From a0aaa3552d53b20a99566ac7116063fbb31b9964 Mon Sep 17 00:00:00 2001 From: xleroy Date: Wed, 3 Jul 2013 15:07:17 +0000 Subject: Treat casts int64 -> float32 as primitive operations instead of two casts int64 -> float64 -> float32. The latter causes double rounding errors. git-svn-id: https://yquem.inria.fr/compcert/svn/compcert/trunk@2290 fca1b0fc-160b-0410-b1d3-a4f43f01ea2e --- cfrontend/Cshmgenproof.v | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'cfrontend/Cshmgenproof.v') diff --git a/cfrontend/Cshmgenproof.v b/cfrontend/Cshmgenproof.v index 11f80118..3d5bbbae 100644 --- a/cfrontend/Cshmgenproof.v +++ b/cfrontend/Cshmgenproof.v @@ -172,12 +172,12 @@ Proof. Qed. Lemma make_floatoflong_correct: - forall a n sg e le m, + forall a n sg sz e le m, eval_expr ge e le m a (Vlong n) -> - eval_expr ge e le m (make_floatoflong a sg) (Vfloat(cast_long_float sg n)). + eval_expr ge e le m (make_floatoflong a sg sz) (Vfloat(cast_long_float sg sz n)). Proof. intros. unfold make_floatoflong, cast_int_long. - destruct sg; econstructor; eauto. + destruct sg; destruct sz; econstructor; eauto. Qed. Lemma make_longoffloat_correct: -- cgit