[PATCH v3 4/5] tpm: retrieve digest size of unknown algorithms with PCR read

[Jarkko Sakkinen]

On Mon, Nov 05, 2018 at 02:09:12PM +0100, Roberto Sassu wrote:
> On 11/5/2018 1:01 PM, Jarkko Sakkinen wrote:
> > On Mon, Nov 05, 2018 at 10:47:19AM +0100, Roberto Sassu wrote:
> > > > Commit 1db15344f874 ("tpm: implement TPM 2.0 capability to get active
> > > > PCR banks") defined active_banks[7].  Subsequently, commit
> > > > 4d23cc323cdb ("tpm: add securityfs support for TPM 2.0 firmware event
> > > > log") defined TPM2_PCR_ACTIVE_BANKS as 3.  I'm not sure which is the
> > > > correct value, but the number of active_banks should not be hard coded
> > > > here.
> > > 
> > > Jarkko, should I change the value of TPM2_PCR_ACTIVE_BANKS, or set the
> > > size of the active_banks array to TPM2_PCR_ACTIVE_BANKS?
> > 
> > Hi, sorry I missed your patch set. Please add me either to 'To' or 'Cc'
> > field of the email if you want a quick response.
> > 
> > I think the implementation is flakky in both places and should be fixed
> > before doing any other changes. Thanks James for pointing out these
> > commits.
> > 
> > What you need to do is to create a prequel commit that reads the number
> > of banks to a variable e.g.
> > 
> >    unsigned int nr_active_banks;
> > 
> > and allocate 'active_banks' dynamically and change the places that
> > James pointed out. I guess it is OK to have a commit with two 'Fixes'
> > tags.
> 
> Ok, then I can remove patch 1/5 if nr_active_banks is included in the
> tpm_chip structure.
> 
> Roberto

Yeah, I think it would be appropriate to have two fixes tags albeit it
is arguable whether those are regressions (probably not, I guess
inconsistency would be a better word) but I don't think they need to be
cc'd to stable at vger.kernel.org.

/Jarkko