From: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Date: Tue, 9 Jan 2024 19:40:33 +0000 (-0300)
Subject: math: Fix test-fenv.c feupdateenv tests
X-Git-Tag: glibc-2.39~83
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=497e4d503025c794a771d2c124123178f557623a;p=thirdparty%2Fglibc.git

math: Fix test-fenv.c feupdateenv tests

The feupdateenv tests added by 802aef27b2 do not restore the floating
point mask, which might keep some floating point exception enabled and
thus make the feupdateenv_single_test raise an unexpected signal.

Checked on x86_64-linux-gnu and aarch64-linux-gnu (on Apple M1 trapping
is supported).
Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
---

diff --git a/math/test-fenv.c b/math/test-fenv.c
index 9db6789e64b..8d39181b14d 100644
--- a/math/test-fenv.c
+++ b/math/test-fenv.c
@@ -659,6 +659,7 @@ static void
 feupdate_single_test (const char *flag_name, int fe_exc)
 {
   feenv_nomask_test (flag_name, fe_exc, feupdateenv);
+  fesetenv (FE_DFL_ENV);
   feenv_mask_test (flag_name, fe_exc, feupdateenv);
 }
 #endif