Jarkko, should there be tpm_chip_start/tpm_chip_stop calls around the tpm1_getcap calls in tpm1_get_timeouts?