Re: [PATCH 1/3] tpm/tpm_crb: implement tpm crb idle state

[Jarkko Sakkinen <jarkko.sakkinen-VuQAYsv1563Yd54FQh9/CA@public.gmane.org>]

On Thu, Sep 08, 2016 at 12:35:48PM +0000, Winkler, Tomas wrote:
> 
> 
> > -----Original Message-----
> > From: Jarkko Sakkinen [mailto:jarkko.sakkinen-VuQAYsv1563Yd54FQh9/CA@public.gmane.org]
> > Sent: Thursday, September 08, 2016 14:18
> > To: Winkler, Tomas <tomas.winkler-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> > Cc: tpmdd-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> > Subject: Re: [PATCH 1/3] tpm/tpm_crb: implement tpm crb idle state
> > 
> > On Thu, Sep 08, 2016 at 02:11:15PM +0300, Jarkko Sakkinen wrote:
> > > On Wed, Sep 07, 2016 at 02:32:31PM +0300, Tomas Winkler wrote:
> > > > The register TPM_CRB_CTRL_REQ_x contains bits goIdle and cmdReady
> > > > for SW to indicate that the device can enter or should exit the idle state.
> > > >
> > > > The legacy ACPI-start (SMI + DMA) based devices do not support these
> > > > bits and the idle state management is not exposed to the host SW.
> > > > Thus, this functionality only is enabled only for a CRB start (MMIO)
> > > > based devices.
> > > >
> > > > We introduce two new callbacks for command ready and go idle for TPM
> > > > CRB device which are called across TPM transactions.
> > > >
> > > > Based on Jarkko Sakkinen <jarkko.sakkinen-VuQAYsv1563Yd54FQh9/CA@public.gmane.org> oringal
> > > > patch
> > > > 'tpm_crb: implement power tpm crb power management'
> > > >
> > > > Signed-off-by: Tomas Winkler <tomas.winkler-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> > > > ---
> > > >  drivers/char/tpm/tpm-interface.c | 21 +++++++++++
> > > >  drivers/char/tpm/tpm_crb.c       | 77
> > ++++++++++++++++++++++++++++++++++++++++
> > > >  include/linux/tpm.h              |  3 +-
> > > >  3 files changed, 100 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/drivers/char/tpm/tpm-interface.c
> > > > b/drivers/char/tpm/tpm-interface.c
> > > > index fd863ff30f79..c78dca5ce7a6 100644
> > > > --- a/drivers/char/tpm/tpm-interface.c
> > > > +++ b/drivers/char/tpm/tpm-interface.c
> > > > @@ -327,6 +327,20 @@ unsigned long tpm_calc_ordinal_duration(struct
> > > > tpm_chip *chip,  }  EXPORT_SYMBOL_GPL(tpm_calc_ordinal_duration);
> > > >
> > > > +static inline int tpm_go_idle(struct tpm_chip *chip) {
> > > > +	if (!chip->ops->idle)
> > > > +		return 0;
> > > > +	return chip->ops->idle(chip);
> > > > +}
> > > > +
> > > > +static inline int tpm_cmd_ready(struct tpm_chip *chip) {
> > > > +	if (!chip->ops->ready)
> > > > +		return 0;
> > > > +	return chip->ops->ready(chip);
> > > > +}
> > > > +
> > > >  /*
> > > >   * Internal kernel interface to transmit TPM commands
> > > >   */
> > > > @@ -353,6 +367,10 @@ ssize_t tpm_transmit(struct tpm_chip *chip,
> > const u8 *buf, size_t bufsiz,
> > > >  	if (!(flags & TPM_TRANSMIT_UNLOCKED))
> > > >  		mutex_lock(&chip->tpm_mutex);
> > > >
> > > > +	rc = tpm_cmd_ready(chip);
> > > > +	if (rc)
> > > > +		goto out;
> > > > +
> > > >  	rc = chip->ops->send(chip, (u8 *) buf, count);
> > > >  	if (rc < 0) {
> > > >  		dev_err(&chip->dev,
> > > > @@ -394,8 +412,11 @@ out_recv:
> > > >  		dev_err(&chip->dev,
> > > >  			"tpm_transmit: tpm_recv: error %zd\n", rc);
> > > >  out:
> > > > +	tpm_go_idle(chip);
> > > > +
> > > >  	if (!(flags & TPM_TRANSMIT_UNLOCKED))
> > > >  		mutex_unlock(&chip->tpm_mutex);
> > > > +
> > > >  	return rc;
> > > >  }
> > > >
> > > > diff --git a/drivers/char/tpm/tpm_crb.c b/drivers/char/tpm/tpm_crb.c
> > > > index 82a3ccd52a3a..98a7fdfe9936 100644
> > > > --- a/drivers/char/tpm/tpm_crb.c
> > > > +++ b/drivers/char/tpm/tpm_crb.c
> > > > @@ -83,6 +83,81 @@ struct crb_priv {
> > > >  	u8 __iomem *rsp;
> > > >  };
> > > >
> > > > +/**
> > > > + * __crb_go_idle - write CRB_CTRL_REQ_GO_IDLE to
> > TPM_CRB_CTRL_REQ
> > > > + *    The device should respond within TIMEOUT_C by clearing the bit.
> > > > + *    Anyhow, we do not wait here as a consequent CMD_READY request
> > > > + *    will be handled correctly even if idle was not completed.
> > > > + *
> > > > + * @dev: tpm device
> > > > + * @priv: crb private context
> > > > + *
> > > > + * Return:  0 always
> > > > + */
> > > > +static int __crb_go_idle(struct device *dev, struct crb_priv *priv)
> > > > +{
> > > > +	if (priv->flags & CRB_FL_ACPI_START)
> > > > +		return 0;
> > > > +	iowrite32(CRB_CTRL_REQ_GO_IDLE, &priv->cca->req);
> > > > +	/* we don't really care when this settles */
> > > > +
> > > > +	return 0;
> > > > +}
> > > > +
> > > > +static int crb_go_idle(struct tpm_chip *chip) {
> > > > +	struct crb_priv *priv = dev_get_drvdata(&chip->dev);
> > > > +
> > > > +	return __crb_go_idle(&chip->dev, priv); }
> > > > +
> > > > +/**
> > > > + * __crb_cmd_ready - write CRB_CTRL_REQ_CMD_READY to
> > TPM_CRB_CTRL_REQ
> > > > + *      and poll till the device acknowledge it by clearing the bit.
> > > > + *      The device should respond within TIMEOUT_C.
> > > > + *
> > > > + *      The function does nothing for devices with ACPI-start method
> > > > + *
> > > > + * @dev: tpm device
> > > > + * @priv: crb private context
> > > > + *
> > > > + * Return:  0 on success -ETIME on timeout;  */ static int
> > > > +__crb_cmd_ready(struct device *dev, struct crb_priv *priv) {
> > > > +	ktime_t stop, start;
> > > > +
> > > > +	if (priv->flags & CRB_FL_ACPI_START)
> > > > +		return 0;
> > > > +
> > > > +	iowrite32(CRB_CTRL_REQ_CMD_READY, &priv->cca->req);
> > > > +
> > > > +	start = ktime_get();
> > > > +	stop = ktime_add(start, ms_to_ktime(TPM2_TIMEOUT_C));
> > > > +	do {
> > > > +		if (!(ioread32(&priv->cca->req) &
> > CRB_CTRL_REQ_CMD_READY)) {
> > > > +			dev_dbg(dev, "cmdReady in %lld usecs\n",
> > > > +				ktime_to_us(ktime_sub(ktime_get(), start)));
> > > > +			return 0;
> > > > +		}
> > > > +		usleep_range(500, 1000);
> > > > +	} while (ktime_before(ktime_get(), stop));
> > >
> > > What's the problem of using wait_for_tpm_stat like:
> > >
> > > http://git.infradead.org/users/jjs/linux-tpmdd.git/commitdiff/7a1172b5
> > > b3cb38083ae931309db216db3c528efe
> > 
> > I'm proponent for my version since it's less intrusive change and adds less
> > ad-hoc code to the CRB driver.
> 
> I'm not sure what is adhoc about this code, this keeps the issue local
> to crb .  I removed the use of wait_for_tpm_stat() for purpose, this
> just abusing this interface for something which is not 'stats'  in
> addition you are setting rubbish from the CA_STATUS as its value is
> not retained  and on the other side  the values of CA_REQUEST is read
> from  edge  (from 1 to 0) so there will be rubbish collected when we
> are not in transition. Last we need to fine tune polling delay for the
> HW we cannot have it in the generic code.

Well we anyway use a synthetized status value for the CRB driver
so I don't see your point. And wait_for_tpm_stat() allows you to
specify any delay.

For me this looks just  more duplicate code to maintain.

/Jarkko

------------------------------------------------------------------------------