Joel Kitching has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/31837 )
Change subject: drivers/tpm: update TPM initialization logic for Intel FSP2.0 ......................................................................
Patch Set 3:
(2 comments)
https://review.coreboot.org/#/c/31837/2//COMMIT_MSG Commit Message:
https://review.coreboot.org/#/c/31837/2//COMMIT_MSG@12 PS2, Line 12: src/drivers/tpm/tpm.c takes care of : initializing TPM (see Kconfig option TPM_INIT).
So, were the devices getting initialized twice currently if VBOOT is not selected -- i.e. […]
I guess we are going to run into more problems with !VBOOT if we leave this code in here... What about the suggestion in the comment below about a new Kconfig flag?
https://review.coreboot.org/#/c/31837/2//COMMIT_MSG@14 PS2, Line 14: TPM will be initialized whenever : verstage is executed, depending on how the device is configured
I am okay if you want to just add a condition saying that RECOVERY_MRC_CACHE hash is maintained in T […]
Thanks for letting me know about this -- I had no idea that RECOVERY_MRC_CACHE triggered TPM read/writes.
Since I'd like to keep some devices working with STARTS_IN_BOOTBLOCK for test purposes, how about fixing it to work? (See latest patchset.)
Alternatively, we could also introduce a more general Kconfig option... I was thinking TPM_EARLY_INIT might be a good pair for TPM_INIT. We can specify that they are mutually exclusive.
* TPM_EARLY_INIT: Something earlier than the ramstage TPM init code starts up the TPM. * TPM_INIT: Ramstage TPM init code starts up the TPM. * [Neither, but with VBOOT]: Vboot takes care of it when neither of these flags are enabled. * [None]: TPM is not enabled.