Let me summarize the history of this patch:
After most platform switched to C_ENVIRONMENT_BOOTBLOCK, I have been
improving this patch following the ORIGINAL way: Initialize the TPM
in bootblock, and measure components of CRTM right after it, but on
some platform the bootblock is too small to link with vboot library
and TPM-Initialization code (recorded in patchset 27~33). These
platforms usually needs a verstage if VBOOT is enabled. I managed to
handle this by letting these platforms initialize the TPM and CRTM on
romstage, like how it is done in patchset 7. The last revision working
in this way is patchset 41.
On the other hand, Julius Werner started to propose his scheme, in
which digests are "cached" in TCPA log and "replayed" to PCR after TPM
is Initialized, since patchset 6, and he advised me again, in patchset
40, after I met the obstacle described above, to follow his scheme, as
done in revisions after patchset 41.
If Julius' scheme breaks the standard, I can revert to patchset 41 or
even much earlier revision as the base for further development, but
now the key problem is, what is the most proper stage in which TPM is
initialized, and CRTM is settled? At least I have no idea of this.
To view, visit change 35077. To unsubscribe, or for help writing mail filters, visit settings.