Intel wireless card issues in N mode

TL;DR: Intel N cards are affected by a nasty microcode bug with newer kernels. See the bug reports below.

I just "upgraded" my ASUS U20A to an Intel 6300. I was really happy with this card until I started using 5GHz N mode and saw huge problems with Tx retries in the thousands and terribly slow upload speeds in general. This is a bit late, but I found some nasty bugs that have yet to be resolved for this card with microcode (other cards have experimental microcode that has seen good results). I'm on 2.6.38-rc2 and still having problems. I also run .37, .36.3 and .35.10. This isn't restricted to just 5GHz, I get problems (but not quite as bad) in 2.4GHz mode. It's bad enough that all of N mode is disabled by default in Ubuntu Maverick. I definitely missed this one before I bought the card. Serves me right for going from an atheros card with nothing proprietary to an intel card with buggy proprietary microcode. If you're in the market for an upgrade, you might want to wait until this is sorted.

Relevant bug reports I've read thru (there may be others too, please post them):
https://bugzilla.kernel.org/show_bug.cgi?id=16691
https://bugs.launchpad.net/ubuntu/maverick/+source/linux-firmware/+bug/630748

 

Would that explain why my 5100 refuses to connect to my router when I set it in N only mode?

 

Yes, if you're running Ubuntu 10.10 (or something based on it). You can change the 11n_disable parameter to 0 in /etc/modprobe.d/intel-5300-iwlagn-disable11n.conf (or just comment out the line with # at the beginning). It might be a different filename; I grabbed that from the bug report. I'm not running Ubuntu myself to know for sure...

If you've already done that, check out syslog (sudo tail -f /var/log/syslog) and look for iwlagn (and network manager) errors or warnings.

There's experimental firmware for the 5000 series cards here. Untar and dump the .ucode file into /lib/firmware/. You probably should move the stock microcode to a different location temporarily. Unload and reload the iwlagn module and see what happens (sudo modprobe -r iwlagn; sudo modprobe iwlagn). Keep an eye on syslog and check for Tx retries with iwconfig. If it's worse with the experimental ucode, move the stock firmware back to /lib/firmware/ and delete the experimental .ucode file before reloading iwlagn.

 

I have an intel 6300 card; let me examine this closely....

 

Another month and there's very little progress from Intel on this (see the bug report). Might be time for me to look for another card... the Atheros AR5008 is another 3 antenna but it only comes in full height (not an issue for me, would be for half height slots).

 

Hmm well I havent done 802.11n yet on my thinkpad and usually dont have issues. But I am having a terrible issue when streaming where if I play a video or stream music on one tab or even separate browser process and I refresh or do anything on another browser or tab it crashes the stream. Its been causing me a ton of heartburn. Thank God for pianobar as it works at least.

 

If you're not on 802.11n this isn't the same problem.

I ditched that intel card anyway shortly after my last post... some people have been waiting since September for intel to fix the microcode.

 

Kernel default wireless drivers suk hell .. they work for most but not fully for all.
You might want to try compile and install compat-wireless, and if needed add patches to it.
ps.:getting aircrack work fully with it

 

Not exactly, this is a firmware problem not a driver problem. It was triggered by some changes in the driver, but those are months and months old now. There's not much left to be said here, except Intel needs to fix their broken microcode.

 

Problem is the hardware works in windows and there seems to be very little motivation to work towards rectifying linux concerns. That goes for all hardware manufacturers btw...

 

In the latest posts to that bug report intel says they're trying to separate out the legacy devices:

I'm not sure how long is fair for Intel, but it's been long enough for me, lol.

 

For anyone looking for an alternative, I finally got the AR5008X. It's running great at 300mbps on 5GHz. The exact model # is AR5BXB72

Keep in mind it is a full size mini pcie card that will not fit in a half height slot (unless you have a lot of extra room and tape it down or something).

 

I would suggest getting a newer card, xb72/5008x is fairly old, 2006 or so.

2x2 9280 (xb92) are widely available in full and half minicard, and cheap.

The 3x3 9380/9390 (xb112/114) are just now starting to show up in full cards. Not sure what the half size will be, might be 2x2 9382/92.

All use the same ath9k driver, but some of the posts on the lists seem to indicate that 5008 and 9001 are less of a target for testing and fixes.

 

If you find a 9390 for sale let me know by PM, I'm interested. The only 3 antenna alternative for sale as of my last post was the 5008X and I have no problems with it.

 

There is a full height 3x3 9380 "apple" card on fleabay right now (AR5BXB112) a little pricey but I'm tempted to try one.
The _90 doesn't really do anything more than _80, basically has some ESD built in and higher temp spec.

Looks like there is an FCC approved design for a 3x3 half height 9380, HB112, maybe someone will make one:

Atheros :: Modems

 

This bug was closed as fixed a few days ago: https://bugzilla.kernel.org/show_bug.cgi?id=16691

However there is a new (somewhat related) bug opened here: https://bugzilla.kernel.org/show_bug.cgi?id=34982