Re: [lwip-users] upgrading 1.3.2 to CVS

[yueyue papa]

I also find the this big change recently. I think the change better than before.  I have not tried to use the latest sys_arch solution.

 

I think it is better to write a sys_arch update guild.  I found

 

1> One structure is changed from *

2> API paramter number is changed, due to this structure define change.

3> Timer out solution changed

 

Lee

On Sat, May 22, 2010 at 2:29 PM, address@hidden <address@hidden> wrote:

Your assumptions are correct: application threads may not use sys_timeout any more (and I hope they didn't, before). The timeout implementation was just broken in a few ways: the biggest bug was that timeouts fired nearly everywhere - you just had to pend on a semaphore or mbox. Now the internal timeouts are handled at a defined state in tcpip_thread (or for NO_SYS==1 when calling the timer-check-function).

Simon

Tyrel Newton wrote:

I just upgraded my application from using 1.3.2 to the latest CVS head.
The biggest change for me was reworking the majority of my
FreeRTOS-based sys_arch layer. The one thing I'm still a little fuzzy on
is the new timer implementation. Previously, I had to store a pointer to
the head sys_timeout structure (or whatever it was called) within the
thread control structure, since previously "timeouts were handled on a
per-thread basis". To me, this means that application-level threads
should no longer use the lwIP-provided timeout mechanisms, but rather
the mechanisms provided directly by the underlying port layer.

A little clarification (or at least expansion on the explanations in the
changelog) on how timeouts are handled in the latest version would be
appreciated.

Thanks,
Tyrel

_______________________________________________
lwip-users mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/lwip-users

 

_______________________________________________
lwip-users mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/lwip-users