From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753191AbcELO04 (ORCPT ); Thu, 12 May 2016 10:26:56 -0400 Received: from foss.arm.com ([217.140.101.70]:59226 "EHLO foss.arm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751901AbcELO0y (ORCPT ); Thu, 12 May 2016 10:26:54 -0400 From: Robin Murphy To: will.deacon@arm.com, catalin.marinas@arm.com, luke.starrett@broadcom.com Cc: linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org, bcm-kernel-feedback-list@broadcom.com Subject: [PATCH] arm64: Implement optimised IP checksum helpers Date: Thu, 12 May 2016 15:26:48 +0100 Message-Id: <6ca77904d1eced60d86145dc5da361f3aefc6fec.1463062758.git.robin.murphy@arm.com> X-Mailer: git-send-email 2.8.1.dirty Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org AArch64 is capable of 128-bit memory accesses without alignment restrictions, which makes it both possible and highly practical to slurp up a typical 20-byte IP header in just 2 loads. Implement our own version of ip_fast_checksum() to take advantage of that, resulting in considerably fewer instructions and memory accesses than the generic version. We can also get more optimal code generation for csum_fold() by defining it a slightly different way round from the generic version, so throw that into the mix too. Suggested-by: Luke Starrett Signed-off-by: Robin Murphy --- Having spent an evening poring over disassembler output, it turns out that there's really no need to drop into asm for this, hurrah! I've taken it for a spin on both big and little-endian, with no ill effects. arch/arm64/include/asm/checksum.h | 49 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 49 insertions(+) create mode 100644 arch/arm64/include/asm/checksum.h diff --git a/arch/arm64/include/asm/checksum.h b/arch/arm64/include/asm/checksum.h new file mode 100644 index 000000000000..7e975ffce837 --- /dev/null +++ b/arch/arm64/include/asm/checksum.h @@ -0,0 +1,49 @@ +/* + * Copyright (C) 2016 ARM Ltd. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ +#ifndef __ASM_CHECKSUM_H +#define __ASM_CHECKSUM_H + +static inline __sum16 csum_fold(__wsum csum) +{ + u32 sum = (__force u32)csum; + sum += (sum >> 16) | (sum << 16); + return ~(__force __sum16)(sum >> 16); +} +#define csum_fold csum_fold + +static inline __sum16 ip_fast_csum(const void *iph, unsigned int ihl) +{ + __uint128_t tmp; + u64 sum; + + tmp = *(const __uint128_t *)iph; + iph += 16; + ihl -= 4; + tmp += ((tmp >> 64) | (tmp << 64)); + sum = tmp >> 64; + do { + sum += *(const u32 *)iph; + iph += 4; + } while (--ihl); + + sum += ((sum >> 32) | (sum << 32)); + return csum_fold(sum >> 32); +} +#define ip_fast_csum ip_fast_csum + +#include + +#endif /* __ASM_CHECKSUM_H */ -- 2.8.1.dirty