From nobody Mon Feb 9 03:52:47 2026 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=gmail.com ARC-Seal: i=1; a=rsa-sha256; t=1769412782; cv=none; d=zohomail.com; s=zohoarc; b=D2kWnvEa+jx0DKWvNeyvDz6PFdiw386FPrONOjtAhdHoX0GAaHhL43NyA3BvwrBViXuKefEPZBOwPovRVvfEUkgEMqHcyqTcKrVYXcnINT0/si2ZprzY3LuC5oZDGIXG5lll4z0SyJ/pgJv/1WGJ0QvPpF3Gq9YJYek9bG/O+vQ= ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=zohomail.com; s=zohoarc; t=1769412782; h=Content-Transfer-Encoding:Cc:Cc:Date:Date:From:From:In-Reply-To:List-Subscribe:List-Post:List-Id:List-Archive:List-Help:List-Unsubscribe:MIME-Version:Message-ID:References:Sender:Subject:Subject:To:To:Message-Id:Reply-To; bh=eKGZrzBEZomEKs9twQKesDOc8wmIyiWzTpm7XRfkzkI=; b=T5OciIQkaoHiOqBWY+G99cWcaKmmg3EG54mYRNpKSKGMFGDeHmazUZU6gfujwJbyOaT+NRpuUwfQqraLMvU8xR1qf7Nvpz14n2qT5ovtpJYWO3AbjA2BDAfRnLoarRa7aLYTHz0eqt6oZOGI3jJtP+XGIc+FyCXH8+6MdQvjQAE= 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 1769412782740102.54242698741325; Sun, 25 Jan 2026 23:33:02 -0800 (PST) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1vkGt7-0004uA-ND; Mon, 26 Jan 2026 02:19:53 -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 1vkGf3-0002rH-2m for qemu-devel@nongnu.org; Mon, 26 Jan 2026 02:05:42 -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 1vkGes-00048e-9F for qemu-devel@nongnu.org; Mon, 26 Jan 2026 02:05:15 -0500 Received: by mail-pj1-x1036.google.com with SMTP id 98e67ed59e1d1-34c718c5481so2278678a91.3 for ; Sun, 25 Jan 2026 23:04:53 -0800 (PST) Received: from brahms.. (fp93c00990.tkyc601.ap.nuro.jp. [147.192.9.144]) by smtp.gmail.com with ESMTPSA id d2e1a72fcca58-823187716ebsm8661487b3a.66.2026.01.25.23.04.48 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Sun, 25 Jan 2026 23:04:51 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1769411092; x=1770015892; darn=nongnu.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:from:to:cc:subject:date :message-id:reply-to; bh=eKGZrzBEZomEKs9twQKesDOc8wmIyiWzTpm7XRfkzkI=; b=hUL0/QkRdtoMkM4N5lwtq68ZLzkrcAv9mobP3ptsoQ+vfQlMRav5SihI4Gb20zVGVj brDMnbeRFnSiYfdFLB48DWqh30lID6D02TeYOGwZR2cXnBvb8nysvRJSshXlq8Y9IR/d Ux764yTTCs3qCI/LJu7mvsxo4IfZW4awJ9VQpHF0kXm5Q+aFVc1VV5xhMqz6yrvJBfEK XISMCThuSiUXN06Fsl9elcqbk+BVox0kIk+KdKzQ6H6FinOhkbWdbQqOZM1EQ/tof5HB z1/ZUFVTe5aegIPgzNjdxLUjmo2rCCSL4lNmWegLd214kDEVipXtbz6bBlC1hE+s/Xz7 wbzw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1769411092; x=1770015892; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-gg:x-gm-message-state:from :to:cc:subject:date:message-id:reply-to; bh=eKGZrzBEZomEKs9twQKesDOc8wmIyiWzTpm7XRfkzkI=; b=fED2wuPE7K/1nteOwyM8FUAj6Dl3xGJs16ZWj+2q/HzudPkMzN4uFmY2NdVVakAxZd icq61Kk29l6qLqY3h8na5+Z5WixRI/IixeAs7PeXieswaEXB0RO8CZRYo9YbhBARCedd 1JUZ2lIasTvMnpZmqb+ggKhfG+IIAsc6q6v7qlmAT2jbRcb7BU+v9JrU/77izD/S0rnK eR/IOSZTNsph+02SnojxeAch5Yt2xaME5e4B/QO251h6FSp9GTJ/jTAyLz0Zxrdh2vCV 576YLeSeWOouj8fOQ9CTfQ/RXFUeIYJHJ6EPllWu2xgp4qhRNRwL8jPP5AAljDQgCZ8C Fe3g== X-Gm-Message-State: AOJu0Yy78EqnqnVcUvZGwxkZYy463k/j4W//JZw+vN/awPOhqkV+t98i vkyeGzRTkxgMGpfg3UssOeY3LB2oh79fOVBobODM9f1AO0XnVyJNyBcNmWvpad9r X-Gm-Gg: AZuq6aJn/nTTodHky3JtsWneV3vvOlyJdKdcjg5bUpMReP4lH5mQKoKYEr/uaSBYrCh kCzIbGoydaoLEWIRpNqFM4zTXIYHYuddedTFn4ccXb3bFIqPvKZ/N5XE0TZ1Ghxb3qBQnJ9BkkO 5KlWLxRrxkDVdkvndNYqKEt4MsofnivTUEhhhWonfoXvugTsJE2GnmFWrtlLU9uI1ckyDIKbFKa zwv+ZA9/tAdrMbtp8UHHp9KP5+cumiCTmrJItybI4x9SfcB00MDPhwCYyMHElN0L07cHHN683GH rgLGXaSE9UzmNe5XduRrahHiiaqmlyfpDc3DNBIokqMOr3TJZos5M2qeiDMM+1auFx/HfhUsiI9 CJuK4p9/x35SEiaALfI9xrj4XL7JhHfM49X+WWjwyazs+fsLsHfCTrZmJ5mShjszT7A5of9bbWY LzIeg4Tk4NYKR2ZwrbaTMgtFF39oB68r7WrzLosRBR4Q== X-Received: by 2002:a17:90b:1b10:b0:34c:75d1:6f90 with SMTP id 98e67ed59e1d1-353c41789c1mr3781364a91.17.1769411092379; Sun, 25 Jan 2026 23:04:52 -0800 (PST) From: Kohei Tokunaga To: qemu-devel@nongnu.org Cc: =?UTF-8?q?Alex=20Benn=C3=A9e?= , =?UTF-8?q?Philippe=20Mathieu-Daud=C3=A9?= , Thomas Huth , Paolo Bonzini , Richard Henderson , =?UTF-8?q?Marc-Andr=C3=A9=20Lureau?= , =?UTF-8?q?Daniel=20P=20=2E=20Berrang=C3=A9?= , WANG Xuerui , Aurelien Jarno , Huacai Chen , Jiaxun Yang , Aleksandar Rikalo , Palmer Dabbelt , Alistair Francis , Stefan Weil , Kohei Tokunaga , qemu-arm@nongnu.org, qemu-riscv@nongnu.org, Stefan Hajnoczi , Pierrick Bouvier Subject: [PATCH v4 10/33] tcg/wasm64: Add load and store instructions Date: Mon, 26 Jan 2026 07:03:23 +0000 Message-ID: <6573ef95bf4d2bd6d454b5cb3fe80026b19350b2.1769407033.git.ktokunaga.mail@gmail.com> X-Mailer: git-send-email 2.43.0 In-Reply-To: References: 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=ktokunaga.mail@gmail.com; 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, FREEMAIL_FROM=0.001, RCVD_IN_DNSWL_NONE=-0.0001, SPF_HELO_NONE=0.001, 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: , 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 @gmail.com) X-ZM-MESSAGEID: 1769412785426154100 Content-Type: text/plain; charset="utf-8" Since Wasm load and store instructions don't support negative offsets, address calculations are performed separately before the memory access. When Emscripten's -sMEMORY64=3D2 is enabled, the address size must be 32bits. So this commit updates the build tools to propagate this flag to the C code via the WASM64_MEMORY64_2 macro. In this case, the emitted code casts pointers to 32bit before memory oprations. Additionally, the declaration of "--wasm64-32bit-address-limit" flag has been moved from the configure script to meson.build. So the flag name is updated to "--enable-wasm64-32bit-address-limit" to follow Meson's naming conventions. TCI instructions are also generated in the same way as the original TCI backend. Signed-off-by: Kohei Tokunaga --- .gitlab-ci.d/buildtest.yml | 2 +- configure | 8 +- meson.build | 4 + meson_options.txt | 3 + scripts/meson-buildoptions.sh | 5 + tcg/wasm64.c | 87 ++++++++++ tcg/wasm64/tcg-target-mo.h | 20 +++ tcg/wasm64/tcg-target-opc.h.inc | 2 + tcg/wasm64/tcg-target.c.inc | 286 ++++++++++++++++++++++++++++++++ 9 files changed, 411 insertions(+), 6 deletions(-) create mode 100644 tcg/wasm64/tcg-target-mo.h diff --git a/.gitlab-ci.d/buildtest.yml b/.gitlab-ci.d/buildtest.yml index e9b5b05e6e..f9a2d4de74 100644 --- a/.gitlab-ci.d/buildtest.yml +++ b/.gitlab-ci.d/buildtest.yml @@ -801,4 +801,4 @@ build-wasm64-32bit: - job: wasm64-emsdk-cross-container variables: IMAGE: emsdk-wasm64-cross - CONFIGURE_ARGS: --static --cpu=3Dwasm64 --wasm64-32bit-address-limit -= -disable-tools --enable-debug --enable-tcg-interpreter + CONFIGURE_ARGS: --static --cpu=3Dwasm64 --enable-wasm64-32bit-address-= limit --disable-tools --enable-debug --enable-tcg-interpreter diff --git a/configure b/configure index e69b3e474e..e3e8e8387e 100755 --- a/configure +++ b/configure @@ -243,7 +243,9 @@ for opt do ;; --without-default-features) default_feature=3D"no" ;; - --wasm64-32bit-address-limit) wasm64_memory64=3D"2" + --enable-wasm64-32bit-address-limit) wasm64_memory64=3D"2" + ;; + --disable-wasm64-32bit-address-limit) wasm64_memory64=3D"1" ;; esac done @@ -760,8 +762,6 @@ for opt do ;; --disable-rust) rust=3Ddisabled ;; - --wasm64-32bit-address-limit) - ;; # everything else has the same name in configure and meson --*) meson_option_parse "$opt" "$optarg" ;; @@ -887,8 +887,6 @@ Advanced options (experts only): --disable-containers don't use containers for cross-building --container-engine=3DTYPE which container engine to use [$container_eng= ine] --gdb=3DGDB-path gdb to use for gdbstub tests [$gdb_bin] - --wasm64-32bit-address-limit Restrict wasm64 address space to 32-bit (de= fault - is to use the whole 64-bit range). EOF meson_options_help cat << EOF diff --git a/meson.build b/meson.build index 3108f01e88..7a81ff58be 100644 --- a/meson.build +++ b/meson.build @@ -371,6 +371,10 @@ elif host_os =3D=3D 'windows' if compiler.get_id() =3D=3D 'clang' and compiler.get_linker_id() !=3D 'l= d.lld' error('On windows, you need to use lld with clang - use msys2 clang64/= clangarm64 env') endif +elif host_os =3D=3D 'emscripten' + if cpu =3D=3D 'wasm64' and get_option('wasm64_32bit_address_limit') + qemu_common_flags +=3D '-DWASM64_MEMORY64_2' + endif endif =20 # Choose instruction set (currently x86-only) diff --git a/meson_options.txt b/meson_options.txt index 2836156257..8b36159a45 100644 --- a/meson_options.txt +++ b/meson_options.txt @@ -390,3 +390,6 @@ option('rust', type: 'feature', value: 'disabled', description: 'Rust support') option('strict_rust_lints', type: 'boolean', value: false, description: 'Enable stricter set of Rust warnings') + +option('wasm64_32bit_address_limit', type: 'boolean', value: false, + description: 'Restrict wasm64 address space to 32-bit (default is t= o use the whole 64-bit range).') diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh index 3d0d132344..04311765be 100644 --- a/scripts/meson-buildoptions.sh +++ b/scripts/meson-buildoptions.sh @@ -56,6 +56,9 @@ meson_options_help() { printf "%s\n" ' dtrace/ftrace/log/nop/simple/s= yslog/ust)' printf "%s\n" ' --enable-tsan enable thread sanitizer' printf "%s\n" ' --enable-ubsan enable undefined behaviour san= itizer' + printf "%s\n" ' --enable-wasm64-32bit-address-limit' + printf "%s\n" ' Restrict wasm64 address space = to 32-bit (default' + printf "%s\n" ' is to use the whole 64-bit ran= ge).' printf "%s\n" ' --firmwarepath=3DVALUES search PATH for firmware fil= es [share/qemu-' printf "%s\n" ' firmware]' printf "%s\n" ' --gdb=3DVALUE Path to GDB' @@ -576,6 +579,8 @@ _meson_option_parse() { --disable-vte) printf "%s" -Dvte=3Ddisabled ;; --enable-vvfat) printf "%s" -Dvvfat=3Denabled ;; --disable-vvfat) printf "%s" -Dvvfat=3Ddisabled ;; + --enable-wasm64-32bit-address-limit) printf "%s" -Dwasm64_32bit_addres= s_limit=3Dtrue ;; + --disable-wasm64-32bit-address-limit) printf "%s" -Dwasm64_32bit_addre= ss_limit=3Dfalse ;; --enable-werror) printf "%s" -Dwerror=3Dtrue ;; --disable-werror) printf "%s" -Dwerror=3Dfalse ;; --enable-whpx) printf "%s" -Dwhpx=3Denabled ;; diff --git a/tcg/wasm64.c b/tcg/wasm64.c index 183dad10a2..fa7413fc1d 100644 --- a/tcg/wasm64.c +++ b/tcg/wasm64.c @@ -20,6 +20,14 @@ =20 #include "qemu/osdep.h" #include "tcg/tcg.h" +#include "tcg/tcg-ldst.h" + +static void tci_args_rl(uint32_t insn, const void *tb_ptr, + TCGReg *r0, void **l1) +{ + *r0 =3D extract32(insn, 8, 4); + *l1 =3D sextract32(insn, 12, 20) + (void *)tb_ptr; +} =20 static void tci_args_rr(uint32_t insn, TCGReg *r0, TCGReg *r1) { @@ -27,6 +35,12 @@ static void tci_args_rr(uint32_t insn, TCGReg *r0, TCGRe= g *r1) *r1 =3D extract32(insn, 12, 4); } =20 +static void tci_args_ri(uint32_t insn, TCGReg *r0, tcg_target_ulong *i1) +{ + *r0 =3D extract32(insn, 8, 4); + *i1 =3D sextract32(insn, 12, 20); +} + static void tci_args_rrr(uint32_t insn, TCGReg *r0, TCGReg *r1, TCGReg *r2) { *r0 =3D extract32(insn, 8, 4); @@ -34,6 +48,13 @@ static void tci_args_rrr(uint32_t insn, TCGReg *r0, TCGR= eg *r1, TCGReg *r2) *r2 =3D extract32(insn, 16, 4); } =20 +static void tci_args_rrs(uint32_t insn, TCGReg *r0, TCGReg *r1, int32_t *i= 2) +{ + *r0 =3D extract32(insn, 8, 4); + *r1 =3D extract32(insn, 12, 4); + *i2 =3D sextract32(insn, 16, 16); +} + static void tci_args_rrbb(uint32_t insn, TCGReg *r0, TCGReg *r1, uint8_t *i2, uint8_t *i3) { @@ -161,9 +182,12 @@ static uintptr_t tcg_qemu_tb_exec_tci(CPUArchState *en= v, const void *v_tb_ptr) uint32_t insn; TCGOpcode opc; TCGReg r0, r1, r2, r3, r4; + tcg_target_ulong t1; uint8_t pos, len; TCGCond condition; uint32_t tmp32; + int32_t ofs; + void *ptr; =20 insn =3D *tb_ptr++; opc =3D extract32(insn, 0, 8); @@ -236,6 +260,69 @@ static uintptr_t tcg_qemu_tb_exec_tci(CPUArchState *en= v, const void *v_tb_ptr) tmp32 =3D tci_compare32(regs[r1], regs[r2], condition); regs[r0] =3D regs[tmp32 ? r3 : r4]; break; + case INDEX_op_tci_movi: + tci_args_ri(insn, &r0, &t1); + regs[r0] =3D t1; + break; + case INDEX_op_tci_movl: + tci_args_rl(insn, tb_ptr, &r0, &ptr); + regs[r0] =3D *(tcg_target_ulong *)ptr; + break; + case INDEX_op_ld: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + regs[r0] =3D *(tcg_target_ulong *)ptr; + break; + case INDEX_op_ld8u: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + regs[r0] =3D *(uint8_t *)ptr; + break; + case INDEX_op_ld8s: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + regs[r0] =3D *(int8_t *)ptr; + break; + case INDEX_op_ld16u: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + regs[r0] =3D *(uint16_t *)ptr; + break; + case INDEX_op_ld16s: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + regs[r0] =3D *(int16_t *)ptr; + break; + case INDEX_op_st: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + *(tcg_target_ulong *)ptr =3D regs[r0]; + break; + case INDEX_op_st8: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + *(uint8_t *)ptr =3D regs[r0]; + break; + case INDEX_op_st16: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + *(uint16_t *)ptr =3D regs[r0]; + break; + case INDEX_op_ld32u: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + regs[r0] =3D *(uint32_t *)ptr; + break; + case INDEX_op_ld32s: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + regs[r0] =3D *(int32_t *)ptr; + break; + case INDEX_op_st32: + tci_args_rrs(insn, &r0, &r1, &ofs); + ptr =3D (void *)(regs[r1] + ofs); + *(uint32_t *)ptr =3D regs[r0]; + break; default: g_assert_not_reached(); } diff --git a/tcg/wasm64/tcg-target-mo.h b/tcg/wasm64/tcg-target-mo.h new file mode 100644 index 0000000000..3ccc56e899 --- /dev/null +++ b/tcg/wasm64/tcg-target-mo.h @@ -0,0 +1,20 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Define target-specific memory model + * + * Based on tci/tcg-target-mo.h + * + * Copyright (c) 2009, 2011 Stefan Weil + */ + +#ifndef TCG_TARGET_MO_H +#define TCG_TARGET_MO_H + +/* + * We could notice __x86_64__ or __s390x__ and reduce the barriers dependi= ng + * on the host. But if you want performance, you use the normal backend. + * We prefer consistency across hosts on this. + */ +#define TCG_TARGET_DEFAULT_MO 0 + +#endif diff --git a/tcg/wasm64/tcg-target-opc.h.inc b/tcg/wasm64/tcg-target-opc.h.= inc index 57274d4569..122b45749a 100644 --- a/tcg/wasm64/tcg-target-opc.h.inc +++ b/tcg/wasm64/tcg-target-opc.h.inc @@ -4,5 +4,7 @@ * * These opcodes for use between the tci generator and interpreter. */ +DEF(tci_movi, 1, 0, 1, TCG_OPF_NOT_PRESENT) +DEF(tci_movl, 1, 0, 1, TCG_OPF_NOT_PRESENT) DEF(tci_setcond32, 1, 2, 1, TCG_OPF_NOT_PRESENT) DEF(tci_movcond32, 1, 2, 1, TCG_OPF_NOT_PRESENT) diff --git a/tcg/wasm64/tcg-target.c.inc b/tcg/wasm64/tcg-target.c.inc index dd75deecd3..6b6d6402f4 100644 --- a/tcg/wasm64/tcg-target.c.inc +++ b/tcg/wasm64/tcg-target.c.inc @@ -143,6 +143,18 @@ typedef enum { OPC_GLOBAL_GET =3D 0x23, OPC_GLOBAL_SET =3D 0x24, =20 + OPC_I64_LOAD =3D 0x29, + OPC_I64_LOAD8_S =3D 0x30, + OPC_I64_LOAD8_U =3D 0x31, + OPC_I64_LOAD16_S =3D 0x32, + OPC_I64_LOAD16_U =3D 0x33, + OPC_I64_LOAD32_S =3D 0x34, + OPC_I64_LOAD32_U =3D 0x35, + OPC_I64_STORE =3D 0x37, + OPC_I64_STORE8 =3D 0x3c, + OPC_I64_STORE16 =3D 0x3d, + OPC_I64_STORE32 =3D 0x3e, + OPC_I32_CONST =3D 0x41, OPC_I64_CONST =3D 0x42, =20 @@ -168,6 +180,7 @@ typedef enum { OPC_I64_GE_S =3D 0x59, OPC_I64_GE_U =3D 0x5a, =20 + OPC_I32_ADD =3D 0x6a, OPC_I32_SHR_S =3D 0x75, OPC_I32_SHR_U =3D 0x76, =20 @@ -421,6 +434,84 @@ static void tcg_wasm_out_sextract(TCGContext *s, TCGRe= g dest, TCGReg arg1, tcg_wasm_out_op_idx(s, OPC_GLOBAL_SET, REG_IDX(dest)); } =20 +/* + * The size of the offset field of Wasm's load/store instruction differs + * depending on the "-sMEMORY64" flag value: 64bit when "-sMEMORY64=3D1" + * and 32bit when "-sMEMORY64=3D2". + */ +#if defined(WASM64_MEMORY64_2) +typedef uint32_t wasm_ldst_offset_t; +#else +typedef uint64_t wasm_ldst_offset_t; +#endif +static void tcg_wasm_out_op_ldst( + TCGContext *s, WasmInsn instr, uint32_t a, wasm_ldst_offset_t o) +{ + tcg_wasm_out_op(s, instr); + tcg_wasm_out_leb128(s, a); + tcg_wasm_out_leb128(s, (wasm_ldst_offset_t)o); +} + +/* + * tcg_wasm_out_norm_ptr emits instructions to adjust the 64bit pointer va= lue + * at the top of the stack to satisfy Wasm's memory addressing requirement= s. + */ +static intptr_t tcg_wasm_out_norm_ptr(TCGContext *s, intptr_t offset) +{ +#if defined(WASM64_MEMORY64_2) + /* + * If Emscripten's "-sMEMORY64=3D2" is enabled, + * the address size is limited to 32bit. + */ + tcg_wasm_out_op(s, OPC_I32_WRAP_I64); +#endif + /* + * Wasm's load/store instructions don't support negative value in + * the offset field. So this function calculates the target address + * using the base and the offset and makes the offset field 0. + */ + if (offset < 0) { +#if defined(WASM64_MEMORY64_2) + tcg_wasm_out_op_const(s, OPC_I32_CONST, offset); + tcg_wasm_out_op(s, OPC_I32_ADD); +#else + tcg_wasm_out_op_const(s, OPC_I64_CONST, offset); + tcg_wasm_out_op(s, OPC_I64_ADD); +#endif + offset =3D 0; + } + return offset; +} + +static void tcg_wasm_out_ld( + TCGContext *s, WasmInsn opc, TCGReg val, TCGReg base, intptr_t offset) +{ + tcg_wasm_out_op_idx(s, OPC_GLOBAL_GET, REG_IDX(base)); + offset =3D tcg_wasm_out_norm_ptr(s, offset); + tcg_wasm_out_op_ldst(s, opc, 0, offset); + tcg_wasm_out_op_idx(s, OPC_GLOBAL_SET, REG_IDX(val)); +} + +static void tcg_wasm_out_st( + TCGContext *s, WasmInsn opc, TCGReg val, TCGReg base, intptr_t offset) +{ + tcg_wasm_out_op_idx(s, OPC_GLOBAL_GET, REG_IDX(base)); + offset =3D tcg_wasm_out_norm_ptr(s, offset); + tcg_wasm_out_op_idx(s, OPC_GLOBAL_GET, REG_IDX(val)); + tcg_wasm_out_op_ldst(s, opc, 0, offset); +} + +static void tcg_out_op_ri(TCGContext *s, TCGOpcode op, TCGReg r0, int32_t = i1) +{ + tcg_insn_unit_tci insn =3D 0; + + tcg_debug_assert(i1 =3D=3D sextract32(i1, 0, 20)); + insn =3D deposit32(insn, 0, 8, op); + insn =3D deposit32(insn, 8, 4, r0); + insn =3D deposit32(insn, 12, 20, i1); + tcg_out32(s, insn); +} + static void tcg_out_op_rr(TCGContext *s, TCGOpcode op, TCGReg r0, TCGReg r= 1) { tcg_insn_unit_tci insn =3D 0; @@ -443,6 +534,19 @@ static void tcg_out_op_rrr(TCGContext *s, TCGOpcode op, tcg_out32(s, insn); } =20 +static void tcg_out_op_rrs(TCGContext *s, TCGOpcode op, + TCGReg r0, TCGReg r1, intptr_t i2) +{ + tcg_insn_unit_tci insn =3D 0; + + tcg_debug_assert(i2 =3D=3D sextract32(i2, 0, 16)); + insn =3D deposit32(insn, 0, 8, op); + insn =3D deposit32(insn, 8, 4, r0); + insn =3D deposit32(insn, 12, 4, r1); + insn =3D deposit32(insn, 16, 16, i2); + tcg_out32(s, insn); +} + static void tcg_out_op_rrbb(TCGContext *s, TCGOpcode op, TCGReg r0, TCGReg r1, uint8_t b2, uint8_t b3) { @@ -669,6 +773,188 @@ static const TCGOutOpMovcond outop_movcond =3D { .out =3D tgen_movcond, }; =20 +static void tcg_tci_out_movi(TCGContext *s, TCGType type, + TCGReg ret, tcg_target_long arg) +{ + switch (type) { + case TCG_TYPE_I32: + arg =3D (int32_t)arg; + /* fall through */ + case TCG_TYPE_I64: + break; + default: + g_assert_not_reached(); + } + + if (arg =3D=3D sextract32(arg, 0, 20)) { + tcg_out_op_ri(s, INDEX_op_tci_movi, ret, arg); + } else { + tcg_insn_unit_tci insn =3D 0; + + new_pool_label(s, arg, 20, s->code_ptr, 0); + insn =3D deposit32(insn, 0, 8, INDEX_op_tci_movl); + insn =3D deposit32(insn, 8, 4, ret); + tcg_out32(s, insn); + } +} + +static void stack_bounds_check(TCGReg base, intptr_t offset) +{ + if (base =3D=3D TCG_REG_CALL_STACK) { + tcg_debug_assert(offset >=3D 0); + tcg_debug_assert(offset < (TCG_STATIC_CALL_ARGS_SIZE + + TCG_STATIC_FRAME_SIZE)); + } +} + +static void tcg_out_ldst(TCGContext *s, TCGOpcode op, TCGReg val, + TCGReg base, intptr_t offset) +{ + stack_bounds_check(base, offset); + if (offset !=3D sextract32(offset, 0, 16)) { + tcg_tci_out_movi(s, TCG_TYPE_PTR, TCG_REG_TMP, offset); + tcg_out_op_rrr(s, INDEX_op_add, TCG_REG_TMP, TCG_REG_TMP, base); + base =3D TCG_REG_TMP; + offset =3D 0; + } + tcg_out_op_rrs(s, op, val, base, offset); +} + +static void tcg_out_ld(TCGContext *s, TCGType type, TCGReg val, TCGReg bas= e, + intptr_t offset) +{ + TCGOpcode op =3D INDEX_op_ld; + WasmInsn wasm_opc =3D OPC_I64_LOAD; + + if (type =3D=3D TCG_TYPE_I32) { + op =3D INDEX_op_ld32u; + wasm_opc =3D OPC_I64_LOAD32_U; + } + tcg_out_ldst(s, op, val, base, offset); + tcg_wasm_out_ld(s, wasm_opc, val, base, offset); +} + +static void tgen_ld8u(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, INDEX_op_ld8u, dest, base, offset); + tcg_wasm_out_ld(s, OPC_I64_LOAD8_U, dest, base, offset); +} + +static const TCGOutOpLoad outop_ld8u =3D { + .base.static_constraint =3D C_O1_I1(r, r), + .out =3D tgen_ld8u, +}; + +static void tgen_ld8s(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, INDEX_op_ld8s, dest, base, offset); + tcg_wasm_out_ld(s, OPC_I64_LOAD8_S, dest, base, offset); +} + +static const TCGOutOpLoad outop_ld8s =3D { + .base.static_constraint =3D C_O1_I1(r, r), + .out =3D tgen_ld8s, +}; + +static void tgen_ld16u(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, INDEX_op_ld16u, dest, base, offset); + tcg_wasm_out_ld(s, OPC_I64_LOAD16_U, dest, base, offset); +} + +static const TCGOutOpLoad outop_ld16u =3D { + .base.static_constraint =3D C_O1_I1(r, r), + .out =3D tgen_ld16u, +}; + +static void tgen_ld16s(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, INDEX_op_ld16s, dest, base, offset); + tcg_wasm_out_ld(s, OPC_I64_LOAD16_S, dest, base, offset); +} + +static const TCGOutOpLoad outop_ld16s =3D { + .base.static_constraint =3D C_O1_I1(r, r), + .out =3D tgen_ld16s, +}; + +static void tgen_ld32u(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, INDEX_op_ld32u, dest, base, offset); + tcg_wasm_out_ld(s, OPC_I64_LOAD32_U, dest, base, offset); +} + +static const TCGOutOpLoad outop_ld32u =3D { + .base.static_constraint =3D C_O1_I1(r, r), + .out =3D tgen_ld32u, +}; + +static void tgen_ld32s(TCGContext *s, TCGType type, TCGReg dest, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, INDEX_op_ld32s, dest, base, offset); + tcg_wasm_out_ld(s, OPC_I64_LOAD32_S, dest, base, offset); +} + +static const TCGOutOpLoad outop_ld32s =3D { + .base.static_constraint =3D C_O1_I1(r, r), + .out =3D tgen_ld32s, +}; + +static void tgen_st8(TCGContext *s, TCGType type, TCGReg data, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, INDEX_op_st8, data, base, offset); + tcg_wasm_out_st(s, OPC_I64_STORE8, data, base, offset); +} + +static const TCGOutOpStore outop_st8 =3D { + .base.static_constraint =3D C_O0_I2(r, r), + .out_r =3D tgen_st8, +}; + +static void tgen_st16(TCGContext *s, TCGType type, TCGReg data, + TCGReg base, ptrdiff_t offset) +{ + tcg_out_ldst(s, INDEX_op_st16, data, base, offset); + tcg_wasm_out_st(s, OPC_I64_STORE16, data, base, offset); +} + +static const TCGOutOpStore outop_st16 =3D { + .base.static_constraint =3D C_O0_I2(r, r), + .out_r =3D tgen_st16, +}; + +static void tcg_out_st(TCGContext *s, TCGType type, TCGReg val, TCGReg bas= e, + intptr_t offset) +{ + TCGOpcode op =3D INDEX_op_st; + WasmInsn wasm_opc =3D OPC_I64_STORE; + + if (type =3D=3D TCG_TYPE_I32) { + op =3D INDEX_op_st32; + wasm_opc =3D OPC_I64_STORE32; + } + tcg_out_ldst(s, op, val, base, offset); + tcg_wasm_out_st(s, wasm_opc, val, base, offset); +} + +static const TCGOutOpStore outop_st =3D { + .base.static_constraint =3D C_O0_I2(r, r), + .out_r =3D tcg_out_st, +}; + +static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, + TCGReg base, intptr_t ofs) +{ + return false; +} + static void tcg_out_tb_start(TCGContext *s) { init_sub_buf(); --=20 2.43.0