This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug tree-optimization/69117] [6 Regression] wrong code at -O1 -fstrict-aliasing
- From: "rguenth at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Thu, 14 Jan 2016 12:42:14 +0000
- Subject: [Bug tree-optimization/69117] [6 Regression] wrong code at -O1 -fstrict-aliasing
- Auto-submitted: auto-generated
- References: <bug-69117-4 at http dot gcc dot gnu dot org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69117
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Priority|P3 |P1
CC| |matz at gcc dot gnu.org
--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ok, making this P1 ... but I don't see a way to handle this right now - need to
think of more.
Btw, similar issues may appear with substituting names with range info
that is not valid at the point we substitute into and then simplification
simplifying the thing to a wrong result.
It might be the only good solution is to only ever substitute (aka valueize)
to available value-numbers. But this would mean the SCC VN algorithm is
flawed in itself and we should ditch it in favor of a regular RPO one.
Wonder if somebody can come up with a testcase for the range-info case ...