Re: [PATCH] ppc: Fix 'calli' when floating-point arguments are passed

[Paulo César Pereira de Andrade]

Em qua., 7 de set. de 2022 às 09:47, Paul Cercueil
<paul@crapouillou.net> escreveu:

[snip]

> >   Just to better understand, how is the:
> >
> >     jmpi 0x20000d0c
> >
> > generated? Just a plain jit_jmpi(ADDR); where ADDR is in a previously
> > generated jit?
>
> Yes that's correct.
>
> >   This must be supported for obvious reasons, but should avoid this
> > pattern as much as possible, as it basically breaks all the logic to
> > keep the live state of registers.
>
> Well, what's my alternative? I have to exit trampolines by jumping to
> an address, and the address I'm jumping to is in a different _jit
> context.

  That is one of the obvious reasons :) It is like, as if in C you were able
to 'goto' to anywhere in the code, including other libraries, not even
needing to 'declare' a label. But it is useful enough for jit generation that
it must be supported.

> -Paul

Thanks,
Paulo