Re: [lwip-users] cs8900 lost packets bug

[Paul]

Hi Leon,

I'll track down the application / technical note for you,
If you can trigger a break point after a transmit fails
then this bug is affecting your system.

It would like to suggest that a processor/system agnostic
cs8900 driver code should be implemented.

Paul Clarke

Leon Woestenberg wrote:

> Hello Paul,
>
> thank you very much for this report.
>
>  
>
> > buffer. This results in  poor performance as there is less memory for
> > incoming packets and thus lost packets.
> >
> > (Does this sound familar to anyone?)
> >
> >    
> Our performance is poor, but that can be accounted to the microprocessor
> being used, non-preemptive round-robin applications and several other
> issues.
>
>  
>
> > The fix is described in the Crystal application notes 
> >
> >    
> Do you know which AN?
>
>  
>
> >    <<< SET THE TRANSMIT BUFFER SIZE TO ZERO TO DEALLOCATE  MEMORY
> >  TXLENGTH = 0;
> >    <<<  USE THE FORCE TRANSMIT COMMAND TO FREE THE MEMORY
> >  TXCMD = 0x01C9U;
> >
> >    
> Do other drivers (NetBSD, Linux) implement this as well?
>
> Regards,
>
> Leon.
>
>