RE: [lwip-users] Abnormal Tx delays

[Ofer Epstein]

 

Hello Nicolas,

 

I am interested in further details regarding the driver problem you have had. I think I also have a MAC driver problem.

It looks like the MAC stops sending frames suddenly and this is causing the lwip stack to deadlock while waiting to TX DONE IRQ.

I managed to force tx restart on the MAC and this "fix" the deadlock but with a big performance penalty, so it's is not a real solution.

I use a TALITY MAC driver.

Please let me know your experience with your MAC driver.

 

Thanks,

Ofer

 

 

 

---

From: address@hidden [mailto:address@hidden On Behalf Of Nicolas Pinault
Sent: Friday, September 21, 2007 8:49 AM
To: Frédéric BERNON; Mailing list for lwIP users
Subject: Re: [lwip-users] Abnormal Tx delays

 

Bonjour Frédéric,

I finaly got it to work correctly. The problem was not in lwip. It was in my MAC driver. It was a mix between a hardware not well documented behavior and the software driver I have written. Behavior was as if request packets were coming in lwip but answer packets were outputed long time after.
If anyone is interrested, I'll give details.

Anyway, your message helped me to better understand lwip and to write a better initialisation function.

Thanks for your help.

Nicolas

Frédéric BERNON a écrit :

I don't got such problems with current code. Since you got the CVS HEAD today, to you have updated your port & driver with all "last" requirements?

 

Take a look to http://lwip.scribblewiki.com/LwIP_version_1.3.0_release_notes. It's not updated "of the day", but most important points are listed.

 

I think to things like sys_mbox_tryfetch, lwip_init (if you use tcpip_init, you don't have to call it directly), the new NETIF_FLAG_ETHARP flag for netifs.

 

There is also last changes in CHANGELOG, like rename of MEMP_TCPIP_MSG to MEMP_TCPIP_MSG_API and the new MEMP_TCPIP_MSG_INPKT pool for input packets.

 

Perhaps the problem is here? I hope it help you...

 

----- Original Message -----

From: "Nicolas Pinault" <address@hidden>

To: "Mailing list for lwIP users" <address@hidden>

Sent: Tuesday, September 18, 2007 5:34 PM

Subject: [lwip-users] Abnormal Tx delays

 

Hi all,

I have a problem with my system.
I use lwip CVS HEAD (got it today).
I use µC/OS-II.
I use ethernet chip KSZ8841 (8 bit mode).
I use tcpip_input to input frame into lwip.
I use several tasks to send/receive data thru lwip.
One of these tasks is a basic WEB server.

At startup, pinging my system (from a WinXP PC) is ok.
Then I use FireFox to browse the WEB of my system.
I refresh several times.
After a random number of refresh, lwip still responds to input packets
but with big delays.
Even pinging my system with a quiet traffic is not ok. Answers come
several seconds (several tens of seconds) after the incoming packets.
See the WireShark capture (a filter has been used to capture only
interresting packets).
Looking at lwip traces, it seems that incoming packets are stored but
not treated immediately. After a random delay, they are treated and
response packet are sent.

I can't figure out where the problem is.

Any help is welcome.

Nicolas

---

> _______________________________________________
> lwip-users mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/lwip-users

_______________________________________________

lwip-users mailing list

address@hidden

http://lists.nongnu.org/mailman/listinfo/lwip-users

______________________________________________________________________
DSP Group LTD. automatically scans all emails and attachments using MessageLabs Email Security System.
______________________________________________________________________