Bug#462831: rt73.ko driver causes load average 2.0

[Ondrej Certik]

On Jan 29, 2008 9:33 PM, Ben Hutchings <ben at decadent.org.uk> wrote:
> On Tue, 2008-01-29 at 19:02 +0100, maximilian attems wrote:
> > On Sun, 27 Jan 2008, Ondrej Certik wrote:
> >
> > >
> > > the rt73.ko driver in Debian and kernel 2.6.22, as well as the latest
> > > upstream cvs version and the kernel 2.6.23 causes
> > > the load average to be 2.0 when the interface up. Doing "ifdown
> > > wlan0", the load average goes to back to 0.00.
> > >
> > > I reported the bug here with more details:
> > >
> > > http://rt2x00.serialmonkey.com/phpBB2/viewtopic.php?t=4568
> > >
> > > But noone responded. Do you have some tips what I can try?

> install 2.6.24 from unstable should have newer version?
> and see how it works..

Currently, I am getting:

$ wajig install linux-headers-2.6.24-1-686
[...]
The following packages have unmet dependencies:
  linux-headers-2.6.24-1-686: Depends: linux-kbuild-2.6.24 but it is
not installable
E: Broken packages

So I'll wait a few days, when this gets resolved and try it then.

(I need the headers to compile the rt73 driver)

> This is not necessarily a serious problem.  Any (kernel or user) thread
> in the "io-wait" state counts toward the load, even though it is not
> taking processor time.  However, to avoid confusion, kernel threads
> should not sleep in a way that puts them in io-wait state.

Yes, it's probably not a serious problem, but still it's imho a bug.

> > install 2.6.24 from unstable should have newer version?
> > and see how it works..
>
> That's actually a different driver (rt73usb), which is also available
> for 2.6.22 and 2.6.23 through the rt2x00-source package.

There is no rt73.ko in the 2.6.24 as far as I know.

BTW, I just noticed that upstream removed my attachement with detailed
logs of the problem from my bug report. I am very disappointed, as it
cost me some time to generate the debug logs properly.

Ondrej