On Mon, 20 May 2019, Richard Biener wrote:
On Sun, May 19, 2019 at 6:22 PM Marc Glisse <marc.gli...@inria.fr> wrote:
Hello,
I noticed this one with BIT_NOT_EXPR a while ago. C++ testcase because I
haven't looked at gimplefe yet...
OK. I wonder if there's any issue with -ftrapv and negate/abs of
INT_MIN containing
vectors (we don't implement -ftrapv optabs for vectors of course...).
We at least
avoid folding those cases I think and sanitizers then may see those
unconditionally?
You are right. I had thought of floating point operations (say sqrt) where
we might not fold because of the rounding mode, but the issue is more
general. I believe I should really do what I was lazily trying to avoid:
fold both operations and only perform the simplification if both returned
constants.
--
Marc Glisse