[ECOS] Re: Network latency

andrew bell allenheath1@googlemail.com
Tue Oct 27 14:12:00 GMT 2009


Hi Grant

Thanks for the info, following your advise we've done the following:

We're using a PowerPC with a FEC, which is directly connected to an
ethernet switch MAC using the MII.

The MAC was set to full duplex, 100MBits, no flow control.

The stock if_fec.c during fec_eth_reset, sets the Transmit control to
"4+0", which is bit 29 - Full Duplex Enable in X_CNTRL.

It also however configures fec->RxControl = RxControl_MII | RXControl_DRT;

The DRT is "Disable receive on transmit".

In the manual for the MPC852T it says to set DRT active for
half-duplex, and disable for Full Duplex.

When I remove the DRT being or'd in to the RxControl register, we get
much better reliability.

I'm now seeing a pause about once an hour, instead of once every
couple of minutes.

Does anyone know why

fec->RxControl = RxControl_MII | RXControl_DRT

contains the DRT portion when it looks like the fec driver is setting
up for full duplex ?

TIA

Andy.

On Mon, Oct 26, 2009 at 4:57 PM, Grant Edwards
<grant.b.edwards@gmail.com> wrote:
> On 2009-10-26, andrew bell <allenheath1@googlemail.com> wrote:
>
>> I?ve gone back to the nc_master / slave tests and monitored
>> the throughput when connecting a PC to one of the boxes
>> running eCos. Initially everything looked fine until I insert
>> a switch between the PC and eCos box.
>>
>> With any switch between the PC / eCos box, there is poor
>> performance when doing an RX test (PC -> eCos box), which
>> shows lots of out of order delivery / delayed ACK / fast
>> rexmit, and the throughout graph looks like a saw tooth.
>
> In my experience that's usually a sign of half/full duplex
> mismatch between one of the PHYs and the Ethernet controller to
> which it is attached.  It could also be due to a mismatch
> between a PHY and the switch, but you've got to try pretty hard
> to make that happen.  Can you try setting the switch ports to
> various different duplicities?  Or try a half-duplex hub/switch
> instead of a full-duplex switch?
>
>> But hey, the PC?s running faster than the eCos box, so maybe
>> this is to be expected.
>
> Nope.
>
> --
> Grant
>
>
>
> --
> Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
> and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss
>
>

--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss



More information about the Ecos-discuss mailing list