[PATCH v4 2/4] tpm: ignore burstcount to improve tpm_tis send() performance

[kgold@linux.vnet.ibm.com (Ken Goldman)]

On 10/20/2017 10:42 AM, Alexander.Steffen at infineon.com wrote:
> 
> This seems to fail reliably with my SPI TPM 2.0. I get EIO when
> trying to send large amounts of data, e.g. with TPM2_Hash, and
> subsequent tests seem to take an unusual amount of time. More
> analysis probably has to wait until November, since I am going to be
> in Prague next week.

I have a guess as to the cause of the failure.  Would it be possible for 
you to test it?

1 - My guess is that EIO is coming from here:

static int tpm_tis_send_data(struct tpm_chip *chip, u8 *buf, size_t len)
...
	/* write last byte */
	rc = tpm_tis_write8(priv, TPM_DATA_FIFO(priv->locality), buf[count]);
	if (rc < 0)
		goto out_err;

	if (wait_for_tpm_stat(chip, TPM_STS_VALID, chip->timeout_c,
				&priv->int_queue, false) < 0) {
		rc = -ETIME;
		goto out_err;
	}
	status = tpm_tis_status(chip);
	if (!itpm && (status & TPM_STS_DATA_EXPECT) != 0) {
		rc = -EIO;
		goto out_err;
	}
...

Can you verify that this is the cause.

2 - If that's the cause, I believe that there is a latent bug.  Expect 
is not guaranteed to become false immediately.  It only occurs after the 
TPM firmware has emptied the FIFO.  Thus, the tpm_tis_status() really 
should be something like "wait_for_tpm_expect_false()", with a sleep loop.

This missing wait has been in the code for a while.  If may just surface 
now because the patch causes data to be written faster, and thus it 
takes longer for the TPM to empty the FIFO and clear Expect.

It also makes sense that it would occur more often on long commands.

--
To unsubscribe from this list: send the line "unsubscribe linux-security-module" in
the body of a message to majordomo at vger.kernel.org
More majordomo info@ http://vger.kernel.org/majordomo-info.html