From mboxrd@z Thu Jan  1 00:00:00 1970
From: Vitaly Kuznetsov <vkuznets@redhat.com>
Subject: RE: [PATCH v2 1/4] x86/hyper-v: move synic/stimer control structures definitions to hyperv-tlfs.h
Date: Tue, 27 Nov 2018 17:32:10 +0100
Message-ID: <87in0ijxbp.fsf@vitty.brq.redhat.com>
References: <20181126154732.23025-1-vkuznets@redhat.com> <20181126154732.23025-2-vkuznets@redhat.com> <20181126200413.GA7852@rkaganb.sw.ru> <87wooyk6na.fsf@vitty.brq.redhat.com> <CY4PR21MB0773225F986528897624BAA6D7D00@CY4PR21MB0773.namprd21.prod.outlook.com>
Mime-Version: 1.0
Content-Type: text/plain
Cc: "kvm\@vger.kernel.org" <kvm@vger.kernel.org>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Radim =?utf-8?B?S3LEjW3DocWZ?= <rkrcmar@redhat.com>,
        "linux-kernel\@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "x86\@kernel.org" <x86@kernel.org>
To: Michael Kelley <mikelley@microsoft.com>,
        Roman Kagan <rkagan@virtuozzo.com>,
        "KY Srinivasan" <kys@microsoft.com>,
        Haiyang Zhang <haiyangz@microsoft.com>,
        Stephen Hemminger <sthemmin@microsoft.com>
Return-path: <linux-kernel-owner@vger.kernel.org>
In-Reply-To: <CY4PR21MB0773225F986528897624BAA6D7D00@CY4PR21MB0773.namprd21.prod.outlook.com>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: kvm.vger.kernel.org

Out of pure curiosity I decided to check what 'gcc -O3' produces when we
use bitfields and masks. As of 'gcc version 8.2.1 20181105 (Red Hat 8.2.1-5) (GCC)'

1) bitfields:

struct abc {
	int enabled:1;
	int _pad:7;
	int vec:8;
};

int is_good(struct abc *s) {
	if (s->enabled)
		return s->vec;
	else 
		return 0;
}

results in
 
is_good:
        xorl    %eax, %eax
        testb   $1, (%rdi)
        je      .L1
        movsbl  1(%rdi), %eax
.L1:
        ret

2) masks

#include <stdint.h>

#define S_ENABLED 1
#define S_VEC_MASK 0xff00
#define S_VEC_SHIFT 8

int is_good(uint16_t *s) {
	if (*s & S_ENABLED)
		return (*s & S_VEC_MASK) >> S_VEC_SHIFT;
	else 
		return 0;
}

results in

is_good:
        movzwl  (%rdi), %edx
        movzbl  %dh, %eax
        andl    $1, %edx
        movl    $0, %edx
        cmove   %edx, %eax
        ret

so bitfields version looks somewhat more efficient. I'm not sure if my
example is too synthetic though.

-- 
Vitaly