Adding OpenWrt support for Xiaomi AX3600

Package available again. Thanks for sharing this info.

1 Like

FWIW, I'm pretty sure I can consistently sysupgrade my AX6 by doing rmmod ath11k_ahb; rmmod ath11k; sysupgrade /tmp/.... Probably not a good solution, but a solution nonetheless.

Update pushed to use the new wolfssl package with Crypto extensions

6 Likes

Looks like IPQ60xx development is being done here https://github.com/robimarko/openwrt/issues/53 in case any one is looking for it. And this is was my missing piece https://git.codelinaro.org/clo/qsdk/oss/kernel/linux-ipq-5.4/-/commit/714e6b9ab711846cd09b94f83436e23932dd2814

1 Like

Ugh, so Kalle pushed 2.7 FW but its as broken as the one under QUIC

need to find time to open the bugzilla and see if they will ignore

Shame, seems we really need QCA then.

Do you think you can maybe open an issue on Bugzilla as Kale asked? Is it a lot of work? Ansuel is trying for almost 3 months without success so far and it seems there is no other solution than asking QCA for help.

Well, I can open a ticket, especially now that its been pushed to Kalle-s repo where stuff is supposed to be backwards compatible

7 Likes

That'd be great. Then we can only hope that QCA is willing so solve this issue after all.

1 Like

it is weird that stuff that doesn't work is pushed ... or perhaps there are some QCA dependencies

I have opened a bugzilla ticket regarding the timeout issue that is bothering us for upgrading:
https://bugzilla.kernel.org/show_bug.cgi?id=216513

It would be great for more logs to be added of prints you get with similar errors once:
service wpad stop is invoked before upgrading

1 Like

After 4+ days of uptime and issuing service wpad stop I am getting this (kernel 5.15.63):

[428131.991651] device wlan1 left promiscuous mode
[428131.991806] br-lan: port 4(wlan1) entered disabled state
[428137.101175] ath11k c000000.wifi: failed to flush transmit queue, data pkts pending 6
[428137.602888] br-lan: port 5(wlan0) entered disabled state
[428137.691912] device wlan0 left promiscuous mode
[428137.691966] br-lan: port 5(wlan0) entered disabled state

Question is if this is driver or maybe something they fixed in later firmwares (2.6 or 2.7) which we cannot run atm?

Well, nobody except for QCA can tell you this.

I encourage everybody to post these in the bugzilla ticket

I'll try to. But can you have a look first to see if it's relevant? I run a slightly older firmware from your github and see the "failed to flush" also long before "wpad stop":

 OpenWrt SNAPSHOT, r0-fdabf5d
 -----------------------------------------------------
root@ax2:~# uptime
 11:22:13 up 1 day,  7:26,  load average: 0.00, 0.00, 0.00
root@ax2:~# uname -a
Linux ax2 5.15.58 #0 SMP Tue Aug 2 15:21:16 2022 aarch64 GNU/Linux
root@ax2:~# service wpad stop
root@ax2:~# dmesg
[    0.000000] Booting Linux on physical CPU 0x0000000000 [0x410fd034]
[    0.000000] Linux version 5.15.58 (runner@fv-az182-446) (aarch64-openwrt-linux-musl-gcc (OpenWrt GCC 11.3.0 r0-fdabf5d) 11.3.0, GNU ld (GNU Binutils) 2.37) #0 SMP Tue Aug 2 15:21:16 2022
[    0.000000] Machine model: Xiaomi AX3600
...
[84771.594623] Rekeying PTK for STA fa:2a:88:c7:00:2b but driver can't safely do that.
[84776.654563] ath11k c000000.wifi: failed to flush transmit queue, data pkts pending 260
[85909.622481] Rekeying PTK for STA fa:2a:88:c7:00:2b but driver can't safely do that.
[85914.652443] ath11k c000000.wifi: failed to flush transmit queue, data pkts pending 274
[92168.340858] Rekeying PTK for STA fa:2a:88:c7:00:2b but driver can't safely do that.
[92173.360795] ath11k c000000.wifi: failed to flush transmit queue, data pkts pending 286
[94285.357052] Rekeying PTK for STA fa:2a:88:c7:00:2b but driver can't safely do that.
[94290.396989] ath11k c000000.wifi: failed to flush transmit queue, data pkts pending 288
[103785.854412] Rekeying PTK for STA fa:2a:88:c7:00:2b but driver can't safely do that.
[103790.864364] ath11k c000000.wifi: failed to flush transmit queue, data pkts pending 305
[113341.252895] device wlan2 left promiscuous mode
[113341.253060] br-lan: port 5(wlan2) entered disabled state
[113346.372617] ath11k c000000.wifi: failed to flush transmit queue, data pkts pending 329
[113346.586626] ath10k_pci 0000:01:00.0: could not get mac80211 beacon, vdev_id: 0 addr: 28:d1:27:4e:56:75
[113346.634517] ath10k_pci 0000:01:00.0: mac flush null vif, drop 0 queues 0xffff
[113346.754534] br-lan: port 6(wlan0) entered disabled state
[113351.652600] ath11k c000000.wifi: failed to flush transmit queue, data pkts pending 21
[113351.703231] br-lan: port 7(wlan1) entered disabled state
[113351.743035] device wlan0 left promiscuous mode
[113351.743087] br-lan: port 6(wlan0) entered disabled state
[113351.843405] device wlan1 left promiscuous mode
[113351.843459] br-lan: port 7(wlan1) entered disabled state

Thats seems to be a similar issue, feel free to post it there

2 Likes

Hi there,

sorry for interrupting the discussion. I just want to ask if there is a possibility to recover the device from a hard-brick with jtag?

I've read this whole thread carefully and there are some hints it could work. But I do not have deeper knowledge howto manage it beside that what I've read about in the last couple of days.

I have a CH341a and I would give it a try to reflash uboot (if possible).

So from my understanding I would need to know howto connect the device with CH341a first.

Above the Flash Chip I've discovered a 2x4 pinout and I've meassured 1,8v on lets say pin 5 (read from the left up to the right above the flash chip). Others are dead respectively ∞ Ω. I'm not a electrican. oO

If this is a dead end/not worth in your eyes just say it loud! :smiley: So I can dump this unit. :frowning:

In theory yes, but in practice nobody has done it.

CH341A is not gonna be useful here at all since this is a NAND only device

2 Likes

Hi,

thanks for answering. :slight_smile:

Well then its not worth I guess. I cannot desolder the whole chip due to lack of hardware and skill ofc. If I'm going to buy all tools what I would need I still could fail to do. Not worth. :confused:

Just for interest: What would I need instead a CH341A for reprogram via JTAG?

Something like a J-Link, but the issue is software.

Nobody has use JTAG on these so far, and even if you get OpenOCD configured you would still need to enable required clocks and then train the DRAM to get U-boot loaded so you could recover

1 Like

Thx for input. That is beyond my skills. I did JTAG in the past and tinkerd with raspberrypi and FT232 (which is able to do jtag also btw.) in connection with OpenOCD but not that deep.

If someone of the developers here is able to resurrect this unit and is located in Germany I would ship it for free. You can keep the unit ofc. Just PM me.