RE: [PATCH] wifi: rtw89: usb: fix TX flow control by tracking in-flight URBs

[Mh_chen <mh_chen@realtek.com>]

+Isaiah,

-----Original Message-----
From: Lucid Duck <lucid_duck@justthetip.ca> 
Sent: Wednesday, January 28, 2026 4:01 AM
To: Bitterblue Smith <rtl8821cerfe2@gmail.com>
Cc: linux-wireless@vger.kernel.org; Ping-Ke Shih <pkshih@realtek.com>; Mh_chen <mh_chen@realtek.com>
Subject: Re: [PATCH] wifi: rtw89: usb: fix TX flow control by tracking in-flight URBs


External mail : This email originated from outside the organization. Do not reply, click links, or open attachments unless you recognize the sender and know the content is safe.



On Mon, 26 Jan 2026, Bitterblue Smith wrote:
> The CH12 checks in the completion handler are unnecessary since there 
> is one in the resource checking function.

You're right. I'll remove the redundant check in v2.

> Is there a reason to add a new counter instead of just using the 
> length of each tx_queue?

Good question. The tx_queue length tracks SKBs queued for submission, but doesn't account for URBs that have been submitted to USB core and are in-flight (between submit and completion callback). An SKB is dequeued before usb_submit_urb(), so skb_queue_len() would undercount actual resource usage.

The atomic counter tracks the full lifecycle: incremented at submit, decremented at completion. This gives mac80211 accurate backpressure even when URBs are pending in the USB subsystem.

Happy to hear if I'm missing something that would make skb_queue_len() work here.

--
Lucid Duck