Re: [lwip-users] PBUF Leak Problems...

[Jonathan Larmour]

Thomas Catalino wrote:
> We are using the sockets API. I agree, there could be an issue with the 
> application not reading the data out of the stack. However, the 
> application is complex and we did not write it, so I'll probably have to 
> investigate this from the bottom up - hopefully I can poke around the 
> stack to find out which sockets are backed up with data waiting to be read.
>
> Is there an easy recipe to locate a socket in the stack that is backed 
> up? A few pointers might be very helpful, we can probably follow our 
> noses from there...

If you're using just the sockets API, you could look at the "sockets" 
array, declared in sockets.c as:

static struct lwip_socket sockets[NUM_SOCKETS];

where struct lwip_socket is:

struct lwip_socket {
  struct netconn *conn;
  struct netbuf *lastdata;
  u16_t lastoffset;
  u16_t rcvevent;
  u16_t sendevent;
  u16_t  flags;
  int err;
};

That gives you a route to finding out your connections. The 'conn' field is 
NULL for unused entries. You could use rcvevent to know whether a receive 
event occured, or more helpfully, look at conn->recvmbox. Each mbox entry 
(in whatever form your OS stores it in the mbox) is a pbuf (for TCP sockets).

Jifl
--
eCosCentric Limited      http://www.eCosCentric.com/     The eCos experts
Barnwell House, Barnwell Drive, Cambridge, UK.       Tel: +44 1223 245571
Registered in England and Wales: Reg No 4422071.
 >>>> Visit us on stand 810 at The Embedded Systems Show 2007, NEC <<<<
 >>>> Oct 17-18 Birmingham, UK http://www.edaexhibitions.com/ess/  <<<<
------["The best things in life aren't things."]------      Opinions==mine