Re: [PATCH v4] tpm: Mask TPM RC in tpm2_start_auth_session()

[Jarkko Sakkinen <jarkko@kernel.org>]

On Mon, Apr 07, 2025 at 09:13:37PM +0300, Jarkko Sakkinen wrote:
> On Mon, Apr 07, 2025 at 03:51:21PM +0200, Stefano Garzarella wrote:
> > On Mon, Apr 07, 2025 at 03:28:05PM +0300, Jarkko Sakkinen wrote:
> > > tpm2_start_auth_session() does not mask TPM RC correctly from the callers:
> > > 
> > > [   28.766528] tpm tpm0: A TPM error (2307) occurred start auth session
> > > 
> > > Process TPM RCs inside tpm2_start_auth_session(), and map them to POSIX
> > > error codes.
> > > 
> > > Cc: stable@vger.kernel.org # v6.10+
> > > Fixes: 699e3efd6c64 ("tpm: Add HMAC session start and end functions")
> > > Reported-by: Herbert Xu <herbert@gondor.apana.org.au>
> > > Closes: https://lore.kernel.org/linux-integrity/Z_NgdRHuTKP6JK--@gondor.apana.org.au/
> > > Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> > > ---
> > > v4:
> > > - tpm_to_ret()
> > > v3:
> > > - rc > 0
> > > v2:
> > > - Investigate TPM rc only after destroying tpm_buf.
> > > ---
> > > drivers/char/tpm/tpm2-sessions.c | 20 ++++++--------------
> > > include/linux/tpm.h              | 21 +++++++++++++++++++++
> > > 2 files changed, 27 insertions(+), 14 deletions(-)
> > > 
> > > diff --git a/drivers/char/tpm/tpm2-sessions.c b/drivers/char/tpm/tpm2-sessions.c
> > > index 3f89635ba5e8..102e099f22c1 100644
> > > --- a/drivers/char/tpm/tpm2-sessions.c
> > > +++ b/drivers/char/tpm/tpm2-sessions.c
> > > @@ -40,11 +40,6 @@
> > >  *
> > >  * These are the usage functions:
> > >  *
> > > - * tpm2_start_auth_session() which allocates the opaque auth structure
> > > - *	and gets a session from the TPM.  This must be called before
> > > - *	any of the following functions.  The session is protected by a
> > > - *	session_key which is derived from a random salt value
> > > - *	encrypted to the NULL seed.
> > >  * tpm2_end_auth_session() kills the session and frees the resources.
> > >  *	Under normal operation this function is done by
> > >  *	tpm_buf_check_hmac_response(), so this is only to be used on
> > > @@ -963,16 +958,13 @@ static int tpm2_load_null(struct tpm_chip *chip, u32 *null_key)
> > > }
> > > 
> > > /**
> > > - * tpm2_start_auth_session() - create a HMAC authentication session with the TPM
> > > - * @chip: the TPM chip structure to create the session with
> > > + * tpm2_start_auth_session() - Create an a HMAC authentication session
> > > + * @chip:	A TPM chip
> > >  *
> > > - * This function loads the NULL seed from its saved context and starts
> > > - * an authentication session on the null seed, fills in the
> > > - * @chip->auth structure to contain all the session details necessary
> > > - * for performing the HMAC, encrypt and decrypt operations and
> > > - * returns.  The NULL seed is flushed before this function returns.
> > > + * Loads the ephemeral key (null seed), and starts an HMAC authenticated
> > > + * session. The null seed is flushed before the return.
> > >  *
> > > - * Return: zero on success or actual error encountered.
> > > + * Returns zero on success, or a POSIX error code.
> > >  */
> > > int tpm2_start_auth_session(struct tpm_chip *chip)
> > > {
> > > @@ -1024,7 +1016,7 @@ int tpm2_start_auth_session(struct tpm_chip *chip)
> > > 	/* hash algorithm for session */
> > > 	tpm_buf_append_u16(&buf, TPM_ALG_SHA256);
> > > 
> > > -	rc = tpm_transmit_cmd(chip, &buf, 0, "start auth session");
> > > +	rc = tpm_to_ret(tpm_transmit_cmd(chip, &buf, 0, "StartAuthSession"));
> > > 	tpm2_flush_context(chip, null_key);
> > > 
> > > 	if (rc == TPM2_RC_SUCCESS)
> > > diff --git a/include/linux/tpm.h b/include/linux/tpm.h
> > > index 6c3125300c00..c826d5a9d894 100644
> > > --- a/include/linux/tpm.h
> > > +++ b/include/linux/tpm.h
> > > @@ -257,8 +257,29 @@ enum tpm2_return_codes {
> > > 	TPM2_RC_TESTING		= 0x090A, /* RC_WARN */
> > > 	TPM2_RC_REFERENCE_H0	= 0x0910,
> > > 	TPM2_RC_RETRY		= 0x0922,
> > > +	TPM2_RC_SESSION_MEMORY	= 0x0903,
> > 
> > nit: the other values are in ascending order, should we keep it or is it not
> > important?
> > 
> > (more a question for me than for the patch)
> 
> nope
> 
> > 
> > > };
> > > 
> > > +/*
> > > + * Convert a return value from tpm_transmit_cmd() to a POSIX return value. The
> > > + * fallback return value is -EFAULT.
> > > + */
> > > +static inline ssize_t tpm_to_ret(ssize_t ret)
> > > +{
> > > +	/* Already a POSIX error: */
> > > +	if (ret < 0)
> > > +		return ret;
> > > +
> > > +	switch (ret) {
> > > +	case TPM2_RC_SUCCESS:
> > > +		return 0;
> > > +	case TPM2_RC_SESSION_MEMORY:
> > > +		return -ENOMEM;
> > > +	default:
> > > +		return -EFAULT;
> > > +	}
> > > +}
> > 
> > I like this and in the future we could reuse it in different places like
> > tpm2_load_context() and tpm2_save_context().
> > 
> > Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
> > 
> > 
> > BTW for my understading, looking at that code (sorry if the answer is
> > obvious, but I'm learning) I'm confused about the use of tpm2_rc_value().
> > 
> > For example in tpm2_load_context() we have:
> > 
> >     	rc = tpm_transmit_cmd(chip, &tbuf, 4, NULL);
> >     	...
> > 	} else if (tpm2_rc_value(rc) == TPM2_RC_HANDLE ||
> > 		   rc == TPM2_RC_REFERENCE_H0) {
> > 
> > While in tpm2_save_context(), we have:
> > 
> > 	rc = tpm_transmit_cmd(chip, &tbuf, 0, NULL);
> > 	...
> > 	} else if (tpm2_rc_value(rc) == TPM2_RC_REFERENCE_H0) {
> > 
> > So to check TPM2_RC_REFERENCE_H0 we are using tpm2_rc_value() only
> > sometimes, what's the reason?
> 
> Good catch, I'll update...
> 
> TPM RC is a struct or bitfield.

Applied to my -next: https://web.git.kernel.org/pub/scm/linux/kernel/git/jarkko/linux-tpmdd.git/log/?h=next

BR, Jarkko