Many upgrades to tpm_lite.

Many upgrades to tpm_lite.
They include:
- support for local and cross-compilation with emerge;
- support for execution on TPM devices (formerly only linked with emulator);
- added ownership detection command;
- more complete test of space creation and write locking;
- added PPWRITE permission to space creation.

[Luigi Semenzato, 2010-03-11 22:34:48]

Hi Gaurav,

more changes are coming, but this is a good checkpointing time.  Mainly, the
library and testsuite (still only one test) now works both with the emulator and
the actual TPM device, both on our 64-bit workstations and in 32-bit Chrome OS.

Thanks!

[gauravsh, 2010-03-12 00:03:38]

http://codereview.chromium.org/870004/diff/1/2
File src/platform/tpm_lite/src/Makefile (right):

http://codereview.chromium.org/870004/diff/1/2#newcode1
src/platform/tpm_lite/src/Makefile:1: # Simple Makefile for local compilation on
the platform.
Nit: copyright.

http://codereview.chromium.org/870004/diff/1/2#newcode4
src/platform/tpm_lite/src/Makefile:4: 
Why not just create a Makefile which has a special platform target. (i.e. make
all does nothing).

http://codereview.chromium.org/870004/diff/1/8
File src/platform/tpm_lite/src/tlcl/tlcl.c (right):

http://codereview.chromium.org/870004/diff/1/8#newcode38
src/platform/tpm_lite/src/tlcl/tlcl.c:38: void printbytes(uint8_t* a, int n) {
make this static. Also PrintBytes()

http://codereview.chromium.org/870004/diff/1/8#newcode100
src/platform/tpm_lite/src/tlcl/tlcl.c:100: void tpm_execute(uint8_t *in,
uint32_t in_len,
nit: TPMExecute()
Also, in and in_len can be const? Why is pout_len called so? why not out_len?

http://codereview.chromium.org/870004/diff/1/8#newcode122
src/platform/tpm_lite/src/tlcl/tlcl.c:122: memmove(out, response, n);
since out and response don't overlap - you can just use memcpy(). It's more
efficient and doesn't use a temporary.

http://codereview.chromium.org/870004/diff/1/8#newcode137
src/platform/tpm_lite/src/tlcl/tlcl.c:137: tpm_execute(request,
TpmCommandSize(request), response, &response_length);
Did you name tpm_execute() to be consistent with the tpmemu_execue()? If so,
ignore my previous comment.

[Luigi Semenzato, 2010-03-12 05:00:45]

Thanks, I made almost all the changes you suggested and clarified a couple of
points.

http://codereview.chromium.org/870004/diff/1/2
File src/platform/tpm_lite/src/Makefile (right):

http://codereview.chromium.org/870004/diff/1/2#newcode4
src/platform/tpm_lite/src/Makefile:4: 
On 2010/03/12 00:03:38, gauravsh wrote:
> Why not just create a Makefile which has a special platform target. (i.e. make
> all does nothing).

Thanks, good idea.

http://codereview.chromium.org/870004/diff/1/8
File src/platform/tpm_lite/src/tlcl/tlcl.c (right):

http://codereview.chromium.org/870004/diff/1/8#newcode100
src/platform/tpm_lite/src/tlcl/tlcl.c:100: void tpm_execute(uint8_t *in,
uint32_t in_len,
On 2010/03/12 00:03:38, gauravsh wrote:
> nit: TPMExecute()
> Also, in and in_len can be const? Why is pout_len called so? why not out_len?

I changed the name to follow the convention of this file.  pout_len has a p
because it's a pointer to a length variable, used as an IN-OUT parameter. 
in_len is int and only IN.

http://codereview.chromium.org/870004/diff/1/8#newcode122
src/platform/tpm_lite/src/tlcl/tlcl.c:122: memmove(out, response, n);
On 2010/03/12 00:03:38, gauravsh wrote:
> since out and response don't overlap - you can just use memcpy(). It's more
> efficient and doesn't use a temporary.

Sure!  I don't even remember what prompted me to use memmove.

http://codereview.chromium.org/870004/diff/1/8#newcode137
src/platform/tpm_lite/src/tlcl/tlcl.c:137: tpm_execute(request,
TpmCommandSize(request), response, &response_length);
On 2010/03/12 00:03:38, gauravsh wrote:
> Did you name tpm_execute() to be consistent with the tpmemu_execue()? If so,
> ignore my previous comment.

Yes, but it's better to be consistent with this file, thanks.  tpmemu_execute is
consistent with the emulator conventions.

[gauravsh, 2010-03-12 18:58:33]

lgtm