[fedora-arm] smsc95xx performance bug: eth vs usb

omalleys at msu.edu omalleys at msu.edu
Wed Jun 1 15:14:51 UTC 2011


Quoting Andy Green <andy at warmcat.com>:

> On 06/01/2011 09:56 AM, Somebody in the thread at some point said:
>
> Hi -
>
>>>> Note: I get crappy performance with the network *up*, as long as
>>>> there's no traffic.
>
>>> I had a go a reproducing this also on a 2.6.38-based kernel on Panda.
>>> I was thinking it sounds a bit like USB suspend (not the same as
>
>>> to see if there was something obvious to hack suspend disabled but
>>> didn't see anything useful.
>>
>> try disabling pm_runtime completely.
>
> Thanks for the idea -->
>
> # CONFIG_PM_RUNTIME is not set
>
> Acts the same
>
> root at linaro:~# dd if=/dev/zero of=/media/FAD8-72F0/dump bs=4096 count=10240
> 10240+0 records in
> 10240+0 records out
> 41943040 bytes (42 MB) copied, 16.045 s, 2.6 MB/s
> root at linaro:~# dd if=/dev/zero of=/media/FAD8-72F0/dump bs=4096 count=10240
> 10240+0 records in
> 10240+0 records out
> 41943040 bytes (42 MB) copied, 17.28 s, 2.4 MB/s
> root at linaro:~# dd if=/dev/zero of=/media/FAD8-72F0/dump bs=4096 count=10240
> 10240+0 records in
> 10240+0 records out
> 41943040 bytes (42 MB) copied, 11.8782 s, 3.5 MB/s
> root at linaro:~# dd if=/dev/zero of=/media/FAD8-72F0/dump bs=4096 count=10240
> 10240+0 records in
> 10240+0 records out
> 41943040 bytes (42 MB) copied, 12.368 s, 3.4 MB/s
>
> CONFIG_OMAP_WATCHDOG blows an imprecise external about with  
> RUNTIME_PM disabled BTW, I guess that peripheral unit isn't clocked  
> without it.

FlexPWR is a feature of the chip so im not sure you -can- disable it.  
I wonder if is kicking in and lowering the voltage on the whole bus  
because 1 of the devices is idle. Or whether it is trying to save  
power after sending the data, waiting for the data to be written on  
the device.

Does it happen with the memory stick and the drive connected and doing  
something like:
dd if=/dev/zero of=/media/FAD8-72F0/dump bs=4096 count=10240 &
dd if=/dev/zero of=/media/memorystick/dump bs=4096 count=10240 &


Or does it happen with reads too?
dd if=/media/FAD8-72F0/dump of=/dev/zero bs=4096 count=10240






More information about the arm mailing list