lwfinger/rtw89

Last update broke wifi

BUGOVER opened this issue · 16 comments

BUGOVER commented

After last update wifi is not working on ubuntu 22.04 and kernel "6.2.0-39"

OakAtsume commented

I've also been experiencing issues with the latest update of rtw89;
with issues ranging from:

  • Unable Connection
  • Reports of Beacon loss from wpa_supplicant every second
Jan 05 13:17:53 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:17:54 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:17:55 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:17:56 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:17:57 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:17:58 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:17:59 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:18:00 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:18:01 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:18:02 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:18:03 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS
Jan 05 13:18:04 [Redacted] wpa_supplicant[828]: wlan0: CTRL-EVENT-BEACON-LOSS

to simply an un-usable connection.

Running EndeavourOS 6.6.9-arch1-1

Allow me to add some more context;
I've attempted some fixed suggested in past issues even disabling 'powersaving' and such but no luck.

[oak@[Redacted] ~]$ cat /etc/modprobe.d/50-rtw89.conf 
options rtw89_core disable_ps_mode=y
options rtw89_core disable_clkreq=y
options rtw89_core disable_aspm_l1=y
options rtw89_core disable_aspm_l1ss=y
lwfinger commented

Have you tried bisecting the set of updates? It works perfectly on my openSUSE Tumbleweed running kenel 6.7-rc8.

OakAtsume commented

I personally haven't (a bit unsure on what you mean by "bisecting"); but I've noticed that this issue is happening to other's as well; aside from this issue.
Where the connection is established but the actual networking is unusable (for my case due to beacons being lost?)

alou-S commented

@lwfinger Problem is caused by commit f247fef at

rtw89/mac80211.c

Lines 1150 to 1155 in 139b147

#if RHEL_RELEASE_CODE > RHEL_RELEASE_VERSION(10, 0)
.bss_info_changed = rtw89_ops_bss_info_changed,
.start_ap = rtw89_ops_start_ap,
.stop_ap = rtw89_ops_stop_ap,
.conf_tx = rtw89_ops_conf_tx,
#endif

rtw89/mac80211.c

Lines 1174 to 1177 in 139b147

#if RHEL_RELEASE_CODE >= RHEL_RELEASE_VERSION(10, 0)
.assign_vif_chanctx = rtw89_ops_assign_vif_chanctx,
.unassign_vif_chanctx = rtw89_ops_unassign_vif_chanctx,
#endif

Where the author added the #if preprocessor which reserves the options under it exclusively to Red Hat Linux builds.

Also I'm pretty sure the commit author meant RHEL_RELEASE_CODE >= RHEL_RELEASE_VERSION(10, 0) and not RHEL_RELEASE_CODE > RHEL_RELEASE_VERSION(10, 0) on line 1150

  • 👍1
  • ❤️1
lwfinger commented

@alou-S - Does that change fix your system?

@OakAtsume - Bisection is a feature of git. If you can identify a commit that works (good) and one that fails (bad), then bisect will find one halfway between good and bad. You then build and test that revision, and report if this is the new good, or bad. Repeat until you find the bad one. See https://stackoverflow.com/questions/4713088/how-do-i-use-git-bisect#4714297 for another explanation.

lwfinger commented

I just fixed some more errors in the RHEL9 patches. Please pull and try again.

  • 👍1
OakAtsume commented

Hello I wanted to confirm that the given patch from @alou-S worked for me.

BUGOVER commented

It works, thanks, I think we can close this issue

weaki233 commented

@lwfinger ,Hi~ I’ve encountered a problem with the rtw89 and rtw_8852ce drivers following a kernel update. The drivers functioned well with kernel version 5.15.0.97, but they no longer work even after I followed the update steps in the README for kernel version 5.15.0.100. Has there been an adaptation for the newer kernel, or is it possible I’ve overlooked something that remain to be done to solve it?

lwfinger commented

I need to see some error messages, or the full output of dmesg to help you. Do the command 'sudo dmesg > dmesg.txt' and attach dmesg.txt to this issue. I build the drivers with 5.15.0 - I have no idea what was done with your system.

weaki233 commented

@lwfinger Thank you for your prompt reply and assistance, here is the .txt file attached.
dmesg.txt

It seems like the NVIDIA taints the kernel but I don't know how to solve it.
The following are the lines that I think may cause the problem or you can see the file in detail.
'''
[ 2.025612] rtw89_8852ce 0000:08:00.0: failed to register hw
[ 2.025625] rtw89_8852ce 0000:08:00.0: failed to register core hw
[ 2.025634] rtw89_8852ce 0000:08:00.0: failed to register core
[ 2.031723] nvidia: module license 'NVIDIA' taints kernel.
[ 2.031725] Disabling lock debugging due to kernel taint
[ 2.050508] usb 3-10: new full-speed USB device number 5 using xhci_hcd
[ 2.079866] rtw89_8852ce: probe of 0000:08:00.0 failed with error -22
[ 2.081240] nvidia-nvlink: Nvlink Core is being initialized, major device number 509
'''

lwfinger commented

You had a problem uploading dmesg.txt. When I click on it, I get a 404 error.

weaki233 commented

You had a problem uploading dmesg.txt. When I click on it, I get a 404 error.

I am sorry about that but I don't know why either. What about this:

dmesg

lwfinger commented

That worked, but with a problem. The first warning shows WARNING: CPU: 10 PID: 476 at net/wireless/core.c:879 wiphy_register+0xa5a/0xa80 [cfg80211]. Before I try to chase this one down, have you done a ;git pull' recently?

Any reason you are stuck at Ubuntu 20.04? There are a lot of improvements in later versions.

weaki233 commented

Yes I have done it following this about 10 hours ago:

When your kernel changes, then you need to do the following:

cd ~/rtw89
git pull
make clean
make
sudo make install

Because I'm now collaborate with the team in my school and they told me to use this version to compatible with their current work.

parithy commented

becon loss issue seems to be present in 6.8.0-31-generic in ubuntu derivative..

2024-05-27T16:21:36.506588-05:00 ##masked## wpa_supplicant[928]: wlp3s0: CTRL-EVENT-BEACON-LOSS
2024-05-27T16:21:36.848136-05:00 ##masked## kernel: rtw89_8852be 0000:03:00.0: timed out to flush queues
2024-05-27T16:21:37.475842-05:00 ##masked## wpa_supplicant[928]: wlp3s0: CTRL-EVENT-DISCONNECTED bssid=#masked## reason=4 locally_generated=1
2024-05-27T16:21:37.481563-05:00 ##masked## wpa_supplicant[928]: wlp3s0: CTRL-EVENT-REGDOM-CHANGE init=CORE type=WORLD
2024-05-27T16:21:37.487342-05:00 ##masked## NetworkManager[927]: [1716844897.4869] device (wlp3s0): supplicant interface state: completed -> disconnected
2024-05-27T16:21:37.487571-05:00 ##masked## NetworkManager[927]: [1716844897.4870] device (p2p-dev-wlp3s0): supplicant management interface state: completed -> disconnected
2024-05-27T16:21:37.704355-05:00 ##masked## NetworkManager[927]: [1716844897.7041] device (wlp3s0): supplicant interface state: disconnected -> scanning
2024-05-27T16:21:37.704501-05:00 ##masked## NetworkManager[927]: [1716844897.7041] device (p2p-dev-wlp3s0): supplicant management interface state: disconnected -> scanning
2024-05-27T16:21:37.896144-05:00 ##masked## wpa_supplicant[928]: wlp3s0: SME: Trying to authenticate with ##masked## (SSID='##masked##' freq=5300 MHz)
2024-05-27T16:21:37.048469-05:00 ##masked## kernel: rtw89_8852be 0000:03:00.0: timed out to flush queues