Video: Linux ath5k Reboot (Part 2)

Yeah, it so deserves a friggin’ sequel.  This was post-reboot. As you can see, the Atheros device wasn’t even detected in dmesg, lspci, etc.

That meant no scanning, no connecting, nothing. What’s the purpose of a netbook if it can’t fucking network?

As I note in the comments, it took me several reboots this time — and yet again in both Linux and Windows — to re-detect the device and be able to network.

Let me make a disclaimer. I’ve followed the bug reports on this from the first time it happened to me. What is it now, nearly six months? I appreciate the serious effort these people are making to make this device class function under Linux. As of my most recent kernel, though, I think it’s proven to still be at a development-stage and still quite unstable. I don’t recommend using Linux on Aspire Ones or other devices using the ath5k driver unless you have a high threshhold for pain and can live with it stopping suddenly and for no apparent reason like this. Maybe they’ll get it functioning better soon. I hope they do.

In the meantime, I’m faced with either changing cards or using Windows rather than Linux. My AA1 is off warranty in a couple months. I may switch cards before then. Or I may spring for Windows 7. The irony of this is, all of this happened while I was making another screencast demonstrating why I was going to ditch CrunchBang/Ubuntu for TinyCore. That’s when the audio from the stream stopped and I realized I was dealing with this mess again.

Caveat emptor. One man’s “free as in freedom” is another man’s quintuple reboot to get the damn thing to work correctly again.

Advertisements

Tags: , , ,

4 Responses to “Video: Linux ath5k Reboot (Part 2)”

  1. kruce Says:

    Which kernel are you using? In the last month, I got my eee (should also have the Atheros chip), and it has a reasonably successful dualboot with Win7 and #! – but I found that the ACPI functions worked “best” with a “netbook” kernel (2.6.29.1) that I got from the eeebuntu repos. But even still, I don’t think I’ve had noticable issues (beyond signal strength, best with 2.6.29.1) with the stock 9.04, updated array.org, and 2.6.30.04 kernels.

    Actually, now that I think of it, I was having odd sound issues as well. I THINK they’re ok with my current kernel, but I’ll have to double check that.

    Didn’t I read somewhere that the ath5k driver is being replaced with a madwifi driver in 2.6.28+ or something?

  2. kruce Says:

    Watched the vid, see the kernel. Disregard.

    • lucky Says:

      Congrats on the Eee. Does lsmod show you’re using ath5k or lspci show an Atheros card? IIRC, Asus has used various cards (including ralinks?) in those things. Maybe not.

      More details: kernel is vanilla 2.6.30.1; the problem has occurred in nearly every distro I’ve run now. I haven’t seen any changes to ath5k since 2.6.30 in the mainline, though there are test patches floating around between developers/testers. I use WPA at home and at work (and also when visiting family); I have a couple VPN accounts. The problem has recurred only at home so far but that’s because I don’t use Linux much elsewhere. I’ve connected via work VPN while running Linux when out of town but that was only for about an hour, which probably isn’t long enough under “normal” circumstances to get a timeout.

      I’ve noticed the timeouts occur more often during periods of heavy Linux use. The night before this event happened, I’d left some things running for longer than five hours before waking up and shutting down. The timeout occurred after I’d been running Linux again at/before lunch for about an hour or so.

      Initially, I thought the problem might be isolated to my particular card — just a crappy one that got past QC. But it hasn’t happened in XP at all so it’s safe to assume it’s Linux-specific. Since I’ve been posting about this issue, I’ve seen a steady trickle of traffic from search engines related to ath5k timeouts. So I know I’m not alone. And as I alluded above, I’ve periodically looked at some of the communication flow between developer(s) and testers and seen that the problem isn’t resolved.

      I also wasn’t sure if the problem was with the driver, with the way the driver uses the 802.11 stack, or with WPA. Following some of the aforementioned development communication, I have questions about if there might be some quirks all the way around.

      I’ll look and see if madwifi is a better solution or even if reverting to ndis would work. Dunno. It’s frustrating, though, that something this buggy would be in mainline if something else actually works better. AFAIK, ath5k is supposed to supercede madwifi. Could be the other way around.

      What kind of sound issues are you having?

  3. kruce Says:

    My eee (1000HEB – a late July Woot! special) looks like its running the AR928 chip, and the ath9k driver, accordingly. Really, the only issues with wifi is that periodically it’ll lose about 30% signal strength on my home network, but I don’t know if its WEP (I still have some older pcmcia cards) or my router (a dd-wrt el cheapo deluxe). l

    As far as sound goes, I generally don’t do much with sound – the occasional Youtube video. So, the only thing I noticed with some kernels is that lack of annoying #! startup sound. But its good with the current kernel.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: