Re: [Madwifi-users] trouble with acktimeout?
Status: Beta
Brought to you by:
otaku
From: Thorsten v. E. <tv...@vo...> - 2004-11-02 18:39:32
|
Greg, as always thank you for your insightful responses! I have a "that can't be true" reaction to one your responses however :-), what am I missing? At 10:49 AM 11/1/2004 -0800, Greg Chesson wrote: >The no-ack flag currently in the driver ONLY affects the transmit side. >It instructs the hw to not expect an ack after the current frame. >However, it does not also instruct the receiver abort the ack generation. >For that, one needs to either globally suppress all ack generation at the >receive end (bad idea) or send qos Data frames. These frames have >an extra two bytes in the header which include an Ack Policy bit. >If set the Ack bit will suppress ack generation at the receiver >for that packet. The qos control field as it is called, is created during >packet encapsulation in the net80211 module and is totally under >sw control - no hal stuff or hw knowledge needed. I understand that the net80211 layer can add the required two bytes without HW/HAL changes. But it's the hardware or HAL on the receiving end that generates the ACKs. Are you saying the HW or HAL already decode these two bytes appropriately? Else changes are necessary there... What is the rough ETA for the driver/hal update to include the QoS stuff? I.e. are we talking days, weeks, or months? Thanks much! Thorsten |