Re: line insn notes cleanups (haifa-sched and modulo-sched)

["Vladimir N. Makarov" <vmakarov at redhat dot com>]

Ian Lance Taylor wrote:

> "Vladimir N. Makarov" <vmakarov@redhat.com> writes:
>
>  
>
> > Riding off the line notes is a good idea and I really like it but they
> > have not removed yet.  As I wrote Tehila,  scheduler should be the
> > last place to remove the code dealing with the line notes because
> > without the code the line notes will misinform and confuse people at
> > least when they look at  dump files after the insn scheduler work.
> >    
>
> I've looked through the code, and I can't see any place where we use
> line notes (i.e., NOTE insns with NOTE_LINE_NUMBER (insn) > 0) to
> generate debugging information.  We use insn locators for that now.
> And the insn locator information is put into the debugging dump files.
>
> So I think this objection is moot.  If you disagree, can you explain
> further what the problem is, perhaps with an example?  Thanks.
>
>  
The problem is in that line notes is still generated.  At least that is 
what I see in dump files.  I think that generation of  line notes should 
be removed first or be a part of the patch.  If it is not done we will 
have line notes after insn scheduling completely in wrong places which 
might confuse people examining dump files.  And we will still just treat 
line notes as other notes in insn scheduling.

So I does not object the idea of locators.  Quite opposite I like it.  
But I don't see reasons for the patch until we remove generation of line 
notes in the first place.

Vlad