Debugging lwIP

From lwIP Wiki

There are a couple of different methods typically used for debugging lwIP, printf-style and external debugging.

[edit] External Debugger

Probably the most familiar method is to fire up a debugger and start single-stepping through the code. However, since lwIP is typically used on embedded targets you can't just light up a debugger and start debugging. There has to be some way to communicate with the embedded target. Under normal circumstances there is a network stack running and gdbserver, or some other debug server daemon, running that enables the debugger to talk to the target debug daemon.

Since the network stack itself is being debugged, the debugger obviously won't be able to use it to talk to the target, so some other route needs to be used. There are a couple of different ways to get around this. Probably the best, in terms of performance, is to run either another stable instance of lwIP, or another network stack, next to the lwIP stack under development. This can be convoluted to setup, depending on the particular architecture of the target, but is doable and is useful for working through problems when doing the initial port. Once the stack is up and running, the debugger is less useful since most of the problems you will run into will be induced by interdependencies with other external components, and there will usually be a time related component to the problem, which eliminates the debugger as a useful tool for tracking the problem down. That's where we get old-school with the built-in printf style debugging.

[edit] printf/built-in debugging

lwIP is developed with some nice support for enabling various debug print options. The debug options are enabled at build time with the use of various xxx_DEBUG options, which are enumerated in the opts.h file. I like to TBD