From patchwork Wed Jun 24 16:21:33 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Joseph Myers X-Patchwork-Id: 488136 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Received: from sourceware.org (server1.sourceware.org [209.132.180.131]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id B054B14030D for ; Thu, 25 Jun 2015 02:21:47 +1000 (AEST) Authentication-Results: ozlabs.org; dkim=pass (1024-bit key; unprotected) header.d=sourceware.org header.i=@sourceware.org header.b=iPibCqTt; dkim-atps=neutral DomainKey-Signature: a=rsa-sha1; c=nofws; d=sourceware.org; h=list-id :list-unsubscribe:list-subscribe:list-archive:list-post :list-help:sender:date:from:to:subject:message-id:mime-version :content-type; q=dns; s=default; b=gcEHJVoQiimlLlwfdiTC1xfjfRVWL NbBNZq4wryBw/vMlp6mD7uIiYNssiGdKbxclRbeiYruXO7V0yCPoBEiKNc0qtejF NuBqD358Qs71Z0mJYA3vcThGZrGrL6D8NTlO9aqwxHicpE9lD/8LG5j98CQAID2H RFamppnZzcFdBc= DKIM-Signature: v=1; a=rsa-sha1; c=relaxed; d=sourceware.org; h=list-id :list-unsubscribe:list-subscribe:list-archive:list-post :list-help:sender:date:from:to:subject:message-id:mime-version :content-type; s=default; bh=oYPQKzbFB/q+c5V4+ftgXfi2JNM=; b=iPi bCqTtLmTQUpsjzsV9zQI70ys6kEoGv7SGGQQV4Fj39bJxx3YRtfrerf9w7nV59H4 D+iFqKJ/zn/mWTNkaqYxWqCV+zWhe5E8NNHWl9cLkEDnPCUK1xp0UBfifqhUS7YK T4VgJc2NY32xE18BLMcHtUgYikhiWcxWtt/fM68I= Received: (qmail 110947 invoked by alias); 24 Jun 2015 16:21:41 -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 110933 invoked by uid 89); 24 Jun 2015 16:21:41 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-2.3 required=5.0 tests=AWL, BAYES_00, RCVD_IN_DNSWL_LOW, SPF_PASS autolearn=ham version=3.3.2 X-HELO: relay1.mentorg.com Date: Wed, 24 Jun 2015 16:21:33 +0000 From: Joseph Myers To: Subject: Fix csin, csinh overflow in directed rounding modes (bug 18593) [committed] Message-ID: User-Agent: Alpine 2.10 (DEB 1266 2009-07-14) MIME-Version: 1.0 csin and csinh can produce bad results when overflowing in directed rounding modes, because a multiplication that can overflow is followed by a possible negation. This patch fixes this by negating one of the arguments of the multiplication before the multiplication instead of negating the result. The new tests for this issue are added to auto-libm-test-in, starting use of that file for csin and csinh. The issue was found in the course of moving existing tests for csin and csinh (existing tests, by being enabled in more cases than previously, showed the issue for float and double but not for long double); that move will now be done separately. Tested for x86_64 and x86 and ulps updated accordingly. Committed. (auto-libm-test-out diffs omitted below.) 2015-06-24 Joseph Myers [BZ #18593] * math/s_csin.c (__csin): Negate before rather than after possibly overflowing multiplication. * math/s_csinf.c (__csinf): Likewise. * math/s_csinh.c (__csinh): Likewise. * math/s_csinhf.c (__csinhf): Likewise. * math/s_csinhl.c (__csinhl): Likewise. * math/s_csinl.c (__csinl): Likewise. * math/auto-libm-test-in: Add some tests of csin and csinh. * math/auto-libm-test-out: Regenerated. * math/libm-test.inc (csin_test_data): Use AUTO_TESTS_c_c. (csinh_test_data): Likewise. * sysdeps/x86_64/fpu/libm-test-ulps: Update. diff --git a/math/auto-libm-test-in b/math/auto-libm-test-in index 6545583..2389c05 100644 --- a/math/auto-libm-test-in +++ b/math/auto-libm-test-in @@ -955,6 +955,16 @@ cpow 0.75 1.25 1.0 1.0 xfail-rounding cpow 0.75 1.25 1.0 0.0 cpow 0.75 1.25 0.0 1.0 +csin 0.75 1e6 +csin 0.75 -1e6 +csin -0.75 1e6 +csin -0.75 -1e6 + +csinh 1e6 0.75 +csinh -1e6 0.75 +csinh 1e6 -0.75 +csinh -1e6 -0.75 + csqrt 0 0 csqrt 0 -0 csqrt -0 0 diff --git a/math/libm-test.inc b/math/libm-test.inc index 03c3831..ba8e8ff 100644 --- a/math/libm-test.inc +++ b/math/libm-test.inc @@ -6466,6 +6466,8 @@ static const struct test_c_c_data csin_test_data[] = TEST_c_c (csin, 0x1p-16434L, 22730, 1.217853148905605987081057582351152052687e4924L, plus_oflow, OVERFLOW_EXCEPTION), #endif + AUTO_TESTS_c_c (csin), + TEST_c_c (csin, min_subnorm_value, min_value, min_subnorm_value, min_value, UNDERFLOW_EXCEPTION), TEST_c_c (csin, min_value, min_subnorm_value, min_value, min_subnorm_value, UNDERFLOW_EXCEPTION), }; @@ -6563,6 +6565,8 @@ static const struct test_c_c_data csinh_test_data[] = TEST_c_c (csinh, 22730, 0x1p-16434L, plus_oflow, 1.217853148905605987081057582351152052687e4924L, OVERFLOW_EXCEPTION), #endif + AUTO_TESTS_c_c (csinh), + TEST_c_c (csinh, min_subnorm_value, min_value, min_subnorm_value, min_value, UNDERFLOW_EXCEPTION), TEST_c_c (csinh, min_value, min_subnorm_value, min_value, min_subnorm_value, UNDERFLOW_EXCEPTION), }; diff --git a/math/s_csin.c b/math/s_csin.c index f1ddb4e..ceee25f 100644 --- a/math/s_csin.c +++ b/math/s_csin.c @@ -52,6 +52,9 @@ __csin (__complex__ double x) cosix = 1.0; } + if (negate) + sinix = -sinix; + if (fabs (__imag__ x) > t) { double exp_t = __ieee754_exp (t); @@ -86,9 +89,6 @@ __csin (__complex__ double x) __imag__ retval = __ieee754_sinh (__imag__ x) * cosix; } - if (negate) - __real__ retval = -__real__ retval; - if (fabs (__real__ retval) < DBL_MIN) { volatile double force_underflow diff --git a/math/s_csinf.c b/math/s_csinf.c index 6c68189..daca689 100644 --- a/math/s_csinf.c +++ b/math/s_csinf.c @@ -52,6 +52,9 @@ __csinf (__complex__ float x) cosix = 1.0f; } + if (negate) + sinix = -sinix; + if (fabsf (__imag__ x) > t) { float exp_t = __ieee754_expf (t); @@ -86,9 +89,6 @@ __csinf (__complex__ float x) __imag__ retval = __ieee754_sinhf (__imag__ x) * cosix; } - if (negate) - __real__ retval = -__real__ retval; - if (fabsf (__real__ retval) < FLT_MIN) { volatile float force_underflow diff --git a/math/s_csinh.c b/math/s_csinh.c index c607c2a..d940c97 100644 --- a/math/s_csinh.c +++ b/math/s_csinh.c @@ -52,6 +52,9 @@ __csinh (__complex__ double x) cosix = 1.0; } + if (negate) + cosix = -cosix; + if (fabs (__real__ x) > t) { double exp_t = __ieee754_exp (t); @@ -86,9 +89,6 @@ __csinh (__complex__ double x) __imag__ retval = __ieee754_cosh (__real__ x) * sinix; } - if (negate) - __real__ retval = -__real__ retval; - if (fabs (__real__ retval) < DBL_MIN) { volatile double force_underflow diff --git a/math/s_csinhf.c b/math/s_csinhf.c index 848f5d1..4e0bc03 100644 --- a/math/s_csinhf.c +++ b/math/s_csinhf.c @@ -52,6 +52,9 @@ __csinhf (__complex__ float x) cosix = 1.0f; } + if (negate) + cosix = -cosix; + if (fabsf (__real__ x) > t) { float exp_t = __ieee754_expf (t); @@ -86,9 +89,6 @@ __csinhf (__complex__ float x) __imag__ retval = __ieee754_coshf (__real__ x) * sinix; } - if (negate) - __real__ retval = -__real__ retval; - if (fabsf (__real__ retval) < FLT_MIN) { volatile float force_underflow diff --git a/math/s_csinhl.c b/math/s_csinhl.c index a437d16..eeb2212 100644 --- a/math/s_csinhl.c +++ b/math/s_csinhl.c @@ -52,6 +52,9 @@ __csinhl (__complex__ long double x) cosix = 1.0; } + if (negate) + cosix = -cosix; + if (fabsl (__real__ x) > t) { long double exp_t = __ieee754_expl (t); @@ -86,9 +89,6 @@ __csinhl (__complex__ long double x) __imag__ retval = __ieee754_coshl (__real__ x) * sinix; } - if (negate) - __real__ retval = -__real__ retval; - if (fabsl (__real__ retval) < LDBL_MIN) { volatile long double force_underflow diff --git a/math/s_csinl.c b/math/s_csinl.c index accaeb5..a4b1eb5 100644 --- a/math/s_csinl.c +++ b/math/s_csinl.c @@ -52,6 +52,9 @@ __csinl (__complex__ long double x) cosix = 1.0; } + if (negate) + sinix = -sinix; + if (fabsl (__imag__ x) > t) { long double exp_t = __ieee754_expl (t); @@ -86,9 +89,6 @@ __csinl (__complex__ long double x) __imag__ retval = __ieee754_sinhl (__imag__ x) * cosix; } - if (negate) - __real__ retval = -__real__ retval; - if (fabsl (__real__ retval) < LDBL_MIN) { volatile long double force_underflow diff --git a/sysdeps/x86_64/fpu/libm-test-ulps b/sysdeps/x86_64/fpu/libm-test-ulps index 2e2722d..101831f 100644 --- a/sysdeps/x86_64/fpu/libm-test-ulps +++ b/sysdeps/x86_64/fpu/libm-test-ulps @@ -1120,9 +1120,9 @@ ildouble: 3 ldouble: 3 Function: Real part of "csin_upward": -double: 1 +double: 2 float: 3 -idouble: 1 +idouble: 2 ifloat: 3 ildouble: 3 ldouble: 3 @@ -1148,9 +1148,9 @@ idouble: 1 ifloat: 1 Function: Real part of "csinh_downward": -double: 1 +double: 2 float: 2 -idouble: 1 +idouble: 2 ifloat: 2 ildouble: 3 ldouble: 3