On Tue, Mar 24, 2026 at 10:03:20PM +0100, Kirill A. Korinsky wrote:
> I think it is strong proof that 160Mhz actually works as expected

This is nice overall. I am happy the Tx side is working well now.

Can we please try to reduce the amount of unrelated changes in the diff,
even if they happen to increase performance in your tests?

Specifically, enabling support for LDPC, STBC, beam-forming, and announcing
support for larger aggregates in VHT-caps are all unrelated changes.

I don't know exactly what the consequences of enabling these features are
and unfortunately I lack the time to dig into them now. But I'm quite sure
more changes will be required than simply flipping feature bits in VHT caps.
These features will trigger new code paths in firmware which the driver
may not yet be prepared to handle, potentially resulting in firmware
crashes or other instabilities in some setups.
For STBC we will likely need to configure related parameters in the rate
selection firmware command, for example. Beamforming might require us
to handle sounding frames perhaps. Not sure about the others, they will
also require some consideration.

I would prefer to enable all of these features one-by-one in isolation
to make it easier to review the changes and spot any potential regressions.