Re: [PATCH v6 0/7] tpm: retrieve digest size of unknown algorithms from TPM

From: Jarkko Sakkinen
Date: Tue Dec 04 2018 - 19:23:06 EST


On Tue, Dec 04, 2018 at 04:21:55PM -0800, Jarkko Sakkinen wrote:
> On Tue, Dec 04, 2018 at 09:21:31AM +0100, Roberto Sassu wrote:
> > Update
> >
> > This version of the patch set includes an additional patch (7/7) which
> > modifies the definition of tpm_pcr_extend() and tpm2_pcr_extend(). The new
> > patch has been included to facilitate the review of the changes to support
> > TPM 2.0 crypto agility for reading/extending PCRs.
> >
> >
> > Original patch set description
> >
> > The TPM driver currently relies on the crypto subsystem to determine the
> > digest size of supported TPM algorithms. In the future, TPM vendors might
> > implement new algorithms in their chips, and those algorithms might not
> > be supported by the crypto subsystem.
> >
> > Usually, vendors provide patches for the new hardware, and likely
> > the crypto subsystem will be updated before the new algorithm is
> > introduced. However, old kernels might be updated later, after patches
> > are included in the mainline kernel. This would leave the opportunity
> > for attackers to misuse PCRs, as PCR banks with an unknown algorithm
> > are not extended.
> >
> > This patch set provides a long term solution for this issue. If a TPM
> > algorithm is not known by the crypto subsystem, the TPM driver retrieves
> > the digest size from the TPM with a PCR read. All the PCR banks are
> > extended, even if the algorithm is not yet supported by the crypto
> > subsystem.
> >
> > PCR bank information (TPM algorithm ID, digest size, crypto subsystem ID)
> > is stored in the tpm_chip structure and available for users of the TPM
> > driver.
> >
> > Changelog
> >
> > v5:
> > - rename digest_struct variable to digest
> > - add _head suffix to tcg_efi_specid_event and tcg_pcr_event2
> > - rename digest_size member of tpm_bank_list to extend_size
> > - change type of alg_id member of tpm_bank list from u8 to u16
> > - add missing semi-colon in pcrlock()
> >
> > v4:
> > - rename active_banks to allocated_banks
> > - replace kmalloc_array() with kcalloc()
> > - increment nr_allocated_banks if at least one PCR in the bank is selected
> > - pass multiple digests to tpm_pcr_extend()
> >
> > v3:
> > - remove end marker change
> > - replace active_banks static array with pointer to dynamic array
> > - remove TPM2_ACTIVE_PCR_BANKS
> >
> > v2:
> > - change the end marker of the active_banks array
> > - check digest size from output of PCR read command
> > - remove count parameter from tpm_pcr_read() and tpm2_pcr_read()
> >
> > v1:
> > - modify definition of tpm_pcr_read()
> > - move hash algorithms and definition of tpm2_digest to include/linux/tpm.h
> >
> > Roberto Sassu (7):
> > tpm: dynamically allocate the allocated_banks array
> > tpm: add _head suffix to tcg_efi_specid_event and tcg_pcr_event2
> > tpm: rename and export tpm2_digest and tpm2_algorithms
> > tpm: modify tpm_pcr_read() definition to pass a TPM hash algorithm
> > tpm: retrieve digest size of unknown algorithms with PCR read
> > tpm: ensure that the output of PCR read contains the correct digest
> > size
> > tpm: pass an array of tpm_bank_list structures to tpm_pcr_extend()
> >
> > drivers/char/tpm/eventlog/tpm2.c | 12 +--
> > drivers/char/tpm/tpm-chip.c | 1 +
> > drivers/char/tpm/tpm-interface.c | 36 +++----
> > drivers/char/tpm/tpm.h | 23 ++---
> > drivers/char/tpm/tpm1-cmd.c | 26 ++++-
> > drivers/char/tpm/tpm2-cmd.c | 154 +++++++++++++++++++++-------
> > include/linux/tpm.h | 43 +++++++-
> > include/linux/tpm_eventlog.h | 19 +---
> > security/integrity/ima/ima_crypto.c | 10 +-
> > security/integrity/ima/ima_queue.c | 5 +-
> > security/keys/trusted.c | 5 +-
> > 11 files changed, 220 insertions(+), 114 deletions(-)
> >
> > --
> > 2.17.1
> >
>
> Some generic stuff I noticed:
>
> * Use SHA1_DIGEST_SIZE, not TPM_DIGEST_SIZE. The latter is just
> uninforming constant that we want to get rid off.

Oops, sent by mistake. Other thing was that if you have an array
in the tail of the struct you can just have [] no subcript zero
needed.

/Jarkko