Add option to use mocked TPM driver

Add option to use mocked TPM driver

Some ARM boards, such as Seaboard, have buggy TPM chip that bricks the borad.
This commit adds a option to use mocked TPM driver.

R=rongchang@chromium.org,rspangler@chromium.org,wad@chromium.org
BUG=chromium-os:14239
TEST=Run verified boot on Seaboard

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=8511f78

[Randall Spangler, 2011-04-19 20:23:35]

If you mock rollback_index.c and tpm_bootmode.c, you don't need to mock tlcl.c
since tlcl.c funcs are only called from those 2 files.  You can remove
mock_tlcl.c entirely.

The tlcl.c mocks for Get functions don't return anything.  This is currently
hidden by the above behavior.  Should either fix this to return known values, or
just don't mock tlcl.c because we don't need to.

Do we have tests which ensure the (real) TPM is properly locked during a normal
boot, so that we won't accidentally ship a firmware with the mock TPM interface?
 Or will we catch that because the firmware wouldn't have called TlclStartup(),
so tpmd will die?

[Che-Liang Chiou, 2011-04-20 02:28:57]

I mock tlcl.c because Rong has ported TPM unit tests to u-boot (functions
defined in tlcl.c are called by u-boot's test drive commands) so that compiler
will not complain about missing symbols. (But running those unit tests on
Seaboards is meaningless.)

On 2011/04/19 20:23:35, Randall Spangler wrote:
> If you mock rollback_index.c and tpm_bootmode.c, you don't need to mock tlcl.c
> since tlcl.c funcs are only called from those 2 files.  You can remove
> mock_tlcl.c entirely.
> 
> The tlcl.c mocks for Get functions don't return anything.  This is currently
> hidden by the above behavior.  Should either fix this to return known values,
or
> just don't mock tlcl.c because we don't need to.
> 
> Do we have tests which ensure the (real) TPM is properly locked during a
normal
> boot, so that we won't accidentally ship a firmware with the mock TPM
interface?
>  Or will we catch that because the firmware wouldn't have called
TlclStartup(),
> so tpmd will die?

[Randall Spangler, 2011-04-20 21:11:42]

The tlcl.c mocks should either provide reasonable sample data or return error.

(That way, if you run the unit tests, you'll see them fail in an expected way,
not a random way.)

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
File firmware/lib/tpm_lite/mocked_tlcl.c (right):

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:40: uint32_t TlclRead(uint32_t index, void*
data, uint32_t length) {
Please fill the returned data with known contents, so you get deterministic
behavior.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:96: uint32_t
TlclGetPermanentFlags(TPM_PERMANENT_FLAGS* pflags) {
ditto.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:100: uint32_t
TlclGetSTClearFlags(TPM_STCLEAR_FLAGS* vflags) {
ditto.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:107: return TPM_SUCCESS;
ditto.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:116: return TPM_SUCCESS;
ditto.  (copy in digest to out digest?)

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:119: uint32_t TlclGetPermissions(uint32_t
index, uint32_t* permissions) {
ditto.

[Che-Liang Chiou, 2011-04-21 11:22:36]

Updated. PTAL.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
File firmware/lib/tpm_lite/mocked_tlcl.c (right):

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:40: uint32_t TlclRead(uint32_t index, void*
data, uint32_t length) {
On 2011/04/20 21:11:43, Randall Spangler wrote:
> Please fill the returned data with known contents, so you get deterministic
> behavior.

Done.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:96: uint32_t
TlclGetPermanentFlags(TPM_PERMANENT_FLAGS* pflags) {
On 2011/04/20 21:11:43, Randall Spangler wrote:
> ditto.

Done.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:100: uint32_t
TlclGetSTClearFlags(TPM_STCLEAR_FLAGS* vflags) {
On 2011/04/20 21:11:43, Randall Spangler wrote:
> ditto.

Done.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:107: return TPM_SUCCESS;
On 2011/04/20 21:11:43, Randall Spangler wrote:
> ditto.

Done.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:116: return TPM_SUCCESS;
On 2011/04/20 21:11:43, Randall Spangler wrote:
> ditto.  (copy in digest to out digest?)

Done.

http://codereview.chromium.org/6883040/diff/1/firmware/lib/tpm_lite/mocked_tl...
firmware/lib/tpm_lite/mocked_tlcl.c:119: uint32_t TlclGetPermissions(uint32_t
index, uint32_t* permissions) {
On 2011/04/20 21:11:43, Randall Spangler wrote:
> ditto.

Done.

http://codereview.chromium.org/6883040/diff/7/Makefile
File Makefile (right):

http://codereview.chromium.org/6883040/diff/7/Makefile#newcode26
Makefile:26: export ARCH=amd64
Change from rebase to ToT.

[Randall Spangler, 2011-04-21 22:56:37]

LGTM, thanks