Subject: Re: [PATCH 7/9] mac80211: add flags for STBC (Space-Time Block
 Coding)
From: Johannes Berg <johannes@sipsolutions.net>
To: Felix Fietkau <nbd@openwrt.org>
Cc: linux-wireless@vger.kernel.org, linville@tuxdriver.com,
	lrodriguez@atheros.com
In-Reply-To: <4BCC2D1E.4000408@openwrt.org>
References: <1271602602-8538-1-git-send-email-nbd@openwrt.org>
	 <1271602602-8538-2-git-send-email-nbd@openwrt.org>
	 <1271602602-8538-3-git-send-email-nbd@openwrt.org>
	 <1271602602-8538-4-git-send-email-nbd@openwrt.org>
	 <1271602602-8538-5-git-send-email-nbd@openwrt.org>
	 <1271602602-8538-6-git-send-email-nbd@openwrt.org>
	 <1271602602-8538-7-git-send-email-nbd@openwrt.org>
	 <1271606009.3873.0.camel@jlt3.sipsolutions.net>
	 <4BCB2DB3.3020700@openwrt.org>
	 <1271656677.3873.5.camel@jlt3.sipsolutions.net>
	 <4BCC1844.4030607@openwrt.org>
	 <1271667582.3873.11.camel@jlt3.sipsolutions.net>
	 <4BCC1D1A.7020705@openwrt.org>
	 <1271668150.3873.17.camel@jlt3.sipsolutions.net>
	 <4BCC2D1E.4000408@openwrt.org>
Content-Type: text/plain; charset="UTF-8"
Date: Mon, 19 Apr 2010 12:19:07 +0200
Message-ID: <1271672347.3873.18.camel@jlt3.sipsolutions.net>
Mime-Version: 1.0
Sender: linux-wireless-owner@vger.kernel.org

On Mon, 2010-04-19 at 12:14 +0200, Felix Fietkau wrote:
> On 2010-04-19 11:09 AM, Johannes Berg wrote:
> > On Mon, 2010-04-19 at 11:06 +0200, Felix Fietkau wrote:
> >> On 2010-04-19 10:59 AM, Johannes Berg wrote:
> >> > On Mon, 2010-04-19 at 10:45 +0200, Felix Fietkau wrote:
> >> >> On 2010-04-19 7:57 AM, Johannes Berg wrote:
> >> >> > On Sun, 2010-04-18 at 18:05 +0200, Felix Fietkau wrote:
> >> >> > 
> >> >> >> >> + * @IEEE80211_TX_CTL_STBC: tells the driver to use Space-Time Block Coding
> >> >> >> >> + *  (STBC) for this frame.
> >> >> >> >>   */
> >> >> >> >>  enum mac80211_tx_control_flags {
> >> >> >> >>  	IEEE80211_TX_CTL_REQ_TX_STATUS		= BIT(0),
> >> >> >> >> @@ -299,6 +301,7 @@ enum mac80211_tx_control_flags {
> >> >> >> >>  	IEEE80211_TX_INTFL_HAS_RADIOTAP		= BIT(20),
> >> >> >> >>  	IEEE80211_TX_INTFL_NL80211_FRAME_TX	= BIT(21),
> >> >> >> >>  	IEEE80211_TX_CTL_LDPC			= BIT(22),
> >> >> >> >> +	IEEE80211_TX_CTL_STBC			= BIT(23),
> >> >> >> > 
> >> >> >> > What if the # of streams is different? That doesn't look sufficient.
> >> >> > 
> >> >> >> Hm, you're right. I initially thought the combination of the MCS index
> >> >> >> and the STBC flag would be enough, but there are still some corner cases.
> >> >> > 
> >> >> > Hm actually I guess that should be sufficient? What corner case are you
> >> >> > thinking of?
> >> >> Support for multi-rate retry and STBC with more than one stream on one
> >> >> side, using rates from both MCS0-7 and MCS8-15 in the rate series.
> >> >> Rx STBC for only one stream on the other side.
> >> > 
> >> > So the flag should be per rate entry instead, no?
> > 
> >> Well, I think if we use two bits in the tx control flags, we don't need
> >> it to be per rate entry.
> > 
> > But then you can't probe stbc properly, can you?
> I'm not sure we even need to probe STBC. In all of the drivers that I've
> looked at, it's always enabled if the peer supports it.
> I'm not aware of any situation where it would make the reception worse,
> aside from hardware damage of course ;)

Ok ... I guess two bits then so you don't have to look up sta capability
when doing the hw programming?

johannes