Sluggish mouse on amd64

Live forum: http://rt2x00.serialmonkey.com/viewtopic.php?t=4778

Fjodor

07-05-2008 00:16:55

I remember back in 2006 when I was using madwifi for an atheros card, that I would get a sluggish mouse, and that it turned out to be a problem with the madwifi driver, specifically on x86_64.

For the longest time now, I've had not only a sluggish mouse, but lagging keyboard too, often with repeating keys (both keyboard and mouse are wireless), with the problem increasing with increasing network load.

Could it be that rt2x00 and x86_64 are doing sort of the same that madwifi did back then?

Also, my load average seems ridiculously high, which could very well explain the sluggishness, however I can't seem to locate any particularly hungry processes, further leading me to believe that it may be rt2x00 eating a lot of juice...

Please advise on how to debug, or tell me if I'm way off...

Best regards and thanks as always,

F

IvD

07-05-2008 13:10:42

Well the most obvious hint would be what happens when rt2x00 isn't loaded?

Fjodor

07-05-2008 15:15:55

Indeed, and as suspected No sluggishness... How on earth would one go about debugging further?

Best regards,

F

IvD

07-05-2008 15:42:31

What version do you currently use?
rt61pci in 2.6.24 is working happily on my x86_64 system, and latest rt2x00.git as well.
Do you have encryption enabled? If it is WEP could you try rt2x00.git with hardware encryption?

Fjodor

07-05-2008 16:49:42

HW crypto would be nice, but the rt2500pci-supported cards don't support that, do they?

Best regards,

F

IvD

07-05-2008 17:45:31

Well the hardware supports it, it is the driver that still needs to be modified. ;)
But although it is on the todo list, the implementation for rt2500 is trickier compared to rt61 (which currently only works for WEP).

IvD

07-05-2008 18:57:55

But what version are you actually using?

Fjodor

08-05-2008 05:07:20

Oh, sorry. rt2500pci ;-)

/F

IvD

08-05-2008 07:41:12

Yes I understood that, but what version?
1) rt2x00.git? from what day?
2) wireless-compat? What datestamp?
3) vanilla kernel? Which kernel version?
4) something else?

Fjodor

08-05-2008 08:55:56

Oh, sorry again -$

Incidentally, I am having some trouble compiling latest git. I imagine you'd like me to see first, if the problem is present in that, so let's look at this first (opening new thread now).

Best regards,

F

Fjodor

08-05-2008 19:53:09

Erm, sorry to report, but with commit 64e4500c9a8df6ca12b295cd24f388ea650b49a7, I get a whole lot off scrolling kernel warnings, and with the version before that, the warnings eventually stopped, and I saw something about a null pointer dereference -(

I don't have time for it now, but I'll see if I can get a net console going tomorrow, so I can capture the warnings...

Best regards,

F

IvD

08-05-2008 20:45:52

I am in the progress up updating rt2x00.git to a more recent kernel version, if you wait 2 days you have the update.
If the null pointer still persist, please post the stacktrace that is printed with it.

Fjodor

09-05-2008 12:15:28

Ok, thanks. I'll wait for the update -)

/F

Fjodor

09-05-2008 17:39:34

Still waiting for the update, though I am confident that you are working as fast as you can. In the meantime, I'd just like to report that I have also had problems at full rate, but since I began running at 36M, the problem with the sluggish mouse seems to have disappeared, leading me to believe that the problem might lie somewhere in code that is triggered in bad reception conditions?

Best regards,

F

IvD

09-05-2008 17:56:14

Well that is possible, but like I mentioned earlier that really depends on the version you are using...

The update is still work in progress, I am reordering patches to prepare for a new release.

IvD

10-05-2008 12:10:20

rt2x00.git has been updated.

Fjodor

10-05-2008 12:36:39

Excellent!

cg-update -f'ing as we speak -)

/F