> On Mon, 2023-03-06 at 18:50 +0100, Lorenzo Bianconi wrote:
> >
> > +struct ieee80211_ops *
> > +mt7921_get_mac80211_ops(struct device *dev, void *drv_data, u8 *fw_features)
> > +{
> > + struct ieee80211_ops *ops;
> > +
> > + ops = devm_kmemdup(dev, &mt7921_ops, sizeof(mt7921_ops), GFP_KERNEL);
>
> It's kind of nice to have static const ops so they can't be a target for
> overwriting and function pointer injection, maybe just declare two
> copies with a macro or so, with and without chanctx, and return the
> appropriate one? It won't even use more memory unless you never run a
> device w/o chanctx ops.
In the current status quo I think it is fine but it does not scale if in the
future we want to add some new fw features that require changing ieee80211_ops
pointer. Honestly I do not know if it is something we should consider nowadays.
What do you think?
Regards,
Lorenzo
>
> johannes
>
Attachment:
signature.asc
Description: PGP signature
