Re: [PATCH 3/3] tpm2-cmd: React correctly to RC_TESTING from self tests

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

On Fri, Aug 18, 2017 at 02:15:32PM +0200, Alexander Steffen wrote:
> The TPM can choose one of two ways to react to the TPM2_SelfTest command.
> It can either run all self tests synchronously and then return RC_SUCCESS
> once all tests were successful. Or it can choose to run the tests
> asynchronously and return RC_TESTING immediately while the self tests still
> execute in the background.
> 
> The previous implementation apparently was not aware of those possibilities
> and attributed RC_TESTING to some prototype chips instead. With this change
> the return code of TPM2_SelfTest is interpreted correctly, i.e. the self
> test result is polled if and only if RC_TESTING is received. Unfortunately,
> the polling cannot be done in the most straightforward way, so the solution
> is explained with a comment in the code.
> 
> Signed-off-by: Alexander Steffen <Alexander.Steffen-d0qZbvYSIPpWk0Htik3J/w@public.gmane.org>
> ---
>  drivers/char/tpm/tpm2-cmd.c | 68 ++++++++++++++++++---------------------------
>  1 file changed, 27 insertions(+), 41 deletions(-)
> 
> diff --git a/drivers/char/tpm/tpm2-cmd.c b/drivers/char/tpm/tpm2-cmd.c
> index e3d4cc3..d09032f 100644
> --- a/drivers/char/tpm/tpm2-cmd.c
> +++ b/drivers/char/tpm/tpm2-cmd.c
> @@ -834,68 +834,54 @@ static const struct tpm_input_header tpm2_selftest_header = {
>  };
>  
>  /**
> - * tpm2_continue_selftest() - start a self test
> - *
> - * @chip: TPM chip to use
> - * @full: test all commands instead of testing only those that were not
> - *        previously tested.
> - *
> - * Return: Same as with tpm_transmit_cmd with exception of RC_TESTING.
> - */
> -static int tpm2_start_selftest(struct tpm_chip *chip, bool full)
> -{
> -	int rc;
> -	struct tpm2_cmd cmd;
> -
> -	cmd.header.in = tpm2_selftest_header;
> -	cmd.params.selftest_in.full_test = full;
> -
> -	rc = tpm_transmit_cmd(chip, NULL, &cmd, TPM2_SELF_TEST_IN_SIZE, 0, 0,
> -			      "continue selftest");
> -
> -	/* At least some prototype chips seem to give RC_TESTING error
> -	 * immediately. This is a workaround for that.
> -	 */
> -	if (rc == TPM2_RC_TESTING) {
> -		dev_warn(&chip->dev, "Got RC_TESTING, ignoring\n");
> -		rc = 0;
> -	}
> -
> -	return rc;
> -}
> -
> -/**
>   * tpm2_do_selftest() - ensure that all self tests have passed
>   *
>   * @chip: TPM chip to use
>   *
>   * Return: Same as with tpm_transmit_cmd.
>   *
> - * During the self test TPM2 commands return with the error code RC_TESTING.
> - * Waiting is done by issuing PCR read until it executes successfully.
> + * The TPM can either run all self tests synchronously and then return
> + * RC_SUCCESS once all tests were successful. Or it can choose to run the tests
> + * asynchronously and return RC_TESTING immediately while the self tests still
> + * execute in the background. This function handles both cases and waits until
> + * all tests have completed.
>   */
>  static int tpm2_do_selftest(struct tpm_chip *chip)
>  {
>  	int rc;
>  	unsigned int delay_msec = 20;
>  	long duration;
> +	struct tpm2_cmd cmd;
>  
>  	duration = jiffies_to_msecs(
>  		tpm2_calc_ordinal_duration(chip, TPM2_CC_SELF_TEST));
>  
> -	rc = tpm2_start_selftest(chip, false);
> -	if (rc)
> -		return rc;
> -
>  	while (duration > 0) {
> -		/* Attempt to read a PCR value */
> -		rc = tpm2_pcr_read(chip, 0, NULL);
> -		if (rc < 0)
> -			break;
> +		cmd.header.in = tpm2_selftest_header;
> +		cmd.params.selftest_in.full_test = 0;
> +
> +		rc = tpm_transmit_cmd(chip, NULL, &cmd, TPM2_SELF_TEST_IN_SIZE,
> +				      0, 0, "continue selftest");
>  
>  		if (rc != TPM2_RC_TESTING)
>  			break;
>  
> +		/* If RC_TESTING is received, ideally the code should now poll
> +		 * the selfTestDone bit in the STS register, as this avoids
> +		 * sending more commands, that might interrupt self tests
> +		 * executing in the background and thus prevent them from ever
> +		 * completing.
> +		 * Unfortunately, it cannot be guaranteed that this bit is
> +		 * correctly implemented for all devices, so the next best thing
> +		 * would be to use TPM2_GetTestResult to query the test result.
> +		 * But the response to that command can be very long, and the
> +		 * code currently lacks the capabilities for efficient
> +		 * unmarshalling, so it is difficult to execute this command.
> +		 * Therefore, we simply run the TPM2_SelfTest command in a loop,
> +		 * which should complete eventually, since we only request the
> +		 * execution of self tests that have not yet been done.
> +		 */

This should be part of the commit message, not part of the code.

/Jarkko

> +
>  		msleep(delay_msec);
>  		duration -= delay_msec;
>  
> -- 
> 2.7.4
> 

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot