On Fri, 2019-02-01 at 15:41 +0200, Jarkko Sakkinen wrote: > On Fri, Feb 01, 2019 at 03:39:49PM +0200, Jarkko Sakkinen wrote: > > On Fri, Feb 01, 2019 at 11:06:41AM +0100, Roberto Sassu wrote: > > > Currently, tpm_pcr_extend() accepts as an input only a SHA1 digest. > > > > > > This patch replaces the hash parameter of tpm_pcr_extend() with an array of > > > tpm_digest structures, so that the caller can provide a digest for each PCR > > > bank currently allocated in the TPM. > > > > > > tpm_pcr_extend() will not extend banks for which no digest was provided, > > > as it happened before this patch, but instead it requires that callers > > > provide the full set of digests. Since the number of digests will always be > > > chip->nr_allocated_banks, the count parameter has been removed. > > > > > > Due to the API change, ima_pcr_extend() and pcrlock() have been modified. > > > Since the number of allocated banks is not known in advance, the memory for > > > the digests must be dynamically allocated. To avoid performance degradation > > > and to avoid that a PCR extend is not done due to lack of memory, the array > > > of tpm_digest structures is allocated by the users of the TPM driver at > > > initialization time. > > > > > > Signed-off-by: Roberto Sassu <roberto.sassu@xxxxxxxxxx> > > > > Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@xxxxxxxxxxxxxxx> > > Tested-by: Jarkko Sakkinen <jarkko.sakkinen@xxxxxxxxxxxxxxx> > > I tested that this does not break TPM. I'd need someone to check that > this does not break IMA. After that, I'm ready to apply this series. Thanks! I just finished compiling, rebooting, and verifying the IMA boot-aggregate matches. While compiling, I saw some messages that "TPM_BUFSIZE is redefined" for tpm_i2c_infineon, tpm_i2c_nuvoton, and st33zp24. Didn't look to see if this is a result of this patch set or not. The IMA boot-aggregate matches the PCRs. Tested-by: Mimi Zohar <zohar@xxxxxxxxxxxxx> (on x86 for TPM 1.2 & PTT TPM 2.0)
