This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Remove redundant checking from may_propagate_copy, allow copyprop through (void *) conversion
On Tue, 4 Sep 2007, Daniel Berlin wrote:
> On 9/4/07, Richard Guenther <rguenther@suse.de> wrote:
> > On Tue, 4 Sep 2007, Daniel Berlin wrote:
> >
> > Changing the prototype of ggc_set_mark to take a (struct typeinfo *)
> > would have fixed the ICE. But I fail to see how the formal parameter
> > type (if pointer) makes a difference at all (for the alias solution).
> >
> > We still somehow try to match *a__.1_1 to *(void *) for call parameters.
> > Why?
> You mean the derefaddr constraint?
> That's necessary in this case as well, because it may get casted later
> to some other type that, when dereferenced, contains pointers (IE a
> struct).
Maybe what I wrote was just confusing ;) The essence is that
the gcc_assert
/* Make sure that the symbol tag has the same alias set as the
pointed-to type. */
gcc_assert (tag_set == get_alias_set (tag));
doesn't make sense if you ask for the alias set of 'void'. That is,
'void' doesn't have an alias set as - err - 'void' isn't anything ;)
Of course get_alias_set happily returns a random number it choses
the first time you feed it 'void'.
So, I was trying to avoid generating a tag for 'void' as well, but
failed or did need to touch too many places.
Richard.