[rt2x00-users] Backport "rt2800pci: fix spurious interrupts generation" to Linux 3.2.x?
Dennis Martin Herbers
dennis.martin.herbers at googlemail.com
Mon Jan 30 05:08:22 EST 2012
Linux 3.2.2 is now in Ubuntu 12.04 development branch and just like 3.3rc1
it did not fix my problem, so maybe I am hitting something else. I looked
through the earlier Ubuntu development kernels and there doesn't seem to be
a kernel that really fixes my issue. I probably just had "luck" at that
time, because sometimes the speed goes down from ~700KB/s to ~75KB/s within
few seconds and sometimes it needs a few minutes. I'd say 30 seconds on
average. I remember that playing with sudo iwconfig wlan0 power off and
then disconnecting/reconnecting increased the speed again until it went
down again after a few seconds, not sure if it's related to power
management though. Where would I see the "irq N: nobody cared" messages,
dmesg? I don't remember seeing those but I'll check back soon.
Also I would be very interested if the fix that went into 3.2.2 and 3.3rc1
actually fixed the slow performance for anyone else (RedHat bugtracker &
Ubuntu Launchpad bug entry). I'll ask the Ubuntu users for new results.
2012/1/23 Stanislaw Gruszka <sgruszka at redhat.com>
> Hi Dennis
> On Fri, Jan 20, 2012 at 11:30:51PM +0100, Dennis Martin Herbers wrote:
> > Unfortunately, I've just tried Linux 3.3rc1 and this does not fix my
> > performance problem (only 10% of the download speed I should have,
> > instead of ~650KB/s). This is werid because this issue seemed to be fixed
> > in one of the Linux 3.2 pre-releases that included a different patch that
> > tried to fix this issue, but which was later pulled and the speed got
> > again. I would have assumed it would be gone now... this is with a RT3090
> > (Lenovo S205).
> Andreas reported one issue with rt2800
> coused by commit f0425beda4d404a6e751439b562100b902ba9c98, you may
> wish to revert it and check if it fix the problem.
> If not, whould be helpfull if you could tell which exact kernel version
> (HEAD commit, or compat-wireless date release) works and which does not
> work. Perhaps we fixed one problem, but on the meantime we introcuded
> antother one.
> Also is not clear if you are hitting the same spourions interrupt
> problem, do you see "irq N: nobody cared" messages?
> Anyway the spurious interrupt fix was just added to stable queue,
> should be relesed with 3.2.2 .
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the users