Assuming you took the latest commit with the initramfs fix yes the single fw change is ready
Can you link it again?
This. Only last 2 thing kernel size is a little lower than what it should be and the partition is not expanded till the real end of the nand. (Have problems with size conversion)
It's bad but apparently without the help of QCA we will never be able to use newer firmware unless somebody is deep into RE and able to find out what crashes the firmware.
Extracted the relevant BDF from the 2.7 commit, repacked it to match the AX6 board (id 658), will test is in a couple minutes.
At this point, what do you think: is it something int the BDF, or something in ath11k?
Fw assume stuff from bdf and bdf have wrong or missing data. Problem is that my idea is that the problem is in the precal bdf and not in board bdf.. if it's like that then they have to fix it.
FW 2.7 with BDF 2.7 yields the same crash. As I tested previous working version (2.4 and 2.5) with their matching BDFs, and they always worked (sometimes not optimally due to the missing board specific coeffs, but not crashing as well), your theory about the precal data sounds very plausible...
1 Like
In ax3600, both 2.7 and 2.6 BDF files work.
I created the new BDF with only the extracted qmi-board-id=658 for AX3600.
@dchard:
See if you are updated, I noticed that Robi interrupted the build and updated his repo (+-2 hours ago).
I also tested ax3600 with the unchanged downloaded board-2.bin file (with all the boards in) and it also works here.
1 Like
You mean you checked and they work properly? Radios coming up, speeds are ok, singnal levels as expected?
By work, I mean: it doesn't crash, radios are up, some new regulatory restrictions are applied, my mobile phone connects and I did a speed test in 5G (got my ISP speed 500/100).
I didn't verify if the new regulatory restrictions are correct.
I saw 2.4 still only goes to chan 11 and max signal is now limited to 20dBm, before it was 30 (incorrect for my country).
I saw I got max signal next to the router.
I don't know what signal lvls are to be expected near and far, so I can't test that.
2 Likes
The new 2.7.0.1-01744 firmware looks good on the dynalink. (with "our" current board file)
AP Mode works, STA mode works and even the crash during sysupgrade, that we saw with the last 2.7.0.1, isn't there any more.
3 Likes
Can you please upload that boardfile and preferably your caldata file as well to wetransfer? You can send the link in private if you want.
1 Like
Can someone please point me where I can find the code for the on-SoC ethernet "switch" (or as much as is avail of this stuff)?
I have a weird problem that if I roam from another AP to the AX3600 (no mesh) the new STA sends a DHCP DISCOVER which hits the DHCP server, a OFFER comes back out on the rack switch port (confirmed by classic old 10Mbit/s HUB in between and SPAN port on the switch) towards the AX3600 but then it shows up nowhere (neither eth0/WAN port, nor the vlan or bridge and neither at the STA) according to tcpdump. After an so far unmeasured amount of time, things simply start working, which seems like a classic ARP table problem on the on-SoC switchy-thingy thinking the node is not local (which it now is). I've tried to disable as much firewall as I could find but I have to admit I am fairly new to OpenWRT.
IPv6-only works just fine, which is why I didn't notice for a while but others using IPv4 did.
2 Likes
@robimarko @dchard
Somehow I only saw the post asking to try the 2.7 BDF and not the previous post asking to test the new 2.7 FW.
The new 2.7 FW also crashes here.
Sorry for the confusion.
New Firmware runs fine on my Dynalink.
[ 9.455274] ath11k c000000.wifi: qmi ignore invalid mem req type 3
[ 9.462814] ath11k c000000.wifi: chip_id 0x0 chip_family 0x0 board_id 0xff soc_id 0xffffffff
[ 9.462869] ath11k c000000.wifi: fw_version 0x270204a5 fw_build_timestamp 2022-08-04 13:05 fw_build_id WLAN.HK.2.7.0.1-01744-QCAHKSWPL_SILICONZ-1
[ 10.142250] EXT4-fs (sda1): mounted filesystem with ordered data mode. Opts: . Quota mode: disabled.
@robimarko is the Dynalink + QNAP to be included on the PR or just the xiaomi ax3600?
Everything that is currently in the PR tree, so that leaves out Dynalink unless somebody can finish it and make a PR
thank you @robimarko. @kirdes your build feels very stable ... I have been running it for over a month now. What is outstanding ? a suggestion probably worth waiting for Robi's PR to be accepted and then a raise a PR out of your repo
I can do a PR for the Dynalink, but that would be single partition only and install via UART and initramfs.
Maybe clayface can add dual partition support later.
4 Likes
thank you @kirdes yes that feels the best and quickest approach to have dynalink on the mainline