From patchwork Wed Jan 29 13:39:34 2014 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Peter Maydell X-Patchwork-Id: 23865 Return-Path: X-Original-To: linaro@patches.linaro.org Delivered-To: linaro@patches.linaro.org Received: from mail-yk0-f197.google.com (mail-yk0-f197.google.com [209.85.160.197]) by ip-10-151-82-157.ec2.internal (Postfix) with ESMTPS id 9887B202FA for ; Wed, 29 Jan 2014 13:56:29 +0000 (UTC) Received: by mail-yk0-f197.google.com with SMTP id 20sf9324198yks.0 for ; Wed, 29 Jan 2014 05:56:28 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:delivered-to:from:to:date :message-id:in-reply-to:references:cc:subject:precedence:list-id :list-unsubscribe:list-archive:list-post:list-help:list-subscribe :errors-to:sender:x-original-sender :x-original-authentication-results:mailing-list; bh=AMSwpAJS6RQxjDR9JNI24I6S83wZGtpUZRAt/V1RpF0=; b=cGcM08PfviuSblb1LwxsnFl0+0q16qhgTes+55LWgYZ6maNnuxrPBdlDReT0xubYm1 KTEzQsJc/TEt8B6Q48f4iYRxuvVH9c7JDjdqJB39eDEIvWudnVWomRaO4LgnhjdrECOi rBG3igBiYXAG9ySIfoT41GuEjGRz3DF2i6OaUyYjVn9ALRAaGjQMgfhXsDB1Fmlt1jE+ zYbYchbuI+ncRh04qN59jqS/e2zQSBJ45hBgeS7svRAYwX+SvrA17zNAejjejsHEkCYy 5or2pIfJ7LAujxlAywocQC9Ta4KcEJSO0tkxJFf0uqNLxGroGXK4ehUyQxm0iIaj/TlD NHow== X-Gm-Message-State: ALoCoQlqGEgQtIxZfzB7+9JqXlZ8Q8aaCs4Jn3Nmo4dMCTGERdyg4I51C+EuCY/JwX4OTtxl0Xwe X-Received: by 10.58.230.72 with SMTP id sw8mr1685679vec.5.1391003788801; Wed, 29 Jan 2014 05:56:28 -0800 (PST) MIME-Version: 1.0 X-BeenThere: patchwork-forward@linaro.org Received: by 10.140.94.129 with SMTP id g1ls106179qge.69.gmail; Wed, 29 Jan 2014 05:56:28 -0800 (PST) X-Received: by 10.52.26.148 with SMTP id l20mr1192112vdg.28.1391003788634; Wed, 29 Jan 2014 05:56:28 -0800 (PST) Received: from mail-ve0-f169.google.com (mail-ve0-f169.google.com [209.85.128.169]) by mx.google.com with ESMTPS id us10si805266vcb.134.2014.01.29.05.56.28 for (version=TLSv1 cipher=ECDHE-RSA-RC4-SHA bits=128/128); Wed, 29 Jan 2014 05:56:28 -0800 (PST) Received-SPF: neutral (google.com: 209.85.128.169 is neither permitted nor denied by best guess record for domain of patch+caf_=patchwork-forward=linaro.org@linaro.org) client-ip=209.85.128.169; Received: by mail-ve0-f169.google.com with SMTP id oy12so1199859veb.14 for ; Wed, 29 Jan 2014 05:56:28 -0800 (PST) X-Received: by 10.220.109.1 with SMTP id h1mr6800154vcp.20.1391003788554; Wed, 29 Jan 2014 05:56:28 -0800 (PST) X-Forwarded-To: patchwork-forward@linaro.org X-Forwarded-For: patch@linaro.org patchwork-forward@linaro.org Delivered-To: patch@linaro.org Received: by 10.220.174.196 with SMTP id u4csp114550vcz; Wed, 29 Jan 2014 05:56:28 -0800 (PST) X-Received: by 10.140.92.213 with SMTP id b79mr11410873qge.108.1391003787962; Wed, 29 Jan 2014 05:56:27 -0800 (PST) Received: from lists.gnu.org (lists.gnu.org. [2001:4830:134:3::11]) by mx.google.com with ESMTPS id fy9si1796538qab.101.2014.01.29.05.56.27 for (version=TLSv1 cipher=RC4-SHA bits=128/128); Wed, 29 Jan 2014 05:56:27 -0800 (PST) Received-SPF: pass (google.com: domain of qemu-devel-bounces+patch=linaro.org@nongnu.org designates 2001:4830:134:3::11 as permitted sender) client-ip=2001:4830:134:3::11; Received: from localhost ([::1]:42470 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1W8Vcx-0006wt-Jl for patch@linaro.org; Wed, 29 Jan 2014 08:56:27 -0500 Received: from eggs.gnu.org ([2001:4830:134:3::10]:49223) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1W8VNo-0001OB-0L for qemu-devel@nongnu.org; Wed, 29 Jan 2014 08:40:49 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1W8VNl-0001Nf-HF for qemu-devel@nongnu.org; Wed, 29 Jan 2014 08:40:47 -0500 Received: from mnementh.archaic.org.uk ([2001:8b0:1d0::1]:45250) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1W8VNl-0001I9-8W for qemu-devel@nongnu.org; Wed, 29 Jan 2014 08:40:45 -0500 Received: from pm215 by mnementh.archaic.org.uk with local (Exim 4.80) (envelope-from ) id 1W8VN7-0006vt-HJ; Wed, 29 Jan 2014 13:40:05 +0000 From: Peter Maydell To: Anthony Liguori Date: Wed, 29 Jan 2014 13:39:34 +0000 Message-Id: <1391002805-26596-8-git-send-email-peter.maydell@linaro.org> X-Mailer: git-send-email 1.7.10.4 In-Reply-To: <1391002805-26596-1-git-send-email-peter.maydell@linaro.org> References: <1391002805-26596-1-git-send-email-peter.maydell@linaro.org> X-detected-operating-system: by eggs.gnu.org: Error: Malformed IPv6 address (bad octet value). X-Received-From: 2001:8b0:1d0::1 Cc: Blue Swirl , qemu-devel@nongnu.org, Aurelien Jarno Subject: [Qemu-devel] [PULL 07/38] target-arm: A64: Add SIMD across-lanes instructions X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: , List-Help: , List-Subscribe: , Errors-To: qemu-devel-bounces+patch=linaro.org@nongnu.org Sender: qemu-devel-bounces+patch=linaro.org@nongnu.org X-Removed-Original-Auth: Dkim didn't pass. X-Original-Sender: peter.maydell@linaro.org X-Original-Authentication-Results: mx.google.com; spf=neutral (google.com: 209.85.128.169 is neither permitted nor denied by best guess record for domain of patch+caf_=patchwork-forward=linaro.org@linaro.org) smtp.mail=patch+caf_=patchwork-forward=linaro.org@linaro.org Mailing-list: list patchwork-forward@linaro.org; contact patchwork-forward+owners@linaro.org X-Google-Group-Id: 836684582541 From: Michael Matz Add support for the SIMD "across lanes" instruction group (C3.6.4). Signed-off-by: Michael Matz [PMM: Updated to current codebase, added fp min/max ops, added unallocated encoding checks] Signed-off-by: Peter Maydell Reviewed-by: Richard Henderson --- target-arm/translate-a64.c | 177 ++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 176 insertions(+), 1 deletion(-) diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c index 59e2a85..de4f518 100644 --- a/target-arm/translate-a64.c +++ b/target-arm/translate-a64.c @@ -4883,6 +4883,29 @@ static void disas_simd_zip_trn(DisasContext *s, uint32_t insn) tcg_temp_free_i64(tcg_resh); } +static void do_minmaxop(DisasContext *s, TCGv_i32 tcg_elt1, TCGv_i32 tcg_elt2, + int opc, bool is_min, TCGv_ptr fpst) +{ + /* Helper function for disas_simd_across_lanes: do a single precision + * min/max operation on the specified two inputs, + * and return the result in tcg_elt1. + */ + if (opc == 0xc) { + if (is_min) { + gen_helper_vfp_minnums(tcg_elt1, tcg_elt1, tcg_elt2, fpst); + } else { + gen_helper_vfp_maxnums(tcg_elt1, tcg_elt1, tcg_elt2, fpst); + } + } else { + assert(opc == 0xf); + if (is_min) { + gen_helper_vfp_mins(tcg_elt1, tcg_elt1, tcg_elt2, fpst); + } else { + gen_helper_vfp_maxs(tcg_elt1, tcg_elt1, tcg_elt2, fpst); + } + } +} + /* C3.6.4 AdvSIMD across lanes * 31 30 29 28 24 23 22 21 17 16 12 11 10 9 5 4 0 * +---+---+---+-----------+------+-----------+--------+-----+------+------+ @@ -4891,7 +4914,159 @@ static void disas_simd_zip_trn(DisasContext *s, uint32_t insn) */ static void disas_simd_across_lanes(DisasContext *s, uint32_t insn) { - unsupported_encoding(s, insn); + int rd = extract32(insn, 0, 5); + int rn = extract32(insn, 5, 5); + int size = extract32(insn, 22, 2); + int opcode = extract32(insn, 12, 5); + bool is_q = extract32(insn, 30, 1); + bool is_u = extract32(insn, 29, 1); + bool is_fp = false; + bool is_min = false; + int esize; + int elements; + int i; + TCGv_i64 tcg_res, tcg_elt; + + switch (opcode) { + case 0x1b: /* ADDV */ + if (is_u) { + unallocated_encoding(s); + return; + } + /* fall through */ + case 0x3: /* SADDLV, UADDLV */ + case 0xa: /* SMAXV, UMAXV */ + case 0x1a: /* SMINV, UMINV */ + if (size == 3 || (size == 2 && !is_q)) { + unallocated_encoding(s); + return; + } + break; + case 0xc: /* FMAXNMV, FMINNMV */ + case 0xf: /* FMAXV, FMINV */ + if (!is_u || !is_q || extract32(size, 0, 1)) { + unallocated_encoding(s); + return; + } + /* Bit 1 of size field encodes min vs max, and actual size is always + * 32 bits: adjust the size variable so following code can rely on it + */ + is_min = extract32(size, 1, 1); + is_fp = true; + size = 2; + break; + default: + unallocated_encoding(s); + return; + } + + esize = 8 << size; + elements = (is_q ? 128 : 64) / esize; + + tcg_res = tcg_temp_new_i64(); + tcg_elt = tcg_temp_new_i64(); + + /* These instructions operate across all lanes of a vector + * to produce a single result. We can guarantee that a 64 + * bit intermediate is sufficient: + * + for [US]ADDLV the maximum element size is 32 bits, and + * the result type is 64 bits + * + for FMAX*V, FMIN*V, ADDV the intermediate type is the + * same as the element size, which is 32 bits at most + * For the integer operations we can choose to work at 64 + * or 32 bits and truncate at the end; for simplicity + * we use 64 bits always. The floating point + * ops do require 32 bit intermediates, though. + */ + if (!is_fp) { + read_vec_element(s, tcg_res, rn, 0, size | (is_u ? 0 : MO_SIGN)); + + for (i = 1; i < elements; i++) { + read_vec_element(s, tcg_elt, rn, i, size | (is_u ? 0 : MO_SIGN)); + + switch (opcode) { + case 0x03: /* SADDLV / UADDLV */ + case 0x1b: /* ADDV */ + tcg_gen_add_i64(tcg_res, tcg_res, tcg_elt); + break; + case 0x0a: /* SMAXV / UMAXV */ + tcg_gen_movcond_i64(is_u ? TCG_COND_GEU : TCG_COND_GE, + tcg_res, + tcg_res, tcg_elt, tcg_res, tcg_elt); + break; + case 0x1a: /* SMINV / UMINV */ + tcg_gen_movcond_i64(is_u ? TCG_COND_LEU : TCG_COND_LE, + tcg_res, + tcg_res, tcg_elt, tcg_res, tcg_elt); + break; + break; + default: + g_assert_not_reached(); + } + + } + } else { + /* Floating point ops which work on 32 bit (single) intermediates. + * Note that correct NaN propagation requires that we do these + * operations in exactly the order specified by the pseudocode. + */ + TCGv_i32 tcg_elt1 = tcg_temp_new_i32(); + TCGv_i32 tcg_elt2 = tcg_temp_new_i32(); + TCGv_i32 tcg_elt3 = tcg_temp_new_i32(); + TCGv_ptr fpst = get_fpstatus_ptr(); + + assert(esize == 32); + assert(elements == 4); + + read_vec_element(s, tcg_elt, rn, 0, MO_32); + tcg_gen_trunc_i64_i32(tcg_elt1, tcg_elt); + read_vec_element(s, tcg_elt, rn, 1, MO_32); + tcg_gen_trunc_i64_i32(tcg_elt2, tcg_elt); + + do_minmaxop(s, tcg_elt1, tcg_elt2, opcode, is_min, fpst); + + read_vec_element(s, tcg_elt, rn, 2, MO_32); + tcg_gen_trunc_i64_i32(tcg_elt2, tcg_elt); + read_vec_element(s, tcg_elt, rn, 3, MO_32); + tcg_gen_trunc_i64_i32(tcg_elt3, tcg_elt); + + do_minmaxop(s, tcg_elt2, tcg_elt3, opcode, is_min, fpst); + + do_minmaxop(s, tcg_elt1, tcg_elt2, opcode, is_min, fpst); + + tcg_gen_extu_i32_i64(tcg_res, tcg_elt1); + tcg_temp_free_i32(tcg_elt1); + tcg_temp_free_i32(tcg_elt2); + tcg_temp_free_i32(tcg_elt3); + tcg_temp_free_ptr(fpst); + } + + tcg_temp_free_i64(tcg_elt); + + /* Now truncate the result to the width required for the final output */ + if (opcode == 0x03) { + /* SADDLV, UADDLV: result is 2*esize */ + size++; + } + + switch (size) { + case 0: + tcg_gen_ext8u_i64(tcg_res, tcg_res); + break; + case 1: + tcg_gen_ext16u_i64(tcg_res, tcg_res); + break; + case 2: + tcg_gen_ext32u_i64(tcg_res, tcg_res); + break; + case 3: + break; + default: + g_assert_not_reached(); + } + + write_fp_dreg(s, rd, tcg_res); + tcg_temp_free_i64(tcg_res); } /* C3.6.5 AdvSIMD copy