Retire -E's "expert" mode. Introduced 23 years ago to avoid
confusing users with FFS attributes that only experts should
fiddle with. Actual use has withered away with functionality
rendered moot or moved elsewhere.

'-e' remains for the truly obscure corner cases.

Simply excise the code for now to see if hidden users/uses are
exposed. Further simplifications are possible if no such
users/uses surface.

ok with sthen@ millert@ kn@ otto@

Pass make flags to kernel and lib builds

Running `make -j4' in /usr/src/distrib/amd64/ramdisk_cd/ et al. executes
make(1) to both build the RAMDISK kernel and build libraries.

Doing so does not propagate the flags specified to the ramdisk_cd
invocation, which in turn means `-j4' for example is ignored and both kernel
and libraries will not be built in parallel.

Pass make(1)'s MFLAGS along to retain relevant flags; make is clever enough
to separate flags, variable assignments and targets from each other and only
pass along things to `MFLAGS' that'd make sense, i.e. `make -C. -j4 foo=bar'
does *not* pass `-C.' to change directories.
(can be easily tested with `make -p ... | grep MFLAGS'.)

This makes hacking on ramdisks/the installer much faster, espescially since
the `bsd' target does `make clean' and therefore builds a new kernel every
time.

OK deraadt

distrib: use ${MACHINE} instead of hardcoded value

ok deraadt@

Change install images called *.fs to *.img. These are UFS filesystem images,
but additionally have a bootblock in the first 8K (since UFS does not use that
space). There are some UEFI direct-from-internet bootloaders that require
the name *.img. So this makes things more convenient for those, while keeping
it consistant in all architectures.
ok kettenis beck kn

The list + list.local approach is only needed if an architecture builds
multiple bsd.rd. hppa builds one. Fold into one file.

Further removal of differences (using install instead of cp, sorting
chunks in a specified order, using a few variables, performing bsd.rd
to bsd.gz conversion as similar as possible)

repair more trivial differences

rm is in /bin, no need to specify

Every Makefile on every architecture had deviated in unique ways.
Eliminate many more differences. When it makes sense, build bsd.rd
and miniroot/ramdisk.fs and cd.iso in the same directory. More steps
coming after this..

remove the abstraction REV on top of abstraction OSrev which is always
the sme.

incorporate GZIPFLAGS into place everywhere

CBIN variable is also pointless, it is always "instbin". Rather than
having two mysterious names, let's settle on one.

Use new vnconfig vnd-auto-allocate mode. Resolve some variation between
architectures, and start removing some crazy junk that has collected
over the years. Being tested on all architectures...
ok various people.

Use new rdsetroot.

drop ${GZIPCMD} and just use 'gzip'

drop ${STRIP}, it's just 'strip'

ok deraadt@

always strip the ctf section from ramdisk kernels

discussed with deraadt@ mpi@
ok deraadt@ kettenis@ mpi@

kernels in compile dirs are readonly, and the bsd.rd we created has
tracked that permission. Make the copy in $RELEASEDIR a+r

Install cert.pem + ftp(1) binary that supports SSL, so the install script
can perform https fetches. While non-critical for sets (which are verified
using signify), it is an uplift for communication towards the ftp.openbsd.org
cgi which provides mirror proposals, entropy seeds, and current time.
[The install script changes are not included in this commit]
ok beck jsing rpe

Use makefs to build bsd.rd; build & boot tested by krw

Use new kernel compile method for ramdisks, de-escalating to $BUILDUSER
as required.
much help from natano and tb.

Append 512 bytes of zero to the kernel image intended to be put into lif.fs,
so that valid bytes exist at the end of the kernel image.

Since the lif image stores files size in 256-byte unit, we can not report EOF
while reading the kernel image until the remainder of the 256-byte sector has
been read. But these zero bytes would cause the gzip readahead machinery in
libsa to believe the kernel is followed by an uncompressed file, and switch to
transparent mode, preventing backwards seeks into the kernel image from working.

This fixes the ``read section header string table failed(0)'' errors when
attempting to boot lif.fs since loadfile_elf.c 1.9.

fix typo

cleanup elfrdsetroot references; ok miod

Turn on the distrib lib bits in crunchgen by setting SRCLIBDIR.
Add the new files to the clean target's rm list.

"why not get it tested?" deraadt@

match latest change by distrib/ramdisk, for extra elfrdsetroot files

Deprecate vnds in favour of svnds.

In effect, this removes the "old" vndX nodes, and renames
the svndX nodes to vndX.

Old svndX nodes will still continue to work though, for now.

Cleanup accordingly.

ok deraadt@, todd@
comments and ok on the man page bits from jmc@

No need to make depend kernels; ok guenther

unify a chunk

Place a smaller spwd.db onto the install media. Be consistant with the naming
of the "additional" list files (for architectures which split the list)
Then... cleanup some TZ which it is clear should not have been placed onto
certain media which don't have the space for them.

Since Aug 2008, the kernel can successfully read the 1 block (512
byte) disklabels even in the midst of a larger sector. E.g. in bytes
512 - 1023 of the first 2048 byte sector on a CD.

On hppa we plopped down such a disklabel on cdXX.iso and installXX.iso.
Once accessible it took precedence over the spoofed label of the CD. Chaos
ensued. Since there is currently no reason for this label on hppa
cd's, eradicate it and its disktab entry. cdXX.iso/installXX.iso can now
be mounted on hppa again.

Also remove a commented out label plopping on Alpha, which references a
non-existant disktab entry.

Problem noted and exact diff creating problem discovered by marco@.

ok deraadt@

further minor unification

For SHA256 hashes to be ready to go onto the install media, we must
build the sets before we build the media. While there we can get rid
of DESTDIR/snapshot too, and simply install straight into RELEASEDIR.
(This also ends up saving an astounding amount of traffic/latency in a nfs
environment)

run make depend for kernel, too; ok miod

do not use the DESTDIR mkboot binary

commit with rdsetroot now accepting the filesystem as a 2nd arg

bsd.rd's physical memory usage collided with the bootblocks... The
bootblocks have been moved, but we want 4.2 to be upgradeable. So
reduce the footprint of the bsd.rd by shrinking the ramdisk filesystem
because it is way too large. tested by sthen and kettenis

cleanup: make -> ${MAKE} consistenly.
zap extra subshells.

okay miod@

Stop using disklabel -r. The disklabel code does not know all the semantics
for pushing disklabels onto the disk (and besides that, it is buggy and
unreadable crap); thanks for help from krw

Update install media generation to the new semantics of the -c flag.
Also replace "newfs -O" with "newfs -O 0" now that -O takes an option.

Add -n to gzip flags. Save a few bytes since we no longer save
the uncompressed filename in the gzip header.

s/GZIP/GZIPCMD/g in `make release` structure to avoid conflict with gzip's
use of a GZIP environment variable for extra command flags.

ok deraadt@, millert@

a quick fix -- pad image w/ extra 64k until boot can cope w/ short reads

never worry about the copyright date on the CD layout again, problem
noticed by drahn

change copyrights on iso images, spotted by miod

for no apparent reason we ran out of inodes

this builds a bootable cdrom w/ a ramdisk kernel. has some problems though

no mdec/boot on the ramdisk. name it chewable for the etc/makefiles

this allows to build a ramdiskNN.lif

Pass make flags to kernel and lib builds

Running `make -j4' in /usr/src/distrib/amd64/ramdisk_cd/ et al. executes
make(1) to both build the RAMDISK kernel and build libraries.

Doing so does not propagate the flags specified to the ramdisk_cd
invocation, which in turn means `-j4' for example is ignored and both kernel
and libraries will not be built in parallel.

Pass make(1)'s MFLAGS along to retain relevant flags; make is clever enough
to separate flags, variable assignments and targets from each other and only
pass along things to `MFLAGS' that'd make sense, i.e. `make -C. -j4 foo=bar'
does *not* pass `-C.' to change directories.
(can be easily tested with `make -p ... | grep MFLAGS'.)

This makes hacking on ramdisks/the installer much faster, espescially since
the `bsd' target does `make clean' and therefore builds a new kernel every
time.

OK deraadt

distrib: use ${MACHINE} instead of hardcoded value

ok deraadt@

Change install images called *.fs to *.img. These are UFS filesystem images,
but additionally have a bootblock in the first 8K (since UFS does not use that
space). There are some UEFI direct-from-internet bootloaders that require
the name *.img. So this makes things more convenient for those, while keeping
it consistant in all architectures.
ok kettenis beck kn

The list + list.local approach is only needed if an architecture builds
multiple bsd.rd. hppa builds one. Fold into one file.

Further removal of differences (using install instead of cp, sorting
chunks in a specified order, using a few variables, performing bsd.rd
to bsd.gz conversion as similar as possible)

repair more trivial differences

rm is in /bin, no need to specify

Every Makefile on every architecture had deviated in unique ways.
Eliminate many more differences. When it makes sense, build bsd.rd
and miniroot/ramdisk.fs and cd.iso in the same directory. More steps
coming after this..

remove the abstraction REV on top of abstraction OSrev which is always
the sme.

incorporate GZIPFLAGS into place everywhere

CBIN variable is also pointless, it is always "instbin". Rather than
having two mysterious names, let's settle on one.

Use new vnconfig vnd-auto-allocate mode. Resolve some variation between
architectures, and start removing some crazy junk that has collected
over the years. Being tested on all architectures...
ok various people.

Use new rdsetroot.

drop ${GZIPCMD} and just use 'gzip'

drop ${STRIP}, it's just 'strip'

ok deraadt@

always strip the ctf section from ramdisk kernels

discussed with deraadt@ mpi@
ok deraadt@ kettenis@ mpi@

kernels in compile dirs are readonly, and the bsd.rd we created has
tracked that permission. Make the copy in $RELEASEDIR a+r

Install cert.pem + ftp(1) binary that supports SSL, so the install script
can perform https fetches. While non-critical for sets (which are verified
using signify), it is an uplift for communication towards the ftp.openbsd.org
cgi which provides mirror proposals, entropy seeds, and current time.
[The install script changes are not included in this commit]
ok beck jsing rpe

Use makefs to build bsd.rd; build & boot tested by krw

Use new kernel compile method for ramdisks, de-escalating to $BUILDUSER
as required.
much help from natano and tb.

Append 512 bytes of zero to the kernel image intended to be put into lif.fs,
so that valid bytes exist at the end of the kernel image.

Since the lif image stores files size in 256-byte unit, we can not report EOF
while reading the kernel image until the remainder of the 256-byte sector has
been read. But these zero bytes would cause the gzip readahead machinery in
libsa to believe the kernel is followed by an uncompressed file, and switch to
transparent mode, preventing backwards seeks into the kernel image from working.

This fixes the ``read section header string table failed(0)'' errors when
attempting to boot lif.fs since loadfile_elf.c 1.9.

fix typo

cleanup elfrdsetroot references; ok miod

Turn on the distrib lib bits in crunchgen by setting SRCLIBDIR.
Add the new files to the clean target's rm list.

"why not get it tested?" deraadt@

match latest change by distrib/ramdisk, for extra elfrdsetroot files

Deprecate vnds in favour of svnds.

In effect, this removes the "old" vndX nodes, and renames
the svndX nodes to vndX.

Old svndX nodes will still continue to work though, for now.

Cleanup accordingly.

ok deraadt@, todd@
comments and ok on the man page bits from jmc@

No need to make depend kernels; ok guenther

unify a chunk

Place a smaller spwd.db onto the install media. Be consistant with the naming
of the "additional" list files (for architectures which split the list)
Then... cleanup some TZ which it is clear should not have been placed onto
certain media which don't have the space for them.

Since Aug 2008, the kernel can successfully read the 1 block (512
byte) disklabels even in the midst of a larger sector. E.g. in bytes
512 - 1023 of the first 2048 byte sector on a CD.

On hppa we plopped down such a disklabel on cdXX.iso and installXX.iso.
Once accessible it took precedence over the spoofed label of the CD. Chaos
ensued. Since there is currently no reason for this label on hppa
cd's, eradicate it and its disktab entry. cdXX.iso/installXX.iso can now
be mounted on hppa again.

Also remove a commented out label plopping on Alpha, which references a
non-existant disktab entry.

Problem noted and exact diff creating problem discovered by marco@.

ok deraadt@

further minor unification

For SHA256 hashes to be ready to go onto the install media, we must
build the sets before we build the media. While there we can get rid
of DESTDIR/snapshot too, and simply install straight into RELEASEDIR.
(This also ends up saving an astounding amount of traffic/latency in a nfs
environment)

run make depend for kernel, too; ok miod

do not use the DESTDIR mkboot binary

commit with rdsetroot now accepting the filesystem as a 2nd arg

bsd.rd's physical memory usage collided with the bootblocks... The
bootblocks have been moved, but we want 4.2 to be upgradeable. So
reduce the footprint of the bsd.rd by shrinking the ramdisk filesystem
because it is way too large. tested by sthen and kettenis

cleanup: make -> ${MAKE} consistenly.
zap extra subshells.

okay miod@

Stop using disklabel -r. The disklabel code does not know all the semantics
for pushing disklabels onto the disk (and besides that, it is buggy and
unreadable crap); thanks for help from krw

Update install media generation to the new semantics of the -c flag.
Also replace "newfs -O" with "newfs -O 0" now that -O takes an option.

Add -n to gzip flags. Save a few bytes since we no longer save
the uncompressed filename in the gzip header.

s/GZIP/GZIPCMD/g in `make release` structure to avoid conflict with gzip's
use of a GZIP environment variable for extra command flags.

ok deraadt@, millert@

a quick fix -- pad image w/ extra 64k until boot can cope w/ short reads

never worry about the copyright date on the CD layout again, problem
noticed by drahn

change copyrights on iso images, spotted by miod

for no apparent reason we ran out of inodes

this builds a bootable cdrom w/ a ramdisk kernel. has some problems though

no mdec/boot on the ramdisk. name it chewable for the etc/makefiles

this allows to build a ramdiskNN.lif

distrib: use ${MACHINE} instead of hardcoded value

ok deraadt@

Change install images called *.fs to *.img. These are UFS filesystem images,
but additionally have a bootblock in the first 8K (since UFS does not use that
space). There are some UEFI direct-from-internet bootloaders that require
the name *.img. So this makes things more convenient for those, while keeping
it consistant in all architectures.
ok kettenis beck kn

The list + list.local approach is only needed if an architecture builds
multiple bsd.rd. hppa builds one. Fold into one file.

Further removal of differences (using install instead of cp, sorting
chunks in a specified order, using a few variables, performing bsd.rd
to bsd.gz conversion as similar as possible)

repair more trivial differences

rm is in /bin, no need to specify

Every Makefile on every architecture had deviated in unique ways.
Eliminate many more differences. When it makes sense, build bsd.rd
and miniroot/ramdisk.fs and cd.iso in the same directory. More steps
coming after this..

remove the abstraction REV on top of abstraction OSrev which is always
the sme.

incorporate GZIPFLAGS into place everywhere

CBIN variable is also pointless, it is always "instbin". Rather than
having two mysterious names, let's settle on one.

Use new vnconfig vnd-auto-allocate mode. Resolve some variation between
architectures, and start removing some crazy junk that has collected
over the years. Being tested on all architectures...
ok various people.

Use new rdsetroot.

drop ${GZIPCMD} and just use 'gzip'

drop ${STRIP}, it's just 'strip'

ok deraadt@

always strip the ctf section from ramdisk kernels

discussed with deraadt@ mpi@
ok deraadt@ kettenis@ mpi@

kernels in compile dirs are readonly, and the bsd.rd we created has
tracked that permission. Make the copy in $RELEASEDIR a+r

Install cert.pem + ftp(1) binary that supports SSL, so the install script
can perform https fetches. While non-critical for sets (which are verified
using signify), it is an uplift for communication towards the ftp.openbsd.org
cgi which provides mirror proposals, entropy seeds, and current time.
[The install script changes are not included in this commit]
ok beck jsing rpe

Use makefs to build bsd.rd; build & boot tested by krw

Use new kernel compile method for ramdisks, de-escalating to $BUILDUSER
as required.
much help from natano and tb.

Append 512 bytes of zero to the kernel image intended to be put into lif.fs,
so that valid bytes exist at the end of the kernel image.

Since the lif image stores files size in 256-byte unit, we can not report EOF
while reading the kernel image until the remainder of the 256-byte sector has
been read. But these zero bytes would cause the gzip readahead machinery in
libsa to believe the kernel is followed by an uncompressed file, and switch to
transparent mode, preventing backwards seeks into the kernel image from working.

This fixes the ``read section header string table failed(0)'' errors when
attempting to boot lif.fs since loadfile_elf.c 1.9.

fix typo

cleanup elfrdsetroot references; ok miod

Turn on the distrib lib bits in crunchgen by setting SRCLIBDIR.
Add the new files to the clean target's rm list.

"why not get it tested?" deraadt@

match latest change by distrib/ramdisk, for extra elfrdsetroot files

Deprecate vnds in favour of svnds.

In effect, this removes the "old" vndX nodes, and renames
the svndX nodes to vndX.

Old svndX nodes will still continue to work though, for now.

Cleanup accordingly.

ok deraadt@, todd@
comments and ok on the man page bits from jmc@

No need to make depend kernels; ok guenther

unify a chunk

Place a smaller spwd.db onto the install media. Be consistant with the naming
of the "additional" list files (for architectures which split the list)
Then... cleanup some TZ which it is clear should not have been placed onto
certain media which don't have the space for them.

Since Aug 2008, the kernel can successfully read the 1 block (512
byte) disklabels even in the midst of a larger sector. E.g. in bytes
512 - 1023 of the first 2048 byte sector on a CD.

On hppa we plopped down such a disklabel on cdXX.iso and installXX.iso.
Once accessible it took precedence over the spoofed label of the CD. Chaos
ensued. Since there is currently no reason for this label on hppa
cd's, eradicate it and its disktab entry. cdXX.iso/installXX.iso can now
be mounted on hppa again.

Also remove a commented out label plopping on Alpha, which references a
non-existant disktab entry.

Problem noted and exact diff creating problem discovered by marco@.

ok deraadt@

further minor unification

For SHA256 hashes to be ready to go onto the install media, we must
build the sets before we build the media. While there we can get rid
of DESTDIR/snapshot too, and simply install straight into RELEASEDIR.
(This also ends up saving an astounding amount of traffic/latency in a nfs
environment)

run make depend for kernel, too; ok miod

do not use the DESTDIR mkboot binary

commit with rdsetroot now accepting the filesystem as a 2nd arg

bsd.rd's physical memory usage collided with the bootblocks... The
bootblocks have been moved, but we want 4.2 to be upgradeable. So
reduce the footprint of the bsd.rd by shrinking the ramdisk filesystem
because it is way too large. tested by sthen and kettenis

cleanup: make -> ${MAKE} consistenly.
zap extra subshells.

okay miod@

Stop using disklabel -r. The disklabel code does not know all the semantics
for pushing disklabels onto the disk (and besides that, it is buggy and
unreadable crap); thanks for help from krw

Update install media generation to the new semantics of the -c flag.
Also replace "newfs -O" with "newfs -O 0" now that -O takes an option.

Add -n to gzip flags. Save a few bytes since we no longer save
the uncompressed filename in the gzip header.

s/GZIP/GZIPCMD/g in `make release` structure to avoid conflict with gzip's
use of a GZIP environment variable for extra command flags.

ok deraadt@, millert@

a quick fix -- pad image w/ extra 64k until boot can cope w/ short reads

never worry about the copyright date on the CD layout again, problem
noticed by drahn

change copyrights on iso images, spotted by miod

for no apparent reason we ran out of inodes

this builds a bootable cdrom w/ a ramdisk kernel. has some problems though

no mdec/boot on the ramdisk. name it chewable for the etc/makefiles

this allows to build a ramdiskNN.lif

Change install images called *.fs to *.img. These are UFS filesystem images,
but additionally have a bootblock in the first 8K (since UFS does not use that
space). There are some UEFI direct-from-internet bootloaders that require
the name *.img. So this makes things more convenient for those, while keeping
it consistant in all architectures.
ok kettenis beck kn

The list + list.local approach is only needed if an architecture builds
multiple bsd.rd. hppa builds one. Fold into one file.

Further removal of differences (using install instead of cp, sorting
chunks in a specified order, using a few variables, performing bsd.rd
to bsd.gz conversion as similar as possible)

repair more trivial differences

rm is in /bin, no need to specify

Every Makefile on every architecture had deviated in unique ways.
Eliminate many more differences. When it makes sense, build bsd.rd
and miniroot/ramdisk.fs and cd.iso in the same directory. More steps
coming after this..

remove the abstraction REV on top of abstraction OSrev which is always
the sme.

incorporate GZIPFLAGS into place everywhere

CBIN variable is also pointless, it is always "instbin". Rather than
having two mysterious names, let's settle on one.

Use new vnconfig vnd-auto-allocate mode. Resolve some variation between
architectures, and start removing some crazy junk that has collected
over the years. Being tested on all architectures...
ok various people.

Use new rdsetroot.

drop ${GZIPCMD} and just use 'gzip'

drop ${STRIP}, it's just 'strip'

ok deraadt@

always strip the ctf section from ramdisk kernels

discussed with deraadt@ mpi@
ok deraadt@ kettenis@ mpi@

kernels in compile dirs are readonly, and the bsd.rd we created has
tracked that permission. Make the copy in $RELEASEDIR a+r

Install cert.pem + ftp(1) binary that supports SSL, so the install script
can perform https fetches. While non-critical for sets (which are verified
using signify), it is an uplift for communication towards the ftp.openbsd.org
cgi which provides mirror proposals, entropy seeds, and current time.
[The install script changes are not included in this commit]
ok beck jsing rpe

Use makefs to build bsd.rd; build & boot tested by krw

Use new kernel compile method for ramdisks, de-escalating to $BUILDUSER
as required.
much help from natano and tb.

Append 512 bytes of zero to the kernel image intended to be put into lif.fs,
so that valid bytes exist at the end of the kernel image.

Since the lif image stores files size in 256-byte unit, we can not report EOF
while reading the kernel image until the remainder of the 256-byte sector has
been read. But these zero bytes would cause the gzip readahead machinery in
libsa to believe the kernel is followed by an uncompressed file, and switch to
transparent mode, preventing backwards seeks into the kernel image from working.

This fixes the ``read section header string table failed(0)'' errors when
attempting to boot lif.fs since loadfile_elf.c 1.9.

fix typo

cleanup elfrdsetroot references; ok miod

Turn on the distrib lib bits in crunchgen by setting SRCLIBDIR.
Add the new files to the clean target's rm list.

"why not get it tested?" deraadt@

match latest change by distrib/ramdisk, for extra elfrdsetroot files

Deprecate vnds in favour of svnds.

In effect, this removes the "old" vndX nodes, and renames
the svndX nodes to vndX.

Old svndX nodes will still continue to work though, for now.

Cleanup accordingly.

ok deraadt@, todd@
comments and ok on the man page bits from jmc@

No need to make depend kernels; ok guenther

unify a chunk

Place a smaller spwd.db onto the install media. Be consistant with the naming
of the "additional" list files (for architectures which split the list)
Then... cleanup some TZ which it is clear should not have been placed onto
certain media which don't have the space for them.

Since Aug 2008, the kernel can successfully read the 1 block (512
byte) disklabels even in the midst of a larger sector. E.g. in bytes
512 - 1023 of the first 2048 byte sector on a CD.

On hppa we plopped down such a disklabel on cdXX.iso and installXX.iso.
Once accessible it took precedence over the spoofed label of the CD. Chaos
ensued. Since there is currently no reason for this label on hppa
cd's, eradicate it and its disktab entry. cdXX.iso/installXX.iso can now
be mounted on hppa again.

Also remove a commented out label plopping on Alpha, which references a
non-existant disktab entry.

Problem noted and exact diff creating problem discovered by marco@.

ok deraadt@

further minor unification

For SHA256 hashes to be ready to go onto the install media, we must
build the sets before we build the media. While there we can get rid
of DESTDIR/snapshot too, and simply install straight into RELEASEDIR.
(This also ends up saving an astounding amount of traffic/latency in a nfs
environment)

run make depend for kernel, too; ok miod

do not use the DESTDIR mkboot binary

commit with rdsetroot now accepting the filesystem as a 2nd arg

bsd.rd's physical memory usage collided with the bootblocks... The
bootblocks have been moved, but we want 4.2 to be upgradeable. So
reduce the footprint of the bsd.rd by shrinking the ramdisk filesystem
because it is way too large. tested by sthen and kettenis

cleanup: make -> ${MAKE} consistenly.
zap extra subshells.

okay miod@

Stop using disklabel -r. The disklabel code does not know all the semantics
for pushing disklabels onto the disk (and besides that, it is buggy and
unreadable crap); thanks for help from krw

Update install media generation to the new semantics of the -c flag.
Also replace "newfs -O" with "newfs -O 0" now that -O takes an option.

Add -n to gzip flags. Save a few bytes since we no longer save
the uncompressed filename in the gzip header.

s/GZIP/GZIPCMD/g in `make release` structure to avoid conflict with gzip's
use of a GZIP environment variable for extra command flags.

ok deraadt@, millert@

a quick fix -- pad image w/ extra 64k until boot can cope w/ short reads

never worry about the copyright date on the CD layout again, problem
noticed by drahn

change copyrights on iso images, spotted by miod

for no apparent reason we ran out of inodes

this builds a bootable cdrom w/ a ramdisk kernel. has some problems though

no mdec/boot on the ramdisk. name it chewable for the etc/makefiles

this allows to build a ramdiskNN.lif

The list + list.local approach is only needed if an architecture builds
multiple bsd.rd. hppa builds one. Fold into one file.

Further removal of differences (using install instead of cp, sorting
chunks in a specified order, using a few variables, performing bsd.rd
to bsd.gz conversion as similar as possible)

repair more trivial differences

rm is in /bin, no need to specify

Every Makefile on every architecture had deviated in unique ways.
Eliminate many more differences. When it makes sense, build bsd.rd
and miniroot/ramdisk.fs and cd.iso in the same directory. More steps
coming after this..

remove the abstraction REV on top of abstraction OSrev which is always
the sme.

incorporate GZIPFLAGS into place everywhere

CBIN variable is also pointless, it is always "instbin". Rather than
having two mysterious names, let's settle on one.

Use new vnconfig vnd-auto-allocate mode. Resolve some variation between
architectures, and start removing some crazy junk that has collected
over the years. Being tested on all architectures...
ok various people.

Use new rdsetroot.

drop ${GZIPCMD} and just use 'gzip'

drop ${STRIP}, it's just 'strip'

ok deraadt@

always strip the ctf section from ramdisk kernels

discussed with deraadt@ mpi@
ok deraadt@ kettenis@ mpi@

kernels in compile dirs are readonly, and the bsd.rd we created has
tracked that permission. Make the copy in $RELEASEDIR a+r

Install cert.pem + ftp(1) binary that supports SSL, so the install script
can perform https fetches. While non-critical for sets (which are verified
using signify), it is an uplift for communication towards the ftp.openbsd.org
cgi which provides mirror proposals, entropy seeds, and current time.
[The install script changes are not included in this commit]
ok beck jsing rpe

Use makefs to build bsd.rd; build & boot tested by krw

Use new kernel compile method for ramdisks, de-escalating to $BUILDUSER
as required.
much help from natano and tb.

Append 512 bytes of zero to the kernel image intended to be put into lif.fs,
so that valid bytes exist at the end of the kernel image.

Since the lif image stores files size in 256-byte unit, we can not report EOF
while reading the kernel image until the remainder of the 256-byte sector has
been read. But these zero bytes would cause the gzip readahead machinery in
libsa to believe the kernel is followed by an uncompressed file, and switch to
transparent mode, preventing backwards seeks into the kernel image from working.

This fixes the ``read section header string table failed(0)'' errors when
attempting to boot lif.fs since loadfile_elf.c 1.9.

fix typo

cleanup elfrdsetroot references; ok miod

Turn on the distrib lib bits in crunchgen by setting SRCLIBDIR.
Add the new files to the clean target's rm list.

"why not get it tested?" deraadt@

match latest change by distrib/ramdisk, for extra elfrdsetroot files

Deprecate vnds in favour of svnds.

In effect, this removes the "old" vndX nodes, and renames
the svndX nodes to vndX.

Old svndX nodes will still continue to work though, for now.

Cleanup accordingly.

ok deraadt@, todd@
comments and ok on the man page bits from jmc@

No need to make depend kernels; ok guenther

unify a chunk

Place a smaller spwd.db onto the install media. Be consistant with the naming
of the "additional" list files (for architectures which split the list)
Then... cleanup some TZ which it is clear should not have been placed onto
certain media which don't have the space for them.

Since Aug 2008, the kernel can successfully read the 1 block (512
byte) disklabels even in the midst of a larger sector. E.g. in bytes
512 - 1023 of the first 2048 byte sector on a CD.

On hppa we plopped down such a disklabel on cdXX.iso and installXX.iso.
Once accessible it took precedence over the spoofed label of the CD. Chaos
ensued. Since there is currently no reason for this label on hppa
cd's, eradicate it and its disktab entry. cdXX.iso/installXX.iso can now
be mounted on hppa again.

Also remove a commented out label plopping on Alpha, which references a
non-existant disktab entry.

Problem noted and exact diff creating problem discovered by marco@.

ok deraadt@

further minor unification

For SHA256 hashes to be ready to go onto the install media, we must
build the sets before we build the media. While there we can get rid
of DESTDIR/snapshot too, and simply install straight into RELEASEDIR.
(This also ends up saving an astounding amount of traffic/latency in a nfs
environment)

run make depend for kernel, too; ok miod

do not use the DESTDIR mkboot binary

commit with rdsetroot now accepting the filesystem as a 2nd arg

bsd.rd's physical memory usage collided with the bootblocks... The
bootblocks have been moved, but we want 4.2 to be upgradeable. So
reduce the footprint of the bsd.rd by shrinking the ramdisk filesystem
because it is way too large. tested by sthen and kettenis

cleanup: make -> ${MAKE} consistenly.
zap extra subshells.

okay miod@

Stop using disklabel -r. The disklabel code does not know all the semantics
for pushing disklabels onto the disk (and besides that, it is buggy and
unreadable crap); thanks for help from krw

Update install media generation to the new semantics of the -c flag.
Also replace "newfs -O" with "newfs -O 0" now that -O takes an option.

Add -n to gzip flags. Save a few bytes since we no longer save
the uncompressed filename in the gzip header.

s/GZIP/GZIPCMD/g in `make release` structure to avoid conflict with gzip's
use of a GZIP environment variable for extra command flags.

ok deraadt@, millert@

a quick fix -- pad image w/ extra 64k until boot can cope w/ short reads

never worry about the copyright date on the CD layout again, problem
noticed by drahn

change copyrights on iso images, spotted by miod

for no apparent reason we ran out of inodes

this builds a bootable cdrom w/ a ramdisk kernel. has some problems though

no mdec/boot on the ramdisk. name it chewable for the etc/makefiles

this allows to build a ramdiskNN.lif

Further removal of differences (using install instead of cp, sorting
chunks in a specified order, using a few variables, performing bsd.rd
to bsd.gz conversion as similar as possible)

repair more trivial differences

rm is in /bin, no need to specify

Every Makefile on every architecture had deviated in unique ways.
Eliminate many more differences. When it makes sense, build bsd.rd
and miniroot/ramdisk.fs and cd.iso in the same directory. More steps
coming after this..

remove the abstraction REV on top of abstraction OSrev which is always
the sme.

incorporate GZIPFLAGS into place everywhere

CBIN variable is also pointless, it is always "instbin". Rather than
having two mysterious names, let's settle on one.

Use new vnconfig vnd-auto-allocate mode. Resolve some variation between
architectures, and start removing some crazy junk that has collected
over the years. Being tested on all architectures...
ok various people.

Use new rdsetroot.

drop ${GZIPCMD} and just use 'gzip'

drop ${STRIP}, it's just 'strip'

ok deraadt@

always strip the ctf section from ramdisk kernels

discussed with deraadt@ mpi@
ok deraadt@ kettenis@ mpi@

kernels in compile dirs are readonly, and the bsd.rd we created has
tracked that permission. Make the copy in $RELEASEDIR a+r

Install cert.pem + ftp(1) binary that supports SSL, so the install script
can perform https fetches. While non-critical for sets (which are verified
using signify), it is an uplift for communication towards the ftp.openbsd.org
cgi which provides mirror proposals, entropy seeds, and current time.
[The install script changes are not included in this commit]
ok beck jsing rpe

Use makefs to build bsd.rd; build & boot tested by krw

Use new kernel compile method for ramdisks, de-escalating to $BUILDUSER
as required.
much help from natano and tb.

Append 512 bytes of zero to the kernel image intended to be put into lif.fs,
so that valid bytes exist at the end of the kernel image.

Since the lif image stores files size in 256-byte unit, we can not report EOF
while reading the kernel image until the remainder of the 256-byte sector has
been read. But these zero bytes would cause the gzip readahead machinery in
libsa to believe the kernel is followed by an uncompressed file, and switch to
transparent mode, preventing backwards seeks into the kernel image from working.

This fixes the ``read section header string table failed(0)'' errors when
attempting to boot lif.fs since loadfile_elf.c 1.9.

fix typo

cleanup elfrdsetroot references; ok miod

Turn on the distrib lib bits in crunchgen by setting SRCLIBDIR.
Add the new files to the clean target's rm list.

"why not get it tested?" deraadt@

match latest change by distrib/ramdisk, for extra elfrdsetroot files

Deprecate vnds in favour of svnds.

In effect, this removes the "old" vndX nodes, and renames
the svndX nodes to vndX.

Old svndX nodes will still continue to work though, for now.

Cleanup accordingly.

ok deraadt@, todd@
comments and ok on the man page bits from jmc@

No need to make depend kernels; ok guenther

unify a chunk

Place a smaller spwd.db onto the install media. Be consistant with the naming
of the "additional" list files (for architectures which split the list)
Then... cleanup some TZ which it is clear should not have been placed onto
certain media which don't have the space for them.

Since Aug 2008, the kernel can successfully read the 1 block (512
byte) disklabels even in the midst of a larger sector. E.g. in bytes
512 - 1023 of the first 2048 byte sector on a CD.

On hppa we plopped down such a disklabel on cdXX.iso and installXX.iso.
Once accessible it took precedence over the spoofed label of the CD. Chaos
ensued. Since there is currently no reason for this label on hppa
cd's, eradicate it and its disktab entry. cdXX.iso/installXX.iso can now
be mounted on hppa again.

Also remove a commented out label plopping on Alpha, which references a
non-existant disktab entry.

Problem noted and exact diff creating problem discovered by marco@.

ok deraadt@

further minor unification

For SHA256 hashes to be ready to go onto the install media, we must
build the sets before we build the media. While there we can get rid
of DESTDIR/snapshot too, and simply install straight into RELEASEDIR.
(This also ends up saving an astounding amount of traffic/latency in a nfs
environment)

run make depend for kernel, too; ok miod

do not use the DESTDIR mkboot binary

commit with rdsetroot now accepting the filesystem as a 2nd arg

bsd.rd's physical memory usage collided with the bootblocks... The
bootblocks have been moved, but we want 4.2 to be upgradeable. So
reduce the footprint of the bsd.rd by shrinking the ramdisk filesystem
because it is way too large. tested by sthen and kettenis

cleanup: make -> ${MAKE} consistenly.
zap extra subshells.

okay miod@

Stop using disklabel -r. The disklabel code does not know all the semantics
for pushing disklabels onto the disk (and besides that, it is buggy and
unreadable crap); thanks for help from krw

Update install media generation to the new semantics of the -c flag.
Also replace "newfs -O" with "newfs -O 0" now that -O takes an option.

Add -n to gzip flags. Save a few bytes since we no longer save
the uncompressed filename in the gzip header.

s/GZIP/GZIPCMD/g in `make release` structure to avoid conflict with gzip's
use of a GZIP environment variable for extra command flags.

ok deraadt@, millert@

a quick fix -- pad image w/ extra 64k until boot can cope w/ short reads

never worry about the copyright date on the CD layout again, problem
noticed by drahn

change copyrights on iso images, spotted by miod

for no apparent reason we ran out of inodes

this builds a bootable cdrom w/ a ramdisk kernel. has some problems though

no mdec/boot on the ramdisk. name it chewable for the etc/makefiles

this allows to build a ramdiskNN.lif

repair more trivial differences

rm is in /bin, no need to specify

Every Makefile on every architecture had deviated in unique ways.
Eliminate many more differences. When it makes sense, build bsd.rd
and miniroot/ramdisk.fs and cd.iso in the same directory. More steps
coming after this..

remove the abstraction REV on top of abstraction OSrev which is always
the sme.

incorporate GZIPFLAGS into place everywhere

CBIN variable is also pointless, it is always "instbin". Rather than
having two mysterious names, let's settle on one.

Use new vnconfig vnd-auto-allocate mode. Resolve some variation between
architectures, and start removing some crazy junk that has collected
over the years. Being tested on all architectures...
ok various people.

Use new rdsetroot.

drop ${GZIPCMD} and just use 'gzip'

drop ${STRIP}, it's just 'strip'

ok deraadt@

always strip the ctf section from ramdisk kernels

discussed with deraadt@ mpi@
ok deraadt@ kettenis@ mpi@

kernels in compile dirs are readonly, and the bsd.rd we created has
tracked that permission. Make the copy in $RELEASEDIR a+r

Install cert.pem + ftp(1) binary that supports SSL, so the install script
can perform https fetches. While non-critical for sets (which are verified
using signify), it is an uplift for communication towards the ftp.openbsd.org
cgi which provides mirror proposals, entropy seeds, and current time.
[The install script changes are not included in this commit]
ok beck jsing rpe

Use makefs to build bsd.rd; build & boot tested by krw

Use new kernel compile method for ramdisks, de-escalating to $BUILDUSER
as required.
much help from natano and tb.

Append 512 bytes of zero to the kernel image intended to be put into lif.fs,
so that valid bytes exist at the end of the kernel image.

Since the lif image stores files size in 256-byte unit, we can not report EOF
while reading the kernel image until the remainder of the 256-byte sector has
been read. But these zero bytes would cause the gzip readahead machinery in
libsa to believe the kernel is followed by an uncompressed file, and switch to
transparent mode, preventing backwards seeks into the kernel image from working.

This fixes the ``read section header string table failed(0)'' errors when
attempting to boot lif.fs since loadfile_elf.c 1.9.

fix typo

cleanup elfrdsetroot references; ok miod

Turn on the distrib lib bits in crunchgen by setting SRCLIBDIR.
Add the new files to the clean target's rm list.

"why not get it tested?" deraadt@

match latest change by distrib/ramdisk, for extra elfrdsetroot files

Deprecate vnds in favour of svnds.

In effect, this removes the "old" vndX nodes, and renames
the svndX nodes to vndX.

Old svndX nodes will still continue to work though, for now.

Cleanup accordingly.

ok deraadt@, todd@
comments and ok on the man page bits from jmc@

No need to make depend kernels; ok guenther

unify a chunk

Place a smaller spwd.db onto the install media. Be consistant with the naming
of the "additional" list files (for architectures which split the list)
Then... cleanup some TZ which it is clear should not have been placed onto
certain media which don't have the space for them.

Since Aug 2008, the kernel can successfully read the 1 block (512
byte) disklabels even in the midst of a larger sector. E.g. in bytes
512 - 1023 of the first 2048 byte sector on a CD.

On hppa we plopped down such a disklabel on cdXX.iso and installXX.iso.
Once accessible it took precedence over the spoofed label of the CD. Chaos
ensued. Since there is currently no reason for this label on hppa
cd's, eradicate it and its disktab entry. cdXX.iso/installXX.iso can now
be mounted on hppa again.

Also remove a commented out label plopping on Alpha, which references a
non-existant disktab entry.

Problem noted and exact diff creating problem discovered by marco@.

ok deraadt@

further minor unification

For SHA256 hashes to be ready to go onto the install media, we must
build the sets before we build the media. While there we can get rid
of DESTDIR/snapshot too, and simply install straight into RELEASEDIR.
(This also ends up saving an astounding amount of traffic/latency in a nfs
environment)

run make depend for kernel, too; ok miod

do not use the DESTDIR mkboot binary

commit with rdsetroot now accepting the filesystem as a 2nd arg

bsd.rd's physical memory usage collided with the bootblocks... The
bootblocks have been moved, but we want 4.2 to be upgradeable. So
reduce the footprint of the bsd.rd by shrinking the ramdisk filesystem
because it is way too large. tested by sthen and kettenis

cleanup: make -> ${MAKE} consistenly.
zap extra subshells.

okay miod@

Stop using disklabel -r. The disklabel code does not know all the semantics
for pushing disklabels onto the disk (and besides that, it is buggy and
unreadable crap); thanks for help from krw

Update install media generation to the new semantics of the -c flag.
Also replace "newfs -O" with "newfs -O 0" now that -O takes an option.

Add -n to gzip flags. Save a few bytes since we no longer save
the uncompressed filename in the gzip header.

s/GZIP/GZIPCMD/g in `make release` structure to avoid conflict with gzip's
use of a GZIP environment variable for extra command flags.

ok deraadt@, millert@

a quick fix -- pad image w/ extra 64k until boot can cope w/ short reads

never worry about the copyright date on the CD layout again, problem
noticed by drahn

change copyrights on iso images, spotted by miod

for no apparent reason we ran out of inodes

this builds a bootable cdrom w/ a ramdisk kernel. has some problems though

no mdec/boot on the ramdisk. name it chewable for the etc/makefiles

this allows to build a ramdiskNN.lif

remove the abstraction REV on top of abstraction OSrev which is always
the sme.

incorporate GZIPFLAGS into place everywhere

CBIN variable is also pointless, it is always "instbin". Rather than
having two mysterious names, let's settle on one.

Use new vnconfig vnd-auto-allocate mode. Resolve some variation between
architectures, and start removing some crazy junk that has collected
over the years. Being tested on all architectures...
ok various people.

Use new rdsetroot.

drop ${GZIPCMD} and just use 'gzip'

drop ${STRIP}, it's just 'strip'

ok deraadt@

always strip the ctf section from ramdisk kernels

discussed with deraadt@ mpi@
ok deraadt@ kettenis@ mpi@

kernels in compile dirs are readonly, and the bsd.rd we created has
tracked that permission. Make the copy in $RELEASEDIR a+r

Install cert.pem + ftp(1) binary that supports SSL, so the install script
can perform https fetches. While non-critical for sets (which are verified
using signify), it is an uplift for communication towards the ftp.openbsd.org
cgi which provides mirror proposals, entropy seeds, and current time.
[The install script changes are not included in this commit]
ok beck jsing rpe

Use makefs to build bsd.rd; build & boot tested by krw

Use new kernel compile method for ramdisks, de-escalating to $BUILDUSER
as required.
much help from natano and tb.

Append 512 bytes of zero to the kernel image intended to be put into lif.fs,
so that valid bytes exist at the end of the kernel image.

Since the lif image stores files size in 256-byte unit, we can not report EOF
while reading the kernel image until the remainder of the 256-byte sector has
been read. But these zero bytes would cause the gzip readahead machinery in
libsa to believe the kernel is followed by an uncompressed file, and switch to
transparent mode, preventing backwards seeks into the kernel image from working.

This fixes the ``read section header string table failed(0)'' errors when
attempting to boot lif.fs since loadfile_elf.c 1.9.

fix typo

cleanup elfrdsetroot references; ok miod

Turn on the distrib lib bits in crunchgen by setting SRCLIBDIR.
Add the new files to the clean target's rm list.

"why not get it tested?" deraadt@

match latest change by distrib/ramdisk, for extra elfrdsetroot files

Deprecate vnds in favour of svnds.

In effect, this removes the "old" vndX nodes, and renames
the svndX nodes to vndX.

Old svndX nodes will still continue to work though, for now.

Cleanup accordingly.

ok deraadt@, todd@
comments and ok on the man page bits from jmc@

No need to make depend kernels; ok guenther

unify a chunk

Place a smaller spwd.db onto the install media. Be consistant with the naming
of the "additional" list files (for architectures which split the list)
Then... cleanup some TZ which it is clear should not have been placed onto
certain media which don't have the space for them.

Since Aug 2008, the kernel can successfully read the 1 block (512
byte) disklabels even in the midst of a larger sector. E.g. in bytes
512 - 1023 of the first 2048 byte sector on a CD.

On hppa we plopped down such a disklabel on cdXX.iso and installXX.iso.
Once accessible it took precedence over the spoofed label of the CD. Chaos
ensued. Since there is currently no reason for this label on hppa
cd's, eradicate it and its disktab entry. cdXX.iso/installXX.iso can now
be mounted on hppa again.

Also remove a commented out label plopping on Alpha, which references a
non-existant disktab entry.

Problem noted and exact diff creating problem discovered by marco@.

ok deraadt@

further minor unification

For SHA256 hashes to be ready to go onto the install media, we must
build the sets before we build the media. While there we can get rid
of DESTDIR/snapshot too, and simply install straight into RELEASEDIR.
(This also ends up saving an astounding amount of traffic/latency in a nfs
environment)

run make depend for kernel, too; ok miod

do not use the DESTDIR mkboot binary

commit with rdsetroot now accepting the filesystem as a 2nd arg

bsd.rd's physical memory usage collided with the bootblocks... The
bootblocks have been moved, but we want 4.2 to be upgradeable. So
reduce the footprint of the bsd.rd by shrinking the ramdisk filesystem
because it is way too large. tested by sthen and kettenis

cleanup: make -> ${MAKE} consistenly.
zap extra subshells.

okay miod@

Stop using disklabel -r. The disklabel code does not know all the semantics
for pushing disklabels onto the disk (and besides that, it is buggy and
unreadable crap); thanks for help from krw

Update install media generation to the new semantics of the -c flag.
Also replace "newfs -O" with "newfs -O 0" now that -O takes an option.

Add -n to gzip flags. Save a few bytes since we no longer save
the uncompressed filename in the gzip header.

s/GZIP/GZIPCMD/g in `make release` structure to avoid conflict with gzip's
use of a GZIP environment variable for extra command flags.

ok deraadt@, millert@

a quick fix -- pad image w/ extra 64k until boot can cope w/ short reads

never worry about the copyright date on the CD layout again, problem
noticed by drahn

change copyrights on iso images, spotted by miod

for no apparent reason we ran out of inodes

this builds a bootable cdrom w/ a ramdisk kernel. has some problems though

no mdec/boot on the ramdisk. name it chewable for the etc/makefiles

this allows to build a ramdiskNN.lif

drop ${GZIPCMD} and just use 'gzip'

drop ${STRIP}, it's just 'strip'

ok deraadt@

always strip the ctf section from ramdisk kernels

discussed with deraadt@ mpi@
ok deraadt@ kettenis@ mpi@

kernels in compile dirs are readonly, and the bsd.rd we created has
tracked that permission. Make the copy in $RELEASEDIR a+r

Install cert.pem + ftp(1) binary that supports SSL, so the install script
can perform https fetches. While non-critical for sets (which are verified
using signify), it is an uplift for communication towards the ftp.openbsd.org
cgi which provides mirror proposals, entropy seeds, and current time.
[The install script changes are not included in this commit]
ok beck jsing rpe

Use makefs to build bsd.rd; build & boot tested by krw

Use new kernel compile method for ramdisks, de-escalating to $BUILDUSER
as required.
much help from natano and tb.

Append 512 bytes of zero to the kernel image intended to be put into lif.fs,
so that valid bytes exist at the end of the kernel image.

Since the lif image stores files size in 256-byte unit, we can not report EOF
while reading the kernel image until the remainder of the 256-byte sector has
been read. But these zero bytes would cause the gzip readahead machinery in
libsa to believe the kernel is followed by an uncompressed file, and switch to
transparent mode, preventing backwards seeks into the kernel image from working.

This fixes the ``read section header string table failed(0)'' errors when
attempting to boot lif.fs since loadfile_elf.c 1.9.

fix typo

cleanup elfrdsetroot references; ok miod

Turn on the distrib lib bits in crunchgen by setting SRCLIBDIR.
Add the new files to the clean target's rm list.

"why not get it tested?" deraadt@

match latest change by distrib/ramdisk, for extra elfrdsetroot files

Deprecate vnds in favour of svnds.

In effect, this removes the "old" vndX nodes, and renames
the svndX nodes to vndX.

Old svndX nodes will still continue to work though, for now.

Cleanup accordingly.

ok deraadt@, todd@
comments and ok on the man page bits from jmc@

No need to make depend kernels; ok guenther

unify a chunk

Place a smaller spwd.db onto the install media. Be consistant with the naming
of the "additional" list files (for architectures which split the list)
Then... cleanup some TZ which it is clear should not have been placed onto
certain media which don't have the space for them.

Since Aug 2008, the kernel can successfully read the 1 block (512
byte) disklabels even in the midst of a larger sector. E.g. in bytes
512 - 1023 of the first 2048 byte sector on a CD.

On hppa we plopped down such a disklabel on cdXX.iso and installXX.iso.
Once accessible it took precedence over the spoofed label of the CD. Chaos
ensued. Since there is currently no reason for this label on hppa
cd's, eradicate it and its disktab entry. cdXX.iso/installXX.iso can now
be mounted on hppa again.

Also remove a commented out label plopping on Alpha, which references a
non-existant disktab entry.

Problem noted and exact diff creating problem discovered by marco@.

ok deraadt@

further minor unification

For SHA256 hashes to be ready to go onto the install media, we must
build the sets before we build the media. While there we can get rid
of DESTDIR/snapshot too, and simply install straight into RELEASEDIR.
(This also ends up saving an astounding amount of traffic/latency in a nfs
environment)

run make depend for kernel, too; ok miod

do not use the DESTDIR mkboot binary

commit with rdsetroot now accepting the filesystem as a 2nd arg

bsd.rd's physical memory usage collided with the bootblocks... The
bootblocks have been moved, but we want 4.2 to be upgradeable. So
reduce the footprint of the bsd.rd by shrinking the ramdisk filesystem
because it is way too large. tested by sthen and kettenis

cleanup: make -> ${MAKE} consistenly.
zap extra subshells.

okay miod@

Stop using disklabel -r. The disklabel code does not know all the semantics
for pushing disklabels onto the disk (and besides that, it is buggy and
unreadable crap); thanks for help from krw

Update install media generation to the new semantics of the -c flag.
Also replace "newfs -O" with "newfs -O 0" now that -O takes an option.

Add -n to gzip flags. Save a few bytes since we no longer save
the uncompressed filename in the gzip header.

s/GZIP/GZIPCMD/g in `make release` structure to avoid conflict with gzip's
use of a GZIP environment variable for extra command flags.

ok deraadt@, millert@

a quick fix -- pad image w/ extra 64k until boot can cope w/ short reads

never worry about the copyright date on the CD layout again, problem
noticed by drahn

change copyrights on iso images, spotted by miod

for no apparent reason we ran out of inodes

this builds a bootable cdrom w/ a ramdisk kernel. has some problems though

no mdec/boot on the ramdisk. name it chewable for the etc/makefiles

this allows to build a ramdiskNN.lif