Re: [lwip-users] Degrading performance and etharp proterr

[Jens Nielsen]

Hi

Regarding the issue/race condition from the devel list, shouldn't it 
only occur once and then at the next interrupt you'd service both the 
frame that was partially received and the one that triggered the second 
interrupt? And how does the issue even occur? Afaik the STM32F4 doesn't 
trigger the interrupt until the full frame is received so if you at some 
point have a partial frame you'd get a new interrupt (and signalled 
semaphore) as soon as it's complete and you can handle it then. Unless 
we're talking about some buffer overflow scenario?

I have used the ST drivers for a long time under heavy load in an 
application where I should have noticed deteriorating performance, I'm 
not saying they're perfect (I haven't really looked at them that hard) 
but at least proven to be pretty stable. At least I don't see that issue 
occuring because of their driver.

Regarding the issue in this topic, it's only related to tx and not 
caused by delays in rx, right? So it should be a completely different issue?

Best regards
Jens


On 2016-04-13 10:52, Freddie Chopin wrote:
> On śro, 2016-04-13 at 10:23 +0200, Francisco Expósito wrote:
> > Hi Freddie,
> >   
> > Thank you very much for your quick answer.
> >   
> > It seems that ST hasn’t corrected yet its EMAC driver in version
> > 1.4.4 (22-January-2016).
> >   
> > Have you managed to find an alternative driver or have you made
> > corrections to the ST one?
> >   
> > I’m a total newbie and I don’t know if I will be able to make the
> > corrections that RAc suggests in this thread…
> >   
> > Thanks again for your help!
> I'm using the driver as provided in the package with HAL for STM32F4,
> but adapted for my own RTOS ( http://distortos.org/ ). I didn't change
> much in the driver, so I guess that if the bug is there, then I will
> experience it sooner or later...
>
> Very soon I'll start some test which will involve long uptime. I'll let
> you know whether I have it too and whether I manage to fix it. If you
> find a good solution, please let me know too.
>
> Regards,
> FCh
>
> _______________________________________________
> lwip-users mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/lwip-users