5.99.11 wpa_supplicant & ath

[Martin S. Weber]

Hi,

first of all, I'm aware of the ral thread.

I've just updated to 5.99.11 from 5.99.3, and my wpa_supplicant stopped working. The

status

20:28:12.315: CTRL-EVENT-SCAN-RESULTS
20:28:12.315: wpa_state=SCANNING

20:28:20.401: CTRL-EVENT-SCAN-RESULTS

20:28:20.401: Trying to associate with <AP-MAC> (SSID='<AP-SSID>' freq=2412 MHz)
20:28:20.401: Associated with <AP-MAC>
20:28:20.401: WPA: Key negotiation completed with <AP-MAC> [PTK=TKIP GTK=TKIP]
20:28:20.401: CTRL-EVENT-CONNECTED - Connection to <AP-MAC> completed (auth) [id=0 id_str=]
20:28:21.412: WPA: Group rekeying completed with <AP-MAC> [GTK=TKIP]
20:28:21.412: CTRL-EVENT-DISCONNECTED - Disconnect event - remove keys
20:28:30.509: CTRL-EVENT-SCAN-RESULTS
20:28:30.509: Trying to associate with <AP-MAC> (SSID='<AP-SSID>' freq=2412 MHz)
20:28:34.552: CTRL-EVENT-SCAN-RESULTS
20:28:40.617: Authentication with <AP-MAC> timed out.
20:28:48.704: CTRL-EVENT-SCAN-RESULTS
20:28:48.704: Trying to associate with <AP-MAC> (SSID='<AP-SSID>' freq=2412 MHz)
20:28:52.747: CTRL-EVENT-SCAN-RESULTS
20:28:58.812: Authentication with <AP-MAC> timed out.
20:29:06.898: CTRL-EVENT-SCAN-RESULTS
20:29:06.898: Trying to associate with <AP-MAC> (SSID='<AP-SSID>' freq=2412 MHz)
20:29:11.952: CTRL-EVENT-SCAN-RESULTS
20:29:17.006: Authentication with <AP-MAC> timed out.
20:29:25.092: CTRL-EVENT-SCAN-RESULTS
20:29:25.092: Trying to associate with <AP-MAC> (SSID='<AP-SSID>' freq=2412 MHz)
20:29:25.092: Associated with <AP-MAC>
20:29:25.092: WPA: Key negotiation completed with <AP-MAC> [PTK=TKIP GTK=TKIP]
20:29:25.092: CTRL-EVENT-CONNECTED - Connection to <AP-MAC> completed (reauth) [id=0 id_str=]
20:29:26.103: WPA: Group rekeying completed with <AP-MAC> [GTK=TKIP]
20:29:26.103: CTRL-EVENT-DISCONNECTED - Disconnect event - remove keys
20:29:35.200: CTRL-EVENT-SCAN-RESULTS
20:29:35.200: Trying to associate with <AP-MAC> (SSID='<AP-SSID>' freq=2412 MHz)

etc. etc. etc. (complete - rekey - disconnect - 3 timeouts - repeat)

I have tried adding the -DNEW_FREEBSD_MLME_ASSOC line to my
usr.sbin/wpa/wpa_supplicant/Makefile, rebuilt, installed that binary and
tried again, with no success.

The ath:

ath0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
ssid "" nwkey 65536:"","","",""
powersave off
address: 00:1f:e1:c7:4d:04
media: IEEE802.11 autoselect (OFDM54 mode 11g)
status: no network
inet6 <junk>%ath0 prefixlen 64 scopeid 0x2

I've changed one sysctl setting, hw.ath0.txintrperiod (=1 from the default
5, during wpa_supplicant it would go down to 1 anyways so I figured I can
start with that) which wasn't a problem until now. I've also verified that
5 isn't working either.

The PSK is still the same.

Any idea what's wrong? Hints?

Thanks in advance...

-Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Roy Marples]

Post by Martin S. Weber
I have tried adding the -DNEW_FREEBSD_MLME_ASSOC line to my
usr.sbin/wpa/wpa_supplicant/Makefile, rebuilt, installed that binary and
tried again, with no success.

It won't work.
You'll need to revert this patch

http://cvsweb.netbsd.org/bsdweb.cgi/src/dist/wpa/src/drivers/driver_bsd.c.diff?r1=1.5&r2=1.6&only_with_tag=MAIN&f=h

Post by Martin S. Weber
Any idea what's wrong? Hints?

Well, I could spend more time seeing if there's any other changes in the
FreeBSD driver I suppose.
I do have an ath card, but it's also my access point so hard to test
wpa_supplicant against it.

We've now had 4 cards reporting success (iwi, ral, rum, wpi) and now one
failure (ath0) so it could be an issue in the ath(4) driver.

Thanks

Roy

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Martin S. Weber]

Post by Roy Marples

Post by Martin S. Weber
I have tried adding the -DNEW_FREEBSD_MLME_ASSOC line to my
usr.sbin/wpa/wpa_supplicant/Makefile, rebuilt, installed that binary and
tried again, with no success.

It won't work.
You'll need to revert this patch
http://cvsweb.netbsd.org/bsdweb.cgi/src/dist/wpa/src/drivers/driver_bsd.c.diff?r1=1.5&r2=1.6&only_with_tag=MAIN&f=h

I've tried with a "yesterday" (cvs up'd and compiled yesterday, installed today)
system, i.e. one that did not have this patch. My sources still contain the lines
with NEW_FREEBSD_MLME_ASSOC (I looked at driver_bsd before adding that line to
my Makefile).

Both with -DNEW_FREEBSD_MLME_ASSOC and without the symptoms are the same (and
it's not working).

Post by Roy Marples
Well, I could spend more time seeing if there's any other changes in the
FreeBSD driver I suppose.
I do have an ath card, but it's also my access point so hard to test
wpa_supplicant against it.

If there's something I can do to help, tell me (although I won't get to it
before tomorrow morning ESDT -- and won't have any internet while trying
things as the ath is my link out atm). [my sources are from "somewhen"
yesterday or maybe even a day older, hmmm]

Post by Roy Marples
We've now had 4 cards reporting success (iwi, ral, rum, wpi) and now one
failure (ath0) so it could be an issue in the ath(4) driver.

I hope not... dumb me hasn't kept the old release sets of the previously
installed (older, but working) system...

Thanks though, so far!

-Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Roy Marples]

Post by Martin S. Weber

Post by Roy Marples

Post by Martin S. Weber
I have tried adding the -DNEW_FREEBSD_MLME_ASSOC line to my
usr.sbin/wpa/wpa_supplicant/Makefile, rebuilt, installed that binary and
tried again, with no success.

It won't work.
You'll need to revert this patch
http://cvsweb.netbsd.org/bsdweb.cgi/src/dist/wpa/src/drivers/driver_bsd.c.diff?r1=1.5&r2=1.6&only_with_tag=MAIN&f=h

I've tried with a "yesterday" (cvs up'd and compiled yesterday, installed today)
system, i.e. one that did not have this patch. My sources still contain the lines
with NEW_FREEBSD_MLME_ASSOC (I looked at driver_bsd before adding that line to
my Makefile).
Both with -DNEW_FREEBSD_MLME_ASSOC and without the symptoms are the same (and
it's not working).

OK, so your isssue has nothing todo with my patch then and sounds like a
kernel/driver issue.

Post by Martin S. Weber
I hope not... dumb me hasn't kept the old release sets of the previously
installed (older, but working) system...

You can always revert to older sources, via cvs.

Thanks

Roy

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Martin S. Weber]

Post by Martin S. Weber

Post by Roy Marples

Post by Martin S. Weber
I have tried adding the -DNEW_FREEBSD_MLME_ASSOC line to my
usr.sbin/wpa/wpa_supplicant/Makefile, rebuilt, installed that binary and
tried again, with no success.

It won't work.
You'll need to revert this patch
http://cvsweb.netbsd.org/bsdweb.cgi/src/dist/wpa/src/drivers/driver_bsd.c.diff?r1=1.5&r2=1.6&only_with_tag=MAIN&f=h

I've tried with a "yesterday" (cvs up'd and compiled yesterday, installed today)
system, i.e. one that did not have this patch. My sources still contain the lines
with NEW_FREEBSD_MLME_ASSOC (I looked at driver_bsd before adding that line to
my Makefile).
Both with -DNEW_FREEBSD_MLME_ASSOC and without the symptoms are the same (and
it's not working).

For the record, I'm back at 5.0 now. Everything's working fine again. I'm not
sure whether I dare going netbsd-5 yet, as mbouyer imported driver_bsd.c 1.6
into netbsd-5. But when going onto netbsd-5 at least could verify that this
change isn't the culprit but something in the driver is. Guess it'll be tossing
a coin each weekend until 'tails' tell me to waste half a day without internet :)

-Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Arnaud Lacombe]

Hi,

Post by Martin S. Weber
For the record, I'm back at 5.0 now. Everything's working fine again. I'm not
sure whether I dare going netbsd-5 yet, as mbouyer imported driver_bsd.c 1.6
into netbsd-5. But when going onto netbsd-5 at least could verify that this
change isn't the culprit but something in the driver is. Guess it'll be tossing
a coin each weekend until 'tails' tell me to waste half a day without internet :)

Are you running a 5.0 kernel or the 5.99.3 kernel you mentioned
previously ? Because the main difference between 5.0 and the -current
branch is the opensource HAL. It seems to me that you're AP is using
WPA-TKIP which is AFAIK broken in -current.

- Arnaud

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Martin S. Weber]

Post by Arnaud Lacombe
Hi,

Post by Martin S. Weber
For the record, I'm back at 5.0 now. Everything's working fine again. I'm not
sure whether I dare going netbsd-5 yet, as mbouyer imported driver_bsd.c 1.6
into netbsd-5. But when going onto netbsd-5 at least could verify that this
change isn't the culprit but something in the driver is. Guess it'll be tossing
a coin each weekend until 'tails' tell me to waste half a day without internet :)

Are you running a 5.0 kernel or the 5.99.3 kernel you mentioned
previously ? Because the main difference between 5.0 and the -current
branch is the opensource HAL. It seems to me that you're AP is using
WPA-TKIP which is AFAIK broken in -current.

I was running 5.99.3 which worked fine and now I'm running 5.0 which works fine.
So if the breakage was introduced in -current, it came after 5.99.3 ...

Regards,

-Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Roy Marples]

Post by Martin S. Weber

Post by Arnaud Lacombe
Hi,

Post by Martin S. Weber
For the record, I'm back at 5.0 now. Everything's working fine again. I'm not
sure whether I dare going netbsd-5 yet, as mbouyer imported driver_bsd.c 1.6
into netbsd-5. But when going onto netbsd-5 at least could verify that this
change isn't the culprit but something in the driver is. Guess it'll be tossing
a coin each weekend until 'tails' tell me to waste half a day without internet :)

Are you running a 5.0 kernel or the 5.99.3 kernel you mentioned
previously ? Because the main difference between 5.0 and the -current
branch is the opensource HAL. It seems to me that you're AP is using
WPA-TKIP which is AFAIK broken in -current.

I was running 5.99.3 which worked fine and now I'm running 5.0 which works fine.
So if the breakage was introduced in -current, it came after 5.99.3 ...

For reference, can you pullup driver_bsd.c 1.6 and see if it works on
your current working system? Keep a backup incase not :)

Thanks

Roy

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Martin S. Weber]

Post by Roy Marples
For reference, can you pullup driver_bsd.c 1.6 and see if it works on
your current working system? Keep a backup incase not :)

Will do tomorrow (still got a big transfer going on which I don't wanna
gamble interrupting).

-Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Martin S. Weber]

Post by Martin S. Weber
Will do tomorrow (still got a big transfer going on which I don't wanna
gamble interrupting).

Ok, things always go slower than planned. In the meantime I have replaced
my /usr/sbin/wpa_supplicant with a build from netbsd-5's head including the
changed content of driver_bsd.c 1.6. I can confirm everything's working fine
with a stock 5.0 GENERIC kernel and full distribution except for that one
file, /usr/sbin/wpa_supplicant (which is up-to-date 5.0_STABLE).

Regards,
-Martin

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Roy Marples]

Post by Martin S. Weber

Post by Martin S. Weber
Will do tomorrow (still got a big transfer going on which I don't wanna
gamble interrupting).

Ok, things always go slower than planned. In the meantime I have replaced
my /usr/sbin/wpa_supplicant with a build from netbsd-5's head including the
changed content of driver_bsd.c 1.6. I can confirm everything's working fine
with a stock 5.0 GENERIC kernel and full distribution except for that one
file, /usr/sbin/wpa_supplicant (which is up-to-date 5.0_STABLE).

Great, thanks!

Roy

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Arnaud Lacombe]

Post by Roy Marples

Post by Martin S. Weber

Post by Martin S. Weber
Will do tomorrow (still got a big transfer going on which I don't wanna
gamble interrupting).

Ok, things always go slower than planned. In the meantime I have replaced
my /usr/sbin/wpa_supplicant with a build from netbsd-5's head including the
changed content of driver_bsd.c 1.6. I can confirm everything's working fine
with a stock 5.0 GENERIC kernel and full distribution except for that one
file, /usr/sbin/wpa_supplicant (which is up-to-date 5.0_STABLE).

Great, thanks!

Well... not really, it still doesn't explain why wpa_supplicant does
not attach on 5.99.11. There should be a regression in ath(4)...

- Arnaud

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Roy Marples]

Post by Arnaud Lacombe

Post by Roy Marples

Post by Martin S. Weber

Post by Martin S. Weber
Will do tomorrow (still got a big transfer going on which I don't wanna
gamble interrupting).

Ok, things always go slower than planned. In the meantime I have replaced
my /usr/sbin/wpa_supplicant with a build from netbsd-5's head including the
changed content of driver_bsd.c 1.6. I can confirm everything's working fine
with a stock 5.0 GENERIC kernel and full distribution except for that one
file, /usr/sbin/wpa_supplicant (which is up-to-date 5.0_STABLE).

Great, thanks!

Well... not really, it still doesn't explain why wpa_supplicant does
not attach on 5.99.11. There should be a regression in ath(4)...

At least it shows that the error is not in wpa_supplicant and in ath(4)
instead of maybe being in either :)

Thanks

Roy


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de