From patchwork Fri Nov 24 22:04:30 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Joseph Myers X-Patchwork-Id: 24509 Received: (qmail 40219 invoked by alias); 24 Nov 2017 22:04:37 -0000 Mailing-List: contact libc-alpha-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-alpha-owner@sourceware.org Delivered-To: mailing list libc-alpha@sourceware.org Received: (qmail 40210 invoked by uid 89); 24 Nov 2017 22:04:37 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-24.3 required=5.0 tests=AWL, BAYES_00, GIT_PATCH_0, GIT_PATCH_1, GIT_PATCH_2, GIT_PATCH_3, KB_WAM_FROM_NAME_SINGLEWORD, RCVD_IN_DNSWL_NONE, SPF_PASS, URIBL_RED autolearn=ham version=3.3.2 spammy= X-HELO: relay1.mentorg.com Date: Fri, 24 Nov 2017 22:04:30 +0000 From: Joseph Myers To: Subject: Make min_of_type_ macros function-like [committed] Message-ID: User-Agent: Alpine 2.20 (DEB 67 2015-01-07) MIME-Version: 1.0 X-ClientProxiedBy: svr-ies-mbx-01.mgc.mentorg.com (139.181.222.1) To SVR-IES-MBX-03.mgc.mentorg.com (139.181.222.3) math_private.h uses __MATH_TG in defining the min_of_type macro used within libm, with min_of_type_ macros for each type. This runs into problems with __MATH_TG expansions used with additional _FloatN and _FloatNx type support, because those can end up macro-expanding the FUNC argument to __MATH_TG before it gets concatenated with a suffix - meaning that min_of_type_ can't simultaneously be the macro name for double, and a prefix to other macro names, since the latter case requires such premature macro expansion not to occur. (This is not a problem for the uses of __MATH_TG in installed headers because FUNC there is a function name in the implementation namespace, and the suffixes themselves don't get macro-expanded.) This patch fixes the problem by making min_of_type_ macros function-like, so no macro expansion occurs when min_of_type_ is expanded on its own as a macro argument, only later when followed by () after expansion. Tested for x86_64, including in conjunction with _Float64x support patches. Committed. 2017-11-24 Joseph Myers * sysdeps/generic/math_private.h (min_of_type_f): Make into a function-like macro. (min_of_type_): Likewise. (min_of_type_l): Likewise. (min_of_type_f128): Likewise. (min_of_type): Pass () as last argument of __MATH_TG. diff --git a/sysdeps/generic/math_private.h b/sysdeps/generic/math_private.h index 80c7c92..f29898c 100644 --- a/sysdeps/generic/math_private.h +++ b/sysdeps/generic/math_private.h @@ -298,12 +298,15 @@ extern void __docos (double __x, double __dx, double __v[]); #define fabs_tg(x) __MATH_TG ((x), (__typeof (x)) __builtin_fabs, (x)) -#define min_of_type_f FLT_MIN -#define min_of_type_ DBL_MIN -#define min_of_type_l LDBL_MIN -#define min_of_type_f128 FLT128_MIN - -#define min_of_type(x) __MATH_TG ((x), (__typeof (x)) min_of_type_, ) +/* These must be function-like macros because some __MATH_TG + implementations macro-expand the function-name argument before + concatenating a suffix to it. */ +#define min_of_type_f() FLT_MIN +#define min_of_type_() DBL_MIN +#define min_of_type_l() LDBL_MIN +#define min_of_type_f128() FLT128_MIN + +#define min_of_type(x) __MATH_TG ((x), (__typeof (x)) min_of_type_, ()) /* If X (which is not a NaN) is subnormal, force an underflow exception. */