[lvc-project] [PATCH] wifi: mac80211: don't ask rate control with zero rate mask if scanning

[Dmitry Antipov]

On 1/18/24 16:46, Johannes Berg wrote:

> On Wed, 2024-01-17 at 15:48 +0300, Dmitry Antipov wrote:
>> If we're scanning and got the control frame with zero rate mask, drop
>> the frame before '__rate_control_send_low()' getting stuck attempting
>> to select supported rate.
> 
> But why drop the frame? I'm still thinking that it just doesn't really
> make sense to apply the rate mask to scanning at all?

Hm. It seems that I'm still missing something important, but I don't
realize why ieee80211_scan_state_set_channel() advances to the (next)
channel even after ieee80211_set_bitrate_mask() resets this channel's
rate mask to 0. Note that the comment in ieee80211_set_bitrate_mask()
explicitly states that there should be at least one usable rate for
the band we're currently operating on. Why this is not applicable to
other band(s) we might probe next?

> The most common use case for this is probably P2P-style things where you
> just don't want to use CCK, but for scanning we have
> NL80211_ATTR_TX_NO_CCK_RATE for this, so there's really no need to apply
> the rate mask?

Does NL80211_ATTR_TX_NO_CCK_RATE makes an effect on bands other than
2.4GHz? Note original reproducer triggers WARN_ONCE() after switching
to 5GHz.

Dmitry