GitLab

That was more involved than I thought...or it was worth, considering that
only old FreeBSD and really old Redhat (think: LILO) images have them.
But that is the kinda guy I am...

Also added option (-F) to write full-image signature (what we do).

This is to facilitate testing again...

To that end, added a new "imagerezip" program which reads in an image
and writes it back out. Right now, this is just to test that reading
and writing images via the new library produces the same result as the
"classic" tools. But it might be useful in the future to repack images
with a different compression level or a different signature (hash function
or blocksize).

Also checked in a hack to imageunzip I did for Xing (-I option) to just
produce a stream of the uncompressed image data (it ignores skips, aka
it doesn't skip on the output device to put the data in the correct place).

Unfortunately, I cannot just compare raw images because even trying to
replicate the chunk packing strategy of the original code, the new library
packs a bit more into each chunk. Not sure why, but anyway, I have to
compare old/new images by dumping their contents and by checking the
generated signatures.

Of course, I have no idea if it is correct...but it is well formed!

New code for managing ranges.

No one has used this code in 10 years and not one will ever use it again,
but it broke my testing...and PISSED ME OFF!

Time for a LILO revival!

Just luuuv string manipulation in C...

The "Emulab configuration" of the mserver will now use uploader_path
if set rather than path.

A slight semantic change goes along with this. If the uploader_path is
set, we assume that it is a temporary file to upload to and we (the mserver)
do not do the dance of uploading to our own temporary and moving it into place.

The -S (sanity check) option adds the ability to do the obvious: validate
an image file using its signature. So what did imagehash do all these years?
Well, its primary job was to generate a signature file that imagezip could
use to generate a delta. Secondarily, it would compare a signature vs. a dis
with an image loaded on it. But now we use imagezip to gen most signatures
while generating the image itself.

Also added -X (xtreme hack) option to have imagehash attempt to check or
generate signature file entries in a way that is consistent with what
imagezip does. The difference is that imagezip can produce hashes for
blocks that span chunk boundaries, imagehash (previously) never did this.

So when validating an image whose signature was generated by imagezip:

   imagehash -SX foo.ndz

To generate a new signature for an image:

   imagehash -cX foo.ndz

Also, if partition is smaller than default zapsize, zap the entire
partition rather than skipping it.

Also some lint.

Required updates to Makefile-linux.sa makefiles which are used
in Ryan's build environment.

Mostly I just wanted to test "the latest" FBSD kernel on Apt
so see if anything done recently would help with the hanging
console problem on the R320s (newer kernels actually only hang
for about 10 seconds and drop around 500 chars when they probe
the ethernet interfaces--this is better than it was).

While in there, I just went ahead and did the client side,
mostly ifdefs in the makefiles.

Finally just had to add a target for the mfs install since there
seemed to be no way to make CLIENT_BINDIR work for both /usr/local/bin
and /etc/testbed. Now Leigh and I can stop overwriting each others' commits.

Uses the MBR/GPT libraries from imagezip, hence the awkward change
to the imagezip Makefile.

With "-X <domname>" capture will track what pty is being exposed by
xenconsoled for a domU console. Capture will doggedly reconnect whenever
it gets disconnected from the pty (which happens during reboots, etc.)

Capture can even be started before the domain exists and it will keep
trying til it does. So betware of typo-ing the domain name, capture won't
care!

Also added "-R <retry-interval-in-ms>" option to determine capture's
"doggedness"--how often it will retry to establish a connection in the
remote/program/xen modes.

Also, print out some of the additional info in debug mode.