[LLVMdev] tail call optimization

[Evan Cheng]

llvm already know if a call is a tail call. There are some backend  
codegen work that needs to be done. It's on my todo list. Just  
haven't gotten to it yet...

Evan
On Jul 15, 2006, at 4:44 PM, Rafael Espíndola wrote:

> compiling:
>
> void f(void);
> void g(void) {
>  f();
> }
>
> With GCC 4.0.3 (-Os) produces the code
>
>        pushl   %ebp
>        movl    %esp, %ebp
>        popl    %ebp
>        jmp     f
>
> With llvm-gcc the result is
>
>        subl $4, %esp
>        call f
>        addl $4, %esp
>        ret
>
> Note that llvm-gcc automatically removes the frame pointer, but
> doesn't use a tail call as GCC does. LLVM has support for a tail
> calling convention. Is it missing a pass that tries to "promote" a
> regular call to a tail call?
>
> Thanks,
> Rafael
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev