From patchwork Thu Mar 5 05:02:04 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Luke Nelson X-Patchwork-Id: 11421193 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 23685138D for ; Thu, 5 Mar 2020 05:02:29 +0000 (UTC) Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id F12A2208C3 for ; Thu, 5 Mar 2020 05:02:28 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=lists.infradead.org header.i=@lists.infradead.org header.b="ufn4Ohrl"; dkim=fail reason="signature verification failed" (1024-bit key) header.d=cs.washington.edu header.i=@cs.washington.edu header.b="EoW/rZKr" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org F12A2208C3 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=cs.washington.edu Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-riscv-bounces+patchwork-linux-riscv=patchwork.kernel.org@lists.infradead.org DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20170209; h=Sender:Cc:List-Subscribe: List-Help:List-Post:List-Archive:List-Unsubscribe:List-Id: Content-Transfer-Encoding:Content-Type:MIME-Version:References:In-Reply-To: Message-Id:Date:Subject:To:From:Reply-To:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=F6xpWJNnviYx0UXiufCowhh03LwABxA+pXEEfc9R5RE=; b=ufn4Ohrlep1CdU cQ8+SVOfTnpR1LK9SNhDj5QxjdsasVhzz9f9g0vVPDzLV9DJZfnVFwHRi8e2YAWnjJ+TNiPTSZIx4 cJ8+Gshqlyl+bWK0xfd0yLSO0pvQHPmEbBxQs5OQJXJRDBaCWKkYSkZDYroQzLrbAPae3SatyM3qd 1Rli3cY5bdlgLp+lM1vsSh0qDx3zs1Ue8mGVkGFrGFOT0GkwBYwGE7WzPTqJorgl2Zi6eBN14v2yn rGKKagMjFYPrd1PgYdyFGfKMmpLKAT8WpxQqfA3lHMicQAy4Y3P03waYG/Fer6YCsRPdR87gLEnGs MX/6uRkSs2jNuIILDTHQ==; Received: from localhost ([127.0.0.1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.92.3 #3 (Red Hat Linux)) id 1j9ieO-0004UW-CK; Thu, 05 Mar 2020 05:02:24 +0000 Received: from mail-pg1-x543.google.com ([2607:f8b0:4864:20::543]) by bombadil.infradead.org with esmtps (Exim 4.92.3 #3 (Red Hat Linux)) id 1j9ieJ-0004RI-M7 for linux-riscv@lists.infradead.org; Thu, 05 Mar 2020 05:02:22 +0000 Received: by mail-pg1-x543.google.com with SMTP id b1so2131783pgm.8 for ; Wed, 04 Mar 2020 21:02:19 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=cs.washington.edu; s=goo201206; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=F6xpWJNnviYx0UXiufCowhh03LwABxA+pXEEfc9R5RE=; b=EoW/rZKrpqiQ8IpaU03NKmhe5xIKgB4WRhdxJbguJbvdWj7jY0ryml+MHJWsOiddHT qM8iN3UkaPIPXJyrJ043THt3tMDRLIivxCPcaDYBL2hNhb5ztR6klvlBtiYH2DQyGwhj nliveVLDE4rTVRr62jdpVMN6LvprUyAKxDc1A= X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=F6xpWJNnviYx0UXiufCowhh03LwABxA+pXEEfc9R5RE=; b=tBBajz57CitBPCAWmeh1T/rDMDNmrLl9YMzE9+IQ4ZsOSQURB0F41NAx9TGs8zNMo1 lbUOi2Q9rvy1MriRRdQyWscOkM2FPL+bWz8t0Wxyy5DTr1/UOd1bF7Z5q4rawuZx06tq gZ/qq4GwmjdM2g42lwmCpqjIln5QtPm8ijIfXP1XIXHJYbFtvwnYU0KH1SvWHz381S2J edq3KZON+Y55QJOz944EGsiDZeJb7IuZOiVEtjYHIIemyJflctiqun9Lpk/P+i7Rm6qw +W6S7MtaUPO9dku6JUi+bpb5BleXZ0TDfxWO240qaO9DybXAC5Oi2xsoTcWwTLiWEG64 mNSg== X-Gm-Message-State: ANhLgQ2t3c4lAHcgb+GRecPFG/AuJi3pz6yxayInwtoWAeweEwG46hD4 WhRiz0HZBbOLp/hDP+drzRak0w== X-Google-Smtp-Source: ADFU+vsHA0frR6iRl6MgoKMNb4yR7jkavunG04ogNwOu3NEiE6yGiFHJ9doeFV+6+MoKZxIsIOPafw== X-Received: by 2002:a62:1456:: with SMTP id 83mr5100200pfu.237.1583384537809; Wed, 04 Mar 2020 21:02:17 -0800 (PST) Received: from ryzen.cs.washington.edu ([2607:4000:200:11:e9fe:faad:3d84:58ea]) by smtp.gmail.com with ESMTPSA id y7sm17820466pfq.15.2020.03.04.21.02.16 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 04 Mar 2020 21:02:17 -0800 (PST) From: Luke Nelson X-Google-Original-From: Luke Nelson To: bpf@vger.kernel.org Subject: [PATCH bpf-next v5 1/4] riscv, bpf: factor common RISC-V JIT code Date: Wed, 4 Mar 2020 21:02:04 -0800 Message-Id: <20200305050207.4159-2-luke.r.nels@gmail.com> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20200305050207.4159-1-luke.r.nels@gmail.com> References: <20200305050207.4159-1-luke.r.nels@gmail.com> MIME-Version: 1.0 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20200304_210219_732093_0C2F593D X-CRM114-Status: GOOD ( 16.26 ) X-Spam-Score: -0.1 (/) X-Spam-Report: SpamAssassin version 3.4.3 on bombadil.infradead.org summary: Content analysis details: (-0.1 points) pts rule name description ---- ---------------------- -------------------------------------------------- -0.0 RCVD_IN_DNSWL_NONE RBL: Sender listed at https://www.dnswl.org/, no trust [2607:f8b0:4864:20:0:0:0:543 listed in] [list.dnswl.org] -0.0 SPF_PASS SPF: sender matches SPF record 0.0 SPF_HELO_NONE SPF: HELO does not publish an SPF Record -0.1 DKIM_VALID Message has at least one valid DKIM or DK signature -0.1 DKIM_VALID_AU Message has a valid DKIM or DK signature from author's domain 0.1 DKIM_SIGNED Message has a DKIM or DK signature, not necessarily valid X-BeenThere: linux-riscv@lists.infradead.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Song Liu , linux-doc@vger.kernel.org, Yonghong Song , Paul Walmsley , Alexei Starovoitov , netdev@vger.kernel.org, Mauro Carvalho Chehab , linux-riscv@lists.infradead.org, Rob Herring , Daniel Borkmann , Jonathan Corbet , Jakub Kicinski , Andrii Nakryiko , Xi Wang , Albert Ou , Luke Nelson , Jonathan Cameron , Andy Shevchenko , =?utf-8?b?QmrDtnJuIFQ=?= =?utf-8?b?w7ZwZWw=?= , linux-kernel@vger.kernel.org, Martin KaFai Lau , Stephen Hemminger , Palmer Dabbelt , Greg Kroah-Hartman , "David S. Miller" Sender: "linux-riscv" Errors-To: linux-riscv-bounces+patchwork-linux-riscv=patchwork.kernel.org@lists.infradead.org This patch factors out code that can be used by both the RV64 and RV32 BPF JITs to a common bpf_jit.h and bpf_jit_core.c. Move struct definitions and macro-like functions to header. Rename rv_sb_insn/rv_uj_insn to rv_b_insn/rv_j_insn to match the RISC-V specification. Move reusable functions emit_body() and bpf_int_jit_compile() to bpf_jit_core.c with minor simplifications. Rename emit_insn() and build_{prologue,epilogue}() to be prefixed with "bpf_jit_" as they are no longer static. Rename bpf_jit_comp.c to bpf_jit_comp64.c to be more explicit. Co-developed-by: Xi Wang Signed-off-by: Xi Wang Signed-off-by: Luke Nelson --- arch/riscv/net/Makefile | 3 +- arch/riscv/net/bpf_jit.h | 466 ++++++++++++++ .../net/{bpf_jit_comp.c => bpf_jit_comp64.c} | 605 +----------------- arch/riscv/net/bpf_jit_core.c | 166 +++++ 4 files changed, 639 insertions(+), 601 deletions(-) create mode 100644 arch/riscv/net/bpf_jit.h rename arch/riscv/net/{bpf_jit_comp.c => bpf_jit_comp64.c} (69%) create mode 100644 arch/riscv/net/bpf_jit_core.c diff --git a/arch/riscv/net/Makefile b/arch/riscv/net/Makefile index ec5b14763316..018074dbf986 100644 --- a/arch/riscv/net/Makefile +++ b/arch/riscv/net/Makefile @@ -1,2 +1,3 @@ # SPDX-License-Identifier: GPL-2.0-only -obj-$(CONFIG_BPF_JIT) += bpf_jit_comp.o + +obj-$(CONFIG_BPF_JIT) += bpf_jit_core.o bpf_jit_comp64.o diff --git a/arch/riscv/net/bpf_jit.h b/arch/riscv/net/bpf_jit.h new file mode 100644 index 000000000000..23c123331f94 --- /dev/null +++ b/arch/riscv/net/bpf_jit.h @@ -0,0 +1,466 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * Common functionality for RV32 and RV64 BPF JIT compilers + * + * Copyright (c) 2019 Björn Töpel + * + */ + +#ifndef _BPF_JIT_H +#define _BPF_JIT_H + +#include +#include +#include + +enum { + RV_REG_ZERO = 0, /* The constant value 0 */ + RV_REG_RA = 1, /* Return address */ + RV_REG_SP = 2, /* Stack pointer */ + RV_REG_GP = 3, /* Global pointer */ + RV_REG_TP = 4, /* Thread pointer */ + RV_REG_T0 = 5, /* Temporaries */ + RV_REG_T1 = 6, + RV_REG_T2 = 7, + RV_REG_FP = 8, /* Saved register/frame pointer */ + RV_REG_S1 = 9, /* Saved register */ + RV_REG_A0 = 10, /* Function argument/return values */ + RV_REG_A1 = 11, /* Function arguments */ + RV_REG_A2 = 12, + RV_REG_A3 = 13, + RV_REG_A4 = 14, + RV_REG_A5 = 15, + RV_REG_A6 = 16, + RV_REG_A7 = 17, + RV_REG_S2 = 18, /* Saved registers */ + RV_REG_S3 = 19, + RV_REG_S4 = 20, + RV_REG_S5 = 21, + RV_REG_S6 = 22, + RV_REG_S7 = 23, + RV_REG_S8 = 24, + RV_REG_S9 = 25, + RV_REG_S10 = 26, + RV_REG_S11 = 27, + RV_REG_T3 = 28, /* Temporaries */ + RV_REG_T4 = 29, + RV_REG_T5 = 30, + RV_REG_T6 = 31, +}; + +struct rv_jit_context { + struct bpf_prog *prog; + u32 *insns; /* RV insns */ + int ninsns; + int epilogue_offset; + int *offset; /* BPF to RV */ + unsigned long flags; + int stack_size; +}; + +struct rv_jit_data { + struct bpf_binary_header *header; + u8 *image; + struct rv_jit_context ctx; +}; + +static inline void bpf_fill_ill_insns(void *area, unsigned int size) +{ + memset(area, 0, size); +} + +static inline void bpf_flush_icache(void *start, void *end) +{ + flush_icache_range((unsigned long)start, (unsigned long)end); +} + +static inline void emit(const u32 insn, struct rv_jit_context *ctx) +{ + if (ctx->insns) + ctx->insns[ctx->ninsns] = insn; + + ctx->ninsns++; +} + +static inline int epilogue_offset(struct rv_jit_context *ctx) +{ + int to = ctx->epilogue_offset, from = ctx->ninsns; + + return (to - from) << 2; +} + +/* Return -1 or inverted cond. */ +static inline int invert_bpf_cond(u8 cond) +{ + switch (cond) { + case BPF_JEQ: + return BPF_JNE; + case BPF_JGT: + return BPF_JLE; + case BPF_JLT: + return BPF_JGE; + case BPF_JGE: + return BPF_JLT; + case BPF_JLE: + return BPF_JGT; + case BPF_JNE: + return BPF_JEQ; + case BPF_JSGT: + return BPF_JSLE; + case BPF_JSLT: + return BPF_JSGE; + case BPF_JSGE: + return BPF_JSLT; + case BPF_JSLE: + return BPF_JSGT; + } + return -1; +} + +static inline bool is_12b_int(long val) +{ + return -(1L << 11) <= val && val < (1L << 11); +} + +static inline int is_12b_check(int off, int insn) +{ + if (!is_12b_int(off)) { + pr_err("bpf-jit: insn=%d 12b < offset=%d not supported yet!\n", + insn, (int)off); + return -1; + } + return 0; +} + +static inline bool is_13b_int(long val) +{ + return -(1L << 12) <= val && val < (1L << 12); +} + +static inline bool is_21b_int(long val) +{ + return -(1L << 20) <= val && val < (1L << 20); +} + +static inline int rv_offset(int insn, int off, struct rv_jit_context *ctx) +{ + int from, to; + + off++; /* BPF branch is from PC+1, RV is from PC */ + from = (insn > 0) ? ctx->offset[insn - 1] : 0; + to = (insn + off > 0) ? ctx->offset[insn + off - 1] : 0; + return (to - from) << 2; +} + +/* Instruction formats. */ + +static inline u32 rv_r_insn(u8 funct7, u8 rs2, u8 rs1, u8 funct3, u8 rd, + u8 opcode) +{ + return (funct7 << 25) | (rs2 << 20) | (rs1 << 15) | (funct3 << 12) | + (rd << 7) | opcode; +} + +static inline u32 rv_i_insn(u16 imm11_0, u8 rs1, u8 funct3, u8 rd, u8 opcode) +{ + return (imm11_0 << 20) | (rs1 << 15) | (funct3 << 12) | (rd << 7) | + opcode; +} + +static inline u32 rv_s_insn(u16 imm11_0, u8 rs2, u8 rs1, u8 funct3, u8 opcode) +{ + u8 imm11_5 = imm11_0 >> 5, imm4_0 = imm11_0 & 0x1f; + + return (imm11_5 << 25) | (rs2 << 20) | (rs1 << 15) | (funct3 << 12) | + (imm4_0 << 7) | opcode; +} + +static inline u32 rv_b_insn(u16 imm12_1, u8 rs2, u8 rs1, u8 funct3, u8 opcode) +{ + u8 imm12 = ((imm12_1 & 0x800) >> 5) | ((imm12_1 & 0x3f0) >> 4); + u8 imm4_1 = ((imm12_1 & 0xf) << 1) | ((imm12_1 & 0x400) >> 10); + + return (imm12 << 25) | (rs2 << 20) | (rs1 << 15) | (funct3 << 12) | + (imm4_1 << 7) | opcode; +} + +static inline u32 rv_u_insn(u32 imm31_12, u8 rd, u8 opcode) +{ + return (imm31_12 << 12) | (rd << 7) | opcode; +} + +static inline u32 rv_j_insn(u32 imm20_1, u8 rd, u8 opcode) +{ + u32 imm; + + imm = (imm20_1 & 0x80000) | ((imm20_1 & 0x3ff) << 9) | + ((imm20_1 & 0x400) >> 2) | ((imm20_1 & 0x7f800) >> 11); + + return (imm << 12) | (rd << 7) | opcode; +} + +static inline u32 rv_amo_insn(u8 funct5, u8 aq, u8 rl, u8 rs2, u8 rs1, + u8 funct3, u8 rd, u8 opcode) +{ + u8 funct7 = (funct5 << 2) | (aq << 1) | rl; + + return rv_r_insn(funct7, rs2, rs1, funct3, rd, opcode); +} + +static inline u32 rv_addi(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 0, rd, 0x13); +} + +static inline u32 rv_andi(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 7, rd, 0x13); +} + +static inline u32 rv_ori(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 6, rd, 0x13); +} + +static inline u32 rv_xori(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 4, rd, 0x13); +} + +static inline u32 rv_slli(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 1, rd, 0x13); +} + +static inline u32 rv_srli(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 5, rd, 0x13); +} + +static inline u32 rv_srai(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(0x400 | imm11_0, rs1, 5, rd, 0x13); +} + +static inline u32 rv_lui(u8 rd, u32 imm31_12) +{ + return rv_u_insn(imm31_12, rd, 0x37); +} + +static inline u32 rv_auipc(u8 rd, u32 imm31_12) +{ + return rv_u_insn(imm31_12, rd, 0x17); +} + +static inline u32 rv_add(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 0, rd, 0x33); +} + +static inline u32 rv_sub(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0x20, rs2, rs1, 0, rd, 0x33); +} + +static inline u32 rv_and(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 7, rd, 0x33); +} + +static inline u32 rv_or(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 6, rd, 0x33); +} + +static inline u32 rv_xor(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 4, rd, 0x33); +} + +static inline u32 rv_sll(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 1, rd, 0x33); +} + +static inline u32 rv_srl(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 5, rd, 0x33); +} + +static inline u32 rv_sra(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0x20, rs2, rs1, 5, rd, 0x33); +} + +static inline u32 rv_mul(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(1, rs2, rs1, 0, rd, 0x33); +} + +static inline u32 rv_divu(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(1, rs2, rs1, 5, rd, 0x33); +} + +static inline u32 rv_remu(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(1, rs2, rs1, 7, rd, 0x33); +} + +static inline u32 rv_jal(u8 rd, u32 imm20_1) +{ + return rv_j_insn(imm20_1, rd, 0x6f); +} + +static inline u32 rv_jalr(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 0, rd, 0x67); +} + +static inline u32 rv_beq(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_b_insn(imm12_1, rs2, rs1, 0, 0x63); +} + +static inline u32 rv_bne(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_b_insn(imm12_1, rs2, rs1, 1, 0x63); +} + +static inline u32 rv_bltu(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_b_insn(imm12_1, rs2, rs1, 6, 0x63); +} + +static inline u32 rv_bgeu(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_b_insn(imm12_1, rs2, rs1, 7, 0x63); +} + +static inline u32 rv_blt(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_b_insn(imm12_1, rs2, rs1, 4, 0x63); +} + +static inline u32 rv_bge(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_b_insn(imm12_1, rs2, rs1, 5, 0x63); +} + +static inline u32 rv_lbu(u8 rd, u16 imm11_0, u8 rs1) +{ + return rv_i_insn(imm11_0, rs1, 4, rd, 0x03); +} + +static inline u32 rv_lhu(u8 rd, u16 imm11_0, u8 rs1) +{ + return rv_i_insn(imm11_0, rs1, 5, rd, 0x03); +} + +static inline u32 rv_sb(u8 rs1, u16 imm11_0, u8 rs2) +{ + return rv_s_insn(imm11_0, rs2, rs1, 0, 0x23); +} + +static inline u32 rv_sh(u8 rs1, u16 imm11_0, u8 rs2) +{ + return rv_s_insn(imm11_0, rs2, rs1, 1, 0x23); +} + +static inline u32 rv_sw(u8 rs1, u16 imm11_0, u8 rs2) +{ + return rv_s_insn(imm11_0, rs2, rs1, 2, 0x23); +} + +static inline u32 rv_amoadd_w(u8 rd, u8 rs2, u8 rs1, u8 aq, u8 rl) +{ + return rv_amo_insn(0, aq, rl, rs2, rs1, 2, rd, 0x2f); +} + +static inline u32 rv_addiw(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 0, rd, 0x1b); +} + +static inline u32 rv_slliw(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 1, rd, 0x1b); +} + +static inline u32 rv_srliw(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(imm11_0, rs1, 5, rd, 0x1b); +} + +static inline u32 rv_sraiw(u8 rd, u8 rs1, u16 imm11_0) +{ + return rv_i_insn(0x400 | imm11_0, rs1, 5, rd, 0x1b); +} + +static inline u32 rv_addw(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 0, rd, 0x3b); +} + +static inline u32 rv_subw(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0x20, rs2, rs1, 0, rd, 0x3b); +} + +static inline u32 rv_sllw(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 1, rd, 0x3b); +} + +static inline u32 rv_srlw(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 5, rd, 0x3b); +} + +static inline u32 rv_sraw(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0x20, rs2, rs1, 5, rd, 0x3b); +} + +static inline u32 rv_mulw(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(1, rs2, rs1, 0, rd, 0x3b); +} + +static inline u32 rv_divuw(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(1, rs2, rs1, 5, rd, 0x3b); +} + +static inline u32 rv_remuw(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(1, rs2, rs1, 7, rd, 0x3b); +} + +static inline u32 rv_ld(u8 rd, u16 imm11_0, u8 rs1) +{ + return rv_i_insn(imm11_0, rs1, 3, rd, 0x03); +} + +static inline u32 rv_lwu(u8 rd, u16 imm11_0, u8 rs1) +{ + return rv_i_insn(imm11_0, rs1, 6, rd, 0x03); +} + +static inline u32 rv_sd(u8 rs1, u16 imm11_0, u8 rs2) +{ + return rv_s_insn(imm11_0, rs2, rs1, 3, 0x23); +} + +static inline u32 rv_amoadd_d(u8 rd, u8 rs2, u8 rs1, u8 aq, u8 rl) +{ + return rv_amo_insn(0, aq, rl, rs2, rs1, 3, rd, 0x2f); +} + +void bpf_jit_build_prologue(struct rv_jit_context *ctx); +void bpf_jit_build_epilogue(struct rv_jit_context *ctx); + +int bpf_jit_emit_insn(const struct bpf_insn *insn, struct rv_jit_context *ctx, + bool extra_pass); + +#endif /* _BPF_JIT_H */ diff --git a/arch/riscv/net/bpf_jit_comp.c b/arch/riscv/net/bpf_jit_comp64.c similarity index 69% rename from arch/riscv/net/bpf_jit_comp.c rename to arch/riscv/net/bpf_jit_comp64.c index 483f4ad7f4dc..cc1985d8750a 100644 --- a/arch/riscv/net/bpf_jit_comp.c +++ b/arch/riscv/net/bpf_jit_comp64.c @@ -7,42 +7,7 @@ #include #include -#include - -enum { - RV_REG_ZERO = 0, /* The constant value 0 */ - RV_REG_RA = 1, /* Return address */ - RV_REG_SP = 2, /* Stack pointer */ - RV_REG_GP = 3, /* Global pointer */ - RV_REG_TP = 4, /* Thread pointer */ - RV_REG_T0 = 5, /* Temporaries */ - RV_REG_T1 = 6, - RV_REG_T2 = 7, - RV_REG_FP = 8, - RV_REG_S1 = 9, /* Saved registers */ - RV_REG_A0 = 10, /* Function argument/return values */ - RV_REG_A1 = 11, /* Function arguments */ - RV_REG_A2 = 12, - RV_REG_A3 = 13, - RV_REG_A4 = 14, - RV_REG_A5 = 15, - RV_REG_A6 = 16, - RV_REG_A7 = 17, - RV_REG_S2 = 18, /* Saved registers */ - RV_REG_S3 = 19, - RV_REG_S4 = 20, - RV_REG_S5 = 21, - RV_REG_S6 = 22, - RV_REG_S7 = 23, - RV_REG_S8 = 24, - RV_REG_S9 = 25, - RV_REG_S10 = 26, - RV_REG_S11 = 27, - RV_REG_T3 = 28, /* Temporaries */ - RV_REG_T4 = 29, - RV_REG_T5 = 30, - RV_REG_T6 = 31, -}; +#include "bpf_jit.h" #define RV_REG_TCC RV_REG_A6 #define RV_REG_TCC_SAVED RV_REG_S6 /* Store A6 in S6 if program do calls */ @@ -73,22 +38,6 @@ enum { RV_CTX_F_SEEN_S6 = RV_REG_S6, }; -struct rv_jit_context { - struct bpf_prog *prog; - u32 *insns; /* RV insns */ - int ninsns; - int epilogue_offset; - int *offset; /* BPF to RV */ - unsigned long flags; - int stack_size; -}; - -struct rv_jit_data { - struct bpf_binary_header *header; - u8 *image; - struct rv_jit_context ctx; -}; - static u8 bpf_to_rv_reg(int bpf_reg, struct rv_jit_context *ctx) { u8 reg = regmap[bpf_reg]; @@ -156,346 +105,11 @@ static u8 rv_tail_call_reg(struct rv_jit_context *ctx) return RV_REG_A6; } -static void emit(const u32 insn, struct rv_jit_context *ctx) -{ - if (ctx->insns) - ctx->insns[ctx->ninsns] = insn; - - ctx->ninsns++; -} - -static u32 rv_r_insn(u8 funct7, u8 rs2, u8 rs1, u8 funct3, u8 rd, u8 opcode) -{ - return (funct7 << 25) | (rs2 << 20) | (rs1 << 15) | (funct3 << 12) | - (rd << 7) | opcode; -} - -static u32 rv_i_insn(u16 imm11_0, u8 rs1, u8 funct3, u8 rd, u8 opcode) -{ - return (imm11_0 << 20) | (rs1 << 15) | (funct3 << 12) | (rd << 7) | - opcode; -} - -static u32 rv_s_insn(u16 imm11_0, u8 rs2, u8 rs1, u8 funct3, u8 opcode) -{ - u8 imm11_5 = imm11_0 >> 5, imm4_0 = imm11_0 & 0x1f; - - return (imm11_5 << 25) | (rs2 << 20) | (rs1 << 15) | (funct3 << 12) | - (imm4_0 << 7) | opcode; -} - -static u32 rv_sb_insn(u16 imm12_1, u8 rs2, u8 rs1, u8 funct3, u8 opcode) -{ - u8 imm12 = ((imm12_1 & 0x800) >> 5) | ((imm12_1 & 0x3f0) >> 4); - u8 imm4_1 = ((imm12_1 & 0xf) << 1) | ((imm12_1 & 0x400) >> 10); - - return (imm12 << 25) | (rs2 << 20) | (rs1 << 15) | (funct3 << 12) | - (imm4_1 << 7) | opcode; -} - -static u32 rv_u_insn(u32 imm31_12, u8 rd, u8 opcode) -{ - return (imm31_12 << 12) | (rd << 7) | opcode; -} - -static u32 rv_uj_insn(u32 imm20_1, u8 rd, u8 opcode) -{ - u32 imm; - - imm = (imm20_1 & 0x80000) | ((imm20_1 & 0x3ff) << 9) | - ((imm20_1 & 0x400) >> 2) | ((imm20_1 & 0x7f800) >> 11); - - return (imm << 12) | (rd << 7) | opcode; -} - -static u32 rv_amo_insn(u8 funct5, u8 aq, u8 rl, u8 rs2, u8 rs1, - u8 funct3, u8 rd, u8 opcode) -{ - u8 funct7 = (funct5 << 2) | (aq << 1) | rl; - - return rv_r_insn(funct7, rs2, rs1, funct3, rd, opcode); -} - -static u32 rv_addiw(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 0, rd, 0x1b); -} - -static u32 rv_addi(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 0, rd, 0x13); -} - -static u32 rv_addw(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0, rs2, rs1, 0, rd, 0x3b); -} - -static u32 rv_add(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0, rs2, rs1, 0, rd, 0x33); -} - -static u32 rv_subw(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0x20, rs2, rs1, 0, rd, 0x3b); -} - -static u32 rv_sub(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0x20, rs2, rs1, 0, rd, 0x33); -} - -static u32 rv_and(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0, rs2, rs1, 7, rd, 0x33); -} - -static u32 rv_or(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0, rs2, rs1, 6, rd, 0x33); -} - -static u32 rv_xor(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0, rs2, rs1, 4, rd, 0x33); -} - -static u32 rv_mulw(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(1, rs2, rs1, 0, rd, 0x3b); -} - -static u32 rv_mul(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(1, rs2, rs1, 0, rd, 0x33); -} - -static u32 rv_divuw(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(1, rs2, rs1, 5, rd, 0x3b); -} - -static u32 rv_divu(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(1, rs2, rs1, 5, rd, 0x33); -} - -static u32 rv_remuw(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(1, rs2, rs1, 7, rd, 0x3b); -} - -static u32 rv_remu(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(1, rs2, rs1, 7, rd, 0x33); -} - -static u32 rv_sllw(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0, rs2, rs1, 1, rd, 0x3b); -} - -static u32 rv_sll(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0, rs2, rs1, 1, rd, 0x33); -} - -static u32 rv_srlw(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0, rs2, rs1, 5, rd, 0x3b); -} - -static u32 rv_srl(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0, rs2, rs1, 5, rd, 0x33); -} - -static u32 rv_sraw(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0x20, rs2, rs1, 5, rd, 0x3b); -} - -static u32 rv_sra(u8 rd, u8 rs1, u8 rs2) -{ - return rv_r_insn(0x20, rs2, rs1, 5, rd, 0x33); -} - -static u32 rv_lui(u8 rd, u32 imm31_12) -{ - return rv_u_insn(imm31_12, rd, 0x37); -} - -static u32 rv_slli(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 1, rd, 0x13); -} - -static u32 rv_andi(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 7, rd, 0x13); -} - -static u32 rv_ori(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 6, rd, 0x13); -} - -static u32 rv_xori(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 4, rd, 0x13); -} - -static u32 rv_slliw(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 1, rd, 0x1b); -} - -static u32 rv_srliw(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 5, rd, 0x1b); -} - -static u32 rv_srli(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 5, rd, 0x13); -} - -static u32 rv_sraiw(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(0x400 | imm11_0, rs1, 5, rd, 0x1b); -} - -static u32 rv_srai(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(0x400 | imm11_0, rs1, 5, rd, 0x13); -} - -static u32 rv_jal(u8 rd, u32 imm20_1) -{ - return rv_uj_insn(imm20_1, rd, 0x6f); -} - -static u32 rv_jalr(u8 rd, u8 rs1, u16 imm11_0) -{ - return rv_i_insn(imm11_0, rs1, 0, rd, 0x67); -} - -static u32 rv_beq(u8 rs1, u8 rs2, u16 imm12_1) -{ - return rv_sb_insn(imm12_1, rs2, rs1, 0, 0x63); -} - -static u32 rv_bltu(u8 rs1, u8 rs2, u16 imm12_1) -{ - return rv_sb_insn(imm12_1, rs2, rs1, 6, 0x63); -} - -static u32 rv_bgeu(u8 rs1, u8 rs2, u16 imm12_1) -{ - return rv_sb_insn(imm12_1, rs2, rs1, 7, 0x63); -} - -static u32 rv_bne(u8 rs1, u8 rs2, u16 imm12_1) -{ - return rv_sb_insn(imm12_1, rs2, rs1, 1, 0x63); -} - -static u32 rv_blt(u8 rs1, u8 rs2, u16 imm12_1) -{ - return rv_sb_insn(imm12_1, rs2, rs1, 4, 0x63); -} - -static u32 rv_bge(u8 rs1, u8 rs2, u16 imm12_1) -{ - return rv_sb_insn(imm12_1, rs2, rs1, 5, 0x63); -} - -static u32 rv_sb(u8 rs1, u16 imm11_0, u8 rs2) -{ - return rv_s_insn(imm11_0, rs2, rs1, 0, 0x23); -} - -static u32 rv_sh(u8 rs1, u16 imm11_0, u8 rs2) -{ - return rv_s_insn(imm11_0, rs2, rs1, 1, 0x23); -} - -static u32 rv_sw(u8 rs1, u16 imm11_0, u8 rs2) -{ - return rv_s_insn(imm11_0, rs2, rs1, 2, 0x23); -} - -static u32 rv_sd(u8 rs1, u16 imm11_0, u8 rs2) -{ - return rv_s_insn(imm11_0, rs2, rs1, 3, 0x23); -} - -static u32 rv_lbu(u8 rd, u16 imm11_0, u8 rs1) -{ - return rv_i_insn(imm11_0, rs1, 4, rd, 0x03); -} - -static u32 rv_lhu(u8 rd, u16 imm11_0, u8 rs1) -{ - return rv_i_insn(imm11_0, rs1, 5, rd, 0x03); -} - -static u32 rv_lwu(u8 rd, u16 imm11_0, u8 rs1) -{ - return rv_i_insn(imm11_0, rs1, 6, rd, 0x03); -} - -static u32 rv_ld(u8 rd, u16 imm11_0, u8 rs1) -{ - return rv_i_insn(imm11_0, rs1, 3, rd, 0x03); -} - -static u32 rv_amoadd_w(u8 rd, u8 rs2, u8 rs1, u8 aq, u8 rl) -{ - return rv_amo_insn(0, aq, rl, rs2, rs1, 2, rd, 0x2f); -} - -static u32 rv_amoadd_d(u8 rd, u8 rs2, u8 rs1, u8 aq, u8 rl) -{ - return rv_amo_insn(0, aq, rl, rs2, rs1, 3, rd, 0x2f); -} - -static u32 rv_auipc(u8 rd, u32 imm31_12) -{ - return rv_u_insn(imm31_12, rd, 0x17); -} - -static bool is_12b_int(s64 val) -{ - return -(1 << 11) <= val && val < (1 << 11); -} - -static bool is_13b_int(s64 val) -{ - return -(1 << 12) <= val && val < (1 << 12); -} - -static bool is_21b_int(s64 val) -{ - return -(1L << 20) <= val && val < (1L << 20); -} - static bool is_32b_int(s64 val) { return -(1L << 31) <= val && val < (1L << 31); } -static int is_12b_check(int off, int insn) -{ - if (!is_12b_int(off)) { - pr_err("bpf-jit: insn=%d 12b < offset=%d not supported yet!\n", - insn, (int)off); - return -1; - } - return 0; -} - static void emit_imm(u8 rd, s64 val, struct rv_jit_context *ctx) { /* Note that the immediate from the add is sign-extended, @@ -535,23 +149,6 @@ static void emit_imm(u8 rd, s64 val, struct rv_jit_context *ctx) emit(rv_addi(rd, rd, lower), ctx); } -static int rv_offset(int insn, int off, struct rv_jit_context *ctx) -{ - int from, to; - - off++; /* BPF branch is from PC+1, RV is from PC */ - from = (insn > 0) ? ctx->offset[insn - 1] : 0; - to = (insn + off > 0) ? ctx->offset[insn + off - 1] : 0; - return (to - from) << 2; -} - -static int epilogue_offset(struct rv_jit_context *ctx) -{ - int to = ctx->epilogue_offset, from = ctx->ninsns; - - return (to - from) << 2; -} - static void __build_epilogue(bool is_tail_call, struct rv_jit_context *ctx) { int stack_adjust = ctx->stack_size, store_offset = stack_adjust - 8; @@ -596,34 +193,6 @@ static void __build_epilogue(bool is_tail_call, struct rv_jit_context *ctx) ctx); } -/* return -1 or inverted cond */ -static int invert_bpf_cond(u8 cond) -{ - switch (cond) { - case BPF_JEQ: - return BPF_JNE; - case BPF_JGT: - return BPF_JLE; - case BPF_JLT: - return BPF_JGE; - case BPF_JGE: - return BPF_JLT; - case BPF_JLE: - return BPF_JGT; - case BPF_JNE: - return BPF_JEQ; - case BPF_JSGT: - return BPF_JSLE; - case BPF_JSLT: - return BPF_JSGE; - case BPF_JSGE: - return BPF_JSLT; - case BPF_JSLE: - return BPF_JSGT; - } - return -1; -} - static void emit_bcc(u8 cond, u8 rd, u8 rs, int rvoff, struct rv_jit_context *ctx) { @@ -855,8 +424,8 @@ static int emit_call(bool fixed, u64 addr, struct rv_jit_context *ctx) return 0; } -static int emit_insn(const struct bpf_insn *insn, struct rv_jit_context *ctx, - bool extra_pass) +int bpf_jit_emit_insn(const struct bpf_insn *insn, struct rv_jit_context *ctx, + bool extra_pass) { bool is64 = BPF_CLASS(insn->code) == BPF_ALU64 || BPF_CLASS(insn->code) == BPF_JMP; @@ -1434,7 +1003,7 @@ static int emit_insn(const struct bpf_insn *insn, struct rv_jit_context *ctx, return 0; } -static void build_prologue(struct rv_jit_context *ctx) +void bpf_jit_build_prologue(struct rv_jit_context *ctx) { int stack_adjust = 0, store_offset, bpf_stack_adjust; @@ -1515,175 +1084,11 @@ static void build_prologue(struct rv_jit_context *ctx) ctx->stack_size = stack_adjust; } -static void build_epilogue(struct rv_jit_context *ctx) +void bpf_jit_build_epilogue(struct rv_jit_context *ctx) { __build_epilogue(false, ctx); } -static int build_body(struct rv_jit_context *ctx, bool extra_pass, int *offset) -{ - const struct bpf_prog *prog = ctx->prog; - int i; - - for (i = 0; i < prog->len; i++) { - const struct bpf_insn *insn = &prog->insnsi[i]; - int ret; - - ret = emit_insn(insn, ctx, extra_pass); - if (ret > 0) { - i++; - if (offset) - offset[i] = ctx->ninsns; - continue; - } - if (offset) - offset[i] = ctx->ninsns; - if (ret) - return ret; - } - return 0; -} - -static void bpf_fill_ill_insns(void *area, unsigned int size) -{ - memset(area, 0, size); -} - -static void bpf_flush_icache(void *start, void *end) -{ - flush_icache_range((unsigned long)start, (unsigned long)end); -} - -bool bpf_jit_needs_zext(void) -{ - return true; -} - -struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) -{ - bool tmp_blinded = false, extra_pass = false; - struct bpf_prog *tmp, *orig_prog = prog; - int pass = 0, prev_ninsns = 0, i; - struct rv_jit_data *jit_data; - unsigned int image_size = 0; - struct rv_jit_context *ctx; - - if (!prog->jit_requested) - return orig_prog; - - tmp = bpf_jit_blind_constants(prog); - if (IS_ERR(tmp)) - return orig_prog; - if (tmp != prog) { - tmp_blinded = true; - prog = tmp; - } - - jit_data = prog->aux->jit_data; - if (!jit_data) { - jit_data = kzalloc(sizeof(*jit_data), GFP_KERNEL); - if (!jit_data) { - prog = orig_prog; - goto out; - } - prog->aux->jit_data = jit_data; - } - - ctx = &jit_data->ctx; - - if (ctx->offset) { - extra_pass = true; - image_size = sizeof(u32) * ctx->ninsns; - goto skip_init_ctx; - } - - ctx->prog = prog; - ctx->offset = kcalloc(prog->len, sizeof(int), GFP_KERNEL); - if (!ctx->offset) { - prog = orig_prog; - goto out_offset; - } - for (i = 0; i < prog->len; i++) { - prev_ninsns += 32; - ctx->offset[i] = prev_ninsns; - } - - for (i = 0; i < 16; i++) { - pass++; - ctx->ninsns = 0; - if (build_body(ctx, extra_pass, ctx->offset)) { - prog = orig_prog; - goto out_offset; - } - build_prologue(ctx); - ctx->epilogue_offset = ctx->ninsns; - build_epilogue(ctx); - - if (ctx->ninsns == prev_ninsns) { - if (jit_data->header) - break; - - image_size = sizeof(u32) * ctx->ninsns; - jit_data->header = - bpf_jit_binary_alloc(image_size, - &jit_data->image, - sizeof(u32), - bpf_fill_ill_insns); - if (!jit_data->header) { - prog = orig_prog; - goto out_offset; - } - - ctx->insns = (u32 *)jit_data->image; - /* Now, when the image is allocated, the image - * can potentially shrink more (auipc/jalr -> - * jal). - */ - } - prev_ninsns = ctx->ninsns; - } - - if (i == 16) { - pr_err("bpf-jit: image did not converge in <%d passes!\n", i); - bpf_jit_binary_free(jit_data->header); - prog = orig_prog; - goto out_offset; - } - -skip_init_ctx: - pass++; - ctx->ninsns = 0; - - build_prologue(ctx); - if (build_body(ctx, extra_pass, NULL)) { - bpf_jit_binary_free(jit_data->header); - prog = orig_prog; - goto out_offset; - } - build_epilogue(ctx); - - if (bpf_jit_enable > 1) - bpf_jit_dump(prog->len, image_size, pass, ctx->insns); - - prog->bpf_func = (void *)ctx->insns; - prog->jited = 1; - prog->jited_len = image_size; - - bpf_flush_icache(jit_data->header, ctx->insns + ctx->ninsns); - - if (!prog->is_func || extra_pass) { -out_offset: - kfree(ctx->offset); - kfree(jit_data); - prog->aux->jit_data = NULL; - } -out: - if (tmp_blinded) - bpf_jit_prog_release_other(prog, prog == orig_prog ? - tmp : orig_prog); - return prog; -} - void *bpf_jit_alloc_exec(unsigned long size) { return __vmalloc_node_range(size, PAGE_SIZE, BPF_JIT_REGION_START, diff --git a/arch/riscv/net/bpf_jit_core.c b/arch/riscv/net/bpf_jit_core.c new file mode 100644 index 000000000000..709b94ece3ed --- /dev/null +++ b/arch/riscv/net/bpf_jit_core.c @@ -0,0 +1,166 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Common functionality for RV32 and RV64 BPF JIT compilers + * + * Copyright (c) 2019 Björn Töpel + * + */ + +#include +#include +#include "bpf_jit.h" + +/* Number of iterations to try until offsets converge. */ +#define NR_JIT_ITERATIONS 16 + +static int build_body(struct rv_jit_context *ctx, bool extra_pass, int *offset) +{ + const struct bpf_prog *prog = ctx->prog; + int i; + + for (i = 0; i < prog->len; i++) { + const struct bpf_insn *insn = &prog->insnsi[i]; + int ret; + + ret = bpf_jit_emit_insn(insn, ctx, extra_pass); + /* BPF_LD | BPF_IMM | BPF_DW: skip the next instruction. */ + if (ret > 0) + i++; + if (offset) + offset[i] = ctx->ninsns; + if (ret < 0) + return ret; + } + return 0; +} + +bool bpf_jit_needs_zext(void) +{ + return true; +} + +struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) +{ + bool tmp_blinded = false, extra_pass = false; + struct bpf_prog *tmp, *orig_prog = prog; + int pass = 0, prev_ninsns = 0, i; + struct rv_jit_data *jit_data; + struct rv_jit_context *ctx; + unsigned int image_size = 0; + + if (!prog->jit_requested) + return orig_prog; + + tmp = bpf_jit_blind_constants(prog); + if (IS_ERR(tmp)) + return orig_prog; + if (tmp != prog) { + tmp_blinded = true; + prog = tmp; + } + + jit_data = prog->aux->jit_data; + if (!jit_data) { + jit_data = kzalloc(sizeof(*jit_data), GFP_KERNEL); + if (!jit_data) { + prog = orig_prog; + goto out; + } + prog->aux->jit_data = jit_data; + } + + ctx = &jit_data->ctx; + + if (ctx->offset) { + extra_pass = true; + image_size = sizeof(u32) * ctx->ninsns; + goto skip_init_ctx; + } + + ctx->prog = prog; + ctx->offset = kcalloc(prog->len, sizeof(int), GFP_KERNEL); + if (!ctx->offset) { + prog = orig_prog; + goto out_offset; + } + for (i = 0; i < prog->len; i++) { + prev_ninsns += 32; + ctx->offset[i] = prev_ninsns; + } + + for (i = 0; i < NR_JIT_ITERATIONS; i++) { + pass++; + ctx->ninsns = 0; + if (build_body(ctx, extra_pass, ctx->offset)) { + prog = orig_prog; + goto out_offset; + } + bpf_jit_build_prologue(ctx); + ctx->epilogue_offset = ctx->ninsns; + bpf_jit_build_epilogue(ctx); + + if (ctx->ninsns == prev_ninsns) { + if (jit_data->header) + break; + + image_size = sizeof(u32) * ctx->ninsns; + jit_data->header = + bpf_jit_binary_alloc(image_size, + &jit_data->image, + sizeof(u32), + bpf_fill_ill_insns); + if (!jit_data->header) { + prog = orig_prog; + goto out_offset; + } + + ctx->insns = (u32 *)jit_data->image; + /* + * Now, when the image is allocated, the image can + * potentially shrink more (auipc/jalr -> jal). + */ + } + prev_ninsns = ctx->ninsns; + } + + if (i == NR_JIT_ITERATIONS) { + pr_err("bpf-jit: image did not converge in <%d passes!\n", i); + bpf_jit_binary_free(jit_data->header); + prog = orig_prog; + goto out_offset; + } + +skip_init_ctx: + pass++; + ctx->ninsns = 0; + + bpf_jit_build_prologue(ctx); + if (build_body(ctx, extra_pass, NULL)) { + bpf_jit_binary_free(jit_data->header); + prog = orig_prog; + goto out_offset; + } + bpf_jit_build_epilogue(ctx); + + if (bpf_jit_enable > 1) + bpf_jit_dump(prog->len, image_size, pass, ctx->insns); + + prog->bpf_func = (void *)ctx->insns; + prog->jited = 1; + prog->jited_len = image_size; + + bpf_flush_icache(jit_data->header, ctx->insns + ctx->ninsns); + + if (!prog->is_func || extra_pass) { +out_offset: + kfree(ctx->offset); + kfree(jit_data); + prog->aux->jit_data = NULL; + } +out: + + if (tmp_blinded) + bpf_jit_prog_release_other(prog, prog == orig_prog ? + tmp : orig_prog); + return prog; +} From patchwork Thu Mar 5 05:02:05 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Luke Nelson X-Patchwork-Id: 11421199 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 4AED5138D for ; Thu, 5 Mar 2020 05:02:38 +0000 (UTC) Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id 2204D2146E for ; Thu, 5 Mar 2020 05:02:38 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=lists.infradead.org header.i=@lists.infradead.org header.b="MDK4GksL"; dkim=fail reason="signature verification failed" (1024-bit key) header.d=cs.washington.edu header.i=@cs.washington.edu header.b="Z7Na1A7/" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 2204D2146E Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=cs.washington.edu Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-riscv-bounces+patchwork-linux-riscv=patchwork.kernel.org@lists.infradead.org DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20170209; h=Sender:Cc:List-Subscribe: List-Help:List-Post:List-Archive:List-Unsubscribe:List-Id: Content-Transfer-Encoding:Content-Type:MIME-Version:References:In-Reply-To: Message-Id:Date:Subject:To:From:Reply-To:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=NV6AEbjaEc8/8IxaZ62CnVxM/2PY2Fk+J8+O53OGrHo=; b=MDK4GksLlOBj6h Q+0bJu5vTu2kkPStByivOZl+P4hHVYL016INMarOvYAiFYacTRpUYMx8lzz9TQZnBe8NhP2jQedGA ZCtBU5HSB06ktPG1gYpLdOYQuLHMz9CPt8WH8CMzPLqwQISae8bsQPmUfvqglhorRd2cFNFrcbRX8 c8sSXPviENzzJaRTtW9jad34saBMqhjIj/w6YAxZfbciVlrx/Jql+RBj+l/UuDKuAqTAKBQcDL0sT YtIyOLgEPnzF1MBAxmlwfqeO6kM7WqcZ4Ny/7HstewHkFLf3ulPxN3puyGXbxZr2wJXRTe3d5ALvq jmwE7Us2ORlELcLNrp0g==; Received: from localhost ([127.0.0.1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.92.3 #3 (Red Hat Linux)) id 1j9ieY-0004gP-K4; Thu, 05 Mar 2020 05:02:34 +0000 Received: from mail-pf1-x442.google.com ([2607:f8b0:4864:20::442]) by bombadil.infradead.org with esmtps (Exim 4.92.3 #3 (Red Hat Linux)) id 1j9ieP-0004Rl-JI for linux-riscv@lists.infradead.org; Thu, 05 Mar 2020 05:02:30 +0000 Received: by mail-pf1-x442.google.com with SMTP id c144so2145667pfb.10 for ; Wed, 04 Mar 2020 21:02:21 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=cs.washington.edu; s=goo201206; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=NV6AEbjaEc8/8IxaZ62CnVxM/2PY2Fk+J8+O53OGrHo=; b=Z7Na1A7/J2rmtSrjfm1Pnn1+QgyH7j0byzrFHmqCPAl3EW0uR+YTrABKQv6ENPWDPP orDAlfo6BEaSBTg2GuxhH4oU7WQ6o7SCF3PoWH+D20ickxMZ+HiJurAluMVbbhMtGQN1 N57ao9BCgDLuadL6RK8iHN84HQE8T6C9IrSJw= X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=NV6AEbjaEc8/8IxaZ62CnVxM/2PY2Fk+J8+O53OGrHo=; b=CEjWkPnygwFbpjU4Y/sPMKVVmNDFLoR53VU7yo/JXaYW/KVAAI2HfuPxBiytQa5b4O 4sPYKiLdiqf5jGY9BSZZ3latQviD4D4kpluln52SlHZ3uqdW6i4MMbJKPvxdmuhwIkMD 888KKIa0vZVXUOXbqIeFVl1lUZSK40g5wTPazRY7vnoPFW2I5YzGmkUB9EJi+dZrOVBr DbziJD6oeFvSrLykPHktICGRYd7mv1YVxcNtWcBZARDQvtkbeQtAULPEbr8cZnIwl6EK vi628i6RpOEqCT1SQt4fhX8avLON/+XRVFViVnUU6gds4ziJGK8D0uhFse7bwtInvq/c sCPw== X-Gm-Message-State: ANhLgQ1x81gWOjglFGHevXEMiXvvTEk9iMdlxa0QJ20daFjqaQRvi9h+ tesjqcwOPXXuDDSx2Ebf97JvvQ== X-Google-Smtp-Source: ADFU+vskwIjVEaOdYxodQPRM/kZw7WAAgHCv+iVAmebCKkcBp2PP5DfZDd7IBll9iaCRz6zkD3rxiw== X-Received: by 2002:a62:7696:: with SMTP id r144mr6498580pfc.177.1583384539643; Wed, 04 Mar 2020 21:02:19 -0800 (PST) Received: from ryzen.cs.washington.edu ([2607:4000:200:11:e9fe:faad:3d84:58ea]) by smtp.gmail.com with ESMTPSA id y7sm17820466pfq.15.2020.03.04.21.02.18 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 04 Mar 2020 21:02:19 -0800 (PST) From: Luke Nelson X-Google-Original-From: Luke Nelson To: bpf@vger.kernel.org Subject: [PATCH bpf-next v5 2/4] riscv, bpf: add RV32G eBPF JIT Date: Wed, 4 Mar 2020 21:02:05 -0800 Message-Id: <20200305050207.4159-3-luke.r.nels@gmail.com> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20200305050207.4159-1-luke.r.nels@gmail.com> References: <20200305050207.4159-1-luke.r.nels@gmail.com> MIME-Version: 1.0 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20200304_210225_729973_5A248E5D X-CRM114-Status: GOOD ( 21.07 ) X-Spam-Score: -0.1 (/) X-Spam-Report: SpamAssassin version 3.4.3 on bombadil.infradead.org summary: Content analysis details: (-0.1 points) pts rule name description ---- ---------------------- -------------------------------------------------- -0.0 RCVD_IN_DNSWL_NONE RBL: Sender listed at https://www.dnswl.org/, no trust [2607:f8b0:4864:20:0:0:0:442 listed in] [list.dnswl.org] -0.0 SPF_PASS SPF: sender matches SPF record 0.0 SPF_HELO_NONE SPF: HELO does not publish an SPF Record -0.1 DKIM_VALID Message has at least one valid DKIM or DK signature -0.1 DKIM_VALID_AU Message has a valid DKIM or DK signature from author's domain 0.1 DKIM_SIGNED Message has a DKIM or DK signature, not necessarily valid X-BeenThere: linux-riscv@lists.infradead.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Song Liu , linux-doc@vger.kernel.org, Yonghong Song , Paul Walmsley , Alexei Starovoitov , netdev@vger.kernel.org, Mauro Carvalho Chehab , linux-riscv@lists.infradead.org, Rob Herring , Daniel Borkmann , Jonathan Corbet , Jakub Kicinski , Andrii Nakryiko , Xi Wang , Albert Ou , Luke Nelson , Jonathan Cameron , Andy Shevchenko , =?utf-8?b?QmrDtnJuIFQ=?= =?utf-8?b?w7ZwZWw=?= , linux-kernel@vger.kernel.org, Martin KaFai Lau , Stephen Hemminger , Palmer Dabbelt , Greg Kroah-Hartman , "David S. Miller" Sender: "linux-riscv" Errors-To: linux-riscv-bounces+patchwork-linux-riscv=patchwork.kernel.org@lists.infradead.org This is an eBPF JIT for RV32G, adapted from the JIT for RV64G and the 32-bit ARM JIT. There are two main changes required for this to work compared to the RV64 JIT. First, eBPF registers are 64-bit, while RV32G registers are 32-bit. BPF registers either map directly to 2 RISC-V registers, or reside in stack scratch space and are saved and restored when used. Second, many 64-bit ALU operations do not trivially map to 32-bit operations. Operations that move bits between high and low words, such as ADD, LSH, MUL, and others must emulate the 64-bit behavior in terms of 32-bit instructions. This patch also makes related changes to bpf_jit.h, such as adding RISC-V instructions required by the RV32 JIT. Supported features: The RV32 JIT supports the same features and instructions as the RV64 JIT, with the following exceptions: - ALU64 DIV/MOD: Requires loops to implement on 32-bit hardware. - BPF_XADD | BPF_DW: There's no 8-byte atomic instruction in RV32. These features are also unsupported on other BPF JITs for 32-bit architectures. Testing: - lib/test_bpf.c test_bpf: Summary: 378 PASSED, 0 FAILED, [349/366 JIT'ed] test_bpf: test_skb_segment: Summary: 2 PASSED, 0 FAILED The tests that are not JITed are all due to use of 64-bit div/mod or 64-bit xadd. - tools/testing/selftests/bpf/test_verifier.c Summary: 1415 PASSED, 122 SKIPPED, 43 FAILED Tested both with and without BPF JIT hardening. This is the same set of tests that pass using the BPF interpreter with the JIT disabled. Verification and synthesis: We developed the RV32 JIT using our automated verification tool, Serval. We have used Serval in the past to verify patches to the RV64 JIT. We also used Serval to superoptimize the resulting code through program synthesis. You can find the tool and a guide to the approach and results here: https://github.com/uw-unsat/serval-bpf/tree/rv32-jit-v5 Co-developed-by: Xi Wang Signed-off-by: Xi Wang Signed-off-by: Luke Nelson --- arch/riscv/Kconfig | 2 +- arch/riscv/net/Makefile | 8 +- arch/riscv/net/bpf_jit.h | 48 ++ arch/riscv/net/bpf_jit_comp32.c | 1310 +++++++++++++++++++++++++++++++ 4 files changed, 1366 insertions(+), 2 deletions(-) create mode 100644 arch/riscv/net/bpf_jit_comp32.c diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig index 73f029eae0cc..e9791717e006 100644 --- a/arch/riscv/Kconfig +++ b/arch/riscv/Kconfig @@ -56,7 +56,7 @@ config RISCV select ARCH_HAS_PTE_SPECIAL select ARCH_HAS_MMIOWB select ARCH_HAS_DEBUG_VIRTUAL - select HAVE_EBPF_JIT if 64BIT + select HAVE_EBPF_JIT select EDAC_SUPPORT select ARCH_HAS_GIGANTIC_PAGE select ARCH_WANT_HUGE_PMD_SHARE if 64BIT diff --git a/arch/riscv/net/Makefile b/arch/riscv/net/Makefile index 018074dbf986..9a1e5f0a94e5 100644 --- a/arch/riscv/net/Makefile +++ b/arch/riscv/net/Makefile @@ -1,3 +1,9 @@ # SPDX-License-Identifier: GPL-2.0-only -obj-$(CONFIG_BPF_JIT) += bpf_jit_core.o bpf_jit_comp64.o +obj-$(CONFIG_BPF_JIT) += bpf_jit_core.o + +ifeq ($(CONFIG_ARCH_RV64I),y) + obj-$(CONFIG_BPF_JIT) += bpf_jit_comp64.o +else + obj-$(CONFIG_BPF_JIT) += bpf_jit_comp32.o +endif diff --git a/arch/riscv/net/bpf_jit.h b/arch/riscv/net/bpf_jit.h index 23c123331f94..20e235d06f66 100644 --- a/arch/riscv/net/bpf_jit.h +++ b/arch/riscv/net/bpf_jit.h @@ -207,6 +207,8 @@ static inline u32 rv_amo_insn(u8 funct5, u8 aq, u8 rl, u8 rs2, u8 rs1, return rv_r_insn(funct7, rs2, rs1, funct3, rd, opcode); } +/* Instructions shared by both RV32 and RV64. */ + static inline u32 rv_addi(u8 rd, u8 rs1, u16 imm11_0) { return rv_i_insn(imm11_0, rs1, 0, rd, 0x13); @@ -262,6 +264,11 @@ static inline u32 rv_sub(u8 rd, u8 rs1, u8 rs2) return rv_r_insn(0x20, rs2, rs1, 0, rd, 0x33); } +static inline u32 rv_sltu(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(0, rs2, rs1, 3, rd, 0x33); +} + static inline u32 rv_and(u8 rd, u8 rs1, u8 rs2) { return rv_r_insn(0, rs2, rs1, 7, rd, 0x33); @@ -297,6 +304,11 @@ static inline u32 rv_mul(u8 rd, u8 rs1, u8 rs2) return rv_r_insn(1, rs2, rs1, 0, rd, 0x33); } +static inline u32 rv_mulhu(u8 rd, u8 rs1, u8 rs2) +{ + return rv_r_insn(1, rs2, rs1, 3, rd, 0x33); +} + static inline u32 rv_divu(u8 rd, u8 rs1, u8 rs2) { return rv_r_insn(1, rs2, rs1, 5, rd, 0x33); @@ -332,21 +344,46 @@ static inline u32 rv_bltu(u8 rs1, u8 rs2, u16 imm12_1) return rv_b_insn(imm12_1, rs2, rs1, 6, 0x63); } +static inline u32 rv_bgtu(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_bltu(rs2, rs1, imm12_1); +} + static inline u32 rv_bgeu(u8 rs1, u8 rs2, u16 imm12_1) { return rv_b_insn(imm12_1, rs2, rs1, 7, 0x63); } +static inline u32 rv_bleu(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_bgeu(rs2, rs1, imm12_1); +} + static inline u32 rv_blt(u8 rs1, u8 rs2, u16 imm12_1) { return rv_b_insn(imm12_1, rs2, rs1, 4, 0x63); } +static inline u32 rv_bgt(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_blt(rs2, rs1, imm12_1); +} + static inline u32 rv_bge(u8 rs1, u8 rs2, u16 imm12_1) { return rv_b_insn(imm12_1, rs2, rs1, 5, 0x63); } +static inline u32 rv_ble(u8 rs1, u8 rs2, u16 imm12_1) +{ + return rv_bge(rs2, rs1, imm12_1); +} + +static inline u32 rv_lw(u8 rd, u16 imm11_0, u8 rs1) +{ + return rv_i_insn(imm11_0, rs1, 2, rd, 0x03); +} + static inline u32 rv_lbu(u8 rd, u16 imm11_0, u8 rs1) { return rv_i_insn(imm11_0, rs1, 4, rd, 0x03); @@ -377,6 +414,15 @@ static inline u32 rv_amoadd_w(u8 rd, u8 rs2, u8 rs1, u8 aq, u8 rl) return rv_amo_insn(0, aq, rl, rs2, rs1, 2, rd, 0x2f); } +/* + * RV64-only instructions. + * + * These instructions are not available on RV32. Wrap them below a #if to + * ensure that the RV32 JIT doesn't emit any of these instructions. + */ + +#if __riscv_xlen == 64 + static inline u32 rv_addiw(u8 rd, u8 rs1, u16 imm11_0) { return rv_i_insn(imm11_0, rs1, 0, rd, 0x1b); @@ -457,6 +503,8 @@ static inline u32 rv_amoadd_d(u8 rd, u8 rs2, u8 rs1, u8 aq, u8 rl) return rv_amo_insn(0, aq, rl, rs2, rs1, 3, rd, 0x2f); } +#endif /* __riscv_xlen == 64 */ + void bpf_jit_build_prologue(struct rv_jit_context *ctx); void bpf_jit_build_epilogue(struct rv_jit_context *ctx); diff --git a/arch/riscv/net/bpf_jit_comp32.c b/arch/riscv/net/bpf_jit_comp32.c new file mode 100644 index 000000000000..302934177760 --- /dev/null +++ b/arch/riscv/net/bpf_jit_comp32.c @@ -0,0 +1,1310 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * BPF JIT compiler for RV32G + * + * Copyright (c) 2020 Luke Nelson + * Copyright (c) 2020 Xi Wang + * + * The code is based on the BPF JIT compiler for RV64G by Björn Töpel and + * the BPF JIT compiler for 32-bit ARM by Shubham Bansal and Mircea Gherzan. + */ + +#include +#include +#include "bpf_jit.h" + +enum { + /* Stack layout - these are offsets from (top of stack - 4). */ + BPF_R6_HI, + BPF_R6_LO, + BPF_R7_HI, + BPF_R7_LO, + BPF_R8_HI, + BPF_R8_LO, + BPF_R9_HI, + BPF_R9_LO, + BPF_AX_HI, + BPF_AX_LO, + /* Stack space for BPF_REG_6 through BPF_REG_9 and BPF_REG_AX. */ + BPF_JIT_SCRATCH_REGS, +}; + +#define STACK_OFFSET(k) (-4 - ((k) * 4)) + +#define TMP_REG_1 (MAX_BPF_JIT_REG + 0) +#define TMP_REG_2 (MAX_BPF_JIT_REG + 1) + +#define RV_REG_TCC RV_REG_T6 +#define RV_REG_TCC_SAVED RV_REG_S7 + +static const s8 bpf2rv32[][2] = { + /* Return value from in-kernel function, and exit value from eBPF. */ + [BPF_REG_0] = {RV_REG_S2, RV_REG_S1}, + /* Arguments from eBPF program to in-kernel function. */ + [BPF_REG_1] = {RV_REG_A1, RV_REG_A0}, + [BPF_REG_2] = {RV_REG_A3, RV_REG_A2}, + [BPF_REG_3] = {RV_REG_A5, RV_REG_A4}, + [BPF_REG_4] = {RV_REG_A7, RV_REG_A6}, + [BPF_REG_5] = {RV_REG_S4, RV_REG_S3}, + /* + * Callee-saved registers that in-kernel function will preserve. + * Stored on the stack. + */ + [BPF_REG_6] = {STACK_OFFSET(BPF_R6_HI), STACK_OFFSET(BPF_R6_LO)}, + [BPF_REG_7] = {STACK_OFFSET(BPF_R7_HI), STACK_OFFSET(BPF_R7_LO)}, + [BPF_REG_8] = {STACK_OFFSET(BPF_R8_HI), STACK_OFFSET(BPF_R8_LO)}, + [BPF_REG_9] = {STACK_OFFSET(BPF_R9_HI), STACK_OFFSET(BPF_R9_LO)}, + /* Read-only frame pointer to access BPF stack. */ + [BPF_REG_FP] = {RV_REG_S6, RV_REG_S5}, + /* Temporary register for blinding constants. Stored on the stack. */ + [BPF_REG_AX] = {STACK_OFFSET(BPF_AX_HI), STACK_OFFSET(BPF_AX_LO)}, + /* + * Temporary registers used by the JIT to operate on registers stored + * on the stack. Save t0 and t1 to be used as temporaries in generated + * code. + */ + [TMP_REG_1] = {RV_REG_T3, RV_REG_T2}, + [TMP_REG_2] = {RV_REG_T5, RV_REG_T4}, +}; + +static s8 hi(const s8 *r) +{ + return r[0]; +} + +static s8 lo(const s8 *r) +{ + return r[1]; +} + +static void emit_imm(const s8 rd, s32 imm, struct rv_jit_context *ctx) +{ + u32 upper = (imm + (1 << 11)) >> 12; + u32 lower = imm & 0xfff; + + if (upper) { + emit(rv_lui(rd, upper), ctx); + emit(rv_addi(rd, rd, lower), ctx); + } else { + emit(rv_addi(rd, RV_REG_ZERO, lower), ctx); + } +} + +static void emit_imm32(const s8 *rd, s32 imm, struct rv_jit_context *ctx) +{ + /* Emit immediate into lower bits. */ + emit_imm(lo(rd), imm, ctx); + + /* Sign-extend into upper bits. */ + if (imm >= 0) + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + else + emit(rv_addi(hi(rd), RV_REG_ZERO, -1), ctx); +} + +static void emit_imm64(const s8 *rd, s32 imm_hi, s32 imm_lo, + struct rv_jit_context *ctx) +{ + emit_imm(lo(rd), imm_lo, ctx); + emit_imm(hi(rd), imm_hi, ctx); +} + +static void __build_epilogue(bool is_tail_call, struct rv_jit_context *ctx) +{ + int stack_adjust = ctx->stack_size, store_offset = stack_adjust - 4; + const s8 *r0 = bpf2rv32[BPF_REG_0]; + + store_offset -= 4 * BPF_JIT_SCRATCH_REGS; + + /* Set return value if not tail call. */ + if (!is_tail_call) { + emit(rv_addi(RV_REG_A0, lo(r0), 0), ctx); + emit(rv_addi(RV_REG_A1, hi(r0), 0), ctx); + } + + /* Restore callee-saved registers. */ + emit(rv_lw(RV_REG_RA, store_offset - 0, RV_REG_SP), ctx); + emit(rv_lw(RV_REG_FP, store_offset - 4, RV_REG_SP), ctx); + emit(rv_lw(RV_REG_S1, store_offset - 8, RV_REG_SP), ctx); + emit(rv_lw(RV_REG_S2, store_offset - 12, RV_REG_SP), ctx); + emit(rv_lw(RV_REG_S3, store_offset - 16, RV_REG_SP), ctx); + emit(rv_lw(RV_REG_S4, store_offset - 20, RV_REG_SP), ctx); + emit(rv_lw(RV_REG_S5, store_offset - 24, RV_REG_SP), ctx); + emit(rv_lw(RV_REG_S6, store_offset - 28, RV_REG_SP), ctx); + emit(rv_lw(RV_REG_S7, store_offset - 32, RV_REG_SP), ctx); + + emit(rv_addi(RV_REG_SP, RV_REG_SP, stack_adjust), ctx); + + if (is_tail_call) { + /* + * goto *(t0 + 4); + * Skips first instruction of prologue which initializes tail + * call counter. Assumes t0 contains address of target program, + * see emit_bpf_tail_call. + */ + emit(rv_jalr(RV_REG_ZERO, RV_REG_T0, 4), ctx); + } else { + emit(rv_jalr(RV_REG_ZERO, RV_REG_RA, 0), ctx); + } +} + +static bool is_stacked(s8 reg) +{ + return reg < 0; +} + +static const s8 *bpf_get_reg64(const s8 *reg, const s8 *tmp, + struct rv_jit_context *ctx) +{ + if (is_stacked(hi(reg))) { + emit(rv_lw(hi(tmp), hi(reg), RV_REG_FP), ctx); + emit(rv_lw(lo(tmp), lo(reg), RV_REG_FP), ctx); + reg = tmp; + } + return reg; +} + +static void bpf_put_reg64(const s8 *reg, const s8 *src, + struct rv_jit_context *ctx) +{ + if (is_stacked(hi(reg))) { + emit(rv_sw(RV_REG_FP, hi(reg), hi(src)), ctx); + emit(rv_sw(RV_REG_FP, lo(reg), lo(src)), ctx); + } +} + +static const s8 *bpf_get_reg32(const s8 *reg, const s8 *tmp, + struct rv_jit_context *ctx) +{ + if (is_stacked(lo(reg))) { + emit(rv_lw(lo(tmp), lo(reg), RV_REG_FP), ctx); + reg = tmp; + } + return reg; +} + +static void bpf_put_reg32(const s8 *reg, const s8 *src, + struct rv_jit_context *ctx) +{ + if (is_stacked(lo(reg))) { + emit(rv_sw(RV_REG_FP, lo(reg), lo(src)), ctx); + if (!ctx->prog->aux->verifier_zext) + emit(rv_sw(RV_REG_FP, hi(reg), RV_REG_ZERO), ctx); + } else if (!ctx->prog->aux->verifier_zext) { + emit(rv_addi(hi(reg), RV_REG_ZERO, 0), ctx); + } +} + +static void emit_jump_and_link(u8 rd, s32 rvoff, bool force_jalr, + struct rv_jit_context *ctx) +{ + s32 upper, lower; + + if (rvoff && is_21b_int(rvoff) && !force_jalr) { + emit(rv_jal(rd, rvoff >> 1), ctx); + return; + } + + upper = (rvoff + (1 << 11)) >> 12; + lower = rvoff & 0xfff; + emit(rv_auipc(RV_REG_T1, upper), ctx); + emit(rv_jalr(rd, RV_REG_T1, lower), ctx); +} + +static void emit_alu_i64(const s8 *dst, s32 imm, + struct rv_jit_context *ctx, const u8 op) +{ + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + const s8 *rd = bpf_get_reg64(dst, tmp1, ctx); + + switch (op) { + case BPF_MOV: + emit_imm32(rd, imm, ctx); + break; + case BPF_AND: + if (is_12b_int(imm)) { + emit(rv_andi(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_and(lo(rd), lo(rd), RV_REG_T0), ctx); + } + if (imm >= 0) + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + break; + case BPF_OR: + if (is_12b_int(imm)) { + emit(rv_ori(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_or(lo(rd), lo(rd), RV_REG_T0), ctx); + } + if (imm < 0) + emit(rv_ori(hi(rd), RV_REG_ZERO, -1), ctx); + break; + case BPF_XOR: + if (is_12b_int(imm)) { + emit(rv_xori(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_xor(lo(rd), lo(rd), RV_REG_T0), ctx); + } + if (imm < 0) + emit(rv_xori(hi(rd), hi(rd), -1), ctx); + break; + case BPF_LSH: + if (imm >= 32) { + emit(rv_slli(hi(rd), lo(rd), imm - 32), ctx); + emit(rv_addi(lo(rd), RV_REG_ZERO, 0), ctx); + } else if (imm == 0) { + /* Do nothing. */ + } else { + emit(rv_srli(RV_REG_T0, lo(rd), 32 - imm), ctx); + emit(rv_slli(hi(rd), hi(rd), imm), ctx); + emit(rv_or(hi(rd), RV_REG_T0, hi(rd)), ctx); + emit(rv_slli(lo(rd), lo(rd), imm), ctx); + } + break; + case BPF_RSH: + if (imm >= 32) { + emit(rv_srli(lo(rd), hi(rd), imm - 32), ctx); + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + } else if (imm == 0) { + /* Do nothing. */ + } else { + emit(rv_slli(RV_REG_T0, hi(rd), 32 - imm), ctx); + emit(rv_srli(lo(rd), lo(rd), imm), ctx); + emit(rv_or(lo(rd), RV_REG_T0, lo(rd)), ctx); + emit(rv_srli(hi(rd), hi(rd), imm), ctx); + } + break; + case BPF_ARSH: + if (imm >= 32) { + emit(rv_srai(lo(rd), hi(rd), imm - 32), ctx); + emit(rv_srai(hi(rd), hi(rd), 31), ctx); + } else if (imm == 0) { + /* Do nothing. */ + } else { + emit(rv_slli(RV_REG_T0, hi(rd), 32 - imm), ctx); + emit(rv_srli(lo(rd), lo(rd), imm), ctx); + emit(rv_or(lo(rd), RV_REG_T0, lo(rd)), ctx); + emit(rv_srai(hi(rd), hi(rd), imm), ctx); + } + break; + } + + bpf_put_reg64(dst, rd, ctx); +} + +static void emit_alu_i32(const s8 *dst, s32 imm, + struct rv_jit_context *ctx, const u8 op) +{ + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + const s8 *rd = bpf_get_reg32(dst, tmp1, ctx); + + switch (op) { + case BPF_MOV: + emit_imm(lo(rd), imm, ctx); + break; + case BPF_ADD: + if (is_12b_int(imm)) { + emit(rv_addi(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_add(lo(rd), lo(rd), RV_REG_T0), ctx); + } + break; + case BPF_SUB: + if (is_12b_int(-imm)) { + emit(rv_addi(lo(rd), lo(rd), -imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_sub(lo(rd), lo(rd), RV_REG_T0), ctx); + } + break; + case BPF_AND: + if (is_12b_int(imm)) { + emit(rv_andi(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_and(lo(rd), lo(rd), RV_REG_T0), ctx); + } + break; + case BPF_OR: + if (is_12b_int(imm)) { + emit(rv_ori(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_or(lo(rd), lo(rd), RV_REG_T0), ctx); + } + break; + case BPF_XOR: + if (is_12b_int(imm)) { + emit(rv_xori(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_xor(lo(rd), lo(rd), RV_REG_T0), ctx); + } + break; + case BPF_LSH: + if (is_12b_int(imm)) { + emit(rv_slli(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_sll(lo(rd), lo(rd), RV_REG_T0), ctx); + } + break; + case BPF_RSH: + if (is_12b_int(imm)) { + emit(rv_srli(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_srl(lo(rd), lo(rd), RV_REG_T0), ctx); + } + break; + case BPF_ARSH: + if (is_12b_int(imm)) { + emit(rv_srai(lo(rd), lo(rd), imm), ctx); + } else { + emit_imm(RV_REG_T0, imm, ctx); + emit(rv_sra(lo(rd), lo(rd), RV_REG_T0), ctx); + } + break; + } + + bpf_put_reg32(dst, rd, ctx); +} + +static void emit_alu_r64(const s8 *dst, const s8 *src, + struct rv_jit_context *ctx, const u8 op) +{ + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + const s8 *tmp2 = bpf2rv32[TMP_REG_2]; + const s8 *rd = bpf_get_reg64(dst, tmp1, ctx); + const s8 *rs = bpf_get_reg64(src, tmp2, ctx); + + switch (op) { + case BPF_MOV: + emit(rv_addi(lo(rd), lo(rs), 0), ctx); + emit(rv_addi(hi(rd), hi(rs), 0), ctx); + break; + case BPF_ADD: + if (rd == rs) { + emit(rv_srli(RV_REG_T0, lo(rd), 31), ctx); + emit(rv_slli(hi(rd), hi(rd), 1), ctx); + emit(rv_or(hi(rd), RV_REG_T0, hi(rd)), ctx); + emit(rv_slli(lo(rd), lo(rd), 1), ctx); + } else { + emit(rv_add(lo(rd), lo(rd), lo(rs)), ctx); + emit(rv_sltu(RV_REG_T0, lo(rd), lo(rs)), ctx); + emit(rv_add(hi(rd), hi(rd), hi(rs)), ctx); + emit(rv_add(hi(rd), hi(rd), RV_REG_T0), ctx); + } + break; + case BPF_SUB: + emit(rv_sub(RV_REG_T1, hi(rd), hi(rs)), ctx); + emit(rv_sltu(RV_REG_T0, lo(rd), lo(rs)), ctx); + emit(rv_sub(hi(rd), RV_REG_T1, RV_REG_T0), ctx); + emit(rv_sub(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_AND: + emit(rv_and(lo(rd), lo(rd), lo(rs)), ctx); + emit(rv_and(hi(rd), hi(rd), hi(rs)), ctx); + break; + case BPF_OR: + emit(rv_or(lo(rd), lo(rd), lo(rs)), ctx); + emit(rv_or(hi(rd), hi(rd), hi(rs)), ctx); + break; + case BPF_XOR: + emit(rv_xor(lo(rd), lo(rd), lo(rs)), ctx); + emit(rv_xor(hi(rd), hi(rd), hi(rs)), ctx); + break; + case BPF_MUL: + emit(rv_mul(RV_REG_T0, hi(rs), lo(rd)), ctx); + emit(rv_mul(hi(rd), hi(rd), lo(rs)), ctx); + emit(rv_mulhu(RV_REG_T1, lo(rd), lo(rs)), ctx); + emit(rv_add(hi(rd), hi(rd), RV_REG_T0), ctx); + emit(rv_mul(lo(rd), lo(rd), lo(rs)), ctx); + emit(rv_add(hi(rd), hi(rd), RV_REG_T1), ctx); + break; + case BPF_LSH: + emit(rv_addi(RV_REG_T0, lo(rs), -32), ctx); + emit(rv_blt(RV_REG_T0, RV_REG_ZERO, 8), ctx); + emit(rv_sll(hi(rd), lo(rd), RV_REG_T0), ctx); + emit(rv_addi(lo(rd), RV_REG_ZERO, 0), ctx); + emit(rv_jal(RV_REG_ZERO, 16), ctx); + emit(rv_addi(RV_REG_T1, RV_REG_ZERO, 31), ctx); + emit(rv_srli(RV_REG_T0, lo(rd), 1), ctx); + emit(rv_sub(RV_REG_T1, RV_REG_T1, lo(rs)), ctx); + emit(rv_srl(RV_REG_T0, RV_REG_T0, RV_REG_T1), ctx); + emit(rv_sll(hi(rd), hi(rd), lo(rs)), ctx); + emit(rv_or(hi(rd), RV_REG_T0, hi(rd)), ctx); + emit(rv_sll(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_RSH: + emit(rv_addi(RV_REG_T0, lo(rs), -32), ctx); + emit(rv_blt(RV_REG_T0, RV_REG_ZERO, 8), ctx); + emit(rv_srl(lo(rd), hi(rd), RV_REG_T0), ctx); + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + emit(rv_jal(RV_REG_ZERO, 16), ctx); + emit(rv_addi(RV_REG_T1, RV_REG_ZERO, 31), ctx); + emit(rv_slli(RV_REG_T0, hi(rd), 1), ctx); + emit(rv_sub(RV_REG_T1, RV_REG_T1, lo(rs)), ctx); + emit(rv_sll(RV_REG_T0, RV_REG_T0, RV_REG_T1), ctx); + emit(rv_srl(lo(rd), lo(rd), lo(rs)), ctx); + emit(rv_or(lo(rd), RV_REG_T0, lo(rd)), ctx); + emit(rv_srl(hi(rd), hi(rd), lo(rs)), ctx); + break; + case BPF_ARSH: + emit(rv_addi(RV_REG_T0, lo(rs), -32), ctx); + emit(rv_blt(RV_REG_T0, RV_REG_ZERO, 8), ctx); + emit(rv_sra(lo(rd), hi(rd), RV_REG_T0), ctx); + emit(rv_srai(hi(rd), hi(rd), 31), ctx); + emit(rv_jal(RV_REG_ZERO, 16), ctx); + emit(rv_addi(RV_REG_T1, RV_REG_ZERO, 31), ctx); + emit(rv_slli(RV_REG_T0, hi(rd), 1), ctx); + emit(rv_sub(RV_REG_T1, RV_REG_T1, lo(rs)), ctx); + emit(rv_sll(RV_REG_T0, RV_REG_T0, RV_REG_T1), ctx); + emit(rv_srl(lo(rd), lo(rd), lo(rs)), ctx); + emit(rv_or(lo(rd), RV_REG_T0, lo(rd)), ctx); + emit(rv_sra(hi(rd), hi(rd), lo(rs)), ctx); + break; + case BPF_NEG: + emit(rv_sub(lo(rd), RV_REG_ZERO, lo(rd)), ctx); + emit(rv_sltu(RV_REG_T0, RV_REG_ZERO, lo(rd)), ctx); + emit(rv_sub(hi(rd), RV_REG_ZERO, hi(rd)), ctx); + emit(rv_sub(hi(rd), hi(rd), RV_REG_T0), ctx); + break; + } + + bpf_put_reg64(dst, rd, ctx); +} + +static void emit_alu_r32(const s8 *dst, const s8 *src, + struct rv_jit_context *ctx, const u8 op) +{ + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + const s8 *tmp2 = bpf2rv32[TMP_REG_2]; + const s8 *rd = bpf_get_reg32(dst, tmp1, ctx); + const s8 *rs = bpf_get_reg32(src, tmp2, ctx); + + switch (op) { + case BPF_MOV: + emit(rv_addi(lo(rd), lo(rs), 0), ctx); + break; + case BPF_ADD: + emit(rv_add(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_SUB: + emit(rv_sub(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_AND: + emit(rv_and(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_OR: + emit(rv_or(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_XOR: + emit(rv_xor(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_MUL: + emit(rv_mul(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_DIV: + emit(rv_divu(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_MOD: + emit(rv_remu(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_LSH: + emit(rv_sll(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_RSH: + emit(rv_srl(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_ARSH: + emit(rv_sra(lo(rd), lo(rd), lo(rs)), ctx); + break; + case BPF_NEG: + emit(rv_sub(lo(rd), RV_REG_ZERO, lo(rd)), ctx); + break; + } + + bpf_put_reg32(dst, rd, ctx); +} + +static int emit_branch_r64(const s8 *src1, const s8 *src2, s32 rvoff, + struct rv_jit_context *ctx, const u8 op) +{ + int e, s = ctx->ninsns; + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + const s8 *tmp2 = bpf2rv32[TMP_REG_2]; + + const s8 *rs1 = bpf_get_reg64(src1, tmp1, ctx); + const s8 *rs2 = bpf_get_reg64(src2, tmp2, ctx); + + /* + * NO_JUMP skips over the rest of the instructions and the + * emit_jump_and_link, meaning the BPF branch is not taken. + * JUMP skips directly to the emit_jump_and_link, meaning + * the BPF branch is taken. + * + * The fallthrough case results in the BPF branch being taken. + */ +#define NO_JUMP(idx) (6 + (2 * (idx))) +#define JUMP(idx) (2 + (2 * (idx))) + + switch (op) { + case BPF_JEQ: + emit(rv_bne(hi(rs1), hi(rs2), NO_JUMP(1)), ctx); + emit(rv_bne(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JGT: + emit(rv_bgtu(hi(rs1), hi(rs2), JUMP(2)), ctx); + emit(rv_bltu(hi(rs1), hi(rs2), NO_JUMP(1)), ctx); + emit(rv_bleu(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JLT: + emit(rv_bltu(hi(rs1), hi(rs2), JUMP(2)), ctx); + emit(rv_bgtu(hi(rs1), hi(rs2), NO_JUMP(1)), ctx); + emit(rv_bgeu(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JGE: + emit(rv_bgtu(hi(rs1), hi(rs2), JUMP(2)), ctx); + emit(rv_bltu(hi(rs1), hi(rs2), NO_JUMP(1)), ctx); + emit(rv_bltu(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JLE: + emit(rv_bltu(hi(rs1), hi(rs2), JUMP(2)), ctx); + emit(rv_bgtu(hi(rs1), hi(rs2), NO_JUMP(1)), ctx); + emit(rv_bgtu(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JNE: + emit(rv_bne(hi(rs1), hi(rs2), JUMP(1)), ctx); + emit(rv_beq(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JSGT: + emit(rv_bgt(hi(rs1), hi(rs2), JUMP(2)), ctx); + emit(rv_blt(hi(rs1), hi(rs2), NO_JUMP(1)), ctx); + emit(rv_bleu(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JSLT: + emit(rv_blt(hi(rs1), hi(rs2), JUMP(2)), ctx); + emit(rv_bgt(hi(rs1), hi(rs2), NO_JUMP(1)), ctx); + emit(rv_bgeu(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JSGE: + emit(rv_bgt(hi(rs1), hi(rs2), JUMP(2)), ctx); + emit(rv_blt(hi(rs1), hi(rs2), NO_JUMP(1)), ctx); + emit(rv_bltu(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JSLE: + emit(rv_blt(hi(rs1), hi(rs2), JUMP(2)), ctx); + emit(rv_bgt(hi(rs1), hi(rs2), NO_JUMP(1)), ctx); + emit(rv_bgtu(lo(rs1), lo(rs2), NO_JUMP(0)), ctx); + break; + case BPF_JSET: + emit(rv_and(RV_REG_T0, hi(rs1), hi(rs2)), ctx); + emit(rv_bne(RV_REG_T0, RV_REG_ZERO, JUMP(2)), ctx); + emit(rv_and(RV_REG_T0, lo(rs1), lo(rs2)), ctx); + emit(rv_beq(RV_REG_T0, RV_REG_ZERO, NO_JUMP(0)), ctx); + break; + } + +#undef NO_JUMP +#undef JUMP + + e = ctx->ninsns; + /* Adjust for extra insns. */ + rvoff -= (e - s) << 2; + emit_jump_and_link(RV_REG_ZERO, rvoff, true, ctx); + return 0; +} + +static int emit_bcc(u8 op, u8 rd, u8 rs, int rvoff, struct rv_jit_context *ctx) +{ + int e, s = ctx->ninsns; + bool far = false; + int off; + + if (op == BPF_JSET) { + /* + * BPF_JSET is a special case: it has no inverse so we always + * treat it as a far branch. + */ + far = true; + } else if (!is_13b_int(rvoff)) { + op = invert_bpf_cond(op); + far = true; + } + + /* + * For a far branch, the condition is negated and we jump over the + * branch itself, and the two instructions from emit_jump_and_link. + * For a near branch, just use rvoff. + */ + off = far ? 6 : (rvoff >> 1); + + switch (op) { + case BPF_JEQ: + emit(rv_beq(rd, rs, off), ctx); + break; + case BPF_JGT: + emit(rv_bgtu(rd, rs, off), ctx); + break; + case BPF_JLT: + emit(rv_bltu(rd, rs, off), ctx); + break; + case BPF_JGE: + emit(rv_bgeu(rd, rs, off), ctx); + break; + case BPF_JLE: + emit(rv_bleu(rd, rs, off), ctx); + break; + case BPF_JNE: + emit(rv_bne(rd, rs, off), ctx); + break; + case BPF_JSGT: + emit(rv_bgt(rd, rs, off), ctx); + break; + case BPF_JSLT: + emit(rv_blt(rd, rs, off), ctx); + break; + case BPF_JSGE: + emit(rv_bge(rd, rs, off), ctx); + break; + case BPF_JSLE: + emit(rv_ble(rd, rs, off), ctx); + break; + case BPF_JSET: + emit(rv_and(RV_REG_T0, rd, rs), ctx); + emit(rv_beq(RV_REG_T0, RV_REG_ZERO, off), ctx); + break; + } + + if (far) { + e = ctx->ninsns; + /* Adjust for extra insns. */ + rvoff -= (e - s) << 2; + emit_jump_and_link(RV_REG_ZERO, rvoff, true, ctx); + } + return 0; +} + +static int emit_branch_r32(const s8 *src1, const s8 *src2, s32 rvoff, + struct rv_jit_context *ctx, const u8 op) +{ + int e, s = ctx->ninsns; + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + const s8 *tmp2 = bpf2rv32[TMP_REG_2]; + + const s8 *rs1 = bpf_get_reg32(src1, tmp1, ctx); + const s8 *rs2 = bpf_get_reg32(src2, tmp2, ctx); + + e = ctx->ninsns; + /* Adjust for extra insns. */ + rvoff -= (e - s) << 2; + + if (emit_bcc(op, lo(rs1), lo(rs2), rvoff, ctx)) + return -1; + + return 0; +} + +static void emit_call(bool fixed, u64 addr, struct rv_jit_context *ctx) +{ + const s8 *r0 = bpf2rv32[BPF_REG_0]; + const s8 *r5 = bpf2rv32[BPF_REG_5]; + u32 upper = ((u32)addr + (1 << 11)) >> 12; + u32 lower = addr & 0xfff; + + /* R1-R4 already in correct registers---need to push R5 to stack. */ + emit(rv_addi(RV_REG_SP, RV_REG_SP, -16), ctx); + emit(rv_sw(RV_REG_SP, 0, lo(r5)), ctx); + emit(rv_sw(RV_REG_SP, 4, hi(r5)), ctx); + + /* Backup TCC. */ + emit(rv_addi(RV_REG_TCC_SAVED, RV_REG_TCC, 0), ctx); + + /* + * Use lui/jalr pair to jump to absolute address. Don't use emit_imm as + * the number of emitted instructions should not depend on the value of + * addr. + */ + emit(rv_lui(RV_REG_T1, upper), ctx); + emit(rv_jalr(RV_REG_RA, RV_REG_T1, lower), ctx); + + /* Restore TCC. */ + emit(rv_addi(RV_REG_TCC, RV_REG_TCC_SAVED, 0), ctx); + + /* Set return value and restore stack. */ + emit(rv_addi(lo(r0), RV_REG_A0, 0), ctx); + emit(rv_addi(hi(r0), RV_REG_A1, 0), ctx); + emit(rv_addi(RV_REG_SP, RV_REG_SP, 16), ctx); +} + +static int emit_bpf_tail_call(int insn, struct rv_jit_context *ctx) +{ + /* + * R1 -> &ctx + * R2 -> &array + * R3 -> index + */ + int tc_ninsn, off, start_insn = ctx->ninsns; + const s8 *arr_reg = bpf2rv32[BPF_REG_2]; + const s8 *idx_reg = bpf2rv32[BPF_REG_3]; + + tc_ninsn = insn ? ctx->offset[insn] - ctx->offset[insn - 1] : + ctx->offset[0]; + + /* max_entries = array->map.max_entries; */ + off = offsetof(struct bpf_array, map.max_entries); + if (is_12b_check(off, insn)) + return -1; + emit(rv_lw(RV_REG_T1, off, lo(arr_reg)), ctx); + + /* + * if (index >= max_entries) + * goto out; + */ + off = (tc_ninsn - (ctx->ninsns - start_insn)) << 2; + emit_bcc(BPF_JGE, lo(idx_reg), RV_REG_T1, off, ctx); + + /* + * if ((temp_tcc = tcc - 1) < 0) + * goto out; + */ + emit(rv_addi(RV_REG_T1, RV_REG_TCC, -1), ctx); + off = (tc_ninsn - (ctx->ninsns - start_insn)) << 2; + emit_bcc(BPF_JSLT, RV_REG_T1, RV_REG_ZERO, off, ctx); + + /* + * prog = array->ptrs[index]; + * if (!prog) + * goto out; + */ + emit(rv_slli(RV_REG_T0, lo(idx_reg), 2), ctx); + emit(rv_add(RV_REG_T0, RV_REG_T0, lo(arr_reg)), ctx); + off = offsetof(struct bpf_array, ptrs); + if (is_12b_check(off, insn)) + return -1; + emit(rv_lw(RV_REG_T0, off, RV_REG_T0), ctx); + off = (tc_ninsn - (ctx->ninsns - start_insn)) << 2; + emit_bcc(BPF_JEQ, RV_REG_T0, RV_REG_ZERO, off, ctx); + + /* + * tcc = temp_tcc; + * goto *(prog->bpf_func + 4); + */ + off = offsetof(struct bpf_prog, bpf_func); + if (is_12b_check(off, insn)) + return -1; + emit(rv_lw(RV_REG_T0, off, RV_REG_T0), ctx); + emit(rv_addi(RV_REG_TCC, RV_REG_T1, 0), ctx); + /* Epilogue jumps to *(t0 + 4). */ + __build_epilogue(true, ctx); + return 0; +} + +static int emit_load_r64(const s8 *dst, const s8 *src, s16 off, + struct rv_jit_context *ctx, const u8 size) +{ + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + const s8 *tmp2 = bpf2rv32[TMP_REG_2]; + const s8 *rd = bpf_get_reg64(dst, tmp1, ctx); + const s8 *rs = bpf_get_reg64(src, tmp2, ctx); + + emit_imm(RV_REG_T0, off, ctx); + emit(rv_add(RV_REG_T0, RV_REG_T0, lo(rs)), ctx); + + switch (size) { + case BPF_B: + emit(rv_lbu(lo(rd), 0, RV_REG_T0), ctx); + if (!ctx->prog->aux->verifier_zext) + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + break; + case BPF_H: + emit(rv_lhu(lo(rd), 0, RV_REG_T0), ctx); + if (!ctx->prog->aux->verifier_zext) + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + break; + case BPF_W: + emit(rv_lw(lo(rd), 0, RV_REG_T0), ctx); + if (!ctx->prog->aux->verifier_zext) + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + break; + case BPF_DW: + emit(rv_lw(lo(rd), 0, RV_REG_T0), ctx); + emit(rv_lw(hi(rd), 4, RV_REG_T0), ctx); + break; + } + + bpf_put_reg64(dst, rd, ctx); + return 0; +} + +static int emit_store_r64(const s8 *dst, const s8 *src, s16 off, + struct rv_jit_context *ctx, const u8 size, + const u8 mode) +{ + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + const s8 *tmp2 = bpf2rv32[TMP_REG_2]; + const s8 *rd = bpf_get_reg64(dst, tmp1, ctx); + const s8 *rs = bpf_get_reg64(src, tmp2, ctx); + + if (mode == BPF_XADD && size != BPF_W) + return -1; + + emit_imm(RV_REG_T0, off, ctx); + emit(rv_add(RV_REG_T0, RV_REG_T0, lo(rd)), ctx); + + switch (size) { + case BPF_B: + emit(rv_sb(RV_REG_T0, 0, lo(rs)), ctx); + break; + case BPF_H: + emit(rv_sh(RV_REG_T0, 0, lo(rs)), ctx); + break; + case BPF_W: + switch (mode) { + case BPF_MEM: + emit(rv_sw(RV_REG_T0, 0, lo(rs)), ctx); + break; + case BPF_XADD: + emit(rv_amoadd_w(RV_REG_ZERO, lo(rs), RV_REG_T0, 0, 0), + ctx); + break; + } + break; + case BPF_DW: + emit(rv_sw(RV_REG_T0, 0, lo(rs)), ctx); + emit(rv_sw(RV_REG_T0, 4, hi(rs)), ctx); + break; + } + + return 0; +} + +static void emit_rev16(const s8 rd, struct rv_jit_context *ctx) +{ + emit(rv_slli(rd, rd, 16), ctx); + emit(rv_slli(RV_REG_T1, rd, 8), ctx); + emit(rv_srli(rd, rd, 8), ctx); + emit(rv_add(RV_REG_T1, rd, RV_REG_T1), ctx); + emit(rv_srli(rd, RV_REG_T1, 16), ctx); +} + +static void emit_rev32(const s8 rd, struct rv_jit_context *ctx) +{ + emit(rv_addi(RV_REG_T1, RV_REG_ZERO, 0), ctx); + emit(rv_andi(RV_REG_T0, rd, 255), ctx); + emit(rv_add(RV_REG_T1, RV_REG_T1, RV_REG_T0), ctx); + emit(rv_slli(RV_REG_T1, RV_REG_T1, 8), ctx); + emit(rv_srli(rd, rd, 8), ctx); + emit(rv_andi(RV_REG_T0, rd, 255), ctx); + emit(rv_add(RV_REG_T1, RV_REG_T1, RV_REG_T0), ctx); + emit(rv_slli(RV_REG_T1, RV_REG_T1, 8), ctx); + emit(rv_srli(rd, rd, 8), ctx); + emit(rv_andi(RV_REG_T0, rd, 255), ctx); + emit(rv_add(RV_REG_T1, RV_REG_T1, RV_REG_T0), ctx); + emit(rv_slli(RV_REG_T1, RV_REG_T1, 8), ctx); + emit(rv_srli(rd, rd, 8), ctx); + emit(rv_andi(RV_REG_T0, rd, 255), ctx); + emit(rv_add(RV_REG_T1, RV_REG_T1, RV_REG_T0), ctx); + emit(rv_addi(rd, RV_REG_T1, 0), ctx); +} + +static void emit_zext64(const s8 *dst, struct rv_jit_context *ctx) +{ + const s8 *rd; + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + + rd = bpf_get_reg64(dst, tmp1, ctx); + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + bpf_put_reg64(dst, rd, ctx); +} + +int bpf_jit_emit_insn(const struct bpf_insn *insn, struct rv_jit_context *ctx, + bool extra_pass) +{ + bool is64 = BPF_CLASS(insn->code) == BPF_ALU64 || + BPF_CLASS(insn->code) == BPF_JMP; + int s, e, rvoff, i = insn - ctx->prog->insnsi; + u8 code = insn->code; + s16 off = insn->off; + s32 imm = insn->imm; + + const s8 *dst = bpf2rv32[insn->dst_reg]; + const s8 *src = bpf2rv32[insn->src_reg]; + const s8 *tmp1 = bpf2rv32[TMP_REG_1]; + const s8 *tmp2 = bpf2rv32[TMP_REG_2]; + + switch (code) { + case BPF_ALU64 | BPF_MOV | BPF_X: + + case BPF_ALU64 | BPF_ADD | BPF_X: + case BPF_ALU64 | BPF_ADD | BPF_K: + + case BPF_ALU64 | BPF_SUB | BPF_X: + case BPF_ALU64 | BPF_SUB | BPF_K: + + case BPF_ALU64 | BPF_AND | BPF_X: + case BPF_ALU64 | BPF_OR | BPF_X: + case BPF_ALU64 | BPF_XOR | BPF_X: + + case BPF_ALU64 | BPF_MUL | BPF_X: + case BPF_ALU64 | BPF_MUL | BPF_K: + + case BPF_ALU64 | BPF_LSH | BPF_X: + case BPF_ALU64 | BPF_RSH | BPF_X: + case BPF_ALU64 | BPF_ARSH | BPF_X: + if (BPF_SRC(code) == BPF_K) { + emit_imm32(tmp2, imm, ctx); + src = tmp2; + } + emit_alu_r64(dst, src, ctx, BPF_OP(code)); + break; + + case BPF_ALU64 | BPF_NEG: + emit_alu_r64(dst, tmp2, ctx, BPF_OP(code)); + break; + + case BPF_ALU64 | BPF_DIV | BPF_X: + case BPF_ALU64 | BPF_DIV | BPF_K: + case BPF_ALU64 | BPF_MOD | BPF_X: + case BPF_ALU64 | BPF_MOD | BPF_K: + goto notsupported; + + case BPF_ALU64 | BPF_MOV | BPF_K: + case BPF_ALU64 | BPF_AND | BPF_K: + case BPF_ALU64 | BPF_OR | BPF_K: + case BPF_ALU64 | BPF_XOR | BPF_K: + case BPF_ALU64 | BPF_LSH | BPF_K: + case BPF_ALU64 | BPF_RSH | BPF_K: + case BPF_ALU64 | BPF_ARSH | BPF_K: + emit_alu_i64(dst, imm, ctx, BPF_OP(code)); + break; + + case BPF_ALU | BPF_MOV | BPF_X: + if (imm == 1) { + /* Special mov32 for zext. */ + emit_zext64(dst, ctx); + break; + } + /* Fallthrough. */ + + case BPF_ALU | BPF_ADD | BPF_X: + case BPF_ALU | BPF_SUB | BPF_X: + case BPF_ALU | BPF_AND | BPF_X: + case BPF_ALU | BPF_OR | BPF_X: + case BPF_ALU | BPF_XOR | BPF_X: + + case BPF_ALU | BPF_MUL | BPF_X: + case BPF_ALU | BPF_MUL | BPF_K: + + case BPF_ALU | BPF_DIV | BPF_X: + case BPF_ALU | BPF_DIV | BPF_K: + + case BPF_ALU | BPF_MOD | BPF_X: + case BPF_ALU | BPF_MOD | BPF_K: + + case BPF_ALU | BPF_LSH | BPF_X: + case BPF_ALU | BPF_RSH | BPF_X: + case BPF_ALU | BPF_ARSH | BPF_X: + if (BPF_SRC(code) == BPF_K) { + emit_imm32(tmp2, imm, ctx); + src = tmp2; + } + emit_alu_r32(dst, src, ctx, BPF_OP(code)); + break; + + case BPF_ALU | BPF_MOV | BPF_K: + case BPF_ALU | BPF_ADD | BPF_K: + case BPF_ALU | BPF_SUB | BPF_K: + case BPF_ALU | BPF_AND | BPF_K: + case BPF_ALU | BPF_OR | BPF_K: + case BPF_ALU | BPF_XOR | BPF_K: + case BPF_ALU | BPF_LSH | BPF_K: + case BPF_ALU | BPF_RSH | BPF_K: + case BPF_ALU | BPF_ARSH | BPF_K: + /* + * mul,div,mod are handled in the BPF_X case since there are + * no RISC-V I-type equivalents. + */ + emit_alu_i32(dst, imm, ctx, BPF_OP(code)); + break; + + case BPF_ALU | BPF_NEG: + /* + * src is ignored---choose tmp2 as a dummy register since it + * is not on the stack. + */ + emit_alu_r32(dst, tmp2, ctx, BPF_OP(code)); + break; + + case BPF_ALU | BPF_END | BPF_FROM_LE: + { + const s8 *rd = bpf_get_reg64(dst, tmp1, ctx); + + switch (imm) { + case 16: + emit(rv_slli(lo(rd), lo(rd), 16), ctx); + emit(rv_srli(lo(rd), lo(rd), 16), ctx); + /* Fallthrough. */ + case 32: + if (!ctx->prog->aux->verifier_zext) + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + break; + case 64: + /* Do nothing. */ + break; + default: + pr_err("bpf-jit: BPF_END imm %d invalid\n", imm); + return -1; + } + + bpf_put_reg64(dst, rd, ctx); + break; + } + + case BPF_ALU | BPF_END | BPF_FROM_BE: + { + const s8 *rd = bpf_get_reg64(dst, tmp1, ctx); + + switch (imm) { + case 16: + emit_rev16(lo(rd), ctx); + if (!ctx->prog->aux->verifier_zext) + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + break; + case 32: + emit_rev32(lo(rd), ctx); + if (!ctx->prog->aux->verifier_zext) + emit(rv_addi(hi(rd), RV_REG_ZERO, 0), ctx); + break; + case 64: + /* Swap upper and lower halves. */ + emit(rv_addi(RV_REG_T0, lo(rd), 0), ctx); + emit(rv_addi(lo(rd), hi(rd), 0), ctx); + emit(rv_addi(hi(rd), RV_REG_T0, 0), ctx); + + /* Swap each half. */ + emit_rev32(lo(rd), ctx); + emit_rev32(hi(rd), ctx); + break; + default: + pr_err("bpf-jit: BPF_END imm %d invalid\n", imm); + return -1; + } + + bpf_put_reg64(dst, rd, ctx); + break; + } + + case BPF_JMP | BPF_JA: + rvoff = rv_offset(i, off, ctx); + emit_jump_and_link(RV_REG_ZERO, rvoff, false, ctx); + break; + + case BPF_JMP | BPF_CALL: + { + bool fixed; + int ret; + u64 addr; + + ret = bpf_jit_get_func_addr(ctx->prog, insn, extra_pass, &addr, + &fixed); + if (ret < 0) + return ret; + emit_call(fixed, addr, ctx); + break; + } + + case BPF_JMP | BPF_TAIL_CALL: + if (emit_bpf_tail_call(i, ctx)) + return -1; + break; + + case BPF_JMP | BPF_JEQ | BPF_X: + case BPF_JMP | BPF_JEQ | BPF_K: + case BPF_JMP32 | BPF_JEQ | BPF_X: + case BPF_JMP32 | BPF_JEQ | BPF_K: + + case BPF_JMP | BPF_JNE | BPF_X: + case BPF_JMP | BPF_JNE | BPF_K: + case BPF_JMP32 | BPF_JNE | BPF_X: + case BPF_JMP32 | BPF_JNE | BPF_K: + + case BPF_JMP | BPF_JLE | BPF_X: + case BPF_JMP | BPF_JLE | BPF_K: + case BPF_JMP32 | BPF_JLE | BPF_X: + case BPF_JMP32 | BPF_JLE | BPF_K: + + case BPF_JMP | BPF_JLT | BPF_X: + case BPF_JMP | BPF_JLT | BPF_K: + case BPF_JMP32 | BPF_JLT | BPF_X: + case BPF_JMP32 | BPF_JLT | BPF_K: + + case BPF_JMP | BPF_JGE | BPF_X: + case BPF_JMP | BPF_JGE | BPF_K: + case BPF_JMP32 | BPF_JGE | BPF_X: + case BPF_JMP32 | BPF_JGE | BPF_K: + + case BPF_JMP | BPF_JGT | BPF_X: + case BPF_JMP | BPF_JGT | BPF_K: + case BPF_JMP32 | BPF_JGT | BPF_X: + case BPF_JMP32 | BPF_JGT | BPF_K: + + case BPF_JMP | BPF_JSLE | BPF_X: + case BPF_JMP | BPF_JSLE | BPF_K: + case BPF_JMP32 | BPF_JSLE | BPF_X: + case BPF_JMP32 | BPF_JSLE | BPF_K: + + case BPF_JMP | BPF_JSLT | BPF_X: + case BPF_JMP | BPF_JSLT | BPF_K: + case BPF_JMP32 | BPF_JSLT | BPF_X: + case BPF_JMP32 | BPF_JSLT | BPF_K: + + case BPF_JMP | BPF_JSGE | BPF_X: + case BPF_JMP | BPF_JSGE | BPF_K: + case BPF_JMP32 | BPF_JSGE | BPF_X: + case BPF_JMP32 | BPF_JSGE | BPF_K: + + case BPF_JMP | BPF_JSGT | BPF_X: + case BPF_JMP | BPF_JSGT | BPF_K: + case BPF_JMP32 | BPF_JSGT | BPF_X: + case BPF_JMP32 | BPF_JSGT | BPF_K: + + case BPF_JMP | BPF_JSET | BPF_X: + case BPF_JMP | BPF_JSET | BPF_K: + case BPF_JMP32 | BPF_JSET | BPF_X: + case BPF_JMP32 | BPF_JSET | BPF_K: + rvoff = rv_offset(i, off, ctx); + if (BPF_SRC(code) == BPF_K) { + s = ctx->ninsns; + emit_imm32(tmp2, imm, ctx); + src = tmp2; + e = ctx->ninsns; + rvoff -= (e - s) << 2; + } + + if (is64) + emit_branch_r64(dst, src, rvoff, ctx, BPF_OP(code)); + else + emit_branch_r32(dst, src, rvoff, ctx, BPF_OP(code)); + break; + + case BPF_JMP | BPF_EXIT: + if (i == ctx->prog->len - 1) + break; + + rvoff = epilogue_offset(ctx); + emit_jump_and_link(RV_REG_ZERO, rvoff, false, ctx); + break; + + case BPF_LD | BPF_IMM | BPF_DW: + { + struct bpf_insn insn1 = insn[1]; + s32 imm_lo = imm; + s32 imm_hi = insn1.imm; + const s8 *rd = bpf_get_reg64(dst, tmp1, ctx); + + emit_imm64(rd, imm_hi, imm_lo, ctx); + bpf_put_reg64(dst, rd, ctx); + return 1; + } + + case BPF_LDX | BPF_MEM | BPF_B: + case BPF_LDX | BPF_MEM | BPF_H: + case BPF_LDX | BPF_MEM | BPF_W: + case BPF_LDX | BPF_MEM | BPF_DW: + if (emit_load_r64(dst, src, off, ctx, BPF_SIZE(code))) + return -1; + break; + + case BPF_ST | BPF_MEM | BPF_B: + case BPF_ST | BPF_MEM | BPF_H: + case BPF_ST | BPF_MEM | BPF_W: + case BPF_ST | BPF_MEM | BPF_DW: + + case BPF_STX | BPF_MEM | BPF_B: + case BPF_STX | BPF_MEM | BPF_H: + case BPF_STX | BPF_MEM | BPF_W: + case BPF_STX | BPF_MEM | BPF_DW: + case BPF_STX | BPF_XADD | BPF_W: + if (BPF_CLASS(code) == BPF_ST) { + emit_imm32(tmp2, imm, ctx); + src = tmp2; + } + + if (emit_store_r64(dst, src, off, ctx, BPF_SIZE(code), + BPF_MODE(code))) + return -1; + break; + + /* No hardware support for 8-byte atomics in RV32. */ + case BPF_STX | BPF_XADD | BPF_DW: + /* Fallthrough. */ + +notsupported: + pr_info_once("bpf-jit: not supported: opcode %02x ***\n", code); + return -EFAULT; + + default: + pr_err("bpf-jit: unknown opcode %02x\n", code); + return -EINVAL; + } + + return 0; +} + +void bpf_jit_build_prologue(struct rv_jit_context *ctx) +{ + /* Make space to save 9 registers: ra, fp, s1--s7. */ + int stack_adjust = 9 * sizeof(u32), store_offset, bpf_stack_adjust; + const s8 *fp = bpf2rv32[BPF_REG_FP]; + const s8 *r1 = bpf2rv32[BPF_REG_1]; + + bpf_stack_adjust = round_up(ctx->prog->aux->stack_depth, 16); + stack_adjust += bpf_stack_adjust; + + store_offset = stack_adjust - 4; + + stack_adjust += 4 * BPF_JIT_SCRATCH_REGS; + + /* + * The first instruction sets the tail-call-counter (TCC) register. + * This instruction is skipped by tail calls. + */ + emit(rv_addi(RV_REG_TCC, RV_REG_ZERO, MAX_TAIL_CALL_CNT), ctx); + + emit(rv_addi(RV_REG_SP, RV_REG_SP, -stack_adjust), ctx); + + /* Save callee-save registers. */ + emit(rv_sw(RV_REG_SP, store_offset - 0, RV_REG_RA), ctx); + emit(rv_sw(RV_REG_SP, store_offset - 4, RV_REG_FP), ctx); + emit(rv_sw(RV_REG_SP, store_offset - 8, RV_REG_S1), ctx); + emit(rv_sw(RV_REG_SP, store_offset - 12, RV_REG_S2), ctx); + emit(rv_sw(RV_REG_SP, store_offset - 16, RV_REG_S3), ctx); + emit(rv_sw(RV_REG_SP, store_offset - 20, RV_REG_S4), ctx); + emit(rv_sw(RV_REG_SP, store_offset - 24, RV_REG_S5), ctx); + emit(rv_sw(RV_REG_SP, store_offset - 28, RV_REG_S6), ctx); + emit(rv_sw(RV_REG_SP, store_offset - 32, RV_REG_S7), ctx); + + /* Set fp: used as the base address for stacked BPF registers. */ + emit(rv_addi(RV_REG_FP, RV_REG_SP, stack_adjust), ctx); + + /* Set up BPF stack pointer. */ + emit(rv_addi(lo(fp), RV_REG_SP, bpf_stack_adjust), ctx); + emit(rv_addi(hi(fp), RV_REG_ZERO, 0), ctx); + + /* Set up context pointer. */ + emit(rv_addi(lo(r1), RV_REG_A0, 0), ctx); + emit(rv_addi(hi(r1), RV_REG_ZERO, 0), ctx); + + ctx->stack_size = stack_adjust; +} + +void bpf_jit_build_epilogue(struct rv_jit_context *ctx) +{ + __build_epilogue(false, ctx); +} From patchwork Thu Mar 5 05:02:06 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Luke Nelson X-Patchwork-Id: 11421195 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 02E381580 for ; Thu, 5 Mar 2020 05:02:33 +0000 (UTC) Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id D274E21556 for ; Thu, 5 Mar 2020 05:02:32 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=lists.infradead.org header.i=@lists.infradead.org header.b="AaK6oRlX"; dkim=fail reason="signature verification failed" (1024-bit key) header.d=cs.washington.edu header.i=@cs.washington.edu header.b="BAwLygGc" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org D274E21556 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=cs.washington.edu Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-riscv-bounces+patchwork-linux-riscv=patchwork.kernel.org@lists.infradead.org DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20170209; h=Sender:Cc:List-Subscribe: List-Help:List-Post:List-Archive:List-Unsubscribe:List-Id: Content-Transfer-Encoding:MIME-Version:References:In-Reply-To:Message-Id:Date :Subject:To:From:Reply-To:Content-Type:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=R4y1bjuDihHx2oTrIgnH3W8hlU6GthTh0USQXogLFnA=; b=AaK6oRlXsxJjRO jO1ypHFL0TR8UJmTuLGQ0A2quX1DyKEsGq1xtnp11jDGYnJ8am5fqPfuVRf4rtzEMtK8bpwvLoigX qkbcfhTbshUD/GhKh3TwxvC1xuPxNZgwiHYx6Htlx08FvPb37HB4lZHdU9CTC2gC/JQf8x2Zdg5II 04NMxyNBqtpnaU0cwd1b76FNfOAtxZ+wx+sxKc2nt0+bVpyQmIKWJRXOTUp/szB8PD/ZFuK0B9qxk 4AawMT0hy8VyVEXop/rRIoHlTmfuBLWweCACu9KQvsGxEOjSdrrz4hxgmkoC/ufEqrpW4p//B/xoq tPOonkl+vVzIHyivIacw==; Received: from localhost ([127.0.0.1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.92.3 #3 (Red Hat Linux)) id 1j9ieR-0004YS-TA; Thu, 05 Mar 2020 05:02:27 +0000 Received: from mail-pf1-x441.google.com ([2607:f8b0:4864:20::441]) by bombadil.infradead.org with esmtps (Exim 4.92.3 #3 (Red Hat Linux)) id 1j9ieM-0004Rn-38 for linux-riscv@lists.infradead.org; Thu, 05 Mar 2020 05:02:23 +0000 Received: by mail-pf1-x441.google.com with SMTP id q9so2155838pfs.5 for ; Wed, 04 Mar 2020 21:02:21 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=cs.washington.edu; s=goo201206; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=R4y1bjuDihHx2oTrIgnH3W8hlU6GthTh0USQXogLFnA=; b=BAwLygGc+cbMJxseyd+EFVvY8q7V30mqYzVTTSkxSv6SEDlPP17rNHZjoHl0VR6psd UjgL8sjesWeib2YlBcV2uLrZdJaV/cOZh2X4xhbs5JlKkp6XmFt+N88DnDe106GJX2zp l+MTCHbtbt8onH+AZd0ufLRY/z2vOKBZswpck= X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=R4y1bjuDihHx2oTrIgnH3W8hlU6GthTh0USQXogLFnA=; b=Ja2sJWhXNFTfy7ClVIFHR5knZAZTfYwXcp6MvclVnBQoyeH/zEKHGwVK6xeyi1W3Os iTLcIMUdoG5cUrsRttZHohOIBhHcwJP14YoLYxovm3qQxXjtJXEPVjl2xrlXqbUKBbGn 3/zYd5SvLxzm5i86w02Bu2prfyNvoSQsHCL6jC795gp/cbzYE5ynLSaryW0HLDwd9MOf 1LFV08C7Cdhfb6wLVpTBaKeKAgdMHlSCUuIRfFI1NZPVdPH/9IRK9+b7uJxNAJ1oQZC8 EV7S2EMurSlNh01I+t+2Hw4OwkYaBjPNTxxd4FyZPw3PMw2kQAQHb9pWpxNvfFCRPTXV OACg== X-Gm-Message-State: ANhLgQ0W4ksssZs3WhEgRhYA5lmUNCbeHUs7Kctu3Takhy1Rm+u0gjaj cD/XFsASEzzPyIHs8/GLq3lQTg== X-Google-Smtp-Source: ADFU+vvWx25e9jYZqzCSqaGAkmRaphEj/PgjffJq3MCSsLDQMbV5G1+yLhiRLui6IqbawgQdywROtw== X-Received: by 2002:a63:441e:: with SMTP id r30mr5741154pga.51.1583384541389; Wed, 04 Mar 2020 21:02:21 -0800 (PST) Received: from ryzen.cs.washington.edu ([2607:4000:200:11:e9fe:faad:3d84:58ea]) by smtp.gmail.com with ESMTPSA id y7sm17820466pfq.15.2020.03.04.21.02.20 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 04 Mar 2020 21:02:20 -0800 (PST) From: Luke Nelson X-Google-Original-From: Luke Nelson To: bpf@vger.kernel.org Subject: [PATCH bpf-next v5 3/4] bpf, doc: add BPF JIT for RV32G to BPF documentation Date: Wed, 4 Mar 2020 21:02:06 -0800 Message-Id: <20200305050207.4159-4-luke.r.nels@gmail.com> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20200305050207.4159-1-luke.r.nels@gmail.com> References: <20200305050207.4159-1-luke.r.nels@gmail.com> MIME-Version: 1.0 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20200304_210222_146724_F2CDC308 X-CRM114-Status: GOOD ( 11.73 ) X-Spam-Score: -0.1 (/) X-Spam-Report: SpamAssassin version 3.4.3 on bombadil.infradead.org summary: Content analysis details: (-0.1 points) pts rule name description ---- ---------------------- -------------------------------------------------- -0.0 RCVD_IN_DNSWL_NONE RBL: Sender listed at https://www.dnswl.org/, no trust [2607:f8b0:4864:20:0:0:0:441 listed in] [list.dnswl.org] -0.0 SPF_PASS SPF: sender matches SPF record 0.0 SPF_HELO_NONE SPF: HELO does not publish an SPF Record -0.1 DKIM_VALID Message has at least one valid DKIM or DK signature -0.1 DKIM_VALID_AU Message has a valid DKIM or DK signature from author's domain 0.1 DKIM_SIGNED Message has a DKIM or DK signature, not necessarily valid X-BeenThere: linux-riscv@lists.infradead.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Song Liu , linux-doc@vger.kernel.org, Yonghong Song , Paul Walmsley , Alexei Starovoitov , netdev@vger.kernel.org, Mauro Carvalho Chehab , linux-riscv@lists.infradead.org, Rob Herring , Daniel Borkmann , Jonathan Corbet , Jakub Kicinski , Andrii Nakryiko , Xi Wang , Albert Ou , Luke Nelson , Jonathan Cameron , Andy Shevchenko , =?utf-8?b?QmrDtnJuIFQ=?= =?utf-8?b?w7ZwZWw=?= , linux-kernel@vger.kernel.org, Martin KaFai Lau , Stephen Hemminger , Palmer Dabbelt , Greg Kroah-Hartman , "David S. Miller" Sender: "linux-riscv" Errors-To: linux-riscv-bounces+patchwork-linux-riscv=patchwork.kernel.org@lists.infradead.org Update filter.txt and admin-guide to mention the BPF JIT for RV32G. Co-developed-by: Xi Wang Signed-off-by: Xi Wang Signed-off-by: Luke Nelson --- Documentation/admin-guide/sysctl/net.rst | 3 ++- Documentation/networking/filter.txt | 2 +- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/Documentation/admin-guide/sysctl/net.rst b/Documentation/admin-guide/sysctl/net.rst index 287b98708a40..e043c9213388 100644 --- a/Documentation/admin-guide/sysctl/net.rst +++ b/Documentation/admin-guide/sysctl/net.rst @@ -67,7 +67,8 @@ two flavors of JITs, the newer eBPF JIT currently supported on: - sparc64 - mips64 - s390x - - riscv + - riscv64 + - riscv32 And the older cBPF JIT supported on the following archs: diff --git a/Documentation/networking/filter.txt b/Documentation/networking/filter.txt index c4a328f2d57a..2f0f8b17dade 100644 --- a/Documentation/networking/filter.txt +++ b/Documentation/networking/filter.txt @@ -606,7 +606,7 @@ before a conversion to the new layout is being done behind the scenes! Currently, the classic BPF format is being used for JITing on most 32-bit architectures, whereas x86-64, aarch64, s390x, powerpc64, -sparc64, arm32, riscv (RV64G) perform JIT compilation from eBPF +sparc64, arm32, riscv64, riscv32 perform JIT compilation from eBPF instruction set. Some core changes of the new internal format: From patchwork Thu Mar 5 05:02:07 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Luke Nelson X-Patchwork-Id: 11421197 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 2186A1580 for ; Thu, 5 Mar 2020 05:02:35 +0000 (UTC) Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id F1CAB21556 for ; Thu, 5 Mar 2020 05:02:34 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=lists.infradead.org header.i=@lists.infradead.org header.b="KTVXvrMm"; dkim=fail reason="signature verification failed" (1024-bit key) header.d=cs.washington.edu header.i=@cs.washington.edu header.b="dAYgGP3C" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org F1CAB21556 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=cs.washington.edu Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-riscv-bounces+patchwork-linux-riscv=patchwork.kernel.org@lists.infradead.org DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20170209; h=Sender:Cc:List-Subscribe: List-Help:List-Post:List-Archive:List-Unsubscribe:List-Id: Content-Transfer-Encoding:Content-Type:MIME-Version:References:In-Reply-To: Message-Id:Date:Subject:To:From:Reply-To:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=sV8q7HarohWqKtnLYGNVE5KFLA+QloKZuMOzSIKtRfg=; b=KTVXvrMmNcv1yH p3Fz8yt4U4reHi5X7bksR4jFN40hE97En9ArApyEwwNA803heQnz9HYwqy0MypM2juUbpCbBfxcK3 /KnAvXD0c4F1dDpdF35kNeGgjWqXgwLoXPdWpByp62ZjmnLSnoJB8o/xNe9ExJIhtqnwIQI54hAzm r9iHPLRb2RDN1gavs+nY5rurcbeBq7whq6O/EClNCBfrCDWVBX46YRD5Z0+dYfS4fjWbbgMAugvZa 77yEf16Frm4/eQLHSPA4Dh+PMO/vWVtNdsREXkPQ4H83lzvBartYJbbpAFB/slvQJSCOozLJilHvK bNsacGQCZX2LmX6164KA==; Received: from localhost ([127.0.0.1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.92.3 #3 (Red Hat Linux)) id 1j9ieV-0004bo-5H; Thu, 05 Mar 2020 05:02:31 +0000 Received: from mail-pg1-x544.google.com ([2607:f8b0:4864:20::544]) by bombadil.infradead.org with esmtps (Exim 4.92.3 #3 (Red Hat Linux)) id 1j9ieO-0004Th-A7 for linux-riscv@lists.infradead.org; Thu, 05 Mar 2020 05:02:25 +0000 Received: by mail-pg1-x544.google.com with SMTP id m5so2146748pgg.0 for ; Wed, 04 Mar 2020 21:02:23 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=cs.washington.edu; s=goo201206; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=sV8q7HarohWqKtnLYGNVE5KFLA+QloKZuMOzSIKtRfg=; b=dAYgGP3CO8Go0Zt3CaMIpp3oxWWrwBLg5KnHPPIq1+VU1MpOpmu2zAFLvsOtpeZ180 NKGu6+L7b2be7/a8CLckxLaKZP1QmM/5QftVRYVbj5sqDFXW03auDeJMHhm2V1Cj1dlX nD37M0rf1FK33a5vwQcWtV+vReeBbTFCqxzas= X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=sV8q7HarohWqKtnLYGNVE5KFLA+QloKZuMOzSIKtRfg=; b=eXu+/4H3ZPjeUW9s4cEfo3RlbjcLAjcr2x/N83KhBswBW2YMpxTfW6OqjZzqjxqjSQ HcADLpW5kB8VixIUZRul6d97axxkx8LrGgwUeiHd19CC/hV4L932WeGOPFlcCU2vknfW HYnBOWw1Qg7fPnEomxQC9AxRjV4UibkMKQC2qS7DR2E+wCnMk0baR/8jt+BjxMyczyF9 g76C9Z4LLd8vsS9tM1tk1OlFJvsenS1BCjVZdmiPPr9A9JOLejGqQfZ+NLC24BEllZv8 RSczHvzFcj/crko4vlRIG1jEc4GTYXK/1bMVHho8rrMoPQ0XIsPvfs0iH9ing7SJKgaQ XlJA== X-Gm-Message-State: ANhLgQ1uUg/BN8Erki0Yj+086nC0MA48nBiasZszfyK9i2hTL068BPAn +8awfcIccOBEId3OeZuwuQE/PQ== X-Google-Smtp-Source: ADFU+vuDUOkuw26JMyYv3YyKm1JJ4tL0PWpDR+kh5XrqfpKiEVBUYAUr39/AOzWyra/fJYd3DntQNg== X-Received: by 2002:a63:1e4f:: with SMTP id p15mr5929375pgm.28.1583384543147; Wed, 04 Mar 2020 21:02:23 -0800 (PST) Received: from ryzen.cs.washington.edu ([2607:4000:200:11:e9fe:faad:3d84:58ea]) by smtp.gmail.com with ESMTPSA id y7sm17820466pfq.15.2020.03.04.21.02.22 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 04 Mar 2020 21:02:22 -0800 (PST) From: Luke Nelson X-Google-Original-From: Luke Nelson To: bpf@vger.kernel.org Subject: [PATCH bpf-next v5 4/4] MAINTAINERS: add entry for RV32G BPF JIT Date: Wed, 4 Mar 2020 21:02:07 -0800 Message-Id: <20200305050207.4159-5-luke.r.nels@gmail.com> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20200305050207.4159-1-luke.r.nels@gmail.com> References: <20200305050207.4159-1-luke.r.nels@gmail.com> MIME-Version: 1.0 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20200304_210224_378222_E21318AD X-CRM114-Status: GOOD ( 10.35 ) X-Spam-Score: -0.1 (/) X-Spam-Report: SpamAssassin version 3.4.3 on bombadil.infradead.org summary: Content analysis details: (-0.1 points) pts rule name description ---- ---------------------- -------------------------------------------------- -0.0 RCVD_IN_DNSWL_NONE RBL: Sender listed at https://www.dnswl.org/, no trust [2607:f8b0:4864:20:0:0:0:544 listed in] [list.dnswl.org] -0.0 SPF_PASS SPF: sender matches SPF record 0.0 SPF_HELO_NONE SPF: HELO does not publish an SPF Record -0.1 DKIM_VALID Message has at least one valid DKIM or DK signature -0.1 DKIM_VALID_AU Message has a valid DKIM or DK signature from author's domain 0.1 DKIM_SIGNED Message has a DKIM or DK signature, not necessarily valid X-BeenThere: linux-riscv@lists.infradead.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Song Liu , linux-doc@vger.kernel.org, Yonghong Song , Paul Walmsley , Alexei Starovoitov , netdev@vger.kernel.org, Mauro Carvalho Chehab , linux-riscv@lists.infradead.org, Rob Herring , Daniel Borkmann , Jonathan Corbet , Jakub Kicinski , Andrii Nakryiko , Xi Wang , Albert Ou , Luke Nelson , Jonathan Cameron , Andy Shevchenko , =?utf-8?b?QmrDtnJuIFQ=?= =?utf-8?b?w7ZwZWw=?= , linux-kernel@vger.kernel.org, Martin KaFai Lau , Stephen Hemminger , Palmer Dabbelt , Greg Kroah-Hartman , "David S. Miller" Sender: "linux-riscv" Errors-To: linux-riscv-bounces+patchwork-linux-riscv=patchwork.kernel.org@lists.infradead.org Add a new entry for the 32-bit RISC-V JIT to MAINTAINERS and change mailing list from netdev to bpf following the guidelines from commit e42da4c62abb ("docs/bpf: Update bpf development Q/A file"). Cc: Björn Töpel Signed-off-by: Xi Wang Signed-off-by: Luke Nelson --- MAINTAINERS | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/MAINTAINERS b/MAINTAINERS index 8f27f40d22bb..ffad63a02d7a 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -3213,11 +3213,20 @@ L: bpf@vger.kernel.org S: Maintained F: arch/powerpc/net/ -BPF JIT for RISC-V (RV64G) +BPF JIT for RISC-V (32-bit) +M: Luke Nelson +M: Xi Wang +L: bpf@vger.kernel.org +S: Maintained +F: arch/riscv/net/ +X: arch/riscv/net/bpf_jit_comp64.c + +BPF JIT for RISC-V (64-bit) M: Björn Töpel -L: netdev@vger.kernel.org +L: bpf@vger.kernel.org S: Maintained F: arch/riscv/net/ +X: arch/riscv/net/bpf_jit_comp32.c BPF JIT for S390 M: Ilya Leoshkevich