Re: [Qemu-devel] [PATCH v14 11/33] target-tilegx: Framework for decoding bundles

[Richard Henderson]

On 08/29/2015 02:08 PM, Peter Maydell wrote:
> On 24 August 2015 at 17:17, Richard Henderson <address@hidden> wrote:
> > Signed-off-by: Richard Henderson <address@hidden>
> > +    if (dc->jmp.cond != TCG_COND_NEVER) {
> > +        if (dc->jmp.cond == TCG_COND_ALWAYS) {
> > +            tcg_gen_mov_i64(cpu_pc, dc->jmp.dest);
> > +        } else {
> > +            TCGv next = tcg_const_i64(dc->pc + TILEGX_BUNDLE_SIZE_IN_BYTES);
> > +            tcg_gen_movcond_i64(dc->jmp.cond, cpu_pc,
> > +                                dc->jmp.val1, load_zero(dc),
> > +                                dc->jmp.dest, next);
> > +            tcg_temp_free_i64(dc->jmp.val1);
> > +            tcg_temp_free_i64(next);
> > +        }
> > +        tcg_temp_free_i64(dc->jmp.dest);
> > +        tcg_gen_exit_tb(0);
> > +        dc->exit_tb = true;
> > +    }
>
> Doing conditional branches with movcond to pc means we'll
> never be able to link TBs which end with conditional branches,
> right?

The structure of the code is such that we could use goto_tb.
I just didn't want to complicate the initial implementation.


r~