usb/101096: USB WLAN occasionally causes kernel-panics during large downloads

[Pascal G. Hofstee]

The following reply was made to PR usb/101096; it has been noted by GNATS.

From: "Pascal G. Hofstee" <caelian at gmail.com>
To: Hans Petter Selasky <hselasky at c2i.net>
Cc: freebsd-usb at freebsd.org, freebsd-gnats-submit at freebsd.org
Subject: Re: usb/101096: USB WLAN occasionally causes kernel-panics during
	large downloads
Date: Mon, 31 Jul 2006 23:57:18 +0200

 On Mon, 2006-07-31 at 16:32 +0200, Hans Petter Selasky wrote:
 > Hi,
 > 
 > I think the new "if_ural" has this problem fixed, and a couple of others. 
 > Could you have tried the following:
 > 
 > I recommend that you fetch all files from SVN. The SVN port is available from 
 > /usr/ports/devel/subversion . After Subversion is installed you run:
 > 
 > svn --username anonsvn --password anonsvn checkout svn://svn.turbocat.net/i4b
 > 
 > cd i4b/trunk/i4b/FreeBSD.usb
 > 
 > make S=../src package
 > make install
 > 
 > Follow the "make help" instructions.
 > 
 > It might be a good idea to build all kernel modules.
 > 
 > Before you build the kernel copy the following files into 
 > "/usr/src/sys/dev/usb/":
 > 
 > i4b/trunk/i4b/src/sys/dev/usb/if_ural.c
 > i4b/trunk/i4b/src/sys/dev/usb/if_uralreg.h
 > i4b/trunk/i4b/src/sys/dev/usb/if_uralvar.h
 > 
 > NOTE: the kernel build will most likely fail, unless you are using 
 > FreeBSD-7-current. If it fails because there are missing product ID defines, 
 > then just remove these lines from the "ural_devs" structure in "if_ural.c". If 
 > it fails because "IEEE80211_FEXT_SWBMISS" is not defined, then change the 
 > failing code to something like this:
 > 
 > #ifdef IEEE80211_FEXT_SWBMISS
 >         /* enable SW bmiss handling in sta mode */
 >         ic->ic_flags_ext |= IEEE80211_FEXT_SWBMISS;
 > #endif
 > 
 > Else please report the error, if you cannot resolve it.
 
 Well .. i had already updated my source-tree to RELENG_6 so i decided to
 follow the instructions provided and built/installed world and kernel
 entirely and am now on 6.1-STABLE.
 
 The kernel did build without problems ... however upon booting the new
 kernel i now get boatloads of acpi failure messages
 
 acpi: bad write to port 0xcf8 (32), val 0x80002084
 acpi: bad read from port 0xcfc (32)
 
 The system does come up fully multi-user and appears to work normally
 though. I'll keep an eye out to see if the kernel panics are now
 resolved.
 
 -- 
   Pascal Hofstee