Re: Two entry points for a function

[Paul Cercueil]

Hi Marc,

I don't think you can have multiple entry points of a function (or 
maybe with jit_indirect()?) but you can cheese things a bit with a 
trampoline.

Here's what I do, more or less:

----

jit_prolog();
jit_tramp(256);

for (i = 0; i < nb_entry_points; i++) {
   jit_addi(JIT_R0, 1);
   jit_align(8);
}

jit_epilog();
jit_prolog();

/* Read JIT_R0, its value give us what entry point was taken */

jit_epilog();

----

In that example my JIT_R0 is always 0 on entry. The jit_align() makes 
sure that each one of your entry points is aligned to 8 bytes. There is 
still some decoding needed (I read a C function pointer from a table at 
the index pointed by JIT_R0), but the decoding is done in that function 
and not in the callers.

Cheers,
-Paul

Le lun., août 8 2022 at 22:16:28 +0200, Marc Nieper-Wißkirchen 
<marc.nieper+gnu@gmail.com> a écrit :
> I am looking for a way to implement a function (pointer) with GNU 
> lightning that has two entry points.
>
> In native assembly, I would code something like this:
>
>         .align 16
> entry1: jmp L0
>         .align 16
> entry2: ...
>         ...
> L0:     ...
>
> The function would be represented by a pointer to entry1.  If I 
> wanted to jump to the second entry point, I would just add 16 to the 
> pointer.
>
> How can a similar thing be achieved with GNU lightning?  So far, the 
> best that has come to my mind is to use one more indirection.  The 
> "function pointer" points to a structure having the addresses of 
> entry1 and entry2 as fields.  (Let us assume we are in a trampoline 
> so that the stack is already set up).  But this is less efficient due 
> to the extra indirection.
>
> Is there a way to emulate the above native code with GNU lightning?  
> And, if not, what would have to be added?
>
> Marc