Re: Iwlwifi broken in Sid - Bug #1953?
Franklin PIAT wrote:
On Fri, 2008-10-03 at 13:58 +0800, Tim Wood wrote:
I'm running Sid on a Compaq NX8420 laptop, Core 2 Duo with the Intel
3945ABG wifi chip. The latest upgrade to version 6 of kernel
2.6.26-1-686 gives the same error as quoted in Bug #1953.
I assume the bug number is actually http://bugs.debian.org/494703
The bug says that upgrading the firmware solves the problem.
Yes that's the one. It's Bugzilla #1953
The bug have been tagged as "pending", which means that the maintainer
have fixed the bug in his development tree, but he hasn't uploaded a
Actually, if you open the firmware-iwlwifi page, then click on it's
source page, then go to the "Source Repository", then you can
notice that the changelog #12280 states "Update Intel Wireless 3945
firmware, version 18.104.22.168. (closes: #494703)"
So you can compile that package yourself.
I thought it might be the firmware so I purged firmware-iwlwifi and
reinstalled it - made no difference.
Sep 30 09:38:27 HP-Debian kernel: [ 64.624510] iwl3945: Microcode SW error detected. Restarting 0x82000008.
Sep 30 09:38:27 HP-Debian kernel: [ 64.624510] iwl3945: Error Reply type 0x00000005 cmd REPLY_SCAN_CMD (0x80) seq 0x441B ser 0x0000004B
Sep 30 09:38:27 HP-Debian kernel: [ 64.624510] iwl3945: Can't stop Rx DMA.
Installing kernel 2.6.26-bpo.1-686 from Backport fixed the problem.
I've just used the standard Debian kernels as they have worked quite
well for me. I'd consider building my own kernel but am not sure if that
would fix the problem, as I have no idea what would need changing from
the standard .config settings, or even if that would fix it as it would
appear to be a basic fault somewhere in the kernel code.
Read the bug report... it's a firmware issue.
This bug was not evident in 2.6.24 or 2.6.25, only appearing in
2.6.26-6. It is not intermittent. Unloading/loading has no effect. Wifi
is unuseable on my system with that kernel.
Presumably there would be no point in filing a Bug as it's been reported
If it wasn't tagged "pending", you could have send a mail to the bug, to
confirm that you have the bug on your system. (Since it's difficult to
reproduce those intermittent bugs)
Any advise on how to get round this would be welcome.
You already had a workaround, right ?
Yes, to the extent that my system works fine with the Backport kernel.
Thanks for the pointer for the compile of the module - didn't think of