Re: [lwip-users] UDP socket thread safe problem

[Darius Babrauskas]

Hello,

 

I found, that 2  threads  A and B  used  same  systimeout. So it was not directly LWIP problem.

After correcting error, all working Ok. Closed.

 

Darius

----- Original Message -----

From: Darius Babrauskas

To: address@hidden

Sent: Thursday, February 23, 2012 11:34 AM

Subject: [lwip-users] UDP socket thread safe problem

Hello,

 

I have old project with LWIP v.1.2. for AVR32.

 

Problem:

 

2  threads  A and B . Each treads have 1 UDP socket with No( 0 and 1 ).  Each thread have self timeout struct.

 

Each socket using lwip_select function for detect receive data and timeouts. A thread using 1 second timeout, B using 10 seconds timeout.

 

 

Problem:

A or B thread  corrupting timeouts other thread.  If working one thread (or other sleeping) then all ok.  But if both tread using lwip_select, then timeouts begun work strange. Example.: B timeout  must be 10 second, but it triger after  1 second  , then 5 times 10seconds, then  9 seconds or 14 seconds. A timeouts receiving real data, but timeout  begun sometimes 0-1ms. So, data not received, because real must about after  60ms.

 

I found like post.

http://lists.gnu.org/archive/html/lwip-users/2009-01/msg00011.html   but no any answer.

 

 

http://savannah.nongnu.org/bugs/?func=detailitem&item_id=20288

"conn->callback is called from api_msg.c in the core-receive-callbacks (recv_raw/recv_udp/recv_tcp) but also from netconn_recv, when data was received by the

application. This is definitively not thread-safe. "

I understud that it old version. Porting last lwip version - possible more problems.

So.  What advise? (maybe need use in one thread raw  udp function ? )

P.S. Thread C using tcp socket with select function, but it working ok.

 

 

 

 

 

 

 

 

 

---

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