On Sun, 10 Feb 2019 at 22:47, Richard Henderson <richard.hender...@linaro.org> wrote: > > On 2/10/19 1:40 PM, Peter Maydell wrote: > >> Actually, we already break the TB here by default. > > > > Do we? I didn't see any code (apart from the handling > > in the DAIFSet/Clear codepaths, which aren't used for TCO). > > At the start of the function: > > /* End the TB by default, chaining is ok. */ > s->base.is_jmp = DISAS_TOO_MANY;
...that's only in the v2 patchset which you hadn't posted yet :-) -- PMM
