From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-help-return-54666-listarch-gcc-help=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 30018 invoked by alias); 14 Jan 2014 14:53:18 -0000
Mailing-List: contact gcc-help-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-help.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-help/>
List-Post: <mailto:gcc-help@gcc.gnu.org>
List-Help: <mailto:gcc-help-help@gcc.gnu.org>
Sender: gcc-help-owner@gcc.gnu.org
Received: (qmail 30009 invoked by uid 89); 14 Jan 2014 14:53:17 -0000
Authentication-Results: sourceware.org; auth=none
X-Virus-Found: No
X-Spam-SWARE-Status: No, score=-2.6 required=5.0 tests=AWL,BAYES_00,FREEMAIL_FROM,RCVD_IN_DNSWL_LOW,SPF_PASS autolearn=ham version=3.3.2
X-HELO: mail-we0-f180.google.com
Received: from mail-we0-f180.google.com (HELO mail-we0-f180.google.com) (74.125.82.180) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with (AES128-SHA encrypted) ESMTPS; Tue, 14 Jan 2014 14:53:16 +0000
Received: by mail-we0-f180.google.com with SMTP id q59so483687wes.39        for <gcc-help@gcc.gnu.org>; Tue, 14 Jan 2014 06:53:13 -0800 (PST)
MIME-Version: 1.0
X-Received: by 10.194.6.42 with SMTP id x10mr27651489wjx.17.1389711193344; Tue, 14 Jan 2014 06:53:13 -0800 (PST)
Received: by 10.216.199.4 with HTTP; Tue, 14 Jan 2014 06:53:13 -0800 (PST)
Date: Tue, 14 Jan 2014 14:53:00 -0000
Message-ID: <CANd1uZmNiFe_vBqKVu+jqp+Y-kgU5QDkWBisDh5iYHB6tdTfzg@mail.gmail.com>
Subject: why does gcc need -fno-trapping-math to generate roundsd instructions?
From: Jay Foad <jay.foad@gmail.com>
To: gcc-help@gcc.gnu.org
Content-Type: text/plain; charset=UTF-8
X-SW-Source: 2014-01/txt/msg00051.txt.bz2

I'm using GCC 4.8.1 with eglibc 2.17 on an Intel CPU with SSE 4.1.

Given this source:

#include <math.h>
double f(double x){return floor x;}

gcc -msse4.1 -O -S generates:

f:
subq $8, %rsp
call floor
addq $8, %rsp
ret

It seems I need to add -fno-trapping-math to get inline code:

f:
roundsd $1, %xmm0, %xmm0
ret

But why? If I run the first version under a debugger, the call to
"floor" eventually resolves to eglibc's __floor_sse41, which just does
this:

roundsd $0x1,%xmm0,%xmm0
retq

So what is -ftrapping-math really buying me? Why can't the compiler
generate roundsd inline in all cases?

Thanks,
Jay.