Re: [lwip-users] Stability of master branch code

[Robert Deschambault]

I have managed to catch error stat being logged in memp.c.  It is a MEMP_TCPIP_MSG_INPKT type.  I don't know if this is because I introduced a breakpoint in the call to memp_malloc() and was able to get the code to break on the error.  Before I did this, I had a breakpoint on the error only, and it never tripped for over a day.

On Wed, May 6, 2015 at 12:50 PM, Robert Deschambault <address@hidden> wrote:

Ok, the code is no longer responsive, but I don't have a way to check the lwip_stats counters because I can't stop the program.  I have breakpoints on the MEM_STATS_INC(err); and MEMP_STATS_INC(err, type); but the software isn't going there.  Where would be a good place to put a breakpoint so I can observe the stats?

On Tue, May 5, 2015 at 2:19 PM, address@hidden <address@hidden> wrote:

Robert Deschambault wrote:

I have enabled LWIP_STATS and LWIP_STATS_DISPLAY. Should I enable things like PBUF_DEBUG?  Will I see any messages on my debug printf display?  Or should I put a breakpoint where lwip stats error number gets incremented?

As a start, I would not change anything but enable LWIP_STATS to prevent changing runtime behaviour. Then you can check the stats counters from the debugger once it hangs.

Only if you really find there are errors reported in the stats it would make sense to dig further in that direction as it could well be that changing the runtime behaviour could cover your problem.

Simon

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

--

Bob Deschambault

--

Bob Deschambault