On Tue, Mar 23, 2010 at 10:19:55AM -0700, Johannes Berg wrote: > On Tue, 2010-03-23 at 13:46 +0100, Stanislaw Gruszka wrote: > > > + * @qos_disabled: whether QoS (aka WMM) should be disabled or not > > Why "disabled" btw, and not enabled? Everything else is usually done in > a positive sense here, I'd say. To emphasize we disable QoS as conf_tx enable it implicitly. > Also, the default would now be enabled, and I think this needs more work > for AP mode. You're implementing generic API, don't fall into the trap > of thinking it's only for iwlwifi's special case. I did not think this is iwlwifi only case, but only for iwlwifi I have bug reports as far. > So at least there it > should be enabled if the hw supports it and the change flag should be > set at least once. It must be enabled conf_tx is called anyway, so I do not think that is needed. Ok, flag that device can handle bss_info_changed(BSS_CHANGEDS_QOS) but that all is overcomplicated for me. > Also, your code sends a weird signal to hw that doesn't support QoS, it > actually enables qos for it ... Not really. It's only used to disable QoS. Well, my first patch which add new disable_qos callback, was like that. If device do not have QoS, driver do not implement callback - wasn't that simple and clean design? In ideal case we want to change conf_tx to configure all queues or disable qos. But since this is exported to userspace additional disable_qos callback is the best implementation for me. Stanislaw -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html