Intel 5100AGN ndis driver problem

[Paul B. Mahol]

On 3/27/09, knowtree at aloha.com <knowtree at aloha.com> wrote:
> On 3/21/09, Ganbold <ganbold at micom.mng.net> wrote:
>> Paul B. Mahol wrote:
>>> On 3/19/09, Ganbold <ganbold at micom.mng.net> wrote:
>>>
>>>> Hi,
>>>>
>>>> I'm trying to use ndis driver for Intel 5100AGN and got
>>>> following error:
>>>>
>>>> beastie# grep ndis /var/log/messages
>>>> Mar 19 23:41:21 beastie kernel: ndis0: <Intel(R) WiFi Link 5100 AGN> mem
>>>> 0xf4300000-0xf4301fff irq 17 at device 0.0 on pci3
>>>> Mar 19 23:41:21 beastie kernel: ndis0: [ITHREAD]
>>>> Mar 19 23:41:21 beastie kernel: ndis0: NDIS API version: 5.1
>>>> Mar 19 23:41:21 beastie kernel: ndis0: NDIS ERROR: 40001b7c (unknown
>>>> error)
>>>>
>>>> Mar 19 23:41:21 beastie kernel: ndis0: NDIS ERROR: 40001b7c (unknown
>>>> error)
>>>> Mar 19 23:41:21 beastie root: Unknown USB device: vendor 0x08ff product
>>>> 0x2810 bus uhub1
>>>> Mar 19 23:41:21 beastie root: Unknown USB device: vendor 0x0a5c product
>>>> 0x2145 bus uhub1
>>>> Mar 19 23:41:21 beastie root: Unknown USB device: vendor 0x17ef product
>>>> 0x1004 bus uhub3
>>>> Mar 19 23:41:53 beastie kernel: ugen1.3: <Lenovo Computer Corp> at
>>>> usbus1 (disconnected)
>>>> Mar 19 23:41:57 beastie kernel: ugen1.3: <Lenovo Computer Corp> at
>>>> usbus1
>>>> Mar 19 23:41:57 beastie root: Unknown USB device: vendor 0x0a5c product
>>>> 0x2145 bus uhub1
>>>> Mar 19 23:45:36 beastie kernel: wlan0: Ethernet address:
>>>> 00:21:6b:9a:23:8e
>>>> Mar 19 23:52:02 beastie kernel: pid 9861 (initial thread) is using
>>>> legacy pty devices
>>>> Mar 20 00:08:06 beastie kernel: ndis0: NDIS ERROR: 40001b7c (unknown
>>>> error)
>>>> ...
>>>> ndis0 at pci0:3:0:0:    class=0x028000 card=0x12118086 chip=0x42378086
>>>> rev=0x00 hdr=0x00
>>>>     vendor     = 'Intel Corporation'
>>>>     class      = network
>>>>
>>>> I'm running:
>>>> beastie# uname -an
>>>> FreeBSD beastie.micom.mng.net 8.0-CURRENT FreeBSD 8.0-CURRENT #5
>>>> r190040M: Thu Mar 19 21:45:37 ULAT 2009
>>>> tsgan at beastie.micom.mng.net:/usr/obj/usr/src/sys/DEVIL_WITNESS  i386
>>>>
>>>>
>>>> Any idea how to fix this issue?
>>>> Please let me know if you need more info.
>>>>
>>>
>>> How you generate ndis module?
>>> Post whole output _after_ is module loaded
>>> _not_ just lines containing ndis.
>>>
>>> also enable debug.ndis sysctl.
>>>
>>I generated module issuing following command:
>>
>>ndisgen NETw5x32.inf NETw5x32.sys
>>
>>After module is generated, loading modules gives (debug.ndis=1):
>>
>>
>>Mar 21 16:18:33 beastie kernel: no match for KeBugCheck
>>Mar 21 16:18:33 beastie kernel: ndis0: <Intel(R) WiFi Link 5100 AGN> mem
>>0xf4300000-0xf4301fff irq 17 at device 0.0 on pci3
>>Mar 21 16:18:33 beastie kernel: ndis0: [ITHREAD]
>>Mar 21 16:18:33 beastie kernel: ndis0: NDIS API version: 5.1
>>Mar 21 16:18:33 beastie kernel: ndis0: NDIS ERROR: 40001b7c (unknown error)
>>Mar 21 16:18:33 beastie root: Unknown USB device: vendor 0x08ff product
>>0x2810 bus uhub1
>>Mar 21 16:18:33 beastie root: Unknown USB device: vendor 0x0a5c product
>>0x2145 bus uhub1
>>Mar 21 16:18:33 beastie root: Unknown USB device: vendor 0x17ef product
>>0x1004 bus uhub3
>>Mar 21 16:18:34 beastie kernel: attach done.
>>Mar 21 16:18:35 beastie kernel: halting done.
>>
>>Ganbold
>
> I am seeing the exact same thing on 7.1-RELEASE, minus the USB errors. I
> have the same Intel chipset in a Fujitsu T1010 laptop. I downloaded the XP
> driver from Intel, had the same file names as above. When I ran ndisgen it
> asked me for the names of the files even through I specified them on the
> command line. Other than that the build process seemed fine. When I used
> kldload to test it I got the unknown error message but the system seemed
> happy so I set it up to load from loader.conf and added an entry for it in
> /etc/rc.conf. When I rebooted the hardware probe hung at the ndis driver.
>
> After I managed to boot without the kernel module I tried again to load
> with kldload. After the NDIS ERROR line I see
>
> WARNING: using obsoleted if_watchdog interface
>
> Next comes the MAC address, then a page fault followed by a reboot. I tried
> it several times with same results.

That panic should be fixed in CURRENT, if not, post backtrace from textdump
for the start.

> Is there anything I can do to help solve this? Again I am running 7.1 not
> HEAD.

Something in NDISulator initialization code looks to be broken, but current
showstopper are error codes that are not documented anywhere.


-- 
Paul