Changes to allow user-signed kernels to be generated.

Changes to allow user-signed kernels to be generated.

Make vbutil_keyblock handle unsigned blocks. Also enable --unpack option and
add tests for it.

Modify vbutil_kernel to allow unsigned keyblocks, correct usage message,
and fix the --debug option which was somehow disabled.

Update load_kernel_test to accept /dev/null for the public key, to test
non-signed kernel keyblocks.

[Randall Spangler, 2010-08-09 23:03:53]

http://codereview.chromium.org/3124004/diff/1/7
File utility/dev_make_keypair (right):

http://codereview.chromium.org/3124004/diff/1/7#newcode6
utility/dev_make_keypair:6: # Generate .vbpubk and .vbprivk pairs for use by
developer builds. These should
should be --> are (right?)

http://codereview.chromium.org/3124004/diff/1/7#newcode8
utility/dev_make_keypair:8: 
Seems like this shares a lot of code with tests/devkeys/create_new_keys.sh.  Can
we move that code into a common sub-script, to help keep the two in sync?

http://codereview.chromium.org/3124004/diff/1/7#newcode14
utility/dev_make_keypair:14: Usage:  ${0##*/} BASENAME ALG
Seems like we could have ALG default to a reasonable value for developers
(RSA2048, SHA256).

http://codereview.chromium.org/3124004/diff/1/7#newcode79
utility/dev_make_keypair:79: }
This function no effect?

http://codereview.chromium.org/3124004/diff/1/8
File utility/load_kernel_test.c (right):

http://codereview.chromium.org/3124004/diff/1/8#newcode101
utility/load_kernel_test.c:101: printf("Unable to read key file %s, not checking
keyblock sig\n",
Seems like it'd be safer to check argc, rather than letting the user specify a
dummy sign_key.

So
1 arg = image_name
2 args = + sign_key
3 args = + boot_flag

That way, if I DID mean to specify a key file, but I typo'd it, I won't get a
false positive.

[Bill Richardson, 2010-08-10 00:23:35]

Okay, take another look.

http://codereview.chromium.org/3124004/diff/1/7
File utility/dev_make_keypair (right):

http://codereview.chromium.org/3124004/diff/1/7#newcode8
utility/dev_make_keypair:8: 
Not without creating some sort of common.sh file that both could source. That's
likely to be worse than this.

[Randall Spangler, 2010-08-10 00:37:54]

lgtm

http://codereview.chromium.org/3124004/diff/1/7
File utility/dev_make_keypair (right):

http://codereview.chromium.org/3124004/diff/1/7#newcode8
utility/dev_make_keypair:8: 
On 2010/08/10 00:23:35, Bill Richardson wrote:
> Not without creating some sort of common.sh file that both could source.
That's
> likely to be worse than this.
> 

I'll buy that.  Perhaps just mention the commonality in a comment, so that
changes are more likely to hand-propagate.