ath12k: kernel BUG at net/core/skbuff.c:209!

[Alexander Tsoy]

В Вс, 11/05/2025 в 22:40 +0530, Vasanthakumar Thiagarajan пишет:
> 
> 
> On 5/11/2025 12:30 AM, Alexander Tsoy wrote:
> > Hello!
> > 
> > I recently got two QCNCM865 (wcn7850) cards and experience kernel
> > panics on different systems and different kernel versions. Card
> > info:
> > 
> > [   13.584230] ath12k_pci 0000:0b:00.0: BAR 0 [mem 0xfc200000-
> > 0xfc3fffff 64bit]: assigned
> > [   13.584252] ath12k_pci 0000:0b:00.0: enabling device (0000 ->
> > 0002)
> > [   13.584636] ath12k_pci 0000:0b:00.0: MSI vectors: 16
> > [   13.584644] ath12k_pci 0000:0b:00.0: Hardware name: wcn7850
> > hw2.0
> > [   14.210605] ath12k_pci 0000:0b:00.0: chip_id 0x2 chip_family 0x4
> > board_id 0xff soc_id 0x40170200
> > [   14.210613] ath12k_pci 0000:0b:00.0: fw_version 0x100301e1
> > fw_build_timestamp 2023-12-06 04:05 fw_build_id
> > QC_IMAGE_VERSION_STRING=WLAN.HMT.1.0.c5-00481-
> > QCAHMTSWPL_V1.0_V2.0_SILICONZ-3
> > 
> > 
> > Kernels I've tried so far: 6.12.24..6.12.28, 6.14.5, 6.15.0-rc5
> > 
> > Synthetic load like pumping traffic via netcat or iperf seems
> > unable to
> > trigger this bug, but something like downloading of Steam games or
> > torrents is triggering this bug very quickly.

I was wrong. iperf3 with multiple parallel streams and bidirectional
mode enabled triggered this panic in several seconds for me:

iperf3 -t 3600 -P 10 --bidir -c ...

> > 
> > Any ideas on what's going on here? I'm ready to test any patch.
> > 
> 
> Can you try with 
> https://git.kernel.org/pub/scm/linux/kernel/git/ath/ath.git/commit/?id=763216fe6c5df95d122c71ef34c342427c987820
>  
> ?

This patch seems to fix the problem. Thank you! I see a Fixes tag, so
hopefully this commit will eventually hit stable trees.

BTW, I also see IOMMU page faults under stress test (iperf3 command
above). This seems to cause device reset, but at least my systems
doesn't crash now.