On 1 March 2017 at 20:32, Alex Bennée <alex.ben...@linaro.org> wrote: > However I ran into another problem. Code generation leads to a tlb_fill > which runs afoul of a nested tb_lock(). I'm pretty sure the front-end is > using the wrong thing to fetch code:
Isn't that the expected (if admittedly kinda dumb) thing where when we fail a code load during translation it causes us to attempt a recursive translation before we finally figure out that it really isn't going to work and longjump right out to the top level? It's silly and we should probably at some point figure out a way to do it more cleanly, but it doesn't actually cause any problem pre-MTTCG... thanks -- PMM
