Sponsored link
Vyatta –
Linux & Open Source
Alternative to Cisco –
Advanced Routing,
Firewall, VPN, QoS..
Free Download ->
| |||||||||||
VectorLinux ReviewVectorLinux ReviewPosted May 21, 2007 22:04 UTC (Mon) by pr1268 (subscriber, #24648)Parent article: VectorLinux SOHO: A better Slackware than Slackware (Linux.com)
"My [Broadcom 43xx] wlan chip is not natively supported by the Linux kernel..." "VectorLinux detected my chipset and automagically inserted the bcm43xx module. This is correct for similar chipsets, but it does not work with mine." Ms. Linton goes on to mention she had to use NDISWrapper and a Windows driver to get her Wifi card working. She even shows a simple one-line shell script command to enable the card using ndiswrapper. I think this is misleading. The Broadcom 43xx chips have had native support since 2.6.17. It seems likely that the issue with Ms. Linton's laptop is that she doesn't have the firmware loaded (which the bcm43xx kernel module needs). I feel slightly qualified to mention this as I've run recent versions of Slackware and that Wifi card, both pre-2.6.17 with NDISWrapper and a Windows driver, and post-2.6.17 with the bcm43xx module and firmware (bcm43xx-fwcutter firmware extraction tool available here). I don't mean to sound overly disapproving of Ms. Lindon's article; rather, I feel that the issues she faced installing a modern Linux distro on contemporary laptop hardware are typical. Thanks for the article.
(Log in to post comments)
VectorLinux Review Posted May 21, 2007 22:31 UTC (Mon) by zlynx (subscriber, #2285) [Link] The bcm43xx kernel driver does not work on my laptop, no matter which firmware I try.
I get the situation where it tries to authenticate and associate over and over and over again, but never actually succeeds.
NDISwrapper works quite well for me.
So, its entirely possible there are other people out there who can't use the bcm43xx driver.
VectorLinux Review Posted May 24, 2007 2:42 UTC (Thu) by lysse (subscriber, #3190) [Link] Are you using wpa_supplicant? I found myself hitting a brick wall with the bcm43xx drivers until I tried that, and was amazed to find it Just Worked. YMMV.
VectorLinux Review Posted May 21, 2007 22:33 UTC (Mon) by salimma (subscriber, #34460) [Link] FWIW, before my bcm4318-equipped laptop died, I've never been able to get it working with the in-kernel bcm43xx driver (on x86_64, so it might work on i386). With earlier kernels, the driver would freeze the kernel when loaded, with later kernels it is more stable, but while it would show the APs it would fail to associate with one.
bcm43xx woes Posted May 21, 2007 22:53 UTC (Mon) by pr1268 (subscriber, #24648) [Link] I stand corrected. It's a shame that the Broadcom 43xx chipset gives Linux users so much frustration so as to force users to resort to NDISWrapper (not meant to impugn the hard work the NDISWrapper developers have put into that project). Curiously, I've never experienced the issues the above two posters have (I used a Linksys WPG54G ver. 3 firmware PCMCIA card), but after I physically broke the card, I had to revert to an old (but reliable) 802.11b card. Yeah, yeah, I know, no one's forcing anyone to use Linux; <snide-comment>we all could use Microsoft Windows instead (and have a perfectly-functioning Wifi card driver). Ballmer loves this kind of arrangement!</snide-comment>
bcm43xx woes Posted May 23, 2007 5:10 UTC (Wed) by salimma (subscriber, #34460) [Link] Right now the only wireless chipset I'd recommend is Intel's. Even Atheros, which has an entire project (well, two: FreeBSD has its own project) dedicated to developing open-source drivers, has had communication problems in the past, with apparently only one developer having NDA access to documentations.
The 802.11 a/b/g/n chip found in the Core 2 Macbooks (and Pro) and the T/Z 60 Lenovo Thinkpads, for example, only started working with the trunk development version of Atheros last month, IIRC. Half a year or so after the hardware is out.
VectorLinux Review Posted May 22, 2007 13:53 UTC (Tue) by ejp (guest, #29753) [Link] I have the Broadcom 4311 chip mentioned in the article in my Inspiron 1501. With the right firmware, it is possible to get it to both come alive and associate. It does hang the kernel occasionally though, and so far only operates at 11Mb/s. Still, it's making rapid progress.
VectorLinux Review Posted May 22, 2007 20:46 UTC (Tue) by drag (subscriber, #31333) [Link] With my old laptop (RIP *sniff* ) I had the broadcom badness. Airport Extreme flavor.
It's non-x86 so there is no way that NDISWrapper would ever work.
For a long time it would associate, but crap out when the DHCP happenned. Apparently it didn't like being taken down and brought back up quickly. So for a long time only static connections worked.
Once it got into the vanilla kernel it settled down and acted reliably. The only problem was that if it got disconnected then sometimes rmmod'ng the module and modprobe'ng it back in would be the only way to bring it back up.
As for the speed of the connection.. 11Mb/s was the default speed. For some reason it is unable to dynamicly adjust to the speed of the network.
Most of the time I actually slowed it down to 2Mb/s by using iwconfig since that would result in more reliable connections. But it had no problem operating at 54Mb/s given a good signal, I just had to manually set it.
iwconfig sta0 rate 2M
Keep in mind that this driver does come in two flavors. The first one, the one in the kernel currently, is the Intel 80211-based driver.
The advanced, final one, is the mac80211-based (aka dscape 802.11 stack) and is the actual goal. mac80211 should be in 2.6.22.
Other drivers are like that. My personal favorite devices are the Ralink devices and they come in two flavors also.. the regular versions then the rt2x00 mac80211 version.
So, although I don't know this for a fact, I bet those of you stuck with devices that won't work with the in-kernel driver would be smart to try out the Wireless-dev branch of the Linux kernel and let the developers know any problems you run into.
I think that one of the things that are holding these drivers back are because users are giving up to easily and resorting to ndiswrapper. I don't blame you for doing that (after all we all use computers for more then just compiling kernels), but I think that developers would probably like as many different testers on as much different hardware as possible.
That way when the drivers do end up making it into the kernel then they are tested and more then likely will work for new users to Linux with much less effort then they currently do.
http://git.kernel.org/?p=linux/kernel/git/linville/wirele...
VectorLinux Review Posted May 23, 2007 1:10 UTC (Wed) by zlynx (subscriber, #2285) [Link] I tried the new bcm43xx driver in the -mm kernel series. It failed to work at all. After asking about it on LKML I was told that it will only run with newer firmware which will only work on newer Broadcom chips with enough RAM.
I suppose that one of these days I'll be getting a newer laptop and I won't care anymore.
VectorLinux Review Posted May 23, 2007 23:41 UTC (Wed) by beoba (guest, #16942) [Link] I have a Powerbook (12" G4 1.33) with a 4306 (Airport Extreme, older revision?). bcm43xx works fine in Ubuntu 7.04.
|
|||||||||||