[En-Nut-Discussion] Fear not, good Sir... TCP might still be saved...

[Marek Pavlu]

Hi, 

Yes, I try find any way, how fix it, because my application including
treatment symptoms for this moment... 

As I say, I am able shoot down rxi5 with infinite event timeout so and with
same symptoms except heap. 

I do not mean directly NutEventPostFromIrq, but some problem with event
management in high load... 

Regards, 
                Marek Pavlu 

//  -----Original Message----- 
//  From: en-nut-discussion-bounces at egnite.de [mailto:en-nut-discussion- 
//  bounces at egnite.de] On Behalf Of Michael Jones 
//  Sent: Wednesday, June 21, 2006 11:00 AM 
//  To: pavlu at HWserver.cz; 'Ethernut User Chat (English)' 
//  Subject: RE: [En-Nut-Discussion] Fear not,good Sir... TCP might still be

//  saved... 
//  
//  
//  Hi, 
//  
//  I've looked at your solution - yet, as this can solve the symptoms it 
//  does 
//  not solve the problem. 
//  
//  I have done further research and can say that the actual 
//  NutEventPostFromIrq 
//  is not the problem. It's a general problem with e.g. One Shot Timers and

//  the 
//  use (abusage) of the heap. Also there seems to be a problem with the 
//  expired 
//  timer cleanup. In the case of the SMSC chip simply setting the timeout 
//  to 0 
//  solves the problem to but that's not the real solution to the real 
//  problem. 
//  
//  But anyway, thanks for your input! 
//  
//  Cu, 
//  Michael 
//  
//  
//  -----Original Message----- 
//  From: en-nut-discussion-bounces at egnite.de 
//  [mailto:en-nut-discussion-bounces at egnite.de] On Behalf Of Marek Pavlu 
//  Sent: Wednesday, June 21, 2006 10:46 AM 
//  To: 'Ethernut User Chat (English)' 
//  Subject: RE: [En-Nut-Discussion] Fear not,good Sir... TCP might still be

//  saved... 
//  
//  Hi, 
//  
//  My opinion is that cause of this problem is excessive exposed 
//  NutEventPostFromIrq. I mean that evil is in event management:). 
//  
//  In case NutEventWait with timeoute you prove that, Michael. 
//  In case NutEventWait with infinite timeoutem it is still question. 
//  
//  Maybe I have an idea, how it quickly solve ;). 
//  One way is modification Nut/OS semaphore for post from irq and do them 
//  more 
//  safety(critical section and overload). In this case many call irq is 
//  only 
//  one call NutEventPostFromIrq and its mean that in case of irq call: 
//  
//  1. rxi5 is waiting  => only one call NutEventPostFromIrq for multiple 
//  irq 2. 
//  rxi5 is working => NutEventPostFromIrq is not call! 
//  
//  This way conserve system resource's and avoid problem in event 
//  management... 
//  
//  
//  
//  For RTL8019(I have only this chip) and 4.0.2.1 ethernut version. 
//  I test this modifications and so far is good:), but more test is 
//  required... 
//  
//  
//  Files is here http://www.hw-group.com/download/supp/ethernut/rxi5.zip 
//  
//  
//  
//  
//  
//  
//  Regards, 
//                  Marek Pavlu 
//  
//  
//  
//  
//    _____ 
//  
//  avast! Antivirus <http://www.avast.com>  : Odchozi zprava cista. 
//  
//  
//  Virova databaze (VPS): 0625-4, 21.06.2006 
//  Testovano: 21.6.2006 10:46:03 
//  avast! - copyright (c) 2000-2006 ALWIL Software. 
//  
//  
//  
//  _______________________________________________ 
//  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 



  _____  

avast! Antivirus <http://www.avast.com>  : Odchozi zprava cista. 


Virova databaze (VPS): 0625-4, 21.06.2006
Testovano: 21.6.2006 11:26:15
avast! - copyright (c) 2000-2006 ALWIL Software.