This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [tree-ssa]: Dominator opts change LHS without changing VDEF's

On Sun, 10 Aug 2003, Diego Novillo wrote:

> On Sun, 2003-08-10 at 17:17, Daniel Berlin wrote:
> > > > This causes the vdef's to changes, because the memory tag for elt isn't
> > > > the same as elt.
> > > >
> > > They aren't?
> > No.
> >
> But, why?  This only occurs with PTA,

> why aren't those variables considered aliased?
Because they aren't?

> They do seem aliased to me.  If they are aliased,
> they will get the same memory tag.  In this case they don't get the same
> memory tag because alias analysis tells us that they aren't aliased, I
> want to know why alias analysis doesn't consider them aliased.
> The very fact that we can copy propagate one into the other means that
> they are aliased.  If alias analysis is saying that they can't alias,
> then alias analysis is wrong.
> bitmap_element_free (struct bitmap_element *elt)
> {
> ....
>   {
>      struct bitmap_element * elt;
>      struct bitmap_head_def * head;
>      head = head;
>      elt = elt;			<-- Here 'elt' points to 'elt'.

No, this statement means elt (lhs) points to whatever elt (rhs) points to.
elt doesn't point to elt.
That would be *elt = elt
In other words, it means that elt has to point to whatever elt points to.
Not that elt points to elt.
elt *doesn't* point to elt.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]