Kieran Mansley wrote:
On Tue, 2010-04-13 at 22:45 +0800, edgar wrote:
The problem just is that: if tcpip_thread received message very fast,
waitting case would not occur while fetching from mbox,
then expire would be enlarged,
OK, I think I see what you're saying - we might not process timeout
handlers appropriately.
Simon: sys_mbox_post() has gone from CVS head I think. Has this problem
gone too, or just moved into the ports? Edgar: it would help if you
could update to the current CVS code and see if your problem is already
solved as this code has changed since the last release.
This is what
task #7244
(Add sys_arch_ticks/jiffies in sys_arch) is about. It has been fixed
for NO_SYS timeouts, but not yet for the tcpip_thread. I hope to fix it
for 1.4.0 (which I'd like to see out in the next months or so, but
that's a different story).
tcp_slowtmr called every 500ms is not true any more.
That shouldn't happen. How are you arranging to call tcp_tmr()?
Well, that could indeed happen, but only if packets are arriving so
fast that the task never waits. This shouldn't happen often, so while
the timers might be delayed a little, they are still getting called,
just with a jitter.
Simon
|