[tree-ssa] Insert on edge comment

Jan Hubicka jh@suse.cz
Sun Jun 1 19:24:00 GMT 2003


> In message <3ED7E4C7.69416EBC@redhat.com>, "Vladimir N. Makarov" writes:
>  >I am ignorant too (never looked at ssa-tree branch).  As I understand the
>  >problem is in phi nodes which are based on dominator frontier analysis.
>  >Insertion of new BB is simple for general IR but for SSA it can significantly
>  >change dominator frontiers for many blocks and as consequence phi-nodes in
>  >the dominator frontiers. So updating phi-nodes is not a trivial task.
> That's certainly true as well.  If you look at how we do this at the
> RTL level, you'll find that we don't split edges while in SSA form.  
> Instead we queue things (such as edge insertions) and split the edge
> after we no longer need dominance info and PHI nodes.

This does not appear to work if you want to have number of SSA based
optimizers in queue.  In general there is CFG related information that
can not be updated at low level (dominance info, profile on RTL case)
and it probably should be left completely on the higher level code that
usually know what is going on.

It is also very error prone, is it easy to write code that verifies
whether SSA graph is in the consistent form?
Will look into this deeper tomorrow after returning from Canada.

Honza
> 
> jeff



More information about the Gcc mailing list