This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Still missing some addr_expr invariant recomputations
On Tuesday 14 September 2004 00:03, Steven Bosscher wrote:
> Anyway, I've bootstrapped with this hacked patch and Dan's checking
> patch. I'll post the checking patch and the fix as a proper patch
> after cleaning it up and testing it on a few targets (ie. tomorrow
> morning).
Hmm, but this extra checking makes gfortran (and probably C with
nested functions) run into a new kind of "bug":
--------------------------------
program nested_with_addr_expr
integer, pointer :: p
integer, target :: t
p => t
if (p .ne. f()) call abort
contains
function f()
integer, pointer :: f
f => t
end function f
end
--------------------------------
./f951 assignment_1.f90 -Wall
MAIN__
f
t.f90: In function `MAIN__':
t.f90:4: error: invariant not recomputed when ADDR_EXPR changed
&FRAME.0D.471;
t.f90:4: internal compiler error: verify_stmts failed.
Rth, what's the deal here - is that checking patch from Dan just
overly pedantic, or could this be a genuine bug? I would think
the former, but I am not sure, and neither was Dan.
Gr.
Steven