Re: [RFC PATCH v1 05/43] tcg: Add helper function dispatcher and hook tcg_gen_callN

[Anton Johansson]

On 22/11/24, Richard Henderson wrote:
> On 11/20/24 19:49, Anton Johansson wrote:
> > Adds a function pointer to the TCGContext which may be set by targets via
> > the TARGET_HELPER_DISPATCHER macro.  The dispatcher is function
> > 
> >    (void *func, TCGTemp *ret, int nargs, TCGTemp **args) -> bool
> > 
> > which allows targets to hook the generation of helper calls in TCG and
> > take over translation.  Specifically, this will be used by helper-to-tcg
> > to replace helper function translation, without having to modify frontends.
> > 
> > Signed-off-by: Anton Johansson <anjo@rev.ng>
> > ---
> >   accel/tcg/translate-all.c | 4 ++++
> >   include/tcg/tcg.h         | 4 ++++
> >   tcg/tcg.c                 | 5 +++++
> >   3 files changed, 13 insertions(+)
> 
> I guess I'll have to read further to understand this, but my first reaction
> is: why would we not modify how the gen_helper_* functions are defined
> instead?

Hmm this might be a better idea, and we could call the generated could
directly without having to go through a massive switch statement.  What
I have in mind is something like

    #if !glue(OVERRIDE_HELPER_, name) \
    #define DEF_HELPER_FLAGS_1(name, flags, ret, t1)                        \
    extern TCGHelperInfo glue(helper_info_, name);                          \
    static inline void glue(gen_helper_, name)(dh_retvar_decl(ret)          \
        dh_arg_decl(t1, 1))                                                 \
    {                                                                       \
        tcg_gen_call1(glue(helper_info_,name).func,                         \
                      &glue(helper_info_,name), dh_retvar(ret),             \
                      dh_arg(t1, 1));                                       \
    }                                                                       \
    #endif

and we could emit gen_helper_* for helpers which are translated and
redefine OVERRIDE_HELPER_* to 1 (would have to be defaulted to 0
somewhere else).

//Anton