3. If, after the TRNG_CFG.MINREFCYC sampling period, the random data buffer is not completely filled, a
new snapshot of the LFSR is taken and stored in the random data buffer (after optional postprocessing). Con-
trol branches back to point step 2 above and a new TRNG_CFG.MINREFCYC sample period starts.
Locking Detection and Prevention
Lock detection in functional mode uses the sampled outputs of the individual FROs. A FRO alarm event is declared
when a repeating pattern (of up to four sample lengths) is detected continuously for the number of samples defined
in the alarm threshold TRNG_ALMCNT.ALMTHRESH bit field. The alarm event is logged by setting the bit that
corresponds to the FRO that caused the alarm in the
sponding bit in the
TRNG_ALMSTP
that FRO. If the
TRNG_ALMMSK
break locking.
The shutdown count field in the alarm count TRNG_ALMCNT.SHDNCNT register monitors the number of FROs
switched off. (It counts the number of 1 bits in the
(TRNG_ALMCNT.SHDNTHRESH) can be configured to generate the shutdown overflow interrupt
(TRNG_STAT.SHDNOVR). When the shutdown count in the TRNG_ALMCNT.SHDNCNT bit field exceeds the
shutdown threshold in the TRNG_ALMCNT.SHDNTHRESH bit field, the shutdown overflow bit
(TRNG_STAT.SHDNOVR) is set to 1 (which can be used to generate an interrupt).
Software can use two strategies for the TRNG operation:
• Monitored Operation. Software checks the
onds). If a bit is set in that register, then the program must also check the
mine if a FRO was shut down due to multiple alarm events. If no FROs are shut down, the program clears the
register to remove the incidental alarm events. If one or more FROs are shut down, the host
TRNG_ALMMSK
processor can modify the delay selection of those FROs using the
ther locking. For this type of operation, the shutdown threshold is normally set to a low value (two, for exam-
ple). The shutdown overflow interrupt can then be used to signal abnormal operation conditions or the break-
down of FROs.
• Unmonitored Operation. Software sets the shutdown threshold to the acceptable number of FROs to be shut
down before taking corrective actions. It then uses the shutdown overflow interrupt to initiate corrective ac-
tions (clearing the
register). The software must monitor the time interval between these interrupts. If they occur too often (for
example, within a minute after each other), this frequency indicates abnormal operating conditions or the
breakdown of FROs.
Run Testing
Run Test
The TRNG block counts the number of consecutive zeros and ones (runs) in the data stream shifted into the main
LFSR. The run length and bit value is then used to increment a specific bucket counter for these values. After
ADSP-SC58x/ADSP-2158x SHARC+ Processor Hardware Reference
register is set. The FRO is switched off to prevent further alarm events from
register bit was not yet set, the FRO is restarted automatically in an attempt to
TRNG_ALMMSK
and
TRNG_ALMSTP
TRNG_ALMMSK
register. If that bit was already set, the corre-
register.) The shutdown threshold field
TRNG_ALMSTP
TRNG_ALMMSK
register at regular intervals (on the order of sec-
TRNG_FRODETUNE
registers, toggling bits in the
TRNG Functional Description
TRNG_ALMSTP
register to deter-
register to prevent fur-
TRNG_FRODETUNE
47–3
Need help?
Do you have a question about the ADSP-SC58 Series and is the answer not in the manual?