From 932544efa469568525455bcb53d4eabeba0eb295 Mon Sep 17 00:00:00 2001 From: Adhemerval Zanella Date: Wed, 3 Apr 2024 16:39:18 -0300 Subject: math: x86 floor traps when FE_INEXACT is enabled (BZ 31601) The implementations of floor functions using x87 floating point (i386 and 86_64 long double only) traps when FE_INEXACT is enabled. Although this is a GNU extension outside the scope of the C standard, other architectures that also support traps do not show this behavior. The fix moves the implementation to a common one that holds any exceptions with a 'fnclex' (libc_feholdexcept_setround_387). Checked on x86_64-linux-gnu and i686-linux-gnu. Reviewed-by: H.J. Lu --- math/Makefile | 2 ++ 1 file changed, 2 insertions(+) (limited to 'math/Makefile') diff --git a/math/Makefile b/math/Makefile index d2a740eebe..121fe2881a 100644 --- a/math/Makefile +++ b/math/Makefile @@ -511,6 +511,7 @@ tests = \ test-fetestexceptflag \ test-fexcept \ test-fexcept-traps \ + test-floor-except-2 \ test-flt-eval-method \ test-fp-ilogb-constants \ test-fp-llogb-constants \ @@ -991,6 +992,7 @@ CFLAGS-test-fe-snans-always-signal.c += $(config-cflags-signaling-nans) CFLAGS-test-nan-const.c += -fno-builtin CFLAGS-test-ceil-except-2.c += -fno-builtin +CFLAGS-test-floor-except-2.c += -fno-builtin include ../Rules -- cgit v1.2.3