i386/95392: Kernel panic loading ndisgen-created device drivers not found before 6.1

[Chris Thielen]

>Number:         95392
>Category:       i386
>Synopsis:       Kernel panic loading ndisgen-created device drivers not found before 6.1
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    freebsd-i386
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Thu Apr 06 05:30:12 GMT 2006
>Closed-Date:
>Last-Modified:
>Originator:     Chris Thielen
>Release:        6.1-beta4
>Organization:
None
>Environment:
$ uname -a
FreeBSD localhost.localdomain 6.1-BETA4 FreeBSD 6.1-BETA4 #0: Tue Mar 14 13:59:38 UTC 2006     root at wv1u.samsco.home:/usr/obj/usr/src/sys/GENERIC  i386

>Description:
Using an eMachines m2350 laptop, with the Broadcom BCM4306 80211/b/g wireless NIC, I have used ndisgen from the Win2K drivers provided by Broadcom. This normally isn't a problem - the .ko file generates successfully in both 6.0 and 6.1beta4. I used the device driver over the course of months without a single problem in 6.0, but in 6.1beta4, with a fresh install and a fresh .ko file from ndisgen, I am seemingly randomly getting kernel panics. On boot, I am simply logging in, su'ing and kldload'ing the ndisgen-created .ko file and sometimes so, sometimes not, I get a complete kernel panic. The information isn't being dumped anywhere as it reports "no dump device specified" and then prompts me to hit a key to reboot. The oddity is that sometimes it works without a hitch, and it is _only_ on driver load. If the driver does load, I do not experience random kernel panics or anything using my wireless NIC for hours. It only occurs on kldload.
>How-To-Repeat:
Boot up, login in, su and simply run kldload on the ndisgen-created .ko device driver for the Broadcom device.
>Fix:
None, although the kernel panic only occurs in about 1 of every 2 cases, and in practice it's usually on a fresh boot so it's not too big a deal to just spend the extra 30 seconds to reboot, though it is a critical bug certainly.
>Release-Note:
>Audit-Trail:
>Unformatted: