cc2538_rf: default to netdev_ieee802154_submac

[jia200x]

Contribution description

Following #15132, this PR makes netdev_ieee802154_submac default for cc2538_rf. The basic driver doesn't support ACK handling and retransmissions, so the SubMAC fixes this lack.

Testing procedure

Compare ping statistics (with 1kb packets) for any cc2538_rf board between current master and this PR.

Issues/PRs references

#15132

[fjmolinas]

If I compile without using docker (newer toolchain) I get the following:

/home/francisco/workspace/RIOT2/sys/net/link_layer/ieee802154/submac.c: In function 'ieee802154_set_state':
/home/francisco/workspace/RIOT2/sys/net/link_layer/ieee802154/submac.c:415:21: error: implicit conversion from 'enum <anonymous>' to 'ieee802154_submac_state_t' [-Werror=enum-conversion]
  415 |                     state == IEEE802154_STATE_IDLE
      |                     ^~~~~
/home/francisco/workspace/RIOT2/sys/net/link_layer/ieee802154/submac.c:419:64: error: implicit conversion from 'ieee802154_submac_state_t' to 'ieee802154_trx_state_t' [-Werror=enum-conversion]
  419 |         if ((res = ieee802154_radio_request_set_trx_state(dev, new_state)) < 0) {
      |                                                                ^~~~~~~~~
cc1: all warnings being treated as errors
make[3]: *** [/home/francisco/workspace/RIOT2/Makefile.base:107: /home/francisco/workspace/RIOT2/examples/gnrc_networking/bin/openmote-b/ieee802154/submac.o] Error 1
make[2]: *** [/home/francisco/workspace/RIOT2/Makefile.base:30: ALL--/home/francisco/workspace/RIOT2/sys/net/link_layer/ieee802154] Error 2
make[2]: *** Waiting for unfinished jobs....

issue with new gcc needs to be resolved

[fjmolinas]

/home/francisco/workspace/RIOT2/sys/net/link_layer/ieee802154/submac.c: In function 'ieee802154_set_state':
/home/francisco/workspace/RIOT2/sys/net/link_layer/ieee802154/submac.c:415:21: error: implicit conversion from 'enum ' to 'ieee802154_submac_state_t' [-Werror=enum-conversion]
415 | state == IEEE802154_STATE_IDLE
| ^~~~~
/home/francisco/workspace/RIOT2/sys/net/link_layer/ieee802154/submac.c:419:64: error: implicit conversion from 'ieee802154_submac_state_t' to 'ieee802154_trx_state_t' [-Werror=enum-conversion]
419 | if ((res = ieee802154_radio_request_set_trx_state(dev, new_state)) < 0) {
| ^~~~~~~~~
cc1: all warnings being treated as errors
make[3]: *** [/home/francisco/workspace/RIOT2/Makefile.base:107: /home/francisco/workspace/RIOT2/examples/gnrc_networking/bin/openmote-b/ieee802154/submac.o] Error 1
make[2]: *** [/home/francisco/workspace/RIOT2/Makefile.base:30: ALL--/home/francisco/workspace/RIOT2/sys/net/link_layer/ieee802154] Error 2
make[2]: *** Waiting for unfinished jobs....

Seems this is unrelated to this PR though and should be fixed in master.

[fjmolinas]

When built in docker and pinging with parameters ping6 -s 1024 -c 10 -i 170

• master

2020-12-08 14:34:17,359 #  ping6 -s 1024 -c 10 -i 170 fe80::204:2519:1801:cca0
2020-12-08 14:34:19,886 #
2020-12-08 14:34:19,887 # --- fe80::204:2519:1801:cca0 PING statistics ---
2020-12-08 14:34:19,903 # 10 packets transmitted, 0 packets received, 100% packet loss

• pr

2020-12-08 14:34:50,511 #  ping6 -s 1024 -c 10 -i 170 fe80::204:2519:1801:cca0
2020-12-08 14:34:50,670 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=0 ttl=64 rssi=191 dBm time=146.890 ms
2020-12-08 14:34:50,831 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=1 ttl=64 rssi=191 dBm time=149.028 ms
2020-12-08 14:34:51,007 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=2 ttl=64 rssi=191 dBm time=147.896 ms
2020-12-08 14:34:51,182 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=3 ttl=64 rssi=191 dBm time=147.283 ms
2020-12-08 14:34:51,342 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=4 ttl=64 rssi=191 dBm time=139.046 ms
2020-12-08 14:34:51,518 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=5 ttl=64 rssi=191 dBm time=142.993 ms
2020-12-08 14:34:51,679 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=6 ttl=64 rssi=191 dBm time=143.986 ms
2020-12-08 14:34:51,855 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=7 ttl=64 rssi=191 dBm time=150.631 ms
2020-12-08 14:34:52,030 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=8 ttl=64 rssi=191 dBm time=145.850 ms
2020-12-08 14:34:52,191 # 1032 bytes from fe80::204:2519:1801:cca0%6: icmp_seq=9 ttl=64 rssi=191 dBm time=137.350 ms
2020-12-08 14:34:52,191 #
2020-12-08 14:34:52,192 # --- fe80::204:2519:1801:cca0 PING statistics ---
2020-12-08 14:34:52,193 # 10 packets transmitted, 10 packets received, 0% packet loss
2020-12-08 14:34:52,206 # round-trip min/avg/max = 137.350/145.095/150.631 ms

So seems situation is for some reason worse than when the cc2538_radio_ops was introduced, more reason to get this in.

[fjmolinas]

#15584 will take care of the compilation error, but for good measure we should wait for that one to get in before.

[fjmolinas]

ACK

testing something with the lavel check

[fjmolinas]

re-ACK.

[fjmolinas]

Gooo! Thanks for the great work @jia200x !!