drivers/kw41zrf: don't re-get UIDs during reset

[benemorius]

Contribution description

This prevents kw41zrf from changing the long and short addresses during ifconfig 7 set state reset.

Testing procedure

• Using gnrc_networking example, run ifconfig and observe addresses.
• Run ifconfig 7 set state reset
• Run ifconfig again and observe whether the long and short addresses have changed.
• Without this PR, the addresses will have changed.

[benpicco]

This problem affects all drivers, I'd rather have a more general solution.

[miri64]

Partly fixes #9656, but I agree with @benpicco, a more holistic solution would be better.

[maribu]

This problem affects all drivers, I'd rather have a more general solution.

👍

I think it is an issue with the architecture of the drivers. IMO, the address should be generated once during <FOO>_init() or <FOO>_setup(). In <FOO>_reset() only the already generated address should be applied to the hardware. In case of IEEE 802.15.4 transceivers, one might be tempted to introduce an netdev_ieee802154_init() to do just this and have the individual 802.15.4 drivers call this during their <FOO>_init() functions. (Ideally consistently e.g. as first line in the function body.)

[benpicco]

#13746 will assign IDs to netdevs, this way we can always request the same ID from luid_custom without having to store the result.

[maribu]

#13746 will assign IDs to netdevs, this way we can always request the same ID from luid_custom without having to store the result.

IMO this not the solution to the problem here. If the user changes the L2 address by hand and triggers a reset (e.g. both could be done via the ifconfig shell command), one would expect the device to still have the assigned address. And the network stack store the L2 addresses in RAM anyway - so there is no cost.

But we shouldn't do something like

void foo_init(foo_dev_t *dev)
{
    /* ... */
    foo_reset(dev);
    /* ... */
}

void foo_reset(foo_dev_t *dev)
{
    static bool is_first_reset = 1;
    /* ... */
    if (is_first_reset) {
        is_first_reset = 0;
        generate_new_addr();
    }
    apply_addr();
}

but instead:

void foo_init(foo_dev_t *dev)
{
    /* ... */
    generate_new_addr();
    foo_reset(dev);
    /* ... */
}

void foo_reset(foo_dev_t *dev)
{
    /* ... */
    apply_addr();
}

(Or have the L2 addr generate in the foo_setup() rather than the foo_init().) The latter would have no cost in terms of RAM or ROM and fixes the issue.

Still, #13746 provides stable L2 addresses (regardless of number and order of network devices). So #13746 solves a problem, but not this.

[maribu]

So #13746 solves a problem, but not this.

Now it does solve this as well :-)

[benpicco]

Should be fixed now.

[fjmolinas]

Tested on master this is no longer an issuse:

2021-11-18 10:40:18,454 # Iface  6  HWaddr: 04:42  Channel: 26  Page: 0  NID: 0x23 
2021-11-18 10:40:18,458 #           Long HWaddr: 42:84:62:A2:5A:ED:6D:A6 
2021-11-18 10:40:18,466 #            TX-Power: 0dBm  State: IDLE  max. Retrans.: 3  CSMA Retries: 4 
2021-11-18 10:40:18,471 #           ACK_REQ  CSMA  L2-PDU:102  MTU:1280  HL:64  RTR  
2021-11-18 10:40:18,472 #           6LO  IPHC  
2021-11-18 10:40:18,476 #           Source address length: 8
2021-11-18 10:40:18,478 #           Link type: wireless
2021-11-18 10:40:18,485 #           inet6 addr: fe80::4084:62a2:5aed:6da6  scope: link  VAL
2021-11-18 10:40:18,487 #           inet6 group: ff02::2
2021-11-18 10:40:18,491 #           inet6 group: ff02::1
2021-11-18 10:40:18,493 #           inet6 group: ff02::1:ffed:6da6
2021-11-18 10:40:18,495 #           
2021-11-18 10:40:18,498 #           Statistics for Layer 2
2021-11-18 10:40:18,501 #             RX packets 0  bytes 0
2021-11-18 10:40:18,506 #             TX packets 2 (Multicast: 2)  bytes 86
2021-11-18 10:40:18,509 #             TX succeeded 2 errors 0
2021-11-18 10:40:18,511 #           Statistics for IPv6
2021-11-18 10:40:18,515 #             RX packets 0  bytes 0
2021-11-18 10:40:18,519 #             TX packets 2 (Multicast: 2)  bytes 128
2021-11-18 10:40:18,521 #             TX succeeded 2 errors 0
2021-11-18 10:40:18,522 # 
> ifconfig 6 set state reset
2021-11-18 10:40:24,170 # ifconfig 6 set state reset
2021-11-18 10:40:24,176 # success: set state of interface 6 to RESET
> ifconfig
2021-11-18 10:40:25,681 # ifconfig
2021-11-18 10:40:25,686 # Iface  6  HWaddr: 04:42  Channel: 26  Page: 0  NID: 0x23 
2021-11-18 10:40:25,691 #           Long HWaddr: 42:84:62:A2:5A:ED:6D:A6 
2021-11-18 10:40:25,698 #            TX-Power: 0dBm  State: IDLE  max. Retrans.: 3  CSMA Retries: 4 
2021-11-18 10:40:25,703 #           ACK_REQ  CSMA  L2-PDU:102  MTU:1280  HL:64  RTR  
2021-11-18 10:40:25,705 #           6LO  IPHC  
2021-11-18 10:40:25,708 #           Source address length: 2
2021-11-18 10:40:25,711 #           Link type: wireless
2021-11-18 10:40:25,717 #           inet6 addr: fe80::4084:62a2:5aed:6da6  scope: link  VAL
2021-11-18 10:40:25,719 #           inet6 group: ff02::2
2021-11-18 10:40:25,722 #           inet6 group: ff02::1
2021-11-18 10:40:25,726 #           inet6 group: ff02::1:ffed:6da6
2021-11-18 10:40:25,727 #           
2021-11-18 10:40:25,730 #           Statistics for Layer 2
2021-11-18 10:40:25,733 #             RX packets 0  bytes 0
2021-11-18 10:40:25,737 #             TX packets 3 (Multicast: 3)  bytes 129
2021-11-18 10:40:25,740 #             TX succeeded 3 errors 0
2021-11-18 10:40:25,743 #           Statistics for IPv6
2021-11-18 10:40:25,746 #             RX packets 0  bytes 0
2021-11-18 10:40:25,751 #             TX packets 3 (Multicast: 3)  bytes 192
2021-11-18 10:40:25,754 #             TX succeeded 3 errors 0
2021-11-18 10:40:25,754 #