From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([209.51.188.92]:58488) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1gk7mK-0000JG-DO for qemu-devel@nongnu.org; Thu, 17 Jan 2019 08:32:19 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1gk7hr-0003ET-SC for qemu-devel@nongnu.org; Thu, 17 Jan 2019 08:27:41 -0500 Received: from mail-wr1-x42d.google.com ([2a00:1450:4864:20::42d]:35354) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1gk7hp-0002Wr-PO for qemu-devel@nongnu.org; Thu, 17 Jan 2019 08:27:39 -0500 Received: by mail-wr1-x42d.google.com with SMTP id 96so11024679wrb.2 for ; Thu, 17 Jan 2019 05:27:10 -0800 (PST) From: =?UTF-8?q?Alex=20Benn=C3=A9e?= Date: Thu, 17 Jan 2019 13:26:59 +0000 Message-Id: <20190117132703.17790-4-alex.bennee@linaro.org> In-Reply-To: <20190117132703.17790-1-alex.bennee@linaro.org> References: <20190117132703.17790-1-alex.bennee@linaro.org> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Subject: [Qemu-devel] [PULL 3/7] softfloat: enforce softfloat if the host's FMA is broken List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: peter.maydell@linaro.org Cc: qemu-devel@nongnu.org, "Emilio G. Cota" , =?UTF-8?q?Alex=20Benn=C3=A9e?= , Aurelien Jarno From: "Emilio G. Cota" The added branch to the FMA ops is marked as unlikely and therefore its impact on performance (measured with fp-bench) is within noise range when measured on an Intel(R) Xeon(R) Gold 6142 CPU @ 2.60GHz. Reported-by: Laurent Desnogues Signed-off-by: Emilio G. Cota Signed-off-by: Alex Bennée Reviewed-by: Richard Henderson diff --git a/fpu/softfloat.c b/fpu/softfloat.c index 59eac97d10..9132d7a0b0 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -1542,6 +1542,8 @@ soft_f64_muladd(float64 a, float64 b, float64 c, int flags, return float64_round_pack_canonical(pr, status); } +static bool force_soft_fma; + float32 QEMU_FLATTEN float32_muladd(float32 xa, float32 xb, float32 xc, int flags, float_status *s) { @@ -1562,6 +1564,11 @@ float32_muladd(float32 xa, float32 xb, float32 xc, int flags, float_status *s) if (unlikely(!f32_is_zon3(ua, ub, uc))) { goto soft; } + + if (unlikely(force_soft_fma)) { + goto soft; + } + /* * When (a || b) == 0, there's no need to check for under/over flow, * since we know the addend is (normal || 0) and the product is 0. @@ -1623,6 +1630,11 @@ float64_muladd(float64 xa, float64 xb, float64 xc, int flags, float_status *s) if (unlikely(!f64_is_zon3(ua, ub, uc))) { goto soft; } + + if (unlikely(force_soft_fma)) { + goto soft; + } + /* * When (a || b) == 0, there's no need to check for under/over flow, * since we know the addend is (normal || 0) and the product is 0. @@ -7974,3 +7986,24 @@ float128 float128_scalbn(float128 a, int n, float_status *status) , status); } + +static void __attribute__((constructor)) softfloat_init(void) +{ + union_float64 ua, ub, uc, ur; + + if (QEMU_NO_HARDFLOAT) { + return; + } + /* + * Test that the host's FMA is not obviously broken. For example, + * glibc < 2.23 can perform an incorrect FMA on certain hosts; see + * https://sourceware.org/bugzilla/show_bug.cgi?id=13304 + */ + ua.s = 0x0020000000000001ULL; + ub.s = 0x3ca0000000000000ULL; + uc.s = 0x0020000000000000ULL; + ur.h = fma(ua.h, ub.h, uc.h); + if (ur.s != 0x0020000000000001ULL) { + force_soft_fma = true; + } +} -- 2.17.1