Adds the necessary changes to tpm_init to allow triggered initialization.

Adds the necessary changes to tpm_init to allow triggered initialization.

Change-Id: I757a112ef011ee9688f695cdd89ad2786c47778b

BUG=3065
TEST=

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=3b9a6ec

[Chris Masone, 2010-10-02 02:05:40]

http://codereview.chromium.org/3475009/diff/13001/10007
File SConstruct (right):

http://codereview.chromium.org/3475009/diff/13001/10007#newcode61
SConstruct:61: LIBS = ['base', 'chromeos', 'crypto', 'rt', 'protobuf',
'pthread', 'tspi'],
Did you also do a CL to add protobuf2 as an RDEPEND and an DEPEND of cryptohomed
in your ebuild?

http://codereview.chromium.org/3475009/diff/13001/10009
File platform.h (right):

http://codereview.chromium.org/3475009/diff/13001/10009#newcode17
platform.h:17: class Platform {
I assume this class is here to allow you to mock out the umask library call? 
Can a brother get a comment?

http://codereview.chromium.org/3475009/diff/13001/10010
File tpm.cc (right):

http://codereview.chromium.org/3475009/diff/13001/10010#newcode482
tpm.cc:482: bool Tpm::LoadFileBytes(const FilePath& path,
I feel like I've read this method before...

http://codereview.chromium.org/3475009/diff/13001/10010#newcode483
tpm.cc:483: chromeos::Blob* blob) {
this can fit on previous line, no?

http://codereview.chromium.org/3475009/diff/13001/10010#newcode510
tpm.cc:510: chromeos::Blob* owner_password) {
indent

http://codereview.chromium.org/3475009/diff/13001/10010#newcode626
tpm.cc:626: TSS_HPCRS pcrs_handle;
What're we doing with the PCRs here?  I see you need the handle below to do the
Seal operation, but I don't get what you're doing up here.

http://codereview.chromium.org/3475009/diff/13001/10010#newcode855
tpm.cc:855: if (!RandomizeOwnerPassword(context_handle, &owner_password)) {
Shouldn't we maybe do this AFTER we've saved off the owner password?  It just
seems like a crash or shutdown after we randomize the PW, but before we save it
off is worse than saving off the new, randomized PW without getting to change it
in the chip.

[fes, 2010-10-04 17:04:05]

Feedback addressed.  PTAL

http://codereview.chromium.org/3475009/diff/13001/10007
File SConstruct (right):

http://codereview.chromium.org/3475009/diff/13001/10007#newcode61
SConstruct:61: LIBS = ['base', 'chromeos', 'crypto', 'rt', 'protobuf',
'pthread', 'tspi'],
On 2010/10/02 02:05:40, Chris Masone wrote:
> Did you also do a CL to add protobuf2 as an RDEPEND and an DEPEND of
cryptohomed
> in your ebuild?

There is now:
http://codereview.chromium.org/3522011

http://codereview.chromium.org/3475009/diff/13001/10009
File platform.h (right):

http://codereview.chromium.org/3475009/diff/13001/10009#newcode17
platform.h:17: class Platform {
On 2010/10/02 02:05:40, Chris Masone wrote:
> I assume this class is here to allow you to mock out the umask library call? 
> Can a brother get a comment?

Yes, until the duplicate functionality in cryptohome/tpm_init is merged through
merge of the two components, or abstracted into a common library.

http://codereview.chromium.org/3475009/diff/13001/10010
File tpm.cc (right):

http://codereview.chromium.org/3475009/diff/13001/10010#newcode482
tpm.cc:482: bool Tpm::LoadFileBytes(const FilePath& path,
On 2010/10/02 02:05:40, Chris Masone wrote:
> I feel like I've read this method before...

Yes, and the duplication will go away in the future.

http://codereview.chromium.org/3475009/diff/13001/10010#newcode483
tpm.cc:483: chromeos::Blob* blob) {
On 2010/10/02 02:05:40, Chris Masone wrote:
> this can fit on previous line, no?

Done.

http://codereview.chromium.org/3475009/diff/13001/10010#newcode510
tpm.cc:510: chromeos::Blob* owner_password) {
On 2010/10/02 02:05:40, Chris Masone wrote:
> indent
> 

Done.

http://codereview.chromium.org/3475009/diff/13001/10010#newcode626
tpm.cc:626: TSS_HPCRS pcrs_handle;
On 2010/10/02 02:05:40, Chris Masone wrote:
> What're we doing with the PCRs here?  I see you need the handle below to do
the
> Seal operation, but I don't get what you're doing up here.

Done.

http://codereview.chromium.org/3475009/diff/13001/10010#newcode855
tpm.cc:855: if (!RandomizeOwnerPassword(context_handle, &owner_password)) {
On 2010/10/02 02:05:40, Chris Masone wrote:
> Shouldn't we maybe do this AFTER we've saved off the owner password?  It just
> seems like a crash or shutdown after we randomize the PW, but before we save
it
> off is worse than saving off the new, randomized PW without getting to change
it
> in the chip.

Done.

[Chris Masone, 2010-10-04 18:13:39]

LGTM