On Tue, 2010-03-02 at 22:25 +0100, [email protected] wrote:
> Do you inject from userspace or kernelspace?
> How do you know that they are sent at 1 Mbit? Do you sniff from another pc?
> My patch was tested only with kernelspace injection, feedback is welcome.
I think that part is wrong:
+ if (sband->bitrates[i].bitrate <=
+ *iterator.this_arg * 5) {
+ info->control.rates[0].idx = i;
+ break;
+ }
If a bitrate is less than the requested one, it would still match.
I think we should require an exact match. Requests to use an
unsupported rate should result in dropping the packet. That applies to
rate 0 as well, unless there is a special interpretation in the
standard. I don't see any: http://www.radiotap.net/defined-fields/Rate
I tried the part of you patch that handles IEEE80211_RADIOTAP_RATE, but
it still would not work. It turns out the rate control would override
the rate. I don't see an easy way to prevent it short of introducing
another flag.
I've made an alternative patch and I'll submit it to linux-wireless
shortly. It actually works for userspace injection.
--
Regards,
Pavel Roskin
_______________________________________________
ath9k-devel mailing list
[email protected]
https://lists.ath9k.org/mailman/listinfo/ath9k-devel
