This is the mail archive of the
mailing list for the GCC project.
Re: [tree-ssa] must-alias fixes...
On Fri, 2003-11-21 at 05:52, Jan Hubicka wrote:
> Oops, sorry, I tought I've commited the setting of 's' in parser.c as
> obvious. Will do so shortly.
> The variable really can stay uninitialized in some code paths.
> > Also, could you elaborate on the changes to find_addressable_vars, how
> > are those complex references getting in?
> We can construct such a complex referenced via folding (they are valid
> gimple operands) and we can promote them into PHI nodes during constant
Aha. OK, thanks. I'm currently testing this together with a change to
not remove TREE_ADDRESSABLE from static variables. Will commit if it
passes all tests.