From nobody Fri May 10 05:17:18 2024 Delivered-To: importer@patchew.org Authentication-Results: mx.zohomail.com; dkim=pass; spf=pass (zohomail.com: domain of gnu.org designates 209.51.188.17 as permitted sender) smtp.mailfrom=qemu-devel-bounces+importer=patchew.org@nongnu.org; dmarc=pass(p=none dis=none) header.from=linaro.org ARC-Seal: i=1; a=rsa-sha256; t=1709656836; cv=none; d=zohomail.com; s=zohoarc; b=Kbivwz0lr36DV96T0mE9K+AKbAj6IUMyGmOHHtoPtn7UUdhQh35oIyHBboHCYfeGO9WUSGKWrXlxqPGLGAK8fHoRzi8Q1B/9PpOkexJ5wXXbF0TMpjd3xO7AkFNET4fTAkmiBL0BXC1w9wf86z99Rc55d5vJjrOgJN5CUDl2bhQ= ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=zohomail.com; s=zohoarc; t=1709656836; h=Content-Transfer-Encoding:Cc:Cc:Date:Date:From:From:List-Subscribe:List-Post:List-Id:List-Archive:List-Help:List-Unsubscribe:MIME-Version:Message-ID:Sender:Subject:Subject:To:To:Message-Id:Reply-To; bh=uGDboQdgatOQ/3wEUF/IZCqMlV4D3UaeoJrF99IpsFg=; b=KBAdmHDyCcQhgVTEh7seW6TsJglmFkTxZyxEWaYn3vFE0TEX19q6pkcnpbEdGwyJKoe48GhrKWMx70AiRQQpNxqZp24wt253mE8GMm/b1ufIHfxjH/mrXG8tHRiyUp87GyByRM7VeQHvLnPBNbfv9MOA8BVCmYH1f9mIXGU2UsE= ARC-Authentication-Results: i=1; mx.zohomail.com; dkim=pass; spf=pass (zohomail.com: domain of gnu.org designates 209.51.188.17 as permitted sender) smtp.mailfrom=qemu-devel-bounces+importer=patchew.org@nongnu.org; dmarc=pass header.from= (p=none dis=none) Return-Path: Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) by mx.zohomail.com with SMTPS id 1709656836806897.1363687973771; Tue, 5 Mar 2024 08:40:36 -0800 (PST) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1rhXpN-0008JN-OU; Tue, 05 Mar 2024 11:39:42 -0500 Received: from eggs.gnu.org ([2001:470:142:3::10]) by lists.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1rhXpL-0008Ih-Fu for qemu-devel@nongnu.org; Tue, 05 Mar 2024 11:39:39 -0500 Received: from mail-pj1-x1036.google.com ([2607:f8b0:4864:20::1036]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_128_GCM_SHA256:128) (Exim 4.90_1) (envelope-from ) id 1rhXpJ-00009u-9D for qemu-devel@nongnu.org; Tue, 05 Mar 2024 11:39:39 -0500 Received: by mail-pj1-x1036.google.com with SMTP id 98e67ed59e1d1-2995fa850ddso5244065a91.0 for ; Tue, 05 Mar 2024 08:39:35 -0800 (PST) Received: from stoup.. (098-147-055-211.res.spectrum.com. [98.147.55.211]) by smtp.gmail.com with ESMTPSA id g10-20020a17090a714a00b0029ab460019asm9703542pjs.1.2024.03.05.08.39.33 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Tue, 05 Mar 2024 08:39:34 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linaro.org; s=google; t=1709656774; x=1710261574; darn=nongnu.org; h=content-transfer-encoding:mime-version:message-id:date:subject:cc :to:from:from:to:cc:subject:date:message-id:reply-to; bh=uGDboQdgatOQ/3wEUF/IZCqMlV4D3UaeoJrF99IpsFg=; b=nNuA1jmd9kjeGc+qE4GBLfxD3almI6GnTlDY6MVh+swbnF5u9dEfflS1MS6TXIG0Sy YBzF2nB0TuL29Gn3Tjxk/Were3paDBG+D2SJWN/XPyTCxUsREYY/ZhHxHJkkkE0C8ySU CCIx9UtMtehLFTjzdVzi63Vsne/6REqWqs9fPtISUI1vhX7sP6XhC+bQBNlYCNO4vOtX XA0ZedXzAn2SXzsYAVz0BelzqNqXYpcH6r8GrNBJkwCJDdi9O3yyoEXAVkBdeCc73waT 56GbdHVGOpfZUq6NbOBx438BkPQr9A7fovPUY1YdcvXOvsiPXXWKDsORisl9E5zNP+Yg Tw6Q== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1709656774; x=1710261574; h=content-transfer-encoding:mime-version:message-id:date:subject:cc :to:from:x-gm-message-state:from:to:cc:subject:date:message-id :reply-to; bh=uGDboQdgatOQ/3wEUF/IZCqMlV4D3UaeoJrF99IpsFg=; b=PbvaYnNEXqwFoLDfB3wUg3kG1VJ9wqfMnhTMv6vHlo0f7c+Hu8AFxgM8JCDCjxAfD2 44qgTrg/HemaKVdOxiDet5/5bwNT6biMyGjHbkMaIuw2wBu+HeaCOTcIHM7pZSTjsngA KePWsVpfy7oykPfnoCeyxfu2o5qRvpt8ujrwYJQGHwtxka1FzJD8qyj8DsnyAA6flQ3m eZwAQfG3lGXI4rIksKDLekGso6uspMk/AfsuXi5jV4bZQ8o10noGKzP2IEEJrIf7pYMy qfs9FFuP3bxMUhjWelmnNmmi7ficf2Z4z7QGr+3rrTZP9riSR29o8ThAeHDRyo9VjIdK UucQ== X-Gm-Message-State: AOJu0YyaHnxKm1tAW7JwV+ouaTCkW3eJ1B/yoPwAqZLKg4TNrAtHLTBi gZPEpkDwAtIBrrp+VOHiD15Z5WsjL3V/DYWJNb8d57MOoP+o7iJxurnif46IlNfZqlSXwp73N8w M X-Google-Smtp-Source: AGHT+IEfE5i/TraOMlGbGApwTbJESf3Qf2CD1uuUqhzXWni88eZC9jSrNZNyWOThvjlx4LuzUA39WA== X-Received: by 2002:a17:90a:c397:b0:29b:789:28d0 with SMTP id h23-20020a17090ac39700b0029b078928d0mr3800125pjt.22.1709656774566; Tue, 05 Mar 2024 08:39:34 -0800 (PST) From: Richard Henderson To: qemu-devel@nongnu.org Cc: qemu-arm@nongnu.org, qemu-stable@nongnu.org Subject: [PATCH v2] target/arm: Fix 32-bit SMOPA Date: Tue, 5 Mar 2024 06:39:31 -1000 Message-Id: <20240305163931.242795-1-richard.henderson@linaro.org> X-Mailer: git-send-email 2.34.1 MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Received-SPF: pass (zohomail.com: domain of gnu.org designates 209.51.188.17 as permitted sender) client-ip=209.51.188.17; envelope-from=qemu-devel-bounces+importer=patchew.org@nongnu.org; helo=lists.gnu.org; Received-SPF: pass client-ip=2607:f8b0:4864:20::1036; envelope-from=richard.henderson@linaro.org; helo=mail-pj1-x1036.google.com X-Spam_score_int: -20 X-Spam_score: -2.1 X-Spam_bar: -- X-Spam_report: (-2.1 / 5.0 requ) BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1, RCVD_IN_DNSWL_NONE=-0.0001, SPF_HELO_NONE=0.001, SPF_PASS=-0.001, T_SCC_BODY_TEXT_LINE=-0.01 autolearn=ham autolearn_force=no X-Spam_action: no action X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+importer=patchew.org@nongnu.org Sender: qemu-devel-bounces+importer=patchew.org@nongnu.org X-ZohoMail-DKIM: pass (identity @linaro.org) X-ZM-MESSAGEID: 1709656838862100003 Content-Type: text/plain; charset="utf-8" While the 8-bit input elements are sequential in the input vector, the 32-bit output elements are not sequential in the output matrix. Do not attempt to compute 2 32-bit outputs at the same time. Cc: qemu-stable@nongnu.org Fixes: 23a5e3859f5 ("target/arm: Implement SME integer outer product") Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2083 Signed-off-by: Richard Henderson Reviewed-by: Philippe Mathieu-Daud=C3=A9 --- v2: Fixed endian issue; double-checked on s390x. --- target/arm/tcg/sme_helper.c | 77 ++++++++++++++++++------------- tests/tcg/aarch64/sme-smopa-1.c | 47 +++++++++++++++++++ tests/tcg/aarch64/sme-smopa-2.c | 54 ++++++++++++++++++++++ tests/tcg/aarch64/Makefile.target | 2 +- 4 files changed, 147 insertions(+), 33 deletions(-) create mode 100644 tests/tcg/aarch64/sme-smopa-1.c create mode 100644 tests/tcg/aarch64/sme-smopa-2.c diff --git a/target/arm/tcg/sme_helper.c b/target/arm/tcg/sme_helper.c index 904bfdac43..e2e0575039 100644 --- a/target/arm/tcg/sme_helper.c +++ b/target/arm/tcg/sme_helper.c @@ -1083,11 +1083,32 @@ void HELPER(sme_bfmopa)(void *vza, void *vzn, void = *vzm, void *vpn, } } =20 -typedef uint64_t IMOPFn(uint64_t, uint64_t, uint64_t, uint8_t, bool); +typedef uint32_t IMOPFn32(uint32_t, uint32_t, uint32_t, uint8_t, bool); +static inline void do_imopa_s(uint32_t *za, uint32_t *zn, uint32_t *zm, + uint8_t *pn, uint8_t *pm, + uint32_t desc, IMOPFn32 *fn) +{ + intptr_t row, col, oprsz =3D simd_oprsz(desc) / 4; + bool neg =3D simd_data(desc); =20 -static inline void do_imopa(uint64_t *za, uint64_t *zn, uint64_t *zm, - uint8_t *pn, uint8_t *pm, - uint32_t desc, IMOPFn *fn) + for (row =3D 0; row < oprsz; ++row) { + uint8_t pa =3D (pn[H1(row >> 1)] >> ((row & 1) * 4)) & 0xf; + uint32_t *za_row =3D &za[tile_vslice_index(row)]; + uint32_t n =3D zn[H4(row)]; + + for (col =3D 0; col < oprsz; ++col) { + uint8_t pb =3D pm[H1(col >> 1)] >> ((col & 1) * 4); + uint32_t *a =3D &za_row[H4(col)]; + + *a =3D fn(n, zm[H4(col)], *a, pa & pb, neg); + } + } +} + +typedef uint64_t IMOPFn64(uint64_t, uint64_t, uint64_t, uint8_t, bool); +static inline void do_imopa_d(uint64_t *za, uint64_t *zn, uint64_t *zm, + uint8_t *pn, uint8_t *pm, + uint32_t desc, IMOPFn64 *fn) { intptr_t row, col, oprsz =3D simd_oprsz(desc) / 8; bool neg =3D simd_data(desc); @@ -1107,25 +1128,16 @@ static inline void do_imopa(uint64_t *za, uint64_t = *zn, uint64_t *zm, } =20 #define DEF_IMOP_32(NAME, NTYPE, MTYPE) \ -static uint64_t NAME(uint64_t n, uint64_t m, uint64_t a, uint8_t p, bool n= eg) \ +static uint32_t NAME(uint32_t n, uint32_t m, uint32_t a, uint8_t p, bool n= eg) \ { = \ - uint32_t sum0 =3D 0, sum1 =3D 0; = \ + uint32_t sum =3D 0; = \ /* Apply P to N as a mask, making the inactive elements 0. */ = \ n &=3D expand_pred_b(p); = \ - sum0 +=3D (NTYPE)(n >> 0) * (MTYPE)(m >> 0); = \ - sum0 +=3D (NTYPE)(n >> 8) * (MTYPE)(m >> 8); = \ - sum0 +=3D (NTYPE)(n >> 16) * (MTYPE)(m >> 16); = \ - sum0 +=3D (NTYPE)(n >> 24) * (MTYPE)(m >> 24); = \ - sum1 +=3D (NTYPE)(n >> 32) * (MTYPE)(m >> 32); = \ - sum1 +=3D (NTYPE)(n >> 40) * (MTYPE)(m >> 40); = \ - sum1 +=3D (NTYPE)(n >> 48) * (MTYPE)(m >> 48); = \ - sum1 +=3D (NTYPE)(n >> 56) * (MTYPE)(m >> 56); = \ - if (neg) { = \ - sum0 =3D (uint32_t)a - sum0, sum1 =3D (uint32_t)(a >> 32) - sum1; = \ - } else { = \ - sum0 =3D (uint32_t)a + sum0, sum1 =3D (uint32_t)(a >> 32) + sum1; = \ - } = \ - return ((uint64_t)sum1 << 32) | sum0; = \ + sum +=3D (NTYPE)(n >> 0) * (MTYPE)(m >> 0); = \ + sum +=3D (NTYPE)(n >> 8) * (MTYPE)(m >> 8); = \ + sum +=3D (NTYPE)(n >> 16) * (MTYPE)(m >> 16); = \ + sum +=3D (NTYPE)(n >> 24) * (MTYPE)(m >> 24); = \ + return neg ? a - sum : a + sum; = \ } =20 #define DEF_IMOP_64(NAME, NTYPE, MTYPE) \ @@ -1151,16 +1163,17 @@ DEF_IMOP_64(umopa_d, uint16_t, uint16_t) DEF_IMOP_64(sumopa_d, int16_t, uint16_t) DEF_IMOP_64(usmopa_d, uint16_t, int16_t) =20 -#define DEF_IMOPH(NAME) \ - void HELPER(sme_##NAME)(void *vza, void *vzn, void *vzm, void *vpn, = \ - void *vpm, uint32_t desc) = \ - { do_imopa(vza, vzn, vzm, vpn, vpm, desc, NAME); } +#define DEF_IMOPH(NAME, S) \ + void HELPER(sme_##NAME##_##S)(void *vza, void *vzn, void *vzm, = \ + void *vpn, void *vpm, uint32_t desc) = \ + { do_imopa_##S(vza, vzn, vzm, vpn, vpm, desc, NAME##_##S); } =20 -DEF_IMOPH(smopa_s) -DEF_IMOPH(umopa_s) -DEF_IMOPH(sumopa_s) -DEF_IMOPH(usmopa_s) -DEF_IMOPH(smopa_d) -DEF_IMOPH(umopa_d) -DEF_IMOPH(sumopa_d) -DEF_IMOPH(usmopa_d) +DEF_IMOPH(smopa, s) +DEF_IMOPH(umopa, s) +DEF_IMOPH(sumopa, s) +DEF_IMOPH(usmopa, s) + +DEF_IMOPH(smopa, d) +DEF_IMOPH(umopa, d) +DEF_IMOPH(sumopa, d) +DEF_IMOPH(usmopa, d) diff --git a/tests/tcg/aarch64/sme-smopa-1.c b/tests/tcg/aarch64/sme-smopa-= 1.c new file mode 100644 index 0000000000..c62d5e0007 --- /dev/null +++ b/tests/tcg/aarch64/sme-smopa-1.c @@ -0,0 +1,47 @@ +#include +#include + +int main() +{ + static const int cmp[4][4] =3D { + { 110, 134, 158, 182 }, + { 390, 478, 566, 654 }, + { 670, 822, 974, 1126 }, + { 950, 1166, 1382, 1598 } + }; + int dst[4][4]; + int *tmp =3D &dst[0][0]; + + asm volatile( + ".arch armv8-r+sme\n\t" + "smstart\n\t" + "index z0.b, #0, #1\n\t" + "movprfx z1, z0\n\t" + "add z1.b, z1.b, #16\n\t" + "ptrue p0.b\n\t" + "smopa za0.s, p0/m, p0/m, z0.b, z1.b\n\t" + "ptrue p0.s, vl4\n\t" + "mov w12, #0\n\t" + "st1w { za0h.s[w12, #0] }, p0, [%0]\n\t" + "add %0, %0, #16\n\t" + "st1w { za0h.s[w12, #1] }, p0, [%0]\n\t" + "add %0, %0, #16\n\t" + "st1w { za0h.s[w12, #2] }, p0, [%0]\n\t" + "add %0, %0, #16\n\t" + "st1w { za0h.s[w12, #3] }, p0, [%0]\n\t" + "smstop" + : "+r"(tmp) : : "memory"); + + if (memcmp(cmp, dst, sizeof(dst)) =3D=3D 0) { + return 0; + } + + /* See above for correct results. */ + for (int i =3D 0; i < 4; ++i) { + for (int j =3D 0; j < 4; ++j) { + printf("%6d", dst[i][j]); + } + printf("\n"); + } + return 1; +} diff --git a/tests/tcg/aarch64/sme-smopa-2.c b/tests/tcg/aarch64/sme-smopa-= 2.c new file mode 100644 index 0000000000..c9f48c3bfc --- /dev/null +++ b/tests/tcg/aarch64/sme-smopa-2.c @@ -0,0 +1,54 @@ +#include +#include + +int main() +{ + static const long cmp[4][4] =3D { + { 110, 134, 158, 182 }, + { 390, 478, 566, 654 }, + { 670, 822, 974, 1126 }, + { 950, 1166, 1382, 1598 } + }; + long dst[4][4]; + long *tmp =3D &dst[0][0]; + long svl; + + /* Validate that we have a wide enough vector for 4 elements. */ + asm(".arch armv8-r+sme-i64\n\trdsvl %0, #1" : "=3Dr"(svl)); + if (svl < 32) { + return 0; + } + + asm volatile( + "smstart\n\t" + "index z0.h, #0, #1\n\t" + "movprfx z1, z0\n\t" + "add z1.h, z1.h, #16\n\t" + "ptrue p0.b\n\t" + "smopa za0.d, p0/m, p0/m, z0.h, z1.h\n\t" + "ptrue p0.d, vl4\n\t" + "mov w12, #0\n\t" + "st1d { za0h.d[w12, #0] }, p0, [%0]\n\t" + "add %0, %0, #32\n\t" + "st1d { za0h.d[w12, #1] }, p0, [%0]\n\t" + "mov w12, #2\n\t" + "add %0, %0, #32\n\t" + "st1d { za0h.d[w12, #0] }, p0, [%0]\n\t" + "add %0, %0, #32\n\t" + "st1d { za0h.d[w12, #1] }, p0, [%0]\n\t" + "smstop" + : "+r"(tmp) : : "memory"); + + if (memcmp(cmp, dst, sizeof(dst)) =3D=3D 0) { + return 0; + } + + /* See above for correct results. */ + for (int i =3D 0; i < 4; ++i) { + for (int j =3D 0; j < 4; ++j) { + printf("%6ld", dst[i][j]); + } + printf("\n"); + } + return 1; +} diff --git a/tests/tcg/aarch64/Makefile.target b/tests/tcg/aarch64/Makefile= .target index cded1d01fc..ea3e232e65 100644 --- a/tests/tcg/aarch64/Makefile.target +++ b/tests/tcg/aarch64/Makefile.target @@ -67,7 +67,7 @@ endif =20 # SME Tests ifneq ($(CROSS_AS_HAS_ARMV9_SME),) -AARCH64_TESTS +=3D sme-outprod1 +AARCH64_TESTS +=3D sme-outprod1 sme-smopa-1 sme-smopa-2 endif =20 # System Registers Tests --=20 2.34.1