Packet loss with QCA9377 (ath10k) on ASUS VivoBook E410KA

Packet loss with QCA9377 (ath10k) on ASUS VivoBook E410KA

[Bandhan Pramanik]

Hello,

I'm helping someone who was having trouble with their Wi-Fi on Asus 
VivoBook E410KA. The device uses the QCA9377 Wireless Chip.

The device suffered from a single packet loss within 33 seconds while 
running a continuous ping. There might be some PCIe-level communication 
problems too.

All the relevant logs can be found here in a redacted form:

First part (ping, hostnamectl, ip addr, lspci): 
https://gist.github.com/BandhanPramanik/7eefd192c3a728d2473687a786fbbc01/raw/e77f0fa1cf35c748730df220b0057399ac78f539/miscellaneous

Second part (cat /proc/interrupts, modinfo ath10k_pci): 
https://gist.github.com/BandhanPramanik/7eefd192c3a728d2473687a786fbbc01/raw/e77f0fa1cf35c748730df220b0057399ac78f539/miscellaneous%25202

dmesg: 
https://gist.github.com/BandhanPramanik/7eefd192c3a728d2473687a786fbbc01/raw/e77f0fa1cf35c748730df220b0057399ac78f539/dmesg

We’re not sure if this is a driver issue, a firmware problem, or maybe 
something hardware-specific.

Has anyone seen something similar with this chip? Any tips on how to 
troubleshoot or fix the packet loss?

Thanks,

Bandhan Pramanik

বন্ধন প্রামাণিক

Re: Packet loss with QCA9377 (ath10k) on ASUS VivoBook E410KA

[Bjorn Helgaas]

[+cc Jeff, ath10k maintainer, Mani]

On Fri, Aug 01, 2025 at 12:09:01AM +0530, Bandhan Pramanik wrote:
> Hello,
> 
> I'm helping someone who was having trouble with their Wi-Fi on Asus VivoBook
> E410KA. The device uses the QCA9377 Wireless Chip.
> 
> The device suffered from a single packet loss within 33 seconds while
> running a continuous ping. There might be some PCIe-level communication
> problems too.
> 
> All the relevant logs can be found here in a redacted form:
> 
> First part (ping, hostnamectl, ip addr, lspci): https://gist.github.com/BandhanPramanik/7eefd192c3a728d2473687a786fbbc01/raw/e77f0fa1cf35c748730df220b0057399ac78f539/miscellaneous
> 
> Second part (cat /proc/interrupts, modinfo ath10k_pci): https://gist.github.com/BandhanPramanik/7eefd192c3a728d2473687a786fbbc01/raw/e77f0fa1cf35c748730df220b0057399ac78f539/miscellaneous%25202
> 
> dmesg: https://gist.github.com/BandhanPramanik/7eefd192c3a728d2473687a786fbbc01/raw/e77f0fa1cf35c748730df220b0057399ac78f539/dmesg
> 
> We’re not sure if this is a driver issue, a firmware problem, or maybe
> something hardware-specific.
> 
> Has anyone seen something similar with this chip? Any tips on how to
> troubleshoot or fix the packet loss?

ath10k mucks directly with ASPM, which it definitely shouldn't do
because ASPM configuration is extremely sensitive, it involves both
ends of the link, and the PCI core needs to know about it.  So I'm a
little suspicious of some issue there.

Mani recently posted a patch series to clean that up a bit, but it's
not fully cooked yet, and I don't know whether it will help this
situation:
https://lore.kernel.org/r/20250716-ath-aspm-fix-v1-0-dd3e62c1b692@oss.qualcomm.com

Bjorn

Re: Packet loss with QCA9377 (ath10k) on ASUS VivoBook E410KA

[Bandhan Pramanik]

I'll tell the guy to clone the ath-next branch then and will inform you pretty soon.

Bandhan

Re: Packet loss with QCA9377 (ath10k) on ASUS VivoBook E410KA

[Manivannan Sadhasivam]

On Thu, Aug 07, 2025 at 11:33:08AM GMT, Bandhan Pramanik wrote:
> I'll tell the guy to clone the ath-next branch then and will inform you pretty soon.

The patches are not in ath-next. You need to manually pick them up from mailing
list.

But even if it doesn't help, try to disable ASPM using cmdline by passing
'pcie_aspm=off'.

- Mani

-- 
மணிவண்ணன் சதாசிவம்

Re: Packet loss with QCA9377 (ath10k) on ASUS VivoBook E410KA

[Bandhan Pramanik]

Thanks, I will update you guys as soon as I get the logs.

Re: Packet loss with QCA9377 (ath10k) on ASUS VivoBook E410KA

[Bandhan Pramanik]

Sorry for the delay.

He executed it and found that packet losses are still prevalent:

1943 packets transmitted, 1942 received, 0.0514668% packet loss, time 
1944829ms
rtt min/avg/max/mdev = 51.186/244.746/7607.252/320.060 ms, pipe 8

----

On a different note, I just saw that we have gone to the v2 of the 
patch. In the meantime, we were actually struggling compile the kernel 
properly using Github Actions or Fedora Copr, that's why it was taking 
so much time.

Bandhan Pramanik

বন্ধন প্রামাণিক