Re: Problems connecting using ahteros based wireless card



On Thu, 2005-01-27 at 21:28 +0100, Stefan Zechmeister wrote:
> NetworkManager: Activation (ath0) started...
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): access point 'ESSID' is
> encrypted, and a key exists.  No new key needed. NetworkManager:
> Activation (ath0/wireless): using essid 'ESSID', with Open System
> authentication. NetworkManager: Activation (ath0/wireless): no hardware
> link to 'ESSID' in Open System mode, trying Shared Key. NetworkManager:
> Activation (ath0/wireless): using essid 'ESSID', with Shared Key
> authentication. NetworkManager: Activation (ath0/wireless): no hardware
> link to 'ESSID' in Shared Key mode, trying another access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Activation (ath0/wireless): waiting for an access point.
> NetworkManager: Caught SIGINT/SIGTERM NetworkManager: Caught terminiation
> signal NetworkManager: nm_device_activation_cancel(ath0): cancelling...
> NetworkManager: nm_device_activation_worker(ath0): activation canceled.
> NetworkManager: nm_device_activation_worker(ath0): activation canceled.
> NetworkManager: Activation (ath0) ended. NetworkManager:
> nm_device_activation_cancel(ath0): cancelled.

Hi,

What you need to do to debug this is the following...  In a separate
window, run the command "watch -n 1 iwconfig ath0" and let that run
while NM is trying to associate with the AP.  Look at the MAC address
that the card reports, and if it changes to FF:FF:FF:FF:FF:FF too often
during the link check, then we need to figure out what to do.

What NM currently does for a link-check is to wait around 6 or 8
seconds, I think, and for every half-second during that time it wakes
up, checks the MAC address the card reports, and if its a valid MAC
address (ie not all 0s, all 4s, or all Fs) it increments a counter.  If
by the end of that full 6 or 8 seconds, the counter is above a certain
value, then we decided that we have a link.

This particular song & dance is necessary for a couple reasons.  Some
cards need to load firmware into the card whenever you set the ESSID on
the card, and that takes a round-trip to user-space hotplug scripts,
then back into the kernel to upload the firmware to the card.  Then, the
card has to boot the firmware, and try to associate.  This takes time.
Second, some cards have more channels that others, specifically a/b/g
cards.  They need more time to scan all channels for the access point
we're trying to find, and to associate with it.  Third, some cards are
just plain flaky and will take a while to stabilize the MAC address of
the associated AP (they scan around and evidentally pick up the MAC
addresses of other APs before they settle on the correct one, I think
the problem child here is Orinoco w/ HermesI chipset).

So, if you could do the "watch" thing, that would be great, and report
teh results, ie do you see the MAC address be FF:FF:FF... for most of
the period when NetworkManager tries to connect to it?  Dos it usually
stay at the access points MAC?

Second, what is the output of "lspci", and what brand/model is your
Atheros card?  I personally use a Netgear WG511T to test out the madwifi
drivers and haven't noticed any problems, but that's because the card is
b/g only most likely, and doesn't do 802.11a.

Thanks!
Dan




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]