Re: [PATCH v3 7/8] ima: Calculate and extend PCR with digests in ima_template_entry

[Mimi Zohar <zohar@linux.ibm.com>]

On Wed, 2020-03-18 at 12:42 +0000, Roberto Sassu wrote:
> > -----Original Message-----
> > From: owner-linux-security-module@vger.kernel.org [mailto:owner-linux-
> > security-module@vger.kernel.org] On Behalf Of Mimi Zohar
> > Sent: Tuesday, March 3, 2020 5:04 AM
> > To: Roberto Sassu <roberto.sassu@huawei.com>;
> > James.Bottomley@HansenPartnership.com;
> > jarkko.sakkinen@linux.intel.com
> > Cc: linux-integrity@vger.kernel.org; linux-security-module@vger.kernel.org;
> > linux-kernel@vger.kernel.org; Silviu Vlasceanu
> > <Silviu.Vlasceanu@huawei.com>
> > Subject: Re: [PATCH v3 7/8] ima: Calculate and extend PCR with digests in
> > ima_template_entry
> > 
> > On Mon, 2020-02-10 at 11:04 +0100, Roberto Sassu wrote:
> > 
> > > @@ -219,6 +214,8 @@ int ima_restore_measurement_entry(struct
> > ima_template_entry *entry)
> > >
> > >  int __init ima_init_digests(void)
> > >  {
> > > +	u16 digest_size;
> > > +	u16 crypto_id;
> > >  	int i;
> > >
> > >  	if (!ima_tpm_chip)
> > > @@ -229,8 +226,17 @@ int __init ima_init_digests(void)
> > >  	if (!digests)
> > >  		return -ENOMEM;
> > >
> > > -	for (i = 0; i < ima_tpm_chip->nr_allocated_banks; i++)
> > > +	for (i = 0; i < ima_tpm_chip->nr_allocated_banks; i++) {
> > >  		digests[i].alg_id = ima_tpm_chip->allocated_banks[i].alg_id;
> > > +		digest_size = ima_tpm_chip->allocated_banks[i].digest_size;
> > > +		crypto_id = ima_tpm_chip->allocated_banks[i].crypto_id;
> > > +
> > > +		/* for unmapped TPM algorithms digest is still a padded
> > SHA1 */
> > > +		if (crypto_id == HASH_ALGO__LAST)
> > > +			digest_size = SHA1_DIGEST_SIZE;
> > > +
> > > +		memset(digests[i].digest, 0xff, digest_size);
> > 
> > Shouldn't the memset here be of the actual digest size even for
> > unmapped TPM algorithms.
> 
> This is consistent with ima_calc_field_array_hash(), so that a verifier
> will always pad the SHA1 digest with zeros to obtain the final PCR value.
> 
> I can set all bytes if you prefer.

My concern is with violations.  The measurement list will be padded
with 0's, but the value being extended into the TPM will only
partially be 0xFF's.  When verifying the measurement list, replacing
all 0x00's with all 0xFF's is simpler.

Mimi