Bug 205749

Summary: Wifi not working after kernel upgrade from 5.3.13 to 5.4
Product: Networking Reporter: Matt (caltubakna)
Component: WirelessAssignee: networking_wireless (networking_wireless)
Status: RESOLVED PATCH_ALREADY_AVAILABLE    
Severity: blocking CC: bugzilla.kernel.org, bugzilla, dec, dyegomb, dylandforbes, graf2242, holishing, kraskden.net, lilydjwg, linux-kernel-org, linuxwifi, scott, tsattler, tseewald, validatedev, victor.trac
Priority: P1    
Hardware: Intel   
OS: Linux   
Kernel Version: 5.4.1 Subsystem:
Regression: Yes Bisected commit-id:
Attachments: "journalctl --user" log
dmesg log on kernel 5.5.0-RC3
full dmesg 5.5.0rc5
lspci -vvnn

Description Matt 2019-12-03 11:32:11 UTC
Created attachment 286159 [details]
"journalctl --user" log

After upgrading my kernel (on ArchLinux) from 4.11 to 5.4, wpa_supplicant and iwd no longer works due to a problem with the iwlwifi kernel driver.
My linux-firmware version : 20191022.2b016af-3
Does anyone have a solution ? I'm not an expert in kernel debugging.
Comment 1 Matt 2019-12-03 11:37:51 UTC
I'm sorry, it's from 5.3.13 to 5.4.1
Comment 2 Carsten 2019-12-06 09:30:36 UTC
@Eliott this seems to be related: https://bugs.archlinux.org/task/64703
I'm having similar issues with my 9462 card, also not working after the upgrade to 5.4.1. I'll try to post a dmesg from the latest kernel and LTS today.
Comment 3 Carsten 2019-12-06 19:10:56 UTC
I actually have a 9560, it's just reported to be a 9462 by lspci.

iwlwifi:
[ 46.839888] iwlwifi 0000:00:14.3: Detected Intel(R) Wireless-AC 9560
160MHz, REV=0x354

I added dmesg and lspci logs in the arch linux bug tracker thread I linked above.
Comment 4 Thomas Sattler 2019-12-06 23:57:34 UTC
same here: 5.3.13 works but 5.4.[12] doesn't. dmesg (under 5.3.13) says I have

iwlwifi 0000:00:0c.0: Detected Intel(R) Dual Band Wireless AC 9462, REV=0x318
Comment 5 Mert Demir 2019-12-29 19:32:40 UTC
Same problem has happened on Linux 5.5.0-RC3 kernel and all Linux 5.4.x versions.

iwlwifi 0000:00:14.3: Detected Intel(R) Dual Band Wireless AC 9560, REV=0x318
Comment 6 Mert Demir 2019-12-29 19:33:38 UTC
Created attachment 286507 [details]
dmesg log on kernel 5.5.0-RC3
Comment 7 kraskden 2020-01-05 12:21:43 UTC
Same issue has happens after upgrade from 5.3.13.1-1 to 5.4.7

00:14.3 Network controller [0280]: Intel Corporation Cannon Point-LP CNVi [Wireless-AC] [8086:9df0] (rev 30)
Comment 8 kraskden 2020-01-05 12:26:24 UTC
I tried to watch devices via 'iw dev' and 'ip link', and one of twos state was happens: 
1) There are no wireless devices (only loopback at `ip link`)
2) Wireless device has been find, but interface is down and `ip link set wlp0s20f3 up` -> returns "rtnetlink answers input/output error"
Comment 9 dyegomb 2020-01-07 00:13:24 UTC
Maybe this could help:
https://bugs.archlinux.org/task/64703
Comment 10 Chris Murphy 2020-01-12 18:48:41 UTC
This comment
https://bugs.archlinux.org/task/64703#comment184511

Suggests bisect showing the regression is due to commit 06eb547c4ae4 ("iwlwifi: mvm: fix scan config command size"). Someone with the actual hardware should confirm/deny that.

And also, this bug (bug 205749) should have complete dmesg attached; as well as 'lscpi -vvnn' attached. Snipped reports are generally less useful.
Comment 11 Tom Seewald 2020-01-12 19:41:13 UTC
(In reply to Chris Murphy from comment #10)
> This comment
> https://bugs.archlinux.org/task/64703#comment184511
> 
> Suggests bisect showing the regression is due to commit 06eb547c4ae4
> ("iwlwifi: mvm: fix scan config command size"). Someone with the actual
> hardware should confirm/deny that.

It would seem that the Intel wireless developers are aware of this and have submitted a patch on December 30th [1][2].

[1] https://www.spinics.net/lists/linux-wireless/msg194167.html
[2] https://patchwork.kernel.org/patch/11313069/
Comment 12 Chris Murphy 2020-01-12 20:00:02 UTC
For whatever reason, it's not in 5.5rc5, nor in next, which would explain why it's not in 5.4.11.
Comment 13 dec 2020-01-12 20:35:12 UTC
I own effected hardware. I've added the output of "journalctl --no-hostname -k" and "lspci -vvnn" from kernel 5.5.0rc5.
Comment 14 dec 2020-01-12 20:36:05 UTC
Created attachment 286771 [details]
full dmesg 5.5.0rc5
Comment 15 dec 2020-01-12 20:36:31 UTC
Created attachment 286773 [details]
lspci -vvnn
Comment 16 Tom Seewald 2020-01-12 21:10:13 UTC
(In reply to Chris Murphy from comment #12)
> For whatever reason, it's not in 5.5rc5, nor in next, which would explain
> why it's not in 5.4.11.

Right, I believe the patch will need to be approved by the Linux wireless maintainer, Kalle Valo, who may be on vacation.
Comment 17 dec 2020-01-12 23:35:13 UTC
(In reply to dec from comment #13)
> I own effected hardware. I've added the output of "journalctl --no-hostname
> -k" and "lspci -vvnn" from kernel 5.5.0rc5.

I applied the patch https://patchwork.kernel.org/patch/11313069/ to 5.5.0rc5 and my AC 9642 wifi is working again.
Comment 18 Scott Brown 2020-01-27 19:22:25 UTC
Have had this device broken in every 5.4 kernel tried from 5.4.7 through 5.4.13. Have not tried any 5.5. Intel 9462 wifi adapter in an Intel NUC7PJYH. Is there a plan to get this issue assigned and the fix pulled in to mainline?
Comment 19 Tom Seewald 2020-01-27 20:40:13 UTC
(In reply to Scott Brown from comment #18)
> Have had this device broken in every 5.4 kernel tried from 5.4.7 through
> 5.4.13. Have not tried any 5.5. Intel 9462 wifi adapter in an Intel
> NUC7PJYH. Is there a plan to get this issue assigned and the fix pulled in
> to mainline?

The revert[1] landed in 5.5[2].

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=205608749e1ef394f513888091e613c5bfccbcca

[2] https://lkml.org/lkml/2020/1/25/264
Comment 20 Scott Brown 2020-01-29 17:26:30 UTC
Noticed fix has also now been backported to 5.4.16, per https://cdn.kernel.org/pub/linux/kernel/v5.x/ChangeLog-5.4.16
Comment 21 Dylan Forbes 2020-03-14 20:53:30 UTC
So can this ticket be closed out?
Comment 22 Dylan Forbes 2020-03-15 18:52:31 UTC
Also are the Intel devs still active? I hate to bump this thread but I filed a bug the other week and haven't gotten a response