linaro-pkg team mailing list archive

[Ming Lei <709245@xxxxxxxxxxxxxxxxxx>]

On Fri, Jul 6, 2012 at 9:57 PM, Emeric Vigier <709245@xxxxxxxxxxxxxxxxxx> wrote:
> I don't think it is the same problem. My ethernet TX gets stuck stuck to
> 10-20KBps (80-160Kbps). When I force some activity (USB mouse events) it
> can reach ~1MBps (8Mbps). Also the interface is supposed to (and says
> to) be 100Mbps...
>
> I opened a thread on various mailing-lists:
> http://comments.gmane.org/gmane.linux.network/234335
>
> I saw your ehci patch on this thread:
> https://patchwork.kernel.org/patch/1113332/
>
> The thing is, my kernel already includes your patch. I also included
> Will Deacon's patch in process.c (cpu_idle), the one you mentioned here.
>
> But I have some reasons to believe that none of the patches are suited in my case because:
> 1. Ethernet TX performance is still very bad with these patches and corresponding configs,
> 2. smsc95xx driver relies on usbnet. I think that usbnet uses interrupt for RX but not for TX. Then flushing the caches with wmb() just before entering idle (WFI in my case) won't be efficient for me as ethernet TX has no way to wake-up the cpu, it has to wait for an external wakeup source. USB Mouse event improves TX performance.
> 3. I cannot state whether your patch in ehci-q.c "should" work for me, because I don't know if (TX) URB submission in usbnet finally goes through this file's functions or another.
>
> Do you know if my statements above make sense? Do you know where I
> should put the memory barrier and test again in my particular case?


OK, I don't suggest that you do some blind test further without knowing
the reason.

Could you provide the trace result according to the guide in
Documentation/usb/usbmon.txt or the link[1]?

[1], http://www.mjmwired.net/kernel/Documentation/usb/usbmon.txt

thanks,
--
Ming Lei

-- 
You received this bug notification because you are a member of Linaro
Maintainers, which is the registrant for Linaro Linux.
https://bugs.launchpad.net/bugs/709245

Title:
  ARM SMP scheduler performance bug

Status in Linaro Ubuntu Evaluation Builds:
  Fix Released
Status in Linaro Linux:
  Fix Committed
Status in OEM Priority Project:
  Fix Released
Status in OEM Priority Project maverick series:
  Fix Released
Status in OEM Priority Project natty series:
  Fix Released
Status in OEM Priority Project oneiric series:
  Fix Released
Status in “linux-ti-omap4” package in Ubuntu:
  Fix Released
Status in “linux-ti-omap4” source package in Maverick:
  Fix Released
Status in “linux-ti-omap4” source package in Natty:
  Fix Released
Status in “linux-ti-omap4” source package in Oneiric:
  Fix Released

Bug description:
  Original Bug name: "panda: USB disk IO slow"

  This bug effects ARM Cortex A9 cores, snowball, nvidia, OMAP 4, and
  other Cortex A9 processors.  Problem is in Fedora ARM builds also so
  not limited to Ubuntu.

  My Panda's USB seems to be significantly slower than a Beagle C4.

  hdparm shows buffered reads as ~12MB/s on the Panda, and about ~20-25MB/s on a Beagle C4 from the same
  external Lacie USB disk.

  Kernel is 2.6.37-1002-linaro-omap

  Disk shows as:

  [    5.170440] scsi 0:0:0:0: Direct-Access     LaCie    d2 quadra             PQ: 0 ANSI: 4
  [    5.172546] sd 0:0:0:0: Attached scsi generic sg0 type 0
  [    5.175415] sd 0:0:0:0: [sda] 976773168 512-byte logical blocks: (500 GB/465 GiB)

  The board is otherwise idle during the test.

  Doing perf_2.6.37-12 record -a dd if=/dev/sda of=/dev/null bs=4096
  count=100000

  shows :
      81.41%         swapper  [kernel.kallsyms]     [k] default_idle
       6.33%              dd  [kernel.kallsyms]     [k] __copy_to_user
       0.94%         swapper  [kernel.kallsyms]     [k] cpu_idle
       0.51%              dd  [kernel.kallsyms]     [k] __make_request
       0.51%  perf_2.6.37-12  [kernel.kallsyms]     [k] __copy_from_user

  which suggests it's not CPU constrained.

  Dave

To manage notifications about this bug go to:
https://bugs.launchpad.net/linaro-ubuntu/+bug/709245/+subscriptions