(old)Puppy Linux Discussion Forum

I'm on a Toshiba Satellite L35 that I've rehabbed with BionicPup32. Very happy with how it runs. Not super fast, but entirely usable.

Wifi, though, is giving me fits.

LSPCI shows the built-in wifi as an Atheros AR2413/AR2414 adapter. Connection wizard chooses ath5k as the driver.

Connects, loads several sites, but starts slowing down. Finally, will grind to a halt, and sometimes the "loading" indicator spins on my browser indefinitely. Sometimes, it just stops, nothing loads, no messages or indication of what's going on. This can last for a couple minutes.

Then, it seemingly recovers and everything will be fine for 5-10 minutes or so, then it repeats.

Now, plug in an old USB Wifi adapter, a Ralink RT2870/RT3070. Puppy recognizes it and loads the rt2800usb module. Switch over to that connection, connects instantly, and everything runs great. No slowdowns, no stoppages, nothing. Runs every bit as reliably as my main laptop.

Problem is, this thing is so old it only has 2 USB ports, so I really can't afford to give one up. I have the Ralink and a USB mouse plugged in and I'm all out of ports.

Any ideas what could be giving me a problem with the Atheros? I already downloaded the Locale fix (where you specify the country code) but that didn't make a difference.

Any help is appreciated.

squashpup wrote:Problem is, this thing is so old it only has 2 USB ports, so I really can't afford to give one up. I have the Ralink and a USB mouse plugged in and I'm all out of ports.

Get a USB hub. Plug it into the second port. Run the mouse and other USB devices off it.

rcrsn51 wrote: Get a USB hub. Plug it into the second port. Run the mouse and other USB devices off it.

I'd rather troubleshoot the on-board wireless first. Will the Atheros wireless not work?

There are settings to play with, 'keep alive' scripts, firmware blobs etc, but I always got what you're experiencing with ath5k. Linux drivers have improved heaps since then. I agree with rcrsn51 - best to use a hub.
Only other alternative is to replace wireless card with a later one i.e. one that uses ath9k - much better driver (ebay? - check socket & antenna fit).

IIRC, ath5k is a g-mode driver while rt2800usb is n-mode. That will make a big difference in their relative performance.

I think I got it.

Something about hardware encryption/decryption. I guess it messes up this particular Atheros wireless card. I read around about it and found some suggestions.

Anywho, what I did was to create a file in /etc/modprobe.d called "ath5k.conf"

And I put this in it:
I've been running it for nearly an hour now with no slowdowns, no disconnects, and no hesitation.

And, access to all of my USB ports! Yay!

Well done! Please come back more often - plenty of users could do with your investigative skills.

I wish I could! I have too many irons in the fire, though...I actually fixed this laptop up to leave in my as-yet-unbuilt-homebuilt-camper, primarily as a host for CHIRP to work with my ham radio when I'm away from home.

You can see from my posting history that I stay pretty busy...something like four posts in 12 years...lol.

Whenever I can get back, though, I'll try to share as much as possible.

I fixed the thread title to indicate that the problem is solved and made the issue more understandable as well.

Have a good one!

Glad to see this is solved. I was/am not sure about this problem, so did not post a "solution". However a few things did come to mind, not answered/asked.

1.) Some of this relates to the kernel in use, and its specific 'ko' file for the ath5k. I do know the AR241x series was/is buggy, but the AR242x series was fixed. The latter version is found in most netbooks 2008-10. Ralink was also a choice in some. It could be that a 2-series kernel does not have a newer corrected version of the AR241x ko file, whereas a 3-series kernel does, or makes the 241x series compatable with the 242x series.

2.) I was also wondering if your siituation occured on an open (non-encrypted) network, or an encrypted one. My AR242x does not like WPA2, with reason being the chip chops off the last (64th) character ( a zero to 63 implementation instead of a 1 to 64 ).

Regards
8Geee