This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug c++/51313] [4.7 Regression][C++0x] ICE: tree check: expected class 'constant', have 'unary' (nop_expr) in null_ptr_cst_p, at cp/call.c:556
- From: "paolo.carlini at oracle dot com" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Fri, 02 Dec 2011 17:23:24 +0000
- Subject: [Bug c++/51313] [4.7 Regression][C++0x] ICE: tree check: expected class 'constant', have 'unary' (nop_expr) in null_ptr_cst_p, at cp/call.c:556
- Auto-submitted: auto-generated
- References: <bug-51313-4@http.gcc.gnu.org/bugzilla/>
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51313
--- Comment #1 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-12-02 17:23:24 UTC ---
This badly requires a reduced testcase, I'm going to prepare one. Then the fix
should be pretty simple: just use STRIP_NOPs unconditionally, ie cxx0x too, in
null_ptr_cst_p?! Anything wrong with that?
The point is that integer_zerop strips the NOPs before doing its work, but then
TREE_OVERFLOW cannot handle a NOP_EXPR. I don't think we want TREE_OVERFLOW_P
and return false when integer_zerop returns true and no overflow is occurring!