Re: problem with if_wpi and wpa_supplicant.

Benjamin Close wrote:
> Jeff Palmer wrote:
>> Hello current_at_,
>>
>>
>> I've got a laptop that uses the intel 3945abg wireless chipset.  I've
>> followed ben's trials and tribulations with the wpi driver for several
>> months.
>> At one point,  I had plain (unsecured) authentication as well as WEP
>> authentication working. (this is several months ago now)
>> However, the last 2 renditions of the driver won't even work if I have
>> WEP enabled.
>>
>> Having seen the recent import into HEAD (8.0-CURRENT) I decided to give
>> it another whirl on my laptop.
>> With the november 2007 driver on the clearchain site,  I can associate
>> to a completely nencrypted AP.  However, using WEP or WPA fails.
>> I can't even associate to a WEP enabled AP if I set everything manually
>> with ifconfig.
>>
>> Details are as follows.
>>
>> Machine: Toshiba Satellite A135-S4517
>> OS: FreeBSD Laptop.totaldiver.net 7.0-BETA3 FreeBSD 7.0-BETA3 #0: Fri
>> Nov 23 21:33:53 EST 2007   
>> root_at_Laptop.totaldiver.net:/usr/obj/usr/src/sys/GENERIC  i386
>> Driver:
>> http://people.freebsd.org/~benjsc/downloads/wpi/20071102-freebsd-wpi.tar.gz
>>
>>
>> Relevant parts of /boot/loader.conf
>> if_wpi_load="YES"
>> legal.intel_wpi.license_ack=1
>>
>> Relevant parts of /etc/wpa_supplicant.conf
>> network={
>>         ssid="TEST"
>>         scan_ssid=1
>>         key_mgmt=WPA-PSK
>>        
>> psk=4a4de711b81e90f30a5ec88682185a74d20eb85c982f11fe035c87726b43c6c5
>> }
>>
>> The only hint I've been able to find so far, comes from using
>> wpa_supplicant in debug mode.
>>
>> Relevant log from:   wpa_supplicant -d -i wpi0 -c
>> /etc/wpa_supplicant.conf
>> Nov 25 00:32:13 Laptop kernel: idx=192 flags=0 type=119 len=4
>> Nov 25 00:32:13 Laptop kernel: cmd notification qid=4 idx=192 flags=0
>> type=WPI_CMD_SET_POWER_MODE len=4
>> Nov 25 00:32:13 Laptop kernel: wpi_cmd 155 size 12 async 0
>> Nov 25 00:32:13 Laptop kernel: notify qid=0 idx=0 flags=0 type=28 len=16
>> Nov 25 00:32:13 Laptop kernel: cmd notification qid=0 idx=0 flags=0
>> type=WPI_CMD_TX_DATA len=16
>> Nov 25 00:32:14 Laptop kernel: wpi0: could not configure bluetooth
>> coexistence
>> Nov 25 00:32:14 Laptop kernel: Scan cancelled
>> Nov 25 00:32:14 Laptop kernel: config chan 6 flags 8005 cck f ofdm 15
>> Nov 25 00:32:14 Laptop kernel: wpi_cmd 16 size 43 async 1
>> Nov 25 00:32:14 Laptop kernel: wpi0: wpi_cmd: cmd 16 not sent, busy
>> Nov 25 00:32:14 Laptop kernel: wpi0: could not configure
>> Nov 25 00:32:14 Laptop kernel: wpi0: could not send authentication
>> request
>> Nov 25 00:32:14 Laptop kernel: NEWSTATE:INIT
>> Nov 25 00:32:23 Laptop kernel: wpi0: timeout resetting Tx ring 1
>> Nov 25 00:32:23 Laptop kernel: wpi0: timeout resetting Tx ring 3
>> Nov 25 00:32:23 Laptop kernel: wpi0: timeout resetting Tx ring 4
>> Nov 25 00:32:23 Laptop kernel: NEWSTATE:INIT
>> Nov 25 00:32:23 Laptop kernel: notify qid=80 idx=0 flags=0 type=1 len=36
>> Nov 25 00:32:23 Laptop kernel: microcode alive notification version
>> 10e02 alive 1
>> Nov 25 00:32:23 Laptop kernel: notify qid=80 idx=0 flags=0 type=1 len=36
>>
>> The part that caught my attention was
>> Nov 25 00:32:14 Laptop kernel: wpi0: could not configure bluetooth
>> coexistence
>>
>> Google seems to offer 0 results for the above. Looking through the code
>> in the wpi driver tarball shows where the error is coming from,  but the
>> part I'm not understanding is: this machine doesn't have bluetooth, nor
>> any bluetooth devices.
>>
>> I'm not sure what other info would be needed,  but I'm willing to
>> provide anything I can to ben or the community at large. I haven't
>> succeeded any producing a log as to why it won't associate to a WEP
>> enabled AP when manually configured using ifconfig.
>>   
> Hi Jeff,
>    From what I can gather, bluetooth is part of the chipset. Having
> said that I've got no specs so perhaps it's not. I also see the
> firmware doesn't complain about the failed command.
> Could you try the following patch and see if it makes a difference.
> Also can you provide the relevant section of a pciconf -lv. I'm
> curious as to whether you have an ABG card or simply a BG card - I
> know they exist but haven't found them yet.
>
> diff --git a/sys/dev/wpi/if_wpi.c b/sys/dev/wpi/if_wpi.c
> index 43d5512..517473e 100644
> --- a/sys/dev/wpi/if_wpi.c
> +++ b/sys/dev/wpi/if_wpi.c
> _at__at_ -2920,8 +2920,7 _at__at_ wpi_config(struct wpi_softc *sc)
>         0);
>     if (error != 0) {
>         device_printf(sc->sc_dev,
> -            "could not configure bluetooth coexistence\n");
> -        return error;
> +            "could not configure bluetooth coexistence, device lacks
> bluetooth support?\n");
>     }
>
>     /* configure adapter */
>
>
> Cheers,
>    Benjamin
> _______________________________________________
> freebsd-current_at_freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to
> "freebsd-current-unsubscribe_at_freebsd.org"
>
> !DSPAM:8,474bd131174311249717914!
>
>
Ben,

I'm using your tarball, not the perforce version.  so I manually applied
your changes manually. Your patch said line 2920,  the tarball version
was near line 2903.  Let me know if you'd rather I use the perforce
version.  I don't know how to obtain a copy of the perforce version, so
I'd have to figure that out. However: on with the details:



#kldload if_wpi0 returns:

Nov 26 03:09:54 Laptop kernel: wpi0: <Intel(R) PRO/Wireless 3945ABG> mem
0xd8000000-0xd8000fff irq 17 at device 0.0 on pci4
Nov 26 03:09:54 Laptop kernel: wpi0: Driver Revision 20071102
Nov 26 03:09:54 Laptop kernel: wpi0: Hardware Revision (0x1)
Nov 26 03:09:54 Laptop kernel: bus_dmamem_alloc failed to align memory
properly.
Nov 26 03:09:54 Laptop last message repeated 30 times
Nov 26 03:09:54 Laptop kernel: adding chan 1 flags=0x2b maxpwr=15, offset 2
<.. snipped for brevity..>

Attempting to associate with the WEP enabled AP now results in:

Nov 27 04:08:10 Laptop kernel: cmd notification qid=0 idx=0 flags=0
type=WPI_CMD_TX_DATA len=16
Nov 27 04:08:11 Laptop kernel: wpi0: could not configure bluetooth
coexistence, device lacks bluetooth support?
Nov 27 04:08:11 Laptop kernel: wpi_cmd 16 size 43 async 0
Nov 27 04:08:11 Laptop kernel: wpi0: wpi_cmd: cmd 16 not sent, busy
Nov 27 04:08:11 Laptop kernel: wpi0: configure command failed
Nov 27 04:08:11 Laptop kernel: Scan cancelled
Nov 27 04:08:11 Laptop kernel: config chan 7 flags 8005 cck f ofdm 15
Nov 27 04:08:11 Laptop kernel: wpi_cmd 16 size 43 async 1
Nov 27 04:08:11 Laptop kernel: wpi0: wpi_cmd: cmd 16 not sent, busy
Nov 27 04:08:11 Laptop kernel: wpi0: could not configure
Nov 27 04:08:11 Laptop kernel: wpi0: could not send authentication request
Nov 27 04:08:11 Laptop kernel: NEWSTATE:INIT
Nov 27 04:08:20 Laptop kernel: wpi0: timeout resetting Tx ring 1
Nov 27 04:08:20 Laptop kernel: wpi0: timeout resetting Tx ring 3
Nov 27 04:08:20 Laptop kernel: wpi0: timeout resetting Tx ring 4
Nov 27 04:08:20 Laptop kernel: NEWSTATE:INIT
Nov 27 04:08:20 Laptop kernel: notify qid=80 idx=0 flags=0 type=1 len=36
Nov 27 04:08:20 Laptop kernel: microcode alive notification version
10e02 alive 1
Nov 27 04:08:20 Laptop kernel: notify qid=80 idx=0 flags=0 type=1 len=36
Nov 27 04:08:20 Laptop kernel: microcode alive notification version
10e02 alive 1
Nov 27 04:08:20 Laptop kernel: Authenticating, not changing channel

and the relevant pciconf -lv:

wpi0_at_pci0:4:0:0:        class=0x028000 card=0x10408086 chip=0x42228086
rev=0x02 hdr=0x00
    vendor     = 'Intel Corporation'
    device     = '10418086 Intel 3945ABG Wireless LAN controller'
    class      = network

Also,  possibly worth noting:  The first time I loaded the modified
if_wpi driver,  I saw this
Nov 26 03:09:54 Laptop kernel: wpi0: Hardware Revision (0x1)

Subsequent kldunloads and re-kldloads show:
Nov 27 04:15:58 Laptop kernel: wpi0: Hardware Revision (0x801)

Those are both exact cut/pastes,  and the hardware revision is
"different" though, it's the same laptop.  Not sure if this means
anything (maybe the driver isn't unloading cleanly or the card isn't
initializing properly on a reload) or not.