[En-Nut-Discussion] 8019 IOCHRDY question

Bengt Florin bengt at florin.se
Thu Jan 22 13:41:37 CET 2004


Now I also tried with your deadly 'ping -l 10000'.
Yes, I get overflows but RTL is not confused.

Lost packets there are, system stable it is.

/bengan

-----Original Message-----
From: en-nut-discussion-bounces at egnite.de
[mailto:en-nut-discussion-bounces at egnite.de]On Behalf Of Bengt Florin
Sent: den 22 januari 2004 13:34
To: 'Ethernut User Chat (English)'
Subject: RE: [En-Nut-Discussion] 8019 IOCHRDY question


Harald,

I haven't tried yet with a heavily polluted network.

However in my own test, with 100 UDP/s and 1 ping/s the IOCHRDY is actived
at every NicGetPacket().
This disappears with the proposed 1us delay.

If you get overflow with original driver the chip is definitely confused.
The overflow handling by getting packets out takes too long time.
Also NicGetPacket() and NicPutPacket() need to be protected from overflow
interrupts by
cbi(EIMSK, RTL_SIGNAL_BIT)/sbi(EIMSK, RTL_SIGNAL_BIT) pair.
Specially when setting other page than 0 and when reading/writing to remote
DMA.

I did that in my tests and overflow handling is much more robust.

I however do continue improving things until rock-solid state is accieved.

Best Regards
Bengan

-----Original Message-----
From: en-nut-discussion-bounces at egnite.de
[mailto:en-nut-discussion-bounces at egnite.de]On Behalf Of Harald Kipp
Sent: den 22 januari 2004 13:09
To: Ethernut User Chat (English)
Subject: Re: [En-Nut-Discussion] 8019 IOCHRDY question


Bengan,

The IOCHRDY problem mainly occurs in high traffic
situations, like on busy networks with a lot of
Win32 machines and all their broadcasting pollution.

I usually simulate this situation by concurrently
sending ping -l 10000 to the Ethernut.

This causes a lot of overflows and, in addition
raises IOCHRDY. Sometimes the RTL becomes completely
confused.

Harald

_______________________________________________
En-Nut-Discussion mailing list
En-Nut-Discussion at egnite.de
http://www.egnite.de/mailman/listinfo.cgi/en-nut-discussion

_______________________________________________
En-Nut-Discussion mailing list
En-Nut-Discussion at egnite.de
http://www.egnite.de/mailman/listinfo.cgi/en-nut-discussion



More information about the En-Nut-Discussion mailing list