serial console handling for alpha, macppc, zaurus. No functional
change for i386/amd64.

All archs should now have automatic serial console configuration.

ok deraadt@

Use new skeleton kbd(8) '-l' to generalize kbd(8) use. Eliminate
fixed list of available maps and machine dependant md_set_term()
functions.

Any ramdisk with /sbin/kbd present (amd64, cats, i386 at the moment)
will now present list of available keyboard encodings.

ok deraadt@

fly on defaults for MDDISKDEVS and MDCDDEVS

Cleanup and shrinkage:

1) Eliminate 'WARNING: Disk xxx has no label.' message. When installing OpenBSD
this is a normal condition not worth commenting on. From some discussion on
hackers/icb.

2) Use consistant verbiage and case statement when checking the disklabel.

3) Consistantly suppress output of 'disklabel -W', reducing duplicate messages
like '# using MBR partition ...' which are issued again when the 'disklabel -f
...' command is executed.

4) Usual code rectifications - eliminate extra {}'s, multiple echos
elimination, etc.

Fix extraneous display of grep output when searching disklabel output for
error information. Previously things like:

disklabel: no disk label
WARNING: Disk wd0 has no disk label. You will be creating a new one.

would be printed. Now the 'disklabel: no disk label' message is correctly
sent to /dev/null.

Take the opportunity to clean up a bit of code and formatting, making all
the md_checkfordisklabel() functions as identical as possible.

Excise unneeded, unused code.

ok mickey@.

this allows to build a ramdiskNN.lif