[lwip-users] Re: [lwip] RFC - new sys_arch for lwIP 0.5

[Adam Dunkels]

Hi all!

I have had a lot of work the week and haven't really had the time to do my 
mail appropriately, hence the delay.

On Monday 19 November 2001 12.45, you wrote:
> just to give you a little feedback, I have ported your new lwIP 0.5 to
> SMSQ/E, using a serial device with SLIP. IP and ICMP seem to work fine. TCP
> works, but only if I include a lot of *debug output* from debug.h :-(

Wierd. Probably the extra delay introduced by the debugging output makes some 
race condition occur less frequently. In what way does TCP not work when you 
run without debugging output? How are your application using TCP? (I.e., raw 
tcp_()-interface or through the sequential API?)

> As this is a new port, and includes assembler stuff for thread creation and
> timed semaphores, the problems are likely to be my own fault. Unfortunately
> there is no source-level debugger for my target, so I'll have a hard time.

:-/

> If someone has a hint, how debug output could hide TCP related problems,
> that would be very much appreciated (99% joke).

You could try to run a TCP application using the raw stack interface. The 
file proj/unixsim/apps/http_noapi.c in the latest CVS could perhaps be a good 
starting point.

> BTW should sys_arch_sem_wait(sem, 0) return the amount of time spent
> waiting for the semaphore to be signaled, or just 1, or don't care?

Don't care.

Thanks, Peter!

/adam
-- 
Adam Dunkels <address@hidden>
http://www.sics.se/~adam
[This message was sent through the lwip discussion list.]