RM520N-GL documentation for AT+QCFG=“clat"

IoT Modules 5G Module
1

Hi, I’ve been testing the clat support on the RM520N-GL module using AT+QCFG=”clat”, which works but doesn’t seem to be fully translating all ICMP traffic according to spec. However, I’m not sure if I’m configuring it correctly, as this command isn’t documented in v1.1 of the AT command manual which I have.

I gather from searching the forum that there might be a specific CLAT application note for this card? I couldn’t find it in your documentation library, but if there is, please could I have a link or a copy so I can check I am configuring it correctly?

I’m running the RM520NGLAAR03A04M4G_A0.301.A0.301 firmware, which I think is the most recent version?

2

May I ask what makes you use CLAT on the modem and not on the host system?

3

At the moment, I’m just comparing the on-card implementation with a couple of possible host-side options.

That said, these cards can run in pcie root-complex mode, i.e. as hosts in their own right, in which case there isn’t a separate host system to delegate the job to.

1 Like
4

Hi @silvia , please would you be able to confirm if there’s some documentation or an application note for AT+QCFG=”clat”, and whether RM520NGLAAR03A04M4G_A0.301.A0.301 is the most recent version of the R03A04 series for this card? Many thanks!

5

Dear @arachsys
I have sent document to you via Message, please check,

Sent the firmware to your email.

6

Many thanks @silvia, that was really helpful.

I’m seeing a bug with your CLAT implementation, where inbound icmp6 time-exceeded messages with source addresses in the PLAT prefix aren’t being correctly translated to the corresponding v4 icmp time-exceeded messages, thus breaking traceroute and mtr through CLAT.

I’ve confirmed that the provider’s PLAT layer is handling these correctly, e.g. traceroute to 64:ff9b::8.8.8.8 (if you’ll excuse my notation) will work fine, with icmp6 ttl-exceeded for the v4 hops being returned from nat64 addresses as expected. I’ve also seen other CLAT implementations work correctly in this situation.

What’s the right route for me to report this bug, or are you able to forward to the right contact for me?

7

PS I should confirm, I’m using clat with the correct <icmpv4_err_mode> = 2 and <icmpv6_err_mode> = 2. (It’s almost behaving as though these weren’t in effect, but AT+QCFG=”clat” confirms they’re right.)