From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id DBB1BC433EF for ; Thu, 2 Dec 2021 09:50:40 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1356609AbhLBJyC (ORCPT ); Thu, 2 Dec 2021 04:54:02 -0500 Received: from mga01.intel.com ([192.55.52.88]:56004 "EHLO mga01.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1356598AbhLBJx7 (ORCPT ); Thu, 2 Dec 2021 04:53:59 -0500 X-IronPort-AV: E=McAfee;i="6200,9189,10185"; a="260668431" X-IronPort-AV: E=Sophos;i="5.87,281,1631602800"; d="scan'208";a="260668431" Received: from orsmga005.jf.intel.com ([10.7.209.41]) by fmsmga101.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 02 Dec 2021 01:50:37 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.87,281,1631602800"; d="scan'208";a="677605115" Received: from ahunter-desktop.fi.intel.com ([10.237.72.76]) by orsmga005.jf.intel.com with ESMTP; 02 Dec 2021 01:50:34 -0800 From: Adrian Hunter To: Masami Hiramatsu Cc: Arnaldo Carvalho de Melo , Jiri Olsa , linux-kernel@vger.kernel.org, Peter Zijlstra , Ingo Molnar , Borislav Petkov , Dave Hansen , x86@kernel.org, H Peter Anvin , chang.seok.bae@intel.com Subject: [PATCH 1/6] perf tests: Add AMX instructions to x86 instruction decoder test Date: Thu, 2 Dec 2021 11:50:24 +0200 Message-Id: <20211202095029.2165714-2-adrian.hunter@intel.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20211202095029.2165714-1-adrian.hunter@intel.com> References: <20211202095029.2165714-1-adrian.hunter@intel.com> MIME-Version: 1.0 Organization: Intel Finland Oy, Registered Address: PL 281, 00181 Helsinki, Business Identity Code: 0357606 - 4, Domiciled in Helsinki Content-Transfer-Encoding: 8bit Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org x86 instruction decoder is used for both kernel instructions and user space instructions (e.g. uprobes, perf tools Intel PT), so it is good to update it with new instructions. Add AMX instructions to x86 instruction decoder test. A subsequent patch adds the instructions to the instruction decoder. Reference: Intel Architecture Instruction Set Extensions and Future Features Programming Reference May 2021 Document Number: 319433-044 Example: $ INSN='ldtilecfg\|sttilecfg\|tdpbf16ps\|tdpbssd\|' $ INSN+='tdpbsud\|tdpbusd\|'tdpbuud\|tileloadd\|' $ INSN+='tileloaddt1\|tilerelease\|tilestored\|tilezero' $ perf test -v "x86 instruction decoder" |& grep -i $INSN Failed to decode: c4 e2 78 49 04 c8 ldtilecfg (%rax,%rcx,8) Failed to decode: c4 c2 78 49 04 c8 ldtilecfg (%r8,%rcx,8) Failed to decode: c4 e2 79 49 04 c8 sttilecfg (%rax,%rcx,8) Failed to decode: c4 c2 79 49 04 c8 sttilecfg (%r8,%rcx,8) Failed to decode: c4 e2 7a 5c d1 tdpbf16ps %tmm0,%tmm1,%tmm2 Failed to decode: c4 e2 7b 5e d1 tdpbssd %tmm0,%tmm1,%tmm2 Failed to decode: c4 e2 7a 5e d1 tdpbsud %tmm0,%tmm1,%tmm2 Failed to decode: c4 e2 79 5e d1 tdpbusd %tmm0,%tmm1,%tmm2 Failed to decode: c4 e2 78 5e d1 tdpbuud %tmm0,%tmm1,%tmm2 Failed to decode: c4 e2 7b 4b 0c c8 tileloadd (%rax,%rcx,8),%tmm1 Failed to decode: c4 c2 7b 4b 14 c8 tileloadd (%r8,%rcx,8),%tmm2 Failed to decode: c4 e2 79 4b 0c c8 tileloaddt1 (%rax,%rcx,8),%tmm1 Failed to decode: c4 c2 79 4b 14 c8 tileloaddt1 (%r8,%rcx,8),%tmm2 Failed to decode: c4 e2 78 49 c0 tilerelease Failed to decode: c4 e2 7a 4b 0c c8 tilestored %tmm1,(%rax,%rcx,8) Failed to decode: c4 c2 7a 4b 14 c8 tilestored %tmm2,(%r8,%rcx,8) Failed to decode: c4 e2 7b 49 c0 tilezero %tmm0 Failed to decode: c4 e2 7b 49 f8 tilezero %tmm7 Signed-off-by: Adrian Hunter --- tools/perf/arch/x86/tests/insn-x86-dat-64.c | 36 ++++++++++++++++++++ tools/perf/arch/x86/tests/insn-x86-dat-src.c | 21 ++++++++++++ 2 files changed, 57 insertions(+) diff --git a/tools/perf/arch/x86/tests/insn-x86-dat-64.c b/tools/perf/arch/x86/tests/insn-x86-dat-64.c index 5da17d41d302..3548565a1cc5 100644 --- a/tools/perf/arch/x86/tests/insn-x86-dat-64.c +++ b/tools/perf/arch/x86/tests/insn-x86-dat-64.c @@ -2459,6 +2459,42 @@ "3e f2 ff a4 c8 78 56 34 12 \tnotrack bnd jmpq *0x12345678(%rax,%rcx,8)",}, {{0x3e, 0xf2, 0x41, 0xff, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "jmp", "indirect", "3e f2 41 ff a4 c8 78 56 34 12 \tnotrack bnd jmpq *0x12345678(%r8,%rcx,8)",}, +{{0xc4, 0xe2, 0x78, 0x49, 0x04, 0xc8, }, 6, 0, "", "", +"c4 e2 78 49 04 c8 \tldtilecfg (%rax,%rcx,8)",}, +{{0xc4, 0xc2, 0x78, 0x49, 0x04, 0xc8, }, 6, 0, "", "", +"c4 c2 78 49 04 c8 \tldtilecfg (%r8,%rcx,8)",}, +{{0xc4, 0xe2, 0x79, 0x49, 0x04, 0xc8, }, 6, 0, "", "", +"c4 e2 79 49 04 c8 \tsttilecfg (%rax,%rcx,8)",}, +{{0xc4, 0xc2, 0x79, 0x49, 0x04, 0xc8, }, 6, 0, "", "", +"c4 c2 79 49 04 c8 \tsttilecfg (%r8,%rcx,8)",}, +{{0xc4, 0xe2, 0x7a, 0x5c, 0xd1, }, 5, 0, "", "", +"c4 e2 7a 5c d1 \ttdpbf16ps %tmm0,%tmm1,%tmm2",}, +{{0xc4, 0xe2, 0x7b, 0x5e, 0xd1, }, 5, 0, "", "", +"c4 e2 7b 5e d1 \ttdpbssd %tmm0,%tmm1,%tmm2",}, +{{0xc4, 0xe2, 0x7a, 0x5e, 0xd1, }, 5, 0, "", "", +"c4 e2 7a 5e d1 \ttdpbsud %tmm0,%tmm1,%tmm2",}, +{{0xc4, 0xe2, 0x79, 0x5e, 0xd1, }, 5, 0, "", "", +"c4 e2 79 5e d1 \ttdpbusd %tmm0,%tmm1,%tmm2",}, +{{0xc4, 0xe2, 0x78, 0x5e, 0xd1, }, 5, 0, "", "", +"c4 e2 78 5e d1 \ttdpbuud %tmm0,%tmm1,%tmm2",}, +{{0xc4, 0xe2, 0x7b, 0x4b, 0x0c, 0xc8, }, 6, 0, "", "", +"c4 e2 7b 4b 0c c8 \ttileloadd (%rax,%rcx,8),%tmm1",}, +{{0xc4, 0xc2, 0x7b, 0x4b, 0x14, 0xc8, }, 6, 0, "", "", +"c4 c2 7b 4b 14 c8 \ttileloadd (%r8,%rcx,8),%tmm2",}, +{{0xc4, 0xe2, 0x79, 0x4b, 0x0c, 0xc8, }, 6, 0, "", "", +"c4 e2 79 4b 0c c8 \ttileloaddt1 (%rax,%rcx,8),%tmm1",}, +{{0xc4, 0xc2, 0x79, 0x4b, 0x14, 0xc8, }, 6, 0, "", "", +"c4 c2 79 4b 14 c8 \ttileloaddt1 (%r8,%rcx,8),%tmm2",}, +{{0xc4, 0xe2, 0x78, 0x49, 0xc0, }, 5, 0, "", "", +"c4 e2 78 49 c0 \ttilerelease ",}, +{{0xc4, 0xe2, 0x7a, 0x4b, 0x0c, 0xc8, }, 6, 0, "", "", +"c4 e2 7a 4b 0c c8 \ttilestored %tmm1,(%rax,%rcx,8)",}, +{{0xc4, 0xc2, 0x7a, 0x4b, 0x14, 0xc8, }, 6, 0, "", "", +"c4 c2 7a 4b 14 c8 \ttilestored %tmm2,(%r8,%rcx,8)",}, +{{0xc4, 0xe2, 0x7b, 0x49, 0xc0, }, 5, 0, "", "", +"c4 e2 7b 49 c0 \ttilezero %tmm0",}, +{{0xc4, 0xe2, 0x7b, 0x49, 0xf8, }, 5, 0, "", "", +"c4 e2 7b 49 f8 \ttilezero %tmm7",}, {{0x0f, 0x01, 0xcf, }, 3, 0, "", "", "0f 01 cf \tencls ",}, {{0x0f, 0x01, 0xd7, }, 3, 0, "", "", diff --git a/tools/perf/arch/x86/tests/insn-x86-dat-src.c b/tools/perf/arch/x86/tests/insn-x86-dat-src.c index c3808e94c46e..7906f7b2ffeb 100644 --- a/tools/perf/arch/x86/tests/insn-x86-dat-src.c +++ b/tools/perf/arch/x86/tests/insn-x86-dat-src.c @@ -1910,6 +1910,27 @@ int main(void) asm volatile("notrack bnd jmpq *0x12345678(%rax,%rcx,8)"); /* Expecting: jmp indirect 0 */ asm volatile("notrack bnd jmpq *0x12345678(%r8,%rcx,8)"); /* Expecting: jmp indirect 0 */ + /* AMX */ + + asm volatile("ldtilecfg (%rax,%rcx,8)"); + asm volatile("ldtilecfg (%r8,%rcx,8)"); + asm volatile("sttilecfg (%rax,%rcx,8)"); + asm volatile("sttilecfg (%r8,%rcx,8)"); + asm volatile("tdpbf16ps %tmm0, %tmm1, %tmm2"); + asm volatile("tdpbssd %tmm0, %tmm1, %tmm2"); + asm volatile("tdpbsud %tmm0, %tmm1, %tmm2"); + asm volatile("tdpbusd %tmm0, %tmm1, %tmm2"); + asm volatile("tdpbuud %tmm0, %tmm1, %tmm2"); + asm volatile("tileloadd (%rax,%rcx,8), %tmm1"); + asm volatile("tileloadd (%r8,%rcx,8), %tmm2"); + asm volatile("tileloaddt1 (%rax,%rcx,8), %tmm1"); + asm volatile("tileloaddt1 (%r8,%rcx,8), %tmm2"); + asm volatile("tilerelease"); + asm volatile("tilestored %tmm1, (%rax,%rcx,8)"); + asm volatile("tilestored %tmm2, (%r8,%rcx,8)"); + asm volatile("tilezero %tmm0"); + asm volatile("tilezero %tmm7"); + #else /* #ifdef __x86_64__ */ /* bound r32, mem (same op code as EVEX prefix) */ -- 2.25.1