Hello Nathan,

> I'm still somewhat puzzled by these failures, as I haven't been able
> to reproduce them on i386 or powerpc, while using GDB interactively
> (ppc seems to have an i-cache sync issue that can foil breakpoints,
> but it only manifests when run quickly, such as inside of GDB's
> dejagnu testsuite). 

I was running it interactively in both cases. Interestingly, if I omit 
the breakpoint on main and just have the one on thread_execution in 
place then it will hit that one..

Setting another one (e.g. on the line of the sleep() call), or even 
stepping using 's' or 'stepi' will cause the program to run to 
completion. Even setting a second breakpoint on the printf() call after 
the pthread_join() in the main thread is not hit.

As before though, all of these operations work fine using the remote 
debugger, which was based on exactly the same code. I will try adding 
some debug to the main gdb code today to see what it is up to.

Rgds,
John...