Re: [PATCH 1/4] target/arm: Add TB flag for "MVE insns not predicated"

[Peter Maydell]

On Tue, 7 Sept 2021 at 15:11, Richard Henderson
<richard.henderson@linaro.org> wrote:
> >   * gen_update_fp_context() -- this function gets called for pretty
> >     much every FP/MVE insn (as part of vfp_access_check), and it
> >     can in rare cases update the FPSCR.LTPSIZE and the VPR. I guess
> >     this means we really do need to end the TB
> >     if (MVE && s->v7m_new_fp_ctxt_needed) (ie the comment in
> >     gen_update_fp_context "We don't need to arrange to end the TB,
> >     because [we don't cache FPSCR in TB flags]" is no longer true).
> >     That seems likely to be painful because some of the insns that
> >     do a vfp_access_check do complicated things with the TB exits
> >     (eg WLSTP, LETP) that are going to override a naive setting of
> >     is_jmp in gen_update_fp_context()...
>
> Well, we could let gen_goto_tb see that is_jmp is already UPDATE_NOCHAIN, and 
> suppress the
> goto_tb in that case.  That would seem to take care of any entry into 
> gen_jmp(_tb).

We actually already have code that sets is_jmp (to DISAS_UPDATE_EXIT)
from gen_preserve_fp_state() -- we do that if we're using icount,
setting DISAS_UPDATE_EXIT to force this to be the last insn in the TB.

Do icount IO instructions need to avoid a possible goto_tb ?
I suppose we do want them to go back to the main loop.

This suggests that we need to make gen_goto_tb() look at
the is_jmp value anyway, regardless of this series.

-- PMM