Re: [Libjit-developers] Crashing when using jit_insn_call_indirect_vtable with JIT_TAIL_CALL

[Alexander Kellett]

On Thu, Nov 04, 2004 at 08:19:59AM +1000, Rhys Weatherley wrote:
> The crash is mainly due to a missing "jit_insn_return" in 
> "build_function_callback" in your test case.  This causes control to drop off 
> the end of the function into random garbage when compiled and executed.  It 
> should be inserted just after the call to "jit_insn_call_indirect_vtable".

ah. doh. thanks, i always get something wrong in the dang testcases :)

> The other problem is that tail calls are not completely implemented yet in 
> both the interpreter and x86 back end.  It will take a little more time to 
> implement this.  Remove the "JIT_CALL_TAIL" flag and it works fine.

yes, this will work fine for the moment in any case, 
just a slightly higher memory usage :)

> And yes, I know that you need tail calls to work, but Rome wasn't built in a 
> day, so please be patient.  It should be possible for you to make progress in 
> other areas even without tail calls.

yeah. don't worry am doing so :P
just wanted to get the bug report out as early as possible.
sorry for the lack of patience :/

thanks *very* much for the response,
Alex