| Message ID | 1417639668-23500-2-git-send-email-thomas.monjalon@6wind.com (mailing list archive) |
|---|---|
| State | Accepted, archived |
| Headers |
Return-Path: <dev-bounces@dpdk.org> X-Original-To: patchwork@dpdk.org Delivered-To: patchwork@dpdk.org Received: from [92.243.14.124] (localhost [IPv6:::1]) by dpdk.org (Postfix) with ESMTP id 29BE7803D; Wed, 3 Dec 2014 21:49:03 +0100 (CET) Received: from mail-wi0-f180.google.com (mail-wi0-f180.google.com [209.85.212.180]) by dpdk.org (Postfix) with ESMTP id F37F6559C for <dev@dpdk.org>; Wed, 3 Dec 2014 21:49:00 +0100 (CET) Received: by mail-wi0-f180.google.com with SMTP id n3so25758531wiv.7 for <dev@dpdk.org>; Wed, 03 Dec 2014 12:49:00 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references; bh=CSp5GWtWkM0EdbDGUbroVvPm+YUvYbS1z/u7ndnUlMo=; b=FesX3PIQUb4hNLuppSZM0ogcQb45B7wVpaYpbNi0Bi/CKTr+tbio+dvIAqQt6Pt9lW M7O/MhvoyqFuPO9jPqhewn66swJNUYJOrtO7XP5Uvh2EDqNcSI87XTYXnHhHtVVadCST zJMLQRPsfT/2UTQnV9blmne95/g9crQJMi6VaY/L6kfxe75z3yWn0l0J8Zs6pho0b6LV kFQI3oHAFjPWInkp56rU//peqbFclKVd+5YkJTR0eFcIk2YD9rhsrE/6I4kv05wfDj8X 9KPIb4D944UXHCs/9d9aNzpd0BfY9GCeHmxaAUFpnKlJsl50vp/Mq8wNF18yzGH7Ufgv rPWA== X-Gm-Message-State: ALoCoQn+ju1acndetFo56lEJAXvjyoVoRNSh1nglLXWFUDkPvJfPypKSR/OuyWyuInt/rMK4OYIF X-Received: by 10.180.84.198 with SMTP id b6mr105069779wiz.41.1417639740869; Wed, 03 Dec 2014 12:49:00 -0800 (PST) Received: from XPS13.localdomain (188.20.90.92.rev.sfr.net. [92.90.20.188]) by mx.google.com with ESMTPSA id hz9sm37761878wjb.17.2014.12.03.12.48.55 for <multiple recipients> (version=TLSv1.2 cipher=ECDHE-RSA-AES128-SHA bits=128/128); Wed, 03 Dec 2014 12:49:00 -0800 (PST) From: Thomas Monjalon <thomas.monjalon@6wind.com> To: dev@dpdk.org Date: Wed, 3 Dec 2014 21:47:47 +0100 Message-Id: <1417639668-23500-2-git-send-email-thomas.monjalon@6wind.com> X-Mailer: git-send-email 2.1.3 In-Reply-To: <1417639668-23500-1-git-send-email-thomas.monjalon@6wind.com> References: <283531301.lWbIahXLyM@xps13> <1417639668-23500-1-git-send-email-thomas.monjalon@6wind.com> Subject: [dpdk-dev] [PATCH 1/2] eal: detect endianness X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK <dev.dpdk.org> List-Unsubscribe: <http://dpdk.org/ml/options/dev>, <mailto:dev-request@dpdk.org?subject=unsubscribe> List-Archive: <http://dpdk.org/ml/archives/dev/> List-Post: <mailto:dev@dpdk.org> List-Help: <mailto:dev-request@dpdk.org?subject=help> List-Subscribe: <http://dpdk.org/ml/listinfo/dev>, <mailto:dev-request@dpdk.org?subject=subscribe> Errors-To: dev-bounces@dpdk.org Sender: "dev" <dev-bounces@dpdk.org> |
Commit Message
Thomas Monjalon
Dec. 3, 2014, 8:47 p.m. UTC
There is no standard to check endianness.
So we need to try different checks.
Previous trials were done in testpmd (see commits
51f694dd40f56 and 64741f237cf29) without full success.
This one is not guaranteed to work everywhere so it could
evolve when exceptions are found.
If endianness is not detected, there is a fallback on x86
to little endian. It could be forced before doing detection
but it would add some arch-dependent code in the generic header.
The option CONFIG_RTE_ARCH_BIG_ENDIAN introduced for IBM Power only
(commit a982ec81d84d53) can be removed. A compile-time check is better.
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>
---
config/defconfig_ppc_64-power8-linuxapp-gcc | 1 -
.../common/include/arch/ppc_64/rte_byteorder.h | 4 ++--
.../common/include/arch/x86/rte_byteorder.h | 4 ++++
.../common/include/generic/rte_byteorder.h | 28 ++++++++++++++++++++++
4 files changed, 34 insertions(+), 3 deletions(-)
Comments
On 12/4/2014 5:26 AM, Thomas Monjalon wrote: > There is no standard to check endianness. > So we need to try different checks. > Previous trials were done in testpmd (see commits > 51f694dd40f56 and 64741f237cf29) without full success. > This one is not guaranteed to work everywhere so it could > evolve when exceptions are found. > > If endianness is not detected, there is a fallback on x86 > to little endian. It could be forced before doing detection > but it would add some arch-dependent code in the generic header. > > The option CONFIG_RTE_ARCH_BIG_ENDIAN introduced for IBM Power only > (commit a982ec81d84d53) can be removed. A compile-time check is better. > > Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com> > --- > config/defconfig_ppc_64-power8-linuxapp-gcc | 1 - > .../common/include/arch/ppc_64/rte_byteorder.h | 4 ++-- > .../common/include/arch/x86/rte_byteorder.h | 4 ++++ > .../common/include/generic/rte_byteorder.h | 28 ++++++++++++++++++++++ > 4 files changed, 34 insertions(+), 3 deletions(-) > > diff --git a/config/defconfig_ppc_64-power8-linuxapp-gcc b/config/defconfig_ppc_64-power8-linuxapp-gcc > index 48018c3..d97a885 100644 > --- a/config/defconfig_ppc_64-power8-linuxapp-gcc > +++ b/config/defconfig_ppc_64-power8-linuxapp-gcc > @@ -34,7 +34,6 @@ CONFIG_RTE_MACHINE="power8" > > CONFIG_RTE_ARCH="ppc_64" > CONFIG_RTE_ARCH_PPC_64=y > -CONFIG_RTE_ARCH_BIG_ENDIAN=y > CONFIG_RTE_ARCH_64=y > > CONFIG_RTE_TOOLCHAIN="gcc" > diff --git a/lib/librte_eal/common/include/arch/ppc_64/rte_byteorder.h b/lib/librte_eal/common/include/arch/ppc_64/rte_byteorder.h > index 1a89051..80436f2 100644 > --- a/lib/librte_eal/common/include/arch/ppc_64/rte_byteorder.h > +++ b/lib/librte_eal/common/include/arch/ppc_64/rte_byteorder.h > @@ -105,7 +105,7 @@ static inline uint64_t rte_arch_bswap64(uint64_t _x) > /* Power 8 have both little endian and big endian mode > * Power 7 only support big endian > */ > -#ifndef RTE_ARCH_BIG_ENDIAN > +#if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN > > #define rte_cpu_to_le_16(x) (x) > #define rte_cpu_to_le_32(x) (x) > @@ -123,7 +123,7 @@ static inline uint64_t rte_arch_bswap64(uint64_t _x) > #define rte_be_to_cpu_32(x) rte_bswap32(x) > #define rte_be_to_cpu_64(x) rte_bswap64(x) > > -#else > +#else /* RTE_BIG_ENDIAN */ > > #define rte_cpu_to_le_16(x) rte_bswap16(x) > #define rte_cpu_to_le_32(x) rte_bswap32(x) > diff --git a/lib/librte_eal/common/include/arch/x86/rte_byteorder.h b/lib/librte_eal/common/include/arch/x86/rte_byteorder.h > index 1aa6985..ffdb6ef 100644 > --- a/lib/librte_eal/common/include/arch/x86/rte_byteorder.h > +++ b/lib/librte_eal/common/include/arch/x86/rte_byteorder.h > @@ -40,6 +40,10 @@ extern "C" { > > #include "generic/rte_byteorder.h" > > +#ifndef RTE_BYTE_ORDER > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > +#endif > + > /* > * An architecture-optimized byte swap for a 16-bit value. > * > diff --git a/lib/librte_eal/common/include/generic/rte_byteorder.h b/lib/librte_eal/common/include/generic/rte_byteorder.h > index 9358136..ea23fdf 100644 > --- a/lib/librte_eal/common/include/generic/rte_byteorder.h > +++ b/lib/librte_eal/common/include/generic/rte_byteorder.h > @@ -44,6 +44,34 @@ > */ > > #include <stdint.h> > +#ifdef RTE_EXEC_ENV_BSDAPP > +#include <sys/endian.h> > +#else > +#include <endian.h> > +#endif > + > +/* > + * Compile-time endianness detection > + */ > +#define RTE_BIG_ENDIAN 1 > +#define RTE_LITTLE_ENDIAN 2 > +#if defined __BYTE_ORDER > +#if __BYTE_ORDER == __BIG_ENDIAN > +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > +#elif __BYTE_ORDER == __LITTLE_ENDIAN > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > +#endif /* __BYTE_ORDER */ > +#elif defined __BYTE_ORDER__ > +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ > +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > +#endif /* __BYTE_ORDER__ */ > +#elif defined __BIG_ENDIAN__ > +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > +#elif defined __LITTLE_ENDIAN__ > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > +#endif What do you think about : +/* + * Compile-time endianness detection + */ +#define RTE_BIG_ENDIAN 1 +#define RTE_LITTLE_ENDIAN 2 +if defined __BYTE_ORDER__ /* Prefer gcc build-in macros */ +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN +#endif /* __BYTE_ORDER__ */ +#else +#if defined RTE_EXEC_ENV_BSDAPP +#include <sys/endian.h> +#else +#include <endian.h> +#endif +#if defined __BYTE_ORDER +#if __BYTE_ORDER == __BIG_ENDIAN +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN +#elif __BYTE_ORDER == __LITTLE_ENDIAN +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN +#endif /* __BYTE_ORDER */ +#elif defined __BIG_ENDIAN__ +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN +#elif defined __LITTLE_ENDIAN__ +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN +#endif +#endif > > /* > * An internal function to swap bytes in a 16-bit value.
2014-12-04 02:28, Qiu, Michael: > On 12/4/2014 5:26 AM, Thomas Monjalon wrote: > > There is no standard to check endianness. > > So we need to try different checks. > > Previous trials were done in testpmd (see commits > > 51f694dd40f56 and 64741f237cf29) without full success. > > This one is not guaranteed to work everywhere so it could > > evolve when exceptions are found. [...] > > #include <stdint.h> > > +#ifdef RTE_EXEC_ENV_BSDAPP > > +#include <sys/endian.h> > > +#else > > +#include <endian.h> > > +#endif > > + > > +/* > > + * Compile-time endianness detection > > + */ > > +#define RTE_BIG_ENDIAN 1 > > +#define RTE_LITTLE_ENDIAN 2 > > +#if defined __BYTE_ORDER > > +#if __BYTE_ORDER == __BIG_ENDIAN > > +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > > +#elif __BYTE_ORDER == __LITTLE_ENDIAN > > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > > +#endif /* __BYTE_ORDER */ > > +#elif defined __BYTE_ORDER__ > > +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ > > +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > > +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ > > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > > +#endif /* __BYTE_ORDER__ */ > > +#elif defined __BIG_ENDIAN__ > > +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > > +#elif defined __LITTLE_ENDIAN__ > > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > > +#endif > > What do you think about : > > +/* > + * Compile-time endianness detection > + */ > +#define RTE_BIG_ENDIAN 1 > +#define RTE_LITTLE_ENDIAN 2 > +if defined __BYTE_ORDER__ /* Prefer gcc build-in macros */ > +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ > +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > +#endif /* __BYTE_ORDER__ */ > +#else > +#if defined RTE_EXEC_ENV_BSDAPP > +#include <sys/endian.h> > +#else > +#include <endian.h> > +#endif > +#if defined __BYTE_ORDER > +#if __BYTE_ORDER == __BIG_ENDIAN > +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > +#elif __BYTE_ORDER == __LITTLE_ENDIAN > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > +#endif /* __BYTE_ORDER */ > +#elif defined __BIG_ENDIAN__ > +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > +#elif defined __LITTLE_ENDIAN__ > +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > +#endif > +#endif Please, could you give more explanations about your proposal? Why not always try to include endian.h? Why giving high priority to __BYTE_ORDER__?
On 12/4/2014 5:01 PM, Thomas Monjalon wrote: > 2014-12-04 02:28, Qiu, Michael: >> On 12/4/2014 5:26 AM, Thomas Monjalon wrote: >>> There is no standard to check endianness. >>> So we need to try different checks. >>> Previous trials were done in testpmd (see commits >>> 51f694dd40f56 and 64741f237cf29) without full success. >>> This one is not guaranteed to work everywhere so it could >>> evolve when exceptions are found. > [...] >>> #include <stdint.h> >>> +#ifdef RTE_EXEC_ENV_BSDAPP >>> +#include <sys/endian.h> >>> +#else >>> +#include <endian.h> >>> +#endif >>> + >>> +/* >>> + * Compile-time endianness detection >>> + */ >>> +#define RTE_BIG_ENDIAN 1 >>> +#define RTE_LITTLE_ENDIAN 2 >>> +#if defined __BYTE_ORDER >>> +#if __BYTE_ORDER == __BIG_ENDIAN >>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >>> +#elif __BYTE_ORDER == __LITTLE_ENDIAN >>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >>> +#endif /* __BYTE_ORDER */ >>> +#elif defined __BYTE_ORDER__ >>> +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ >>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >>> +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ >>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >>> +#endif /* __BYTE_ORDER__ */ >>> +#elif defined __BIG_ENDIAN__ >>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >>> +#elif defined __LITTLE_ENDIAN__ >>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >>> +#endif >> What do you think about : >> >> +/* >> + * Compile-time endianness detection >> + */ >> +#define RTE_BIG_ENDIAN 1 >> +#define RTE_LITTLE_ENDIAN 2 >> +if defined __BYTE_ORDER__ /* Prefer gcc build-in macros */ >> +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ >> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >> +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ >> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >> +#endif /* __BYTE_ORDER__ */ >> +#else >> +#if defined RTE_EXEC_ENV_BSDAPP >> +#include <sys/endian.h> >> +#else >> +#include <endian.h> >> +#endif >> +#if defined __BYTE_ORDER >> +#if __BYTE_ORDER == __BIG_ENDIAN >> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >> +#elif __BYTE_ORDER == __LITTLE_ENDIAN >> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >> +#endif /* __BYTE_ORDER */ >> +#elif defined __BIG_ENDIAN__ >> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >> +#elif defined __LITTLE_ENDIAN__ >> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >> +#endif >> +#endif > Please, could you give more explanations about your proposal? > Why not always try to include endian.h? I assume that if gcc can handler why we need include that file? Also it seems that only old version could have this issue, newer versions has build in this marcos. So that's why I prefer "__BYTE_ORDER__" for high priority. Thanks, Michael > Why giving high priority to __BYTE_ORDER__? >
2014-12-04 10:28, Qiu, Michael: > On 12/4/2014 5:01 PM, Thomas Monjalon wrote: > > 2014-12-04 02:28, Qiu, Michael: > >> On 12/4/2014 5:26 AM, Thomas Monjalon wrote: > >>> There is no standard to check endianness. > >>> So we need to try different checks. > >>> Previous trials were done in testpmd (see commits > >>> 51f694dd40f56 and 64741f237cf29) without full success. > >>> This one is not guaranteed to work everywhere so it could > >>> evolve when exceptions are found. > > [...] > >>> #include <stdint.h> > >>> +#ifdef RTE_EXEC_ENV_BSDAPP > >>> +#include <sys/endian.h> > >>> +#else > >>> +#include <endian.h> > >>> +#endif > >>> + > >>> +/* > >>> + * Compile-time endianness detection > >>> + */ > >>> +#define RTE_BIG_ENDIAN 1 > >>> +#define RTE_LITTLE_ENDIAN 2 > >>> +#if defined __BYTE_ORDER > >>> +#if __BYTE_ORDER == __BIG_ENDIAN > >>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > >>> +#elif __BYTE_ORDER == __LITTLE_ENDIAN > >>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > >>> +#endif /* __BYTE_ORDER */ > >>> +#elif defined __BYTE_ORDER__ > >>> +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ > >>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > >>> +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ > >>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > >>> +#endif /* __BYTE_ORDER__ */ > >>> +#elif defined __BIG_ENDIAN__ > >>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > >>> +#elif defined __LITTLE_ENDIAN__ > >>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > >>> +#endif > >> What do you think about : > >> > >> +/* > >> + * Compile-time endianness detection > >> + */ > >> +#define RTE_BIG_ENDIAN 1 > >> +#define RTE_LITTLE_ENDIAN 2 > >> +if defined __BYTE_ORDER__ /* Prefer gcc build-in macros */ > >> +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ > >> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > >> +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ > >> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > >> +#endif /* __BYTE_ORDER__ */ > >> +#else > >> +#if defined RTE_EXEC_ENV_BSDAPP > >> +#include <sys/endian.h> > >> +#else > >> +#include <endian.h> > >> +#endif > >> +#if defined __BYTE_ORDER > >> +#if __BYTE_ORDER == __BIG_ENDIAN > >> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > >> +#elif __BYTE_ORDER == __LITTLE_ENDIAN > >> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > >> +#endif /* __BYTE_ORDER */ > >> +#elif defined __BIG_ENDIAN__ > >> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN > >> +#elif defined __LITTLE_ENDIAN__ > >> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN > >> +#endif > >> +#endif > > > > Please, could you give more explanations about your proposal? > > Why not always try to include endian.h? > > I assume that if gcc can handler why we need include that file? Separating include on top is easier to read, and I'm not sure it won't be needed for __BYTE_ORDER__ with some toolchains. > Also it seems that only old version could have this issue, newer > versions has build in this marcos. > > So that's why I prefer "__BYTE_ORDER__" for high priority. I have no problem with reversing this priority. > > Why giving high priority to __BYTE_ORDER__? Any other comment? May I apply with above change?
On 12/4/2014 8:20 PM, Thomas Monjalon wrote: > 2014-12-04 10:28, Qiu, Michael: >> On 12/4/2014 5:01 PM, Thomas Monjalon wrote: >>> 2014-12-04 02:28, Qiu, Michael: >>>> On 12/4/2014 5:26 AM, Thomas Monjalon wrote: >>>>> There is no standard to check endianness. >>>>> So we need to try different checks. >>>>> Previous trials were done in testpmd (see commits >>>>> 51f694dd40f56 and 64741f237cf29) without full success. >>>>> This one is not guaranteed to work everywhere so it could >>>>> evolve when exceptions are found. >>> [...] >>>>> #include <stdint.h> >>>>> +#ifdef RTE_EXEC_ENV_BSDAPP >>>>> +#include <sys/endian.h> >>>>> +#else >>>>> +#include <endian.h> >>>>> +#endif >>>>> + >>>>> +/* >>>>> + * Compile-time endianness detection >>>>> + */ >>>>> +#define RTE_BIG_ENDIAN 1 >>>>> +#define RTE_LITTLE_ENDIAN 2 >>>>> +#if defined __BYTE_ORDER >>>>> +#if __BYTE_ORDER == __BIG_ENDIAN >>>>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >>>>> +#elif __BYTE_ORDER == __LITTLE_ENDIAN >>>>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >>>>> +#endif /* __BYTE_ORDER */ >>>>> +#elif defined __BYTE_ORDER__ >>>>> +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ >>>>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >>>>> +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ >>>>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >>>>> +#endif /* __BYTE_ORDER__ */ >>>>> +#elif defined __BIG_ENDIAN__ >>>>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >>>>> +#elif defined __LITTLE_ENDIAN__ >>>>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >>>>> +#endif >>>> What do you think about : >>>> >>>> +/* >>>> + * Compile-time endianness detection >>>> + */ >>>> +#define RTE_BIG_ENDIAN 1 >>>> +#define RTE_LITTLE_ENDIAN 2 >>>> +if defined __BYTE_ORDER__ /* Prefer gcc build-in macros */ >>>> +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ >>>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >>>> +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ >>>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >>>> +#endif /* __BYTE_ORDER__ */ >>>> +#else >>>> +#if defined RTE_EXEC_ENV_BSDAPP >>>> +#include <sys/endian.h> >>>> +#else >>>> +#include <endian.h> >>>> +#endif >>>> +#if defined __BYTE_ORDER >>>> +#if __BYTE_ORDER == __BIG_ENDIAN >>>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >>>> +#elif __BYTE_ORDER == __LITTLE_ENDIAN >>>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >>>> +#endif /* __BYTE_ORDER */ >>>> +#elif defined __BIG_ENDIAN__ >>>> +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN >>>> +#elif defined __LITTLE_ENDIAN__ >>>> +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN >>>> +#endif >>>> +#endif >>> Please, could you give more explanations about your proposal? >>> Why not always try to include endian.h? >> I assume that if gcc can handler why we need include that file? > Separating include on top is easier to read, and I'm not sure it won't > be needed for __BYTE_ORDER__ with some toolchains. > >> Also it seems that only old version could have this issue, newer >> versions has build in this marcos. >> >> So that's why I prefer "__BYTE_ORDER__" for high priority. > I have no problem with reversing this priority. > >>> Why giving high priority to __BYTE_ORDER__? > Any other comment? May I apply with above change? Acked-by: Michael Qiu <michael.qiu@intel.com>
diff --git a/config/defconfig_ppc_64-power8-linuxapp-gcc b/config/defconfig_ppc_64-power8-linuxapp-gcc index 48018c3..d97a885 100644 --- a/config/defconfig_ppc_64-power8-linuxapp-gcc +++ b/config/defconfig_ppc_64-power8-linuxapp-gcc @@ -34,7 +34,6 @@ CONFIG_RTE_MACHINE="power8" CONFIG_RTE_ARCH="ppc_64" CONFIG_RTE_ARCH_PPC_64=y -CONFIG_RTE_ARCH_BIG_ENDIAN=y CONFIG_RTE_ARCH_64=y CONFIG_RTE_TOOLCHAIN="gcc" diff --git a/lib/librte_eal/common/include/arch/ppc_64/rte_byteorder.h b/lib/librte_eal/common/include/arch/ppc_64/rte_byteorder.h index 1a89051..80436f2 100644 --- a/lib/librte_eal/common/include/arch/ppc_64/rte_byteorder.h +++ b/lib/librte_eal/common/include/arch/ppc_64/rte_byteorder.h @@ -105,7 +105,7 @@ static inline uint64_t rte_arch_bswap64(uint64_t _x) /* Power 8 have both little endian and big endian mode * Power 7 only support big endian */ -#ifndef RTE_ARCH_BIG_ENDIAN +#if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN #define rte_cpu_to_le_16(x) (x) #define rte_cpu_to_le_32(x) (x) @@ -123,7 +123,7 @@ static inline uint64_t rte_arch_bswap64(uint64_t _x) #define rte_be_to_cpu_32(x) rte_bswap32(x) #define rte_be_to_cpu_64(x) rte_bswap64(x) -#else +#else /* RTE_BIG_ENDIAN */ #define rte_cpu_to_le_16(x) rte_bswap16(x) #define rte_cpu_to_le_32(x) rte_bswap32(x) diff --git a/lib/librte_eal/common/include/arch/x86/rte_byteorder.h b/lib/librte_eal/common/include/arch/x86/rte_byteorder.h index 1aa6985..ffdb6ef 100644 --- a/lib/librte_eal/common/include/arch/x86/rte_byteorder.h +++ b/lib/librte_eal/common/include/arch/x86/rte_byteorder.h @@ -40,6 +40,10 @@ extern "C" { #include "generic/rte_byteorder.h" +#ifndef RTE_BYTE_ORDER +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN +#endif + /* * An architecture-optimized byte swap for a 16-bit value. * diff --git a/lib/librte_eal/common/include/generic/rte_byteorder.h b/lib/librte_eal/common/include/generic/rte_byteorder.h index 9358136..ea23fdf 100644 --- a/lib/librte_eal/common/include/generic/rte_byteorder.h +++ b/lib/librte_eal/common/include/generic/rte_byteorder.h @@ -44,6 +44,34 @@ */ #include <stdint.h> +#ifdef RTE_EXEC_ENV_BSDAPP +#include <sys/endian.h> +#else +#include <endian.h> +#endif + +/* + * Compile-time endianness detection + */ +#define RTE_BIG_ENDIAN 1 +#define RTE_LITTLE_ENDIAN 2 +#if defined __BYTE_ORDER +#if __BYTE_ORDER == __BIG_ENDIAN +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN +#elif __BYTE_ORDER == __LITTLE_ENDIAN +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN +#endif /* __BYTE_ORDER */ +#elif defined __BYTE_ORDER__ +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN +#endif /* __BYTE_ORDER__ */ +#elif defined __BIG_ENDIAN__ +#define RTE_BYTE_ORDER RTE_BIG_ENDIAN +#elif defined __LITTLE_ENDIAN__ +#define RTE_BYTE_ORDER RTE_LITTLE_ENDIAN +#endif /* * An internal function to swap bytes in a 16-bit value.