WCN7850: DFS broken since 176f3009ae598d0523b2

[Jeff Johnson]

On 7/17/2025 12:05 PM, Mihai Moldovan wrote:
> Hi
> 
> 
> In 176f3009ae598d0523b267db319fe16f69577231[0], radar detect widths were removed 
> for single pdev devices, fully disabling DFS on WCN7850.
> 
> I do understand that running DFS on two channels simultaneously is (currently?) 
> not supported, but this effectively breaks operations on all DFS channels, even 
> though only one is being used.
> 
> 
> I've tried re-enabling the radar detect widths, but that only lead to an 
> ieee80211 subsystem error (which I haven't further debugged) and made the card 
> unusable because booting it up failed, which required a full power shutdown of 
> the machine to reset it from this broken state (literally disconnecting power 
> physically for a few moments).
> 
> Setting num_different_channels to 1 for single pdev devices and re-enabling the 
> radar detect widths works to re-enable DFS, but naturally that disables MLO.
> 
> I'm probably fine with not having MLO capabilities (at least currently), but 
> having to modify the source code to get DFS support back certainly isn't ideal.
> 
> Are there any plans to fix DFS support for single pdev devices?
> 
> 
> Mihai
> 
> [0] 
> https://patch.msgid.link/20250409-ath12k-wcn7850-mlo-support-v2-9-3801132ca2c3@quicinc.com

Can you please open a ticket in Bugzilla so this can be tracked?

https://wireless.docs.kernel.org/en/latest/en/users/drivers/ath12k/bugreport.html