> So at a high level, we have the cache, that's per function, right?  I just 
> want to make sure we don't have to deal with whether or not the cache needs 
> to be invalidated because a function has an optimize attribute that perhaps 
> changes vector configuration info.  While I don't immediately see a problem 
> with that, it's a slight worry.

Yeah, we reset in ira per function.  But I just realized that I should
rather put it in lra's init.  Doing that in v2.

> By attaching the filter the pseudo we effectively are incorporating the 
> filter aspect on a global basis for that pseudo.  Do we have concerns 
> that the filters accumulated through the various use points could 
> ultimately result in a pseudo that we can't allocate to a hard 
> register?  I guess that's always been a concern and we're just exposing 
> it in a different way with your patches in this space.

Definitely a concern but my hope is that if we hit that case it's rather a real 
bug in defining constraints, rather than due to the way we allocate.  Just a 
hope, though, and no definite proof.

> Do you need to manage free-ing dependent_filter_entries?  Or is that hidden 
> in the bowels of the hash table template code?

Yep, I outsourced this to

  : free_ptr_hash <dependent_filter_entry>


-- 
Regards
 Robin

Reply via email to