Bug 214455 - Kernel 5.14.5 ath11k/mhi regression: Wifi stops working
Summary: Kernel 5.14.5 ath11k/mhi regression: Wifi stops working
Status: RESOLVED CODE_FIX
Alias: None
Product: Drivers
Classification: Unclassified
Component: network-wireless (show other bugs)
Hardware: All Linux
: P1 blocking
Assignee: drivers_network-wireless@kernel-bugs.osdl.org
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2021-09-17 12:33 UTC by dominik.foerderer
Modified: 2021-10-01 08:39 UTC (History)
10 users (show)

See Also:
Kernel Version: 5.14.5, 5.13.18
Tree: Mainline
Regression: Yes


Attachments

Description dominik.foerderer 2021-09-17 12:33:38 UTC
Previously Known Working Version: 5.14.2
Hardware: Dell XPS 13 9310, Qualcomm Atheros QCA6390 (Killer AX500) Wi-Fi
Bios Version: 3.1.0 (latest as of today)
Bios Settings: Virtualization and VT-d are enabled

After updating to 5.14.5 wifi stops working on Dell XPS 9310 with QCA6390. There are two errors in dmesg:

1. mhi:

[   26.118208] qcom_mhi_qrtr mhi0_IPCR: invalid ipcrouter packet

2. ath11k

[   31.309383] ath11k_pci 0000:72:00.0: failed to register fw indication: -110
[   31.309387] ath11k_pci 0000:72:00.0: failed to send qmi firmware indication: -110

The ath11k fail is probably caused by the mhi/qrtr error before.
Comment 1 dominik.foerderer 2021-09-17 13:22:25 UTC
Starting my system with Kernel 5.13.18 shows, that this one is also affected with the same error message. The last known working 5.13 Kernel was 5.13.15 i guess.
Comment 2 dominik.foerderer 2021-09-17 14:43:52 UTC
I found out that the regression is caused by commit 	

3f38cc1ba1e289306f52cb7238406f38ecbfa21a net: qrtr: make checks in qrtr_endpoint_post() stricter

After reverting this commit in 5.14.5 or 5.13.18 everything is fine and Wifi is working.
Comment 3 Jan 2021-09-18 13:00:25 UTC
I ran into the same problem on my Dell XPS 9500, which also has the Qualcomm Atheros QCA6390 (Killer AX500) Wi-Fi card installed.

I can confirm that reverting the commit 3f38cc1ba1e289306f52cb7238406f38ecbfa21a fixes the error and makes the ath11k driver work again.
Comment 4 Artem S. Tashkinov 2021-09-19 11:55:13 UTC
Dan Carpenter,

it's your patch, could you check it please?
Comment 6 Ivo 2021-09-20 01:33:48 UTC
Don't want to recompile the kernel. 

Is it going to be fixed in the next release?
5.14.6 is still not good.

Cheers!
Ivo
Comment 7 Kalle Valo 2021-09-21 11:54:33 UTC
Linus' tree is ok, this is a problem with stable kernels. Someone please report this to stable tree maintainers so that they include the fix to stable trees.

Closing the bug report as upstream is ok.
Comment 8 Philip Müller 2021-09-21 13:57:42 UTC
Will be fixed in the upcoming 5.14.7 release. 5.13 series won't get that fix. Distros might need to backport if they continue that EOL marked kernel series:

https://git.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git/tree/queue-5.14/net-qrtr-revert-check-in-qrtr_endpoint_post.patch
Comment 9 Yichao Yu 2021-09-21 19:54:56 UTC
Not sure if this is the expected behavior of the issue but I've also observed that the card is missing from the PCIe bus on the next boot after this happens.
Resetting and removing the bridge connecting to the network card seems to work for me to make the device appear again. (On XPS 15 9500, the network card is on 6c:00.0 and the bridge to it is at 00.1c.5).
Comment 10 Ivo 2021-09-25 11:20:42 UTC
(In reply to Philip Müller from comment #8)
> Will be fixed in the upcoming 5.14.7 release. 5.13 series won't get that
> fix. Distros might need to backport if they continue that EOL marked kernel
> series:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git/tree/
> queue-5.14/net-qrtr-revert-check-in-qrtr_endpoint_post.patch

Thanks. 
I don't think this same issue exists in 5.13 series.
All trees still need to reserve memory range explicitly in the kern param.
Comment 11 TJ 2021-09-29 08:16:20 UTC
I stumbled across this topic when my AX500 wifi card on the dell XPS13 9310 stopped working after an update on OpenSUSE tumbleweed. Similar to the above comments, the card no longer appeared in lspci output and revering to the older kernel did not work immediately. I had to disable the chip in bios, reboot, and re-enable before the card would be recognized by the system again.
Comment 12 Kalle Valo 2021-10-01 08:39:06 UTC
(In reply to Yichao Yu from comment #9)
> Not sure if this is the expected behavior of the issue but I've also
> observed that the card is missing from the PCIe bus on the next boot after
> this happens.
> Resetting and removing the bridge connecting to the network card seems to
> work for me to make the device appear again. (On XPS 15 9500, the network
> card is on 6c:00.0 and the bridge to it is at 00.1c.5).

One issue per bug report, please. Otherwise bugzilla becomes a nightmare to maintain. Please file a new bug:

https://wireless.wiki.kernel.org/en/users/drivers/ath11k/bugreport

Note You need to log in before you can comment on or make changes to this bug.