Re: Fix more of C/fortran canonical type issues

[Richard Biener <rguenther@suse.de>]

On Mon, 8 Jun 2015, Jan Hubicka wrote:

> > On Mon, 8 Jun 2015, Richard Biener wrote:
> > 
> > > On Mon, 8 Jun 2015, Joseph Myers wrote:
> > > 
> > > > On Mon, 8 Jun 2015, Richard Biener wrote:
> > > > 
> > > > > I'm not sure the C standard mandates compatibility between
> > > > > 
> > > > > struct { int i; } and struct { unsigned i; } for purposes of TBAA.
> > > > > Joseph?
> > > > 
> > > > I don't think they are necessarily compatible for TBAA.
> > > 
> > > Ok, but as int and unsigned are reading either structs element
> > > via a pointer to int or a pointer to unsigned must be supported?
> > 
> > Yes.  The questionable case would be taking an object of one of those 
> > structure types, casting a pointer to it to point to the other structure 
> > type and then dereferencing.
> 
> Are "struct { int i; }" and "struct { unsigned i; }" compatible when one is defined
> in one unit and other in another?
> In any cae, I suppose if int and unsigned int pointers can be used interchangeably,
> we want to ignore TYPE_UNSGINED for purposes of canonical type calculation for LTO.
> So is the second variant of patch OK with a comment update that this is also required
> by C?

I think we should instead work towards eliminating the get_alias_set
langhook first.  The LTO langhook variant contains the same handling, btw,
so just inline that into get_alias_set and see what remains?

Richard.

> Honza
> > 
> > -- 
> > Joseph S. Myers
> > joseph@codesourcery.com
> 
> 

-- 
Richard Biener <rguenther@suse.de>
SUSE LINUX GmbH, GF: Felix Imendoerffer, Jane Smithard, Dilip Upmanyu, Graham Norton, HRB 21284 (AG Nuernberg)