Re: [PATCH v5 14/17] tpm: remove TPM_TRANSMIT_UNLOCKED flag

From: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
To: Stefan Berger <stefanb@linux.ibm.com>
Cc: linux-integrity@vger.kernel.org,
	linux-security-module@vger.kernel.org,
	James Bottomley <James.Bottomley@HansenPartnership.com>,
	Tomas Winkler <tomas.winkler@intel.com>,
	Tadeusz Struk <tadeusz.struk@intel.com>,
	Stefan Berger <stefanb@linux.vnet.ibm.com>,
	Nayna Jain <nayna@linux.ibm.com>, Peter Huewe <peterhuewe@gmx.de>,
	Jason Gunthorpe <jgg@ziepe.ca>, Arnd Bergmann <arnd@arndb.de>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	open list <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v5 14/17] tpm: remove TPM_TRANSMIT_UNLOCKED flag
Date: Thu, 8 Nov 2018 21:02:34 +0200	[thread overview]
Message-ID: <20181108190234.GD20608@linux.intel.com> (raw)
In-Reply-To: <5a0628cb-84e2-6fab-8fce-a77dc7adceb2@linux.ibm.com>

On Thu, Nov 08, 2018 at 10:43:46AM -0500, Stefan Berger wrote:
> > diff --git a/drivers/char/tpm/tpm2-cmd.c b/drivers/char/tpm/tpm2-cmd.c
> > index f2b0e5c52a57..d6abc964ef66 100644
> > --- a/drivers/char/tpm/tpm2-cmd.c
> > +++ b/drivers/char/tpm/tpm2-cmd.c
> > @@ -652,17 +652,12 @@ int tpm2_unseal_trusted(struct tpm_chip *chip,
> >   	u32 blob_handle;
> >   	int rc;
> > 
> > -	mutex_lock(&chip->tpm_mutex);
> > -	rc = tpm2_load_cmd(chip, payload, options, &blob_handle,
> > -			   TPM_TRANSMIT_UNLOCKED);
> > +	rc = tpm2_load_cmd(chip, payload, options, &blob_handle, 0);
> >   	if (rc)
> > -		goto out;
> > +		return rc;
> > 
> > -	rc = tpm2_unseal_cmd(chip, payload, options, blob_handle,
> > -			     TPM_TRANSMIT_UNLOCKED);
> > -	tpm2_flush_context_cmd(chip, blob_handle, TPM_TRANSMIT_UNLOCKED);
> > -out:
> > -	mutex_unlock(&chip->tpm_mutex);
> > +	rc = tpm2_unseal_cmd(chip, payload, options, blob_handle, 0);
> > +	tpm2_flush_context_cmd(chip, blob_handle, 0);
> 
> 
> This lock was covering quite a few commands from being interfered by others.
> Is this still guaranteed to work after or can different subsystems like
> trusted keys and IMA and /dev/tpm0 users interfere with this previous atomic
> sequence ?

One way to keep this sequence atomic would be to use a TPM space. Then
you can release the lock in-between the sequence.

/Jarkko