{"id":2234679,"url":"http://patchwork.ozlabs.org/api/1.2/patches/2234679/?format=json","web_url":"http://patchwork.ozlabs.org/project/qemu-devel/patch/20260507234413.643512-23-richard.henderson@linaro.org/","project":{"id":14,"url":"http://patchwork.ozlabs.org/api/1.2/projects/14/?format=json","name":"QEMU Development","link_name":"qemu-devel","list_id":"qemu-devel.nongnu.org","list_email":"qemu-devel@nongnu.org","web_url":"","scm_url":"","webscm_url":"","list_archive_url":"","list_archive_url_format":"","commit_url_format":""},"msgid":"<20260507234413.643512-23-richard.henderson@linaro.org>","list_archive_url":null,"date":"2026-05-07T23:43:35","name":"[v4,22/60] target/arm: Implement BF1CVTL, BF1CVTL2, BF2CVTL, BF2CVTL2 for AdvSIMD","commit_ref":null,"pull_url":null,"state":"new","archived":false,"hash":"0c92a2732c92a51688f6539ac47d7676299b75aa","submitter":{"id":72104,"url":"http://patchwork.ozlabs.org/api/1.2/people/72104/?format=json","name":"Richard Henderson","email":"richard.henderson@linaro.org"},"delegate":null,"mbox":"http://patchwork.ozlabs.org/project/qemu-devel/patch/20260507234413.643512-23-richard.henderson@linaro.org/mbox/","series":[{"id":503296,"url":"http://patchwork.ozlabs.org/api/1.2/series/503296/?format=json","web_url":"http://patchwork.ozlabs.org/project/qemu-devel/list/?series=503296","date":"2026-05-07T23:43:14","name":"target/arm: Implement FEAT_FP8","version":4,"mbox":"http://patchwork.ozlabs.org/series/503296/mbox/"}],"comments":"http://patchwork.ozlabs.org/api/patches/2234679/comments/","check":"pending","checks":"http://patchwork.ozlabs.org/api/patches/2234679/checks/","tags":{},"related":[],"headers":{"Return-Path":"","X-Original-To":"incoming@patchwork.ozlabs.org","Delivered-To":"patchwork-incoming@legolas.ozlabs.org","Authentication-Results":["legolas.ozlabs.org;\n\tdkim=pass (2048-bit key;\n unprotected) header.d=linaro.org header.i=@linaro.org header.a=rsa-sha256\n header.s=google header.b=dEpsVfqP;\n\tdkim-atps=neutral","legolas.ozlabs.org;\n spf=pass (sender SPF authorized) smtp.mailfrom=nongnu.org\n (client-ip=209.51.188.17; helo=lists1p.gnu.org;\n envelope-from=qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org;\n receiver=patchwork.ozlabs.org)"],"Received":["from lists1p.gnu.org (lists1p.gnu.org [209.51.188.17])\n\t(using TLSv1.2 with cipher ECDHE-ECDSA-AES256-GCM-SHA384 (256/256 bits))\n\t(No client certificate requested)\n\tby legolas.ozlabs.org (Postfix) with ESMTPS id 4gBTWZ2Fyrz1yCg\n\tfor ; Fri, 08 May 2026 09:50:02 +1000 (AEST)","from localhost ([::1] helo=lists1p.gnu.org)\n\tby lists1p.gnu.org with esmtp (Exim 4.90_1)\n\t(envelope-from )\n\tid 1wL8OX-0000Ts-Vo; Thu, 07 May 2026 19:44:42 -0400","from eggs.gnu.org ([2001:470:142:3::10])\n by lists1p.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256)\n (Exim 4.90_1) (envelope-from )\n id 1wL8OU-0000Pl-Kh\n for qemu-devel@nongnu.org; Thu, 07 May 2026 19:44:38 -0400","from mail-ot1-x329.google.com ([2607:f8b0:4864:20::329])\n by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_128_GCM_SHA256:128)\n (Exim 4.90_1) (envelope-from )\n id 1wL8OS-000166-8h\n for qemu-devel@nongnu.org; Thu, 07 May 2026 19:44:38 -0400","by mail-ot1-x329.google.com with SMTP id\n 46e09a7af769-7dbd23bc684so908807a34.2\n for ; Thu, 07 May 2026 16:44:35 -0700 (PDT)","from stoup.attlocal.net ([2600:381:c938:6375:9641:bbb2:a93a:bb4c])\n by smtp.gmail.com with ESMTPSA id\n 46e09a7af769-7e367d8feb1sm84320a34.23.2026.05.07.16.44.34\n (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256);\n Thu, 07 May 2026 16:44:34 -0700 (PDT)"],"DKIM-Signature":"v=1; a=rsa-sha256; c=relaxed/relaxed;\n d=linaro.org; s=google; t=1778197475; x=1778802275; darn=nongnu.org;\n h=content-transfer-encoding:mime-version:references:in-reply-to\n :message-id:date:subject:cc:to:from:from:to:cc:subject:date\n :message-id:reply-to;\n bh=TSBFCaQ/FzZ2EJFRAwhSYWznbxEVEf4wc9Dmae+R1i8=;\n b=dEpsVfqPHX63FVyMmxkPGigb1TycGahDCISXW26s7wjjSksFO8ITfK/M5ZCGwRA2mX\n CjrQAi0rdsSwkOc4VYb6UqoOnJNMGNUDKLRPV5ADwh0g2nnETsgHvLhwkGl/4/Pdg5sx\n sj/lRURCsfsw1sO5+DSlDDyd4XFQrVnxhNFRLyl1cp9jOb+Ug62LK4k2ZA2FYc8UkkTo\n M3bl+o7Vs9wM6fMsDKsfNyxK7Kn527WaTQnMmT+H8SyBsYa2y8bD0Ml+oCHSCdimeSUD\n jRf8jdR1QUhrHEish/betPdqZtyrdhw5f1IGuxQ5gC+6Z6TYl6KYXlY5mUSWgk80qRy7\n pIow==","X-Google-DKIM-Signature":"v=1; a=rsa-sha256; c=relaxed/relaxed;\n d=1e100.net; s=20251104; t=1778197475; x=1778802275;\n h=content-transfer-encoding:mime-version:references:in-reply-to\n :message-id:date:subject:cc:to:from:x-gm-gg:x-gm-message-state:from\n :to:cc:subject:date:message-id:reply-to;\n bh=TSBFCaQ/FzZ2EJFRAwhSYWznbxEVEf4wc9Dmae+R1i8=;\n b=LsS2PnC4FByaAIS6kdYiG5d6KHi8Vm8H6M/1Q5iWIUtkuievHimd35LnpJwAP4RPz9\n QsxZ+UlCnZicDV/cTLr095Zr/ZJMmLwwEaM/CR2k0YurSqYTOPdgJbYAZ5jOnMaQNzvZ\n /03rlClmkUan8RuyhHFifcGv8yGOAhuyiOdq/P26U+F9WJdte+iE/TI8gWwPGeaxHTVJ\n /E0tJMvOBhPSSvTGQpWEU5EZt2rJm5J1KjU9KrwNRQpZyNkMoIWohFwfl8I4jn6MLr/p\n Uy/PO921MxrNBsINE+9fo4/JlVM5rq3dQkNTPomHkEedzx865KasIPSdZtR+23L/UwkY\n 3nLQ==","X-Gm-Message-State":"AOJu0YwvOeWMIHGr89TTLjRwOwkxfzk52L77Zd9k3Roei23OlexRyxCB\n bDLqZRoltx6LAfhgBpZm2rr7glp/Mn03F/zSickX88Ejxw8WREPYxpeoxFs3+ypxRXCRTzwo2zk\n 8Fg6v","X-Gm-Gg":"AeBDiesc40ue12LW8CSEKKGFW0VkCU8jlCNome3x8VMkcznPJOpallkCxjRB5qfM/4A\n +f3sRpTZJAtwFj7odoNRBiTrRlNCDsdG0+InDPq+bsxW+ARaO660Ajr/8Pgg1AbnDyfIKBMj6kM\n v5wj2etuwlCPsa17DyWwC4pVJnj1J+VfcKDPsARHdde30WGlKZ/qJEVrcn3BbHLk9LgvXLBGoh0\n uF/UnalJ7gMvNYpHJYqctnW/aMSVzkjL07lklzKROJmF8HoR9yg68c1bct44Rfzmpndq4WBhKLq\n PrwBlxq/RbJOWzoKzhMMFSLVrmj0qitnsq1q0Mp5FwrqSyCyIJHC80I9rizYEB9saGp0VQoiqdC\n oQR4NSAbP05ZhhPRP0ovK446Hdl7b5Ky69XMQTpE5guVey1LQQO/vEfcdez8lEIJ/xY/LB+vzWp\n zaW+W3z36MDqbWK7ryIbGjfsC8ZMS0QcCjKLmcoUS4iOc99g==","X-Received":"by 2002:a05:6830:3592:b0:7db:b5c9:2d87 with SMTP id\n 46e09a7af769-7e1def36698mr6491350a34.11.1778197475057;\n Thu, 07 May 2026 16:44:35 -0700 (PDT)","From":"Richard Henderson ","To":"qemu-devel@nongnu.org","Cc":"qemu-arm@nongnu.org","Subject":"[PATCH v4 22/60] target/arm: Implement BF1CVTL, BF1CVTL2, BF2CVTL,\n BF2CVTL2 for AdvSIMD","Date":"Thu, 7 May 2026 18:43:35 -0500","Message-ID":"<20260507234413.643512-23-richard.henderson@linaro.org>","X-Mailer":"git-send-email 2.43.0","In-Reply-To":"<20260507234413.643512-1-richard.henderson@linaro.org>","References":"<20260507234413.643512-1-richard.henderson@linaro.org>","MIME-Version":"1.0","Content-Transfer-Encoding":"8bit","Received-SPF":"pass client-ip=2607:f8b0:4864:20::329;\n envelope-from=richard.henderson@linaro.org; helo=mail-ot1-x329.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,\n DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1,\n RCVD_IN_DNSWL_NONE=-0.0001, SPF_HELO_NONE=0.001,\n SPF_PASS=-0.001 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":"qemu development ","List-Unsubscribe":",\n ","List-Archive":"","List-Post":"","List-Help":"","List-Subscribe":",\n ","Errors-To":"qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org","Sender":"qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org"},"content":"Signed-off-by: Richard Henderson \n---\n target/arm/helper-fp8.h | 14 ++++\n target/arm/tcg/helper-fp8-defs.h | 6 ++\n target/arm/tcg/translate-a64.h | 1 +\n target/arm/tcg/fp8_helper.c | 126 +++++++++++++++++++++++++++++++\n target/arm/tcg/translate-a64.c | 34 +++++++++\n target/arm/tcg/a64.decode | 3 +\n target/arm/tcg/meson.build | 1 +\n 7 files changed, 185 insertions(+)\n create mode 100644 target/arm/helper-fp8.h\n create mode 100644 target/arm/tcg/helper-fp8-defs.h\n create mode 100644 target/arm/tcg/fp8_helper.c","diff":"diff --git a/target/arm/helper-fp8.h b/target/arm/helper-fp8.h\nnew file mode 100644\nindex 0000000000..c45211ba22\n--- /dev/null\n+++ b/target/arm/helper-fp8.h\n@@ -0,0 +1,14 @@\n+/* SPDX-License-Identifier: GPL-2.0-or-later */\n+\n+#ifndef HELPER_FP8_H\n+#define HELPER_FP8_H\n+\n+#include \"exec/helper-proto-common.h\"\n+#include \"exec/helper-gen-common.h\"\n+\n+#define HELPER_H \"tcg/helper-fp8-defs.h\"\n+#include \"exec/helper-proto.h.inc\"\n+#include \"exec/helper-gen.h.inc\"\n+#undef HELPER_H\n+\n+#endif /* HELPER_FP8_H */\ndiff --git a/target/arm/tcg/helper-fp8-defs.h b/target/arm/tcg/helper-fp8-defs.h\nnew file mode 100644\nindex 0000000000..0caaf63749\n--- /dev/null\n+++ b/target/arm/tcg/helper-fp8-defs.h\n@@ -0,0 +1,6 @@\n+/*\n+ * AArch64 FP8 helper definitions\n+ * SPDX-License-Identifier: GPL-2.0-or-later\n+ */\n+\n+DEF_HELPER_FLAGS_4(advsimd_bfcvtl, TCG_CALL_NO_RWG, void, ptr, ptr, env, i32)\ndiff --git a/target/arm/tcg/translate-a64.h b/target/arm/tcg/translate-a64.h\nindex 9c45f89305..35f8d4f82e 100644\n--- a/target/arm/tcg/translate-a64.h\n+++ b/target/arm/tcg/translate-a64.h\n@@ -25,6 +25,7 @@ TCGv_i64 read_cpu_reg_sp(DisasContext *s, int reg, int sf);\n void write_fp_dreg(DisasContext *s, int reg, TCGv_i64 v);\n bool logic_imm_decode_wmask(uint64_t *result, unsigned int immn,\n unsigned int imms, unsigned int immr);\n+bool fpmr_access_check(DisasContext *s);\n bool sve_access_check(DisasContext *s);\n bool sme_enabled_check(DisasContext *s);\n bool sme_enabled_check_with_svcr(DisasContext *s, unsigned);\ndiff --git a/target/arm/tcg/fp8_helper.c b/target/arm/tcg/fp8_helper.c\nnew file mode 100644\nindex 0000000000..7c8c4d6e06\n--- /dev/null\n+++ b/target/arm/tcg/fp8_helper.c\n@@ -0,0 +1,126 @@\n+/*\n+ * AArch64 FP8 Operations\n+ * SPDX-License-Identifier: GPL-2.0-or-later\n+ */\n+\n+#include \"qemu/osdep.h\"\n+#include \"cpu.h\"\n+#include \"internals.h\"\n+#include \"tcg/tcg-gvec-desc.h\"\n+#include \"fpu/softfloat.h\"\n+#include \"fpu/softfloat-parts.h\"\n+#include \"helper-fp8.h\"\n+#include \"vec_internal.h\"\n+\n+#define HELPER_H \"tcg/helper-fp8-defs.h\"\n+#include \"exec/helper-info.c.inc\"\n+\n+typedef enum FPMRType {\n+ OFP8_E5M2 = 0,\n+ OFP8_E4M3 = 1,\n+ Unsupp2 = 2,\n+ Unsupp3 = 3,\n+ Unsupp4 = 4,\n+ Unsupp5 = 5,\n+ Unsupp6 = 6,\n+ Unsupp7 = 7,\n+} FPMRType;\n+\n+typedef struct FP8Context {\n+ float_status stat;\n+ ARMFPStatusFlavour fpst;\n+ FPMRType f8fmt;\n+ int scale;\n+ bool high;\n+} FP8Context;\n+\n+static FP8Context fp8_start(CPUARMState *env, uint32_t desc,\n+ FPMRType f8fmt, int scale)\n+{\n+ ARMFPStatusFlavour fpst = extract32(desc, SIMD_DATA_SHIFT + 2, 4);\n+\n+ FP8Context ret = {\n+ .stat = env->vfp.fp_status[fpst],\n+ .fpst = fpst,\n+ .f8fmt = f8fmt,\n+ .scale = scale,\n+ .high = extract32(desc, SIMD_DATA_SHIFT + 1, 1),\n+ };\n+\n+ set_flush_to_zero(0, &ret.stat);\n+ set_flush_inputs_to_zero(0, &ret.stat);\n+ set_default_nan_mode(true, &ret.stat);\n+ set_float_rounding_mode(float_round_nearest_even, &ret.stat);\n+\n+ return ret;\n+}\n+\n+static void fp8_finish(CPUARMState *env, FP8Context *c)\n+{\n+ int new_flags = get_float_exception_flags(&c->stat);\n+\n+ new_flags &= ~float_flag_input_denormal_used;\n+ float_raise(new_flags, &env->vfp.fp_status[c->fpst]);\n+}\n+\n+static FP8Context fp8_src_start(CPUARMState *env, uint32_t desc, int scale_mask)\n+{\n+ bool issrc2 = extract32(desc, SIMD_DATA_SHIFT, 1);\n+ uint64_t fpmr = env->vfp.fpmr;\n+ FPMRType f8fmt = (issrc2\n+ ? FIELD_EX64(fpmr, FPMR, F8S2)\n+ : FIELD_EX64(fpmr, FPMR, F8S1));\n+ int scale;\n+\n+ scale = fpmr >> (issrc2 ? R_FPMR_LSCALE2_SHIFT : R_FPMR_LSCALE_SHIFT);\n+ scale = -(scale & scale_mask);\n+\n+ return fp8_start(env, desc, f8fmt, scale);\n+}\n+\n+\n+static FloatParts64 fp8_invalid_input(uint8_t x, float_status *s)\n+{\n+ /*\n+ * Invalid input format is treated as snan, then one of the uses\n+ * will convert to default nan and raise invalid.\n+ */\n+ float_raise(float_flag_invalid | float_flag_invalid_snan, s);\n+ return parts64_default_nan(s);\n+}\n+\n+typedef FloatParts64 fp8_input_fn(uint8_t x, float_status *s);\n+\n+static fp8_input_fn * const fp8_input_fmt[8] = {\n+ [0 ... 7] = fp8_invalid_input,\n+ [OFP8_E5M2] = float8_e5m2_unpack_canonical,\n+ [OFP8_E4M3] = float8_e4m3_unpack_canonical,\n+};\n+\n+static bfloat16 fcvt_fp8_to_b16(uint8_t x, fp8_input_fn *f8fmt,\n+ int scale, float_status *s)\n+{\n+ FloatParts64 p = f8fmt(x, s);\n+ p = parts64_scalbn(&p, scale, s);\n+ return bfloat16_round_pack_canonical(&p, s);\n+}\n+\n+void HELPER(advsimd_bfcvtl)(void *vd, void *vn, CPUARMState *env, uint32_t desc)\n+{\n+ FP8Context ctx = fp8_src_start(env, desc, 0x3f);\n+ fp8_input_fn *input_fmt = fp8_input_fmt[ctx.f8fmt];\n+ uint8_t *n = vn, scratch[16];\n+ bfloat16 *d = vd;\n+\n+ if (vd == vn) {\n+ n = memcpy(scratch, vn, 16);\n+ }\n+ n += ctx.high * 8;\n+\n+ for (size_t i = 0; i < 8; ++i) {\n+ d[H2(i)] = fcvt_fp8_to_b16(n[H1(i)], input_fmt, ctx.scale, &ctx.stat);\n+ }\n+\n+ fp8_finish(env, &ctx);\n+ clear_tail(vd, 16, simd_maxsz(desc));\n+}\ndiff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c\nindex ac18ceeeab..085e7e3b95 100644\n--- a/target/arm/tcg/translate-a64.c\n+++ b/target/arm/tcg/translate-a64.c\n@@ -22,6 +22,7 @@\n #include \"helper-a64.h\"\n #include \"helper-sme.h\"\n #include \"helper-sve.h\"\n+#include \"helper-fp8.h\"\n #include \"translate.h\"\n #include \"translate-a64.h\"\n #include \"tcg/tcg-op.h\"\n@@ -1457,6 +1458,24 @@ static bool fp_access_check(DisasContext *s)\n return fp_access_check_only(s) && nonstreaming_check(s);\n }\n \n+/*\n+ * Check that FPMR access is enabled, for an indirect reference by a\n+ * vector instruction. See CheckFPMREnabled().\n+ */\n+bool fpmr_access_check(DisasContext *s)\n+{\n+ if (s->fpmr_el) {\n+ /*\n+ * While denied direct access to the FPMR raises SystemRegisterTrap\n+ * and targets a specific EL, denied indirect access to the FPMR\n+ * results in a simple UNDEFINED to the default exception level.\n+ */\n+ unallocated_encoding(s);\n+ return false;\n+ }\n+ return true;\n+}\n+\n /*\n * Return <0 for non-supported element sizes, with MO_16 controlled by\n * FEAT_FP16; return 0 for fp disabled; otherwise return >0 for success.\n@@ -10612,6 +10631,21 @@ static bool trans_FCVTL_v(DisasContext *s, arg_qrr_e *a)\n return true;\n }\n \n+static bool do_f8cvt(DisasContext *s, arg_qrr_e *a,\n+ gen_helper_gvec_2_ptr *fn, bool issrc2)\n+{\n+ if (fpmr_access_check(s) && fp_access_check(s)) {\n+ tcg_gen_gvec_2_ptr(vec_full_reg_offset(s, a->rd),\n+ vec_full_reg_offset(s, a->rn),\n+ tcg_env, 16, vec_full_reg_size(s),\n+ issrc2 | (a->q << 1) | (FPST_A64 << 2), fn);\n+ }\n+ return true;\n+}\n+\n+TRANS_FEAT(BF1CVTL, aa64_f8cvt, do_f8cvt, a, gen_helper_advsimd_bfcvtl, false)\n+TRANS_FEAT(BF2CVTL, aa64_f8cvt, do_f8cvt, a, gen_helper_advsimd_bfcvtl, true)\n+\n static bool trans_OK(DisasContext *s, arg_OK *a)\n {\n return true;\ndiff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode\nindex 02c7264cb9..b7aac148f2 100644\n--- a/target/arm/tcg/a64.decode\n+++ b/target/arm/tcg/a64.decode\n@@ -1910,6 +1910,9 @@ URSQRTE_v 0.10 1110 101 00001 11001 0 ..... ..... @qrr_s\n \n FCVTL_v 0.00 1110 0.1 00001 01111 0 ..... ..... @qrr_sd\n \n+BF1CVTL 0.10 1110 101 00001 01111 0 ..... ..... @qrr_h\n+BF2CVTL 0.10 1110 111 00001 01111 0 ..... ..... @qrr_h\n+\n &fcvt_q rd rn esz q shift\n @fcvtq_h . q:1 . ...... 001 .... ...... rn:5 rd:5 \\\n &fcvt_q esz=1 shift=%fcvt_f_sh_h\ndiff --git a/target/arm/tcg/meson.build b/target/arm/tcg/meson.build\nindex d2364aa39c..56be383189 100644\n--- a/target/arm/tcg/meson.build\n+++ b/target/arm/tcg/meson.build\n@@ -46,6 +46,7 @@ arm_ss.add(when: 'TARGET_AARCH64', if_true: files(\n 'sme_helper.c',\n 'sve_helper.c',\n 'vec_helper64.c',\n+ 'fp8_helper.c',\n ))\n \n arm_common_system_ss.add(when: 'CONFIG_ARM_V7M', if_true: files('cpu-v7m.c'))\n","prefixes":["v4","22/60"]}