On Mon, 2023-08-21 at 18:36 +0800, Wen Gong wrote:
> On 8/21/2023 5:06 PM, Johannes Berg wrote:
> > On Mon, 2023-08-21 at 10:59 +0200, Johannes Berg wrote:
> > > From: Johannes Berg <johannes.berg@xxxxxxxxx>
> > >
> > > This reverts commit cb751b7a57e5 ("mac80211: add parse regulatory info
> > > in 6 GHz operation information") which added a station type bss_conf
> > > assignment in a parsing helper function, which will corrupt mesh data.
> > >
> > Ah crap this won't work, rtw89 already uses this.
> >
> > Wen please send a fix for this ASAP.
> >
> > johannes
>
> Hi Johannes,
>
> I looked the patch some times, but I do not know how it corrupt mesh data,
>
> Is there any clue for me?
Hah, no, I'm wrong ... I looked at it and for some reason thought of
u.mgd instead of vif.bss_conf. Sorry!
Still it's not correct though to write to vif.bss_conf in this function
because it's called from mesh_matches_local() to see if it's even
compatible, for example, so the mere calling a "give me the 6 GHz
chandef" function doesn't indicate we actually are going to use it now.
Also it's wrong for multi-link anyway, so maybe it should have an
optional pointer to the bss_conf to fill it in, or just do it outside of
the function.
johannes
