Re: [RFT] p54: implement MRR

[Johannes Berg <johannes@sipsolutions.net>]

[-- Attachment #1: Type: text/plain, Size: 898 bytes --]

On Sun, 2008-10-12 at 01:58 +0200, Johannes Berg wrote:

> +	/*
> +	 * We support at most 8 tries no matter which rate they're at,
> +	 * we cannot support max_rates * max_rate_tries as we set it
> +	 * here, but setting it correctly to 4/2 or so would limit us
> +	 * artificially if the RC algorithm wants just two rates, so
> +	 * let's say the max of rates and three tries each, and clip
> +	 * it to a total of 8.
> +	 */
> +	dev->max_rates = IEEE80211_TX_MAX_RATES;
> +	dev->max_rate_tries = 3;

Ideally, of course, we'd set max_rates to 8 and max_rate_tries to 1, but
we don't have enough space for that in the tx info (no 8 rates
possible). Hence, the next best thing that matches the HW well would be
4/2 and have the rate control algorithms program the same rate more than
once if it wants more than two tries. Same in mac80211. Thoughts? Is
that feasible?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]