RTS/CTS not fully disabled?

[ahuguet at cttc.es]

Thanks for your reply, Jouni

> Which STA firmware version are the cards using?

They're using 1.4.9 version.

> Is it different between
> the STAs that use RTS/CTS and that don't?

No.

> If I remember correctly, there
> were some changes to rate control algorithm in the firmware and it may
> have involved use of RTS/CTS in some cases. If the firmware versions are
> different, I would assume this would be reason for the behavior you are
> seeing.

>From what I've been able to read in the mailing list, it indeed seems to
point to a rate control issue.
It seems that, if the card fails for the 3rd time (IIRC) to send a packet
(that is, doesn't receive an ACK for said packet) it tries to adapt its
transmission rate, to the value immediately lower.

However, in my experiment this shouldn't be the expected card behaviour.
IMHO it should, even if failed 3 times:
Not call for rate control mechanism.
Keep resending the packet, until the retry limit (set with iwconfig) is
surpassed.
Keep the same transfer rate (if the user set it that way through iwconfig,
without the "auto" option)

Currently, it seems as if the card enters a loop, having found a "non
expected" situation.

I tried performing the test with 6 stations instead of 10 trying to send
the packet, and it worked then flawlessly.
That somehow shows that the RTS/CTS attempts are related to failed
transmissions (with 6 nodes the collision probability is lower, so that a
card fails to deliver three times the same message is quite unexpected
event)

I also tried, back to the 10 stations case, not setting the rate fixed,
but using the "auto" option, but the RTS/CTS kept appearing anyway, even
if the RTS threshold kept at off at all times.

Right now, then, it seems I've only 2 chances left:

1) Download a new STA firmware version for the cards.
http://linux.junsun.net/intersil-prism/ shows 1.4.9 as not very recent
firmware, and maybe trying 1.7.4, which is somehow "recommended" might
solve this "bug". However, it seems nowhere is specified what each
firmware update adds in terms of functionality or problem solving.
In that sense, are hostap-utils still meant to be used?
They appear in the website as "older stable release" thus I'm not entirely
sure weather or not is still recommended to use those tools, or if users
are prompted to use the prism2_srec to flash right away their cards?


2) Try to decrease the collision chances. To that end, the thing I thought
of, would be increasing the contention window. However, I've not found any
setting in iwconfig, nor iwpriv, as to allow me to enlarge the maximum
number that a station could take as back-off.
Does anyone know how to set a larger contention window, than the one the
cards are using by default?

Thanks in advance.