From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: by sourceware.org (Postfix, from userid 1778) id 14F9B385801A; Mon, 3 Oct 2022 21:55:44 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 14F9B385801A DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=sourceware.org; s=default; t=1664834144; bh=4laKYOl2pIXYmA6RMF1W1yhtpeVIl3qevdYyIiQowhY=; h=From:To:Subject:Date:From; b=FmIKwLcX9f9kXImTb4CcGv+ul8XXY39+qA+lhRHzCNFn068IevplQhBmD9V/Ol5Ey Gw007RGzMxYw+/wT1SAbrJeqamNLvO4hhiZTPLrLjddsfyXDS10+HJ0CosZdJhQEbl UC7n94nyHWN/Lw/2bNXPvHFEH8cs+YRRejgdBQ2M= Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit From: Aurelien Jarno To: glibc-cvs@sourceware.org Subject: [glibc/release/2.35/master] x86-64: Require BMI2 for AVX2 str(n)casecmp implementations X-Act-Checkin: glibc X-Git-Author: Aurelien Jarno X-Git-Refname: refs/heads/release/2.35/master X-Git-Oldrev: 97ff96b0e12975ebb409f1668559651f30d13041 X-Git-Newrev: 0161ebf15ba5bd98cf1b47d53c079540cd7871d8 Message-Id: <20221003215544.14F9B385801A@sourceware.org> Date: Mon, 3 Oct 2022 21:55:44 +0000 (GMT) List-Id: https://sourceware.org/git/gitweb.cgi?p=glibc.git;h=0161ebf15ba5bd98cf1b47d53c079540cd7871d8 commit 0161ebf15ba5bd98cf1b47d53c079540cd7871d8 Author: Aurelien Jarno Date: Mon Oct 3 23:16:46 2022 +0200 x86-64: Require BMI2 for AVX2 str(n)casecmp implementations The AVX2 str(n)casecmp implementations use the 'bzhi' instruction, which belongs to the BMI2 CPU feature. NB: It also uses the 'tzcnt' BMI1 instruction, but it is executed as BSF as BSF if the CPU doesn't support TZCNT, and produces the same result for non-zero input. Partially fixes: b77b06e0e296 ("x86: Optimize strcmp-avx2.S") Partially resolves: BZ #29611 Reviewed-by: Noah Goldstein (cherry picked from commit 10f79d3670b036925da63dc532b122d27ce65ff8) Diff: --- sysdeps/x86_64/multiarch/ifunc-impl-list.c | 28 ++++++++++++++++++++-------- sysdeps/x86_64/multiarch/ifunc-strcasecmp.h | 1 + 2 files changed, 21 insertions(+), 8 deletions(-) diff --git a/sysdeps/x86_64/multiarch/ifunc-impl-list.c b/sysdeps/x86_64/multiarch/ifunc-impl-list.c index e97218f624..2d8c87a42e 100644 --- a/sysdeps/x86_64/multiarch/ifunc-impl-list.c +++ b/sysdeps/x86_64/multiarch/ifunc-impl-list.c @@ -404,13 +404,16 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array, IFUNC_IMPL (i, name, strcasecmp, IFUNC_IMPL_ADD (array, i, strcasecmp, (CPU_FEATURE_USABLE (AVX512VL) - && CPU_FEATURE_USABLE (AVX512BW)), + && CPU_FEATURE_USABLE (AVX512BW) + && CPU_FEATURE_USABLE (BMI2)), __strcasecmp_evex) IFUNC_IMPL_ADD (array, i, strcasecmp, - CPU_FEATURE_USABLE (AVX2), + (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2)), __strcasecmp_avx2) IFUNC_IMPL_ADD (array, i, strcasecmp, (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2) && CPU_FEATURE_USABLE (RTM)), __strcasecmp_avx2_rtm) IFUNC_IMPL_ADD (array, i, strcasecmp, @@ -425,13 +428,16 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array, IFUNC_IMPL (i, name, strcasecmp_l, IFUNC_IMPL_ADD (array, i, strcasecmp, (CPU_FEATURE_USABLE (AVX512VL) - && CPU_FEATURE_USABLE (AVX512BW)), + && CPU_FEATURE_USABLE (AVX512BW) + && CPU_FEATURE_USABLE (BMI2)), __strcasecmp_l_evex) IFUNC_IMPL_ADD (array, i, strcasecmp, - CPU_FEATURE_USABLE (AVX2), + (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2)), __strcasecmp_l_avx2) IFUNC_IMPL_ADD (array, i, strcasecmp, (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2) && CPU_FEATURE_USABLE (RTM)), __strcasecmp_l_avx2_rtm) IFUNC_IMPL_ADD (array, i, strcasecmp_l, @@ -560,13 +566,16 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array, IFUNC_IMPL (i, name, strncasecmp, IFUNC_IMPL_ADD (array, i, strncasecmp, (CPU_FEATURE_USABLE (AVX512VL) - && CPU_FEATURE_USABLE (AVX512BW)), + && CPU_FEATURE_USABLE (AVX512BW) + && CPU_FEATURE_USABLE (BMI2)), __strncasecmp_evex) IFUNC_IMPL_ADD (array, i, strncasecmp, - CPU_FEATURE_USABLE (AVX2), + (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2)), __strncasecmp_avx2) IFUNC_IMPL_ADD (array, i, strncasecmp, (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2) && CPU_FEATURE_USABLE (RTM)), __strncasecmp_avx2_rtm) IFUNC_IMPL_ADD (array, i, strncasecmp, @@ -582,13 +591,16 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array, IFUNC_IMPL (i, name, strncasecmp_l, IFUNC_IMPL_ADD (array, i, strncasecmp, (CPU_FEATURE_USABLE (AVX512VL) - && CPU_FEATURE_USABLE (AVX512BW)), + & CPU_FEATURE_USABLE (AVX512BW) + && CPU_FEATURE_USABLE (BMI2)), __strncasecmp_l_evex) IFUNC_IMPL_ADD (array, i, strncasecmp, - CPU_FEATURE_USABLE (AVX2), + (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2)), __strncasecmp_l_avx2) IFUNC_IMPL_ADD (array, i, strncasecmp, (CPU_FEATURE_USABLE (AVX2) + && CPU_FEATURE_USABLE (BMI2) && CPU_FEATURE_USABLE (RTM)), __strncasecmp_l_avx2_rtm) IFUNC_IMPL_ADD (array, i, strncasecmp_l, diff --git a/sysdeps/x86_64/multiarch/ifunc-strcasecmp.h b/sysdeps/x86_64/multiarch/ifunc-strcasecmp.h index 766539c241..e201f6e4d8 100644 --- a/sysdeps/x86_64/multiarch/ifunc-strcasecmp.h +++ b/sysdeps/x86_64/multiarch/ifunc-strcasecmp.h @@ -32,6 +32,7 @@ IFUNC_SELECTOR (void) const struct cpu_features* cpu_features = __get_cpu_features (); if (CPU_FEATURE_USABLE_P (cpu_features, AVX2) + && CPU_FEATURE_USABLE_P (cpu_features, BMI2) && CPU_FEATURES_ARCH_P (cpu_features, AVX_Fast_Unaligned_Load)) { if (CPU_FEATURE_USABLE_P (cpu_features, AVX512VL)