RE: RTL8922AE driver issues

[Ping-Ke Shih <pkshih@realtek.com>]

Samuel Reyes <zohrlaffz@gmail.com> wrote:
> On Wed, May 7, 2025 at 7:23 PM Ping-Ke Shih <pkshih@realtek.com> wrote:
> >
> > Samuel Reyes <zohrlaffz@gmail.com> wrote:
> > > On Mon, May 5, 2025 at 8:11 PM Ping-Ke Shih <pkshih@realtek.com> wrote:
> > > >
> > > > Samuel Reyes <zohrlaffz@gmail.com> wrote:
> > > > > Big update, I updated my motherboard to the newest BIOS (vF4) and it
> > > > > has helped dramatically with the amount of errors in dmesg. But the
> > > > > issue with the connection dropping after some time remains. I added
> > > > > amd_iommu=off to my kernel parameters and ran sudo iw wlan0 set
> > > > > power_save off. I did NOT apply the kernel patch you included yet. I
> > > > > ran dmesg when my connection dropped and saw a bunch of 'failed to
> > > > > update XX RXBD info: -11' errors again.
> > > >
> > > > I don't have clear idea now. Please try my patch first.
> > > > I will check internally and will need your help to collect more data.
> > > >
> > > > >
> > > > > I managed to capture the connection drop in a recording so you can see
> > > > > what it looks like on my end. My connection will remain 'broken'
> > > > > indefinitely until I toggle airplane mode.
> > > > >
> > > > > Video: https://youtu.be/BbVlTU8K9Hg
> > > > >
> > > >
> > > > Can you open a terminal to show kernel log next time? I would like to
> > > > know disconnection happens along 'failed to update XX RXBD info: -11'.
> > > >
> > > > By the way, please not top-posting in wireless mailing.
> > > >
> > >
> > > Hi!
> > >
> > > Thanks again for your assistance.
> > >
> > > I spent all day yesterday and this morning trying to figure out how to
> > > apply your patch and wasn't able to do it. I'm by no means an expert
> > > and quickly found that patching a kernel is no easy feat.
> >
> > What are the problems you met? Below is the rough steps to build driver:
> > 1. grab kernel source
> >    git clone git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
> > 2. switch to tag you are using
> >    eg. git checkout v6.14.2
> > 3. install toolchain/kernel header by 'apt install'
> >    sudo apt install build-essential
> >    (I don't remember how to install kernel header. Please google it yourself.)
> > 4. make driver
> >    e.g. make -C /lib/modules/6.14.2-061402-generic/build
> M=/$(KERNEL_SOURCE_PATH)/drivers/net/wireless/realtek/rtw89/
> > 5. rmmod/insmod
> >    ```lsmod | grep rtw89``` can see all rtw89 modules
> >
> >
> > > I wanted to
> > > record a before and after video (before your patch and after) of the
> > > download, but since I couldn't figure out how to patch my kernel I
> > > only have a before video. It has the terminal running so you can see
> > > the errors pop up in real time. It's pretty long since I recorded it
> > > from the very start of the download. The crash happens at around 11:30
> > > into the video: https://youtu.be/2nrLOuY6Pwk
> >
> > I guess the cause is that RXD can't be recognized and then
> > "failed to release TX skbs" can't free TX WD properly, so run out of
> > TX resource. Therefore, let's try the patch to see if we can resolve
> > 'failed to update XX RXBD info: -11' problem first.
> >
> > >
> > > P.S. I hope this is what you meant about top-posting, I also had to
> > > research what that meant and how to avoid it :P
> >
> > Yes, it is. :)
> >
> 
> The obstacle was unloading the module which was in use to load in the
> patched one. I did it though, I had to boot from an Arch live
> environment on a USB.

That will be easier after this success. :)

> 
> Anyways the patch works!! dmesg is clear of those pesky RXBD messages
> and my download completed without failure. This is the first time I've
> been able to make a large download like that without the connection
> dropping :)

Good to hear the good news. I will make a patch for it. 

> 
> You'll see the download speed dip a few times (e.g. around 4:45
> minutes in the video) over the course of the download, but it always
> recovers successfully. In the past those dips were a sign of the
> connection dropping.
> 

For me, these dips seems to be acceptable. If you want to dig further,
I should check PHY rate first via debugfs:

   cd /sys/kernel/debug/ieee80211/phy0/rtw89
   watch cat phy_info

And, capture video as before. 

I admit this is only first step to see if the symptom is caused by
PHY rate. If yes, I need air sniffer to know how it happens. If not,
maybe system enters power save mode or others.