[ECOS] Regarding delays in UDP (ethernet) ISR

Andrew Lunn andrew.lunn@ascom.ch
Wed Mar 12 08:16:00 GMT 2003


If you are loosing UDP packets, its because your application cannot
keep up. This is probably because your network stack has higher
priority. So you are spending more time with the network stack
receiving packets from the Ethernet device than the application
processing packets from the stack. 

1.3 is very old. The FreeBSD stack in anoncvs is supposed to be a bit
faster. I suggest you try updating. If that does not give the
performance boost you need, try using the profiler thats part of
anoncvs. You can then find out what is taking all the CPU.

         Andrew

On Wed, Mar 12, 2003 at 10:38:30AM +0500, Rajan Srivastava wrote:
> Dear Sir,
> 
> We are using eCos for PowerPC cpu.
> We are observing that net ethernet (10Mbps)
> bandwidth is some 800Kbps.
> 
> With default ecos.ecc's settings, our bandwidth was
> some 300Kbps.
> 
> We tried with more number of BDs for the SCC, more mbuf area, more
> cluster area. We even configured more BDs than ecos.ecc's limit. We
> increased network thread's priority. And then we arrived at 800Kbps 
> only. We observe that ethernet controller (in our
> case it's PowerPC's SCC) delivers packet to eCos, but
> eCos is not delivering all packets to application.
> Some UDP packets are just lost. We don't have any ethernet collisions.
> 
> Can you tell what could be troubling?
> 
> We are using eCos 1.3, can newer version help us?
> 
> Thanks & regards
> Rajan
> --------
> Rajan Srivastava
> Co-ordinating Engineer (VMS/UMS)
> Room No: 926
> C-DOT, New Delhi.
> Ph: 24678974 (Direct)
>     24677525 Extn-226
> 
> 
> 
> 
> -- 
> Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
> and search the list archive: http://sources.redhat.com/ml/ecos-discuss
> 

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



More information about the Ecos-discuss mailing list