I'm having the exact same situation with an Intel 3945 chip: set up wireless properly, and it works 2 out of 10 times. I'm going to try using the ipw3945 drivers from Slackware 12 (Slackware 12.1 supports this chip with iwl3945 kernel modules) and see if anything changes.

Actually, I've been using the iwl3945 module for a few months under Gentoo. It wasn't as good as ipw3945 drivers (I had random disconnections), but it wasn't this bad. This and the fact that Slackware's kernels run my hard drive at 1/10 its normal speed even after tweaking the hell out of it with hdparm (and yes, I do know what's going on) mean that I'm already downloading Ubuntu for a plan B.

About the hard drive: I have a SATA drive on an ICH7 controller. Slackware has ATA support compiled into the kernel. So when Slackware starts up, the kernel's ATA subsystem starts managing the drive controller. Most other distros compile ATA support as a module, and in those the kernel's SATA subsystem gets the drive controller first. The SATA subsystem runs my HD 10 times faster than the ATA system does. At the very least, this means installation took 10 times longer than it should have.

EDIT: I can't install the deprecated ipw3945 drivers. Going to try wicd...

EDIT AGAIN: Tried wicd. Worked perfectly.

Try booting with "hda=noprobe" and see if that helps. Also, try reading the documentation provided with Slackware before complaining.

Thanks. I did read the changes and hints file. I just missed that paragraph :).

dugan, did booting with that option fix the problem? (If so you should edit /etc/lilo.conf to get it to boot that way permanently, if you haven't already, by using a 'append="hda=noprobe"' line).

It did fix the problem, and I wish I'd read that file more carefully.

Well. wireless works flawlessly on Ubuntu Hardy Heron:

I have not had a good look at the actual seup in Hardy but the kernel reports:

I feel like a bit of a loser dumping slackware so fast on this laptop but I can afford time for in depth computing on only one computer at a time and this is my desktop at the moment.

Would this different kernel be the answer? Under this kernel and Hardy the wifi is absolutely flawless.

Andrew

That chip also relies on mac80211.

From what I keep seeing, both here on LQ and also while looking for info on the Broadcom drivers, it looks like mac80211 has trouble automatically associating with access points. Some people here have had good luck playing with the timing. And running wpa_supplicant again fixes the problem, which would make sense, right? Basically, that just asks it to reconnect.

(I have no idea how Ubuntu works. Do they use Network Manager?)

There are SO MANY Broadcom threads in here lately...


Somebody with a BC chipset can maybe combine the various solutions? (I've seen 'add wait 3' to a line somewhere...Sorry, can't remember where exactly, but the BCM seems to need a slight deley to get the dhcp address going...)

Just tried changing "sleep 1" to "sleep 3" in rc.inet1, which bstricks recommended in the wireless problems on -current thread. Didn't help me (rc.inet1 still wouldn't connect most of the time).

EDIT: Just tried wpa_supplicant. Editing wpa_supplicant.conf caused wicd to stop working (sigh) but wpa_supplicant worked, or at least it did the first time. And the second time. And the third time. Took the laptop back to its usual room and it did not work, then I rebooted and it started working again. I can accept this. If it breaks again, then I'm going to retry the ipw3945 drivers (there's a working kernel patch at http://james.colannino.org/downloads.html).

When you're using new kernel drivers, it might be necessary to update
your kernel as new changes are taking place such as in the b43 driver.

Hi,

As the OP I would just like to add that I have successfully managed to get wireless and pretty much everything else working on my Latitude D520 with Slackware 12.1.

The final solution was a little more experience with Slackware and the use of wicd, as several people recommended. Thanks for all the guidance on this one!


Andrew