TP-Link Archer C50 (EU) V6.0 - 5 GHz interface not present

[samba2]

Hi,

I bought an TP-Link Archer C50 (EU) V6.0 router. Since there is currently no dedicated firmware version (device seems to be very recent), I went with the instructions for v5 as described here: https://openwrt.org/toh/tp-link/archer-c50#v4_and_v5_instructions
I decided to manually build the firmware (dd if=...).

Flashing via TFTP succeeded. However, after logging in I am only able to see the 2.4 Ghz device.

I've added some debugging output below. Any idea if and how I can get the 5 Ghz interface running?

Many thanks,

Maik

Some debugging output:

# iwinfo 
wlan0     ESSID: unknown
          Access Point: 90:9A:4A:11:AA:70
          Mode: Client  Channel: unknown (unknown)
          Tx-Power: 22 dBm  Link Quality: unknown/70
          Signal: unknown  Noise: unknown
          Bit Rate: unknown
          Encryption: unknown
          Type: nl80211  HW Mode(s): 802.11bgn
          Hardware: unknown [Generic MAC80211]
          TX power offset: unknown
          Frequency offset: unknown
          Supports VAPs: yes  PHY name: phy0

lsmod:

...
mac80211              442336  5 ath10k_core,mt76x2e,mt76x02_lib,mt7603e,mt76
mt76                   30752  4 mt76x2e,mt76x2_common,mt76x02_lib,mt7603e
mt7603e                36896  0 
mt76x02_lib            38464  2 mt76x2e,mt76x2_common
mt76x2_common          10784  1 mt76x2e
mt76x2e                 9728  0 
...

dmesg:

...
[   11.059794] mt76_wmac 10300000.wmac: ASIC revision: 76280001
[   11.141189] random: crng init done
[   11.144651] random: 6 urandom warning(s) missed due to ratelimiting
[   12.099074] mt76_wmac 10300000.wmac: Firmware Version: 20151201
[   12.105095] mt76_wmac 10300000.wmac: Build Time: 20151201183641
[   12.146729] mt76_wmac 10300000.wmac: firmware init done
[   12.313224] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[   12.420442] PPP generic driver version 2.4.2
[   12.437762] NET: Registered protocol family 24
[   13.389965] kmodloader: done loading kernel modules from /etc/modules.d/*
[   18.657349] rt3050-esw 10110000.esw: link changed 0x00
...

[tp-linkuser]

I was waiting for someone to answer you.

I think that SoC is not same in v6 as in v4, v5.
MediaTek MT7628A (N same)
MediaTek MT7612E (AC not same)

[samba2]

Tnx for respondig!

As I am new to the OpenWrt game - what does "soc" mean? And can I do something to get 5 Ghz working? If e.g. building a custom OpenWrt snapshot would be an option please give me some hints. Generally I am not scared of build systems .

[tp-linkuser]

I don't know about the topic.
I can only help you with this.

rm -f /etc/config/wireless
wifi config

Source:
https://openwrt.org/docs/guide-user/network/wifi/basic#regenerate_configuration

[samba2]

I am wondering how I could detect the hardware which is actually build in. Or should I try all available kernel modules which support 5Ghz?

[tp-linkuser]

I think it is better to go back to the original firmware if you do not need something specific like sqm.

[samba2]

You mean ditching OpenWrt and return to the TP-Link provided firmware?

[Stefan1]

Well, you are basically doing frontrunner work here by building a firmware for a hw version that is not yet supported. Maybe you should seek help in the developer section of the forum. Depending on the built in hw they might tell you which modules / drivers to use.

[tchargui]

Hey
Would you please send the instruction you follow in order to flash openwork. i don't want to brick it and if you can provide a version that work on your router.

Thanks

[tp-linkuser]

it only works apparently with EU version. I have the ES version and I can't install it.

Install: https://openwrt.org/toh/tp-link/archer-c50#v4_and_v5_instructions

with

https://openwrt.org/toh/tp-link/archer-c50#specific_values_needed_for_tftp_on_v3_v4

[Marssl]

Hello, there!

I also recently got V6 of this Router and installed 19.07.7 on it by following the installation instructions you mentioned (during these steps for cutting boot loader etc from vendor fw I used tp-link fw for V6). Then I played around with several mt76-based drivers for no luck, so I went back to stock fw. What was really interesting: All the settings I had done in the stock firmware were there, nothing was erased! The OEM-WebGUI-password was set, the time synchronisation settings (using my primary router as ntp-server), everything I checked was exactly like before. As I usually don't play around much with vendor fw besides having a short glance over stock functionality I can't really tell, if this is a real difference to V4, as I never reverted my V4 back to stock.

Then I searched for TechData about V6 in all the well-known sources, but was not able to find out what 5-GHz-Radio is built into V6. Not even an FCC-ID (also searched for an probably imaginary A5 v6 in an act of pure guessing)... So I opened the router and made some photos. I found an MT7628AN + MT7613BEN. The latter ist different from V4/V5, so I checked for devices with this in the TOH and found some in SNAPSHOT and some hints in the forum that indicated, that mt7615e was the driver to use for this. So I decided to have another try with 21.02.0-rc3, driver mt7615e and Firmware mt7663-firmware-ap. With this combination the 5-GHz became accessible. I made an experimental image using imagebuilder with profile for V4 including luci.

Obvious issue with these images so far seems to be the WLAN-LEDs, they just won't stop blinking after WLAN is activated. Also I cannot tell, if DFS ist working for 5-GHz. And of course the device reports itself as V4, as I didn't create an extra profile for V6.

If you wanna try @samba2 @tchargui you're welcome!

THIS is a link to a DropBox-Folder with 4 Files in it:

• experimental-archer-c50-eu-v6-factory.bin - for use in TP-Link WebGUI, based on 21.02.0-rc3 for C50v4
• archer-c50-eu-v6-sysupgrade-back2stock.bin - for going back to TP-Link-FW via LUCI / SSH-Sysupgrade
• experimental-archer-c50-eu-v6-factory-for-tftp(name tp_recovery).bin - same as first, but for use with TFTP-method
• archer-c50-eu-v6-tftp-back2stock(name tp_recovery).bin - for going back to stock via TFTP

[Marssl]

Hi, again!

I managed to solder some pins to the serial header, located between the blue WAN port and the flash chip. I only connected the left 3 pins Rx,Tx and Gnd. I guess, the outer right should be 3v3, but I didn't connect any cable to it. I used 115200 - 8N1 settings. I posted the OEM Bootlog at PasteBin.

[Marssl]

@Marssl

Obvious issue with these images so far seems to be the WLAN-LEDs, they just won't stop blinking after WLAN is activated. Also I cannot tell, if DFS ist working for 5-GHz. And of course the device reports itself as V4, as I didn't create an extra profile for V6.

Hi again!
The WLAN-LEDs stop blinking after a while and get lit solidly. I guess, they might work correctly and the blinking is for some good reason which I just don't see... Maybe DFS-sensing or something like that So tending to call it "works as it should".
After some reading of bootlogs from V4 and my own V6 I see same flash partitions and started creating a hardware profile locally using the DTS-file of V4, as I expected GPIOs/LEDs and Flash to act just identically, which seems to be the case (besides V6 has just 5 LEDs, WPS-LED ist missing). I can change color of the WAN-LED between orange and green, WAN, LAN, Power, WLAN-5 and WLAN-2 seem to work as expected. WLAN- and RESET-Buttons work, too. I created this profile on top of OpenWRT 21.02.0-RC4 with updated mt76-driver. In some files (platform.sh, etc.) where there are case decisions made for networking, leds etc. I added the V6 to be handled the same ways as V4. The MAC Address of BR-LAN matches the one at the label of the bottom of the device. WLAN-2 an WAN also have this same MAC, only WLAN-5 is different. Is this expected? If not, what should / can I do about this? I must admit, that this kind of work is not one of my talents, but I'll do my very best.

My concerns regarding DFS on the 5GHz mainly foot on
this thread in the forum. Reading the syslog I noticed that after configuring the 5GHz it wouldn't start due to initialization failure related do DFS, while showing ENABLED state in LuCIs wireless section. WPS didn't work then either (using wpad-openssl). Restarting the wireless device couldn't resolve this, but rebooting the whole router did. So after configuring the AP it is a good idea to reboot the router. Indeed I tend to call this a MT7613-driver issue, not a "create-support-for-C50-V6"-issue... When the 5GHz ist up, it works quite well for my "home use", meaning: I did some speedtests, but no excessive stress tests with Iperf for several hours or something like that.

What can I do next to help creating Support for the V6? What has to be reviewed, tested, handed over to somebody with more knowledge and competence?

This thread in developer's part of the forum is closed, so I cannot provide these information there to possibly draw somebodys attention.

Best regards!

Edit:
Today I needed several reboots to get 5GHz up and running. Syslog looked like this:

Mon Aug  9 12:39:12 2021 daemon.notice hostapd: wlan1: ACS-COMPLETED freq=5660 channel=132
Mon Aug  9 12:39:12 2021 daemon.notice hostapd: wlan1: interface state ACS->HT_SCAN
Mon Aug  9 12:39:13 2021 daemon.notice hostapd: wlan1: interface state HT_SCAN->DFS
Mon Aug  9 12:39:13 2021 daemon.notice hostapd: wlan1: DFS-CAC-START freq=5660 chan=132 sec_chan=1, width=1, seg0=138, seg1=0, cac_time=60s
Mon Aug  9 12:39:13 2021 daemon.err hostapd: DFS start_dfs_cac() failed, -1
Mon Aug  9 12:39:13 2021 daemon.err hostapd: Interface initialization failed
Mon Aug  9 12:39:13 2021 daemon.notice hostapd: wlan1: interface state DFS->DISABLED
Mon Aug  9 12:39:13 2021 daemon.notice hostapd: wlan1: AP-DISABLED

Edit 2:
Seems like channel 149 is more likely to succeed than others. I might change to a manually chosen channel instead of AUTO...

Mon Aug  9 12:43:16 2021 daemon.notice hostapd: wlan1: ACS-COMPLETED freq=5745 channel=149
Mon Aug  9 12:43:16 2021 daemon.notice hostapd: wlan1: interface state ACS->HT_SCAN
Mon Aug  9 12:43:17 2021 daemon.notice hostapd: WPS: Converting push_button to virtual_push_button for WPS 2.0 compliance
Mon Aug  9 12:43:17 2021 kern.info kernel: [  303.910064] IPv6: ADDRCONF(NETDEV_CHANGE): wlan1: link becomes ready
Mon Aug  9 12:43:17 2021 kern.info kernel: [  303.916821] br-lan: port 2(wlan1) entered blocking state
Mon Aug  9 12:43:17 2021 kern.info kernel: [  303.922260] br-lan: port 2(wlan1) entered forwarding state
Mon Aug  9 12:43:17 2021 daemon.notice netifd: Network device 'wlan1' link is up
Mon Aug  9 12:43:17 2021 daemon.notice hostapd: wlan1: interface state HT_SCAN->ENABLED
Mon Aug  9 12:43:17 2021 daemon.notice hostapd: wlan1: AP-ENABLED

[TestingForSolutions8]

Been lurking the topic of c50 v6 support for a couple of months and can attest that the build from @Marssl is closest to an expected experience I've tried.

I can also second the fact that the 5ghz drops out for all available clients after a certain time, and a reboot of the system seeming to be the easiest fix.

The weird thing is that the same symptoms manifests in the 2.4ghz radio too

So essentially both radios cuts out at the same time, but I have yet to determine any specific culprit such as load on the system or the like.

[Marssl]

Thank you for commenting on my build @TestingForSolutions8 .

I just added 2 more files to the dropbox folder:

• owrt21.02.0rc4-experimental-c50-v6-sysupgrade.bin
• owrt21.02.0rc4-experimental-sqm-c50-v6-sysupgrade.bin

That is for upgrading to 21.02.0-RC4 which includes a MT7615e driver update. I think it improves the 5GHz quite a bit. You have to force update the firmware using this file as I created a V6 hardware profile for this, while the previous build thinks of itself as a modified V4. I also didn't include the alternative firmware for wlan-client-mode, as most users won't need it and it uses up almost 0.8MB of flash space. The second one includes SQM.

HERE is the link again, for simplicity

[TestingForSolutions8]

Thank you for providing a second test file @Marssl I have it updated now and once again deployed as the main router of the household to test in a real world scenario.

Will report back my findings after seeing an organic flow of users and devices having been through the network

I'm also going to throw in an intentional stress test on a mix of 2.4- and 5hz devices later when I have the time.

[hjsimoes]

Hi @Marssl .. i´m testing it also... Had some issues with vlan bridging with wireless in the last version.. but i´m new to version 21 so it could be me..

Will follow up later on.
Thanks

[TestingForSolutions8]

So I have an update after having tested for a few days. The router works splendidly with all the features one can expect at this point, however it does experience what I might call hard stop issues at times. These being the following:

1. Router has restarted itself after using the 5ghz radio in certain scenarios, with no discernable difference to the type of load. This has not yet occurred using the 2.4ghz radio, but I wouldn't rule it out as never being able to happen, however at this point it does seem more unlikely on the 2.4ghz radio.
2. Router has gone into a state total unresponsiveness. The LEDs where off, and wasn't reachable from either wifi or ethernet. Only solution being a restart by physically pushing the power button on the router. My guess in this case would be a kernel panic, but I can't back that up with hard data.
3. Both the wireless radios does at some point after an uptime period of about 2 days exhibit odd behaviour. Latency tolerant operations work fine when the behaviour occurs, however lower latency operations like sending input data over local network does not (This being for instance sending kb+m inputs). This issue isn't fixed by restarting the radios or disconnecting clients, but rather the only solution being a restart of the router itself.

[psql30]

Hi @Marssl any possibility of starting an official TP-Link Archer C50 V6 page like the ones setup for V5 here: https://openwrt.org/toh/tp-link/archer-c50. You could put details there to help everyone with the newer V6 version of this router.

[TestingForSolutions8]

@TestingForSolutions8

3. Both the wireless radios does at some point after an uptime period of about 2 days exhibit odd behaviour. Latency tolerant operations work fine when the behaviour occurs, however lower latency operations like sending input data over local network does not (This being for instance sending kb+m inputs). This issue isn't fixed by restarting the radios or disconnecting clients, but rather the only solution being a restart of the router itself.

I think I've found the clue to the symptoms for this specific issue with this snippet from my syslog:

Tue Aug  3 13:24:31 2021 kern.warn kernel: [244103.956537] sched: RT throttling activated

This always pops up right before the system starts halting in bad ways, and as far I understand it it's a feature of the linux kernel that is triggered as mechanism to protect against another fault in the system. Here is is little snippet on RT throttling from the linuxfoundation docs page:

Programming failures in real-time applications can cause the entire system to hang. Such a failure could act like a call of a while(true){} loop. When the real-time application has the highest possible priority and is scheduled with SCHED_FIFO policy, no other task can preempt it. This leads to the system blocking all other tasks and scheduling this loop with a CPU load of 100 percent. Real-time throttling is a mechanism to avoid such situations by limiting the execution time of real-time tasks per period.

Direct link https://wiki.linuxfoundation.org/realtime/documentation/technical_basics/sched_rt_throttling

However this only tells me that the kernel went into this fail-safe mode triggered by some other issue, but the logs at the time didn't have a anything peculiar before or after the snippet I posted. However I do have a more recent syslog and it's accompanying kernel log with the same RT throttle trigger.
Syslog:

 
Thu Aug 19 12:44:42 2021 daemon.notice hostapd: nl80211: nl80211_recv_beacons->nl_recvmsgs failed: -5
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.169264] mt76-tx phy1: page allocation failure: order:0, mode:0xa20(GFP_ATOMIC), nodemask=(null)
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.178523] CPU: 0 PID: 700 Comm: mt76-tx phy1 Not tainted 5.4.137 #0
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.185071] Stack : 80640000 805e1420 00000000 00000000 805e0518 82c09aec 83d6b8a0 8061ed63
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.193555]         80587874 000002bc 807733bc 00000000 02596182 00000001 82c09aa0 96391d5c
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.202036]         00000000 00000000 807a0000 00000000 30232037 000000d8 65746e69 2e352064
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.210527]         00000000 0000000c 00000000 0002b95b 00000000 8058d6f0 00000000 00000000
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.219008]         00000000 00000000 02596182 00000a20 00000000 802bbabc 00000000 80770000
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.227490]         ...
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.229965] Call Trace:
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.232474] [<80009c90>] show_stack+0x30/0x100
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.237006] [<801125f0>] warn_alloc+0xb8/0x130
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.241523] [<80112fe8>] __alloc_pages_nodemask+0x980/0xcc8
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.247194] [<801135ec>] page_frag_alloc+0x1a0/0x1cc
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.252242] [<80310628>] fe_poll+0x4d0/0x8e8
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.256596] [<8035b09c>] __napi_poll+0x3c/0x10c
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.261199] [<8035b2f8>] net_rx_action+0xfc/0x270
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.265989] [<804f9048>] __do_softirq+0x110/0x298
Thu Aug 19 12:44:51 2021 kern.warn kernel: [  257.270778] [<80026c18>] do_softirq.part.14+0x60/0x84
Thu Aug 19 12:44:51 2

kernel log:

[  163.139596] sched: RT throttling activated
[  257.169264] mt76-tx phy1: page allocation failure: order:0, mode:0xa20(GFP_ATOMIC), nodemask=(null)
[  257.178523] CPU: 0 PID: 700 Comm: mt76-tx phy1 Not tainted 5.4.137 #0
[  257.185071] Stack : 80640000 805e1420 00000000 00000000 805e0518 82c09aec 83d6b8a0 8061ed63
[  257.193555]         80587874 000002bc 807733bc 00000000 02596182 00000001 82c09aa0 96391d5c
[  257.202036]         00000000 00000000 807a0000 00000000 30232037 000000d8 65746e69 2e352064
[  257.210527]         00000000 0000000c 00000000 0002b95b 00000000 8058d6f0 00000000 00000000
[  257.219008]         00000000 00000000 02596182 00000a20 00000000 802bbabc 00000000 80770000
[  257.227490]         ...
[  257.229965] Call Trace:
[  257.232474] [<80009c90>] show_stack+0x30/0x100
[  257.237006] [<801125f0>] warn_alloc+0xb8/0x130
[  257.241523] [<80112fe8>] __alloc_pages_nodemask+0x980/0xcc8
[  257.247194] [<801135ec>] page_frag_alloc+0x1a0/0x1cc
[  257.252242] [<80310628>] fe_poll+0x4d0/0x8e8
[  257.256596] [<8035b09c>] __napi_poll+0x3c/0x10c
[  257.261199] [<8035b2f8>] net_rx_action+0xfc/0x270
[  257.265989] [<804f9048>] __do_softirq+0x110/0x298
[  257.270778] [<80026c18>] do_softirq.part.14+0x60/0x84
[  257.275908] [<80026cfc>] __local_bh_enable_ip+0xc0/0xc8
[  257.281363] [<83731d04>] ieee80211_next_txq+0x174/0x28c [mac80211]
[  257.287742] [<835165ac>] mt76_tx+0x268/0x56c [mt76]
[  257.292713] Mem-Info:
[  257.295038] active_anon:399 inactive_anon:8 isolated_anon:0
[  257.295038]  active_file:1712 inactive_file:1335 isolated_file:5
[  257.295038]  unevictable:0 dirty:0 writeback:0 unstable:0
[  257.295038]  slab_reclaimable:379 slab_unreclaimable:2263
[  257.295038]  mapped:915 shmem:23 pagetables:72 bounce:0
[  257.295038]  free:766 free_pcp:13 free_cma:0
[  257.327292] Node 0 active_anon:1596kB inactive_anon:32kB active_file:6848kB inactive_file:5340kB unevictable:0kB isolated(anon):0kB isolated(file):20kB mapped:3660kB dirty:0kB writeback:0kB shmem:92kB writeback_tmp:0kB unstable:0kB all_unreclaimable? no
[  257.350043] Normal free:3064kB min:16384kB low:18432kB high:20480kB active_anon:1596kB inactive_anon:32kB active_file:6848kB inactive_file:5340kB unevictable:0kB writepending:0kB present:65536kB managed:58176kB mlocked:0kB kernel_stack:392kB pagetables:288kB bounce:0kB free_pcp:52kB local_pcp:52kB free_cma:0kB
[  257.377873] lowmem_reserve[]: 0 0
[  257.381240] Normal: 88*4kB (UME) 39*8kB (UME) 16*16kB (UM) 17*32kB (UM) 5*64kB (U) 0*128kB 1*256kB (U) 0*512kB 1*1024kB (U) 0*2048kB 0*4096kB = 3064kB
[  257.394947] 3072 total pagecache pages
[  257.398751] 0 pages in swap cache
[  257.402106] Swap cache stats: add 0, delete 0, find 0/0
[  257.407404] Free swap  = 0kB
[  257.410329] Total swap = 0kB
[  257.413243] 16384 pages RAM
[  257.416081] 0 pages HighMem/MovableOnly
[  257.419970] 1840 pages reserved

From glancing through it it does seem like some sort of memory issue related to the wifi, but I'm not all that adept at gleaming issues from logs so take that with grain of salt.

This was all running the router as the the primary router/gateway/access point of my entire household using the build with sqm.

To round of this post I have between now and when the logs were taken reinstalled openwrt on the device with the non sqm build from @Marssl and now only running at as an access point to test the wireless radios in more of a vacuum and I haven't had any issues yet using it in this configuration.

The particulars of what I've experienced so far could of course be because of a bad install performed by me. Are you @Marssl and @hjsimoes experiencing something similar in your testing?