Merge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next

Daniel Borkmann says:

====================
pull-request: bpf-next 2020-03-13

The following pull-request contains BPF updates for your *net-next* tree.

We've added 86 non-merge commits during the last 12 day(s) which contain
a total of 107 files changed, 5771 insertions(+), 1700 deletions(-).

The main changes are:

1) Add modify_return attach type which allows to attach to a function via
   BPF trampoline and is run after the fentry and before the fexit programs
   and can pass a return code to the original caller, from KP Singh.

2) Generalize BPF's kallsyms handling and add BPF trampoline and dispatcher
   objects to be visible in /proc/kallsyms so they can be annotated in
   stack traces, from Jiri Olsa.

3) Extend BPF sockmap to allow for UDP next to existing TCP support in order
   in order to enable this for BPF based socket dispatch, from Lorenz Bauer.

4) Introduce a new bpftool 'prog profile' command which attaches to existing
   BPF programs via fentry and fexit hooks and reads out hardware counters
   during that period, from Song Liu. Example usage:

   bpftool prog profile id 337 duration 3 cycles instructions llc_misses

        4228 run_cnt
     3403698 cycles                                              (84.08%)
     3525294 instructions   #  1.04 insn per cycle               (84.05%)
          13 llc_misses     #  3.69 LLC misses per million isns  (83.50%)

5) Batch of improvements to libbpf, bpftool and BPF selftests. Also addition
   of a new bpf_link abstraction to keep in particular BPF tracing programs
   attached even when the applicaion owning them exits, from Andrii Nakryiko.

6) New bpf_get_current_pid_tgid() helper for tracing to perform PID filtering
   and which returns the PID as seen by the init namespace, from Carlos Neira.

7) Refactor of RISC-V JIT code to move out common pieces and addition of a
   new RV32G BPF JIT compiler, from Luke Nelson.

8) Add gso_size context member to __sk_buff in order to be able to know whether
   a given skb is GSO or not, from Willem de Bruijn.

9) Add a new bpf_xdp_output() helper which reuses XDP's existing perf RB output
   implementation but can be called from tracepoint programs, from Eelco Chaudron.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

6 files changed, 2004 insertions, 602 deletions

diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
index 1a3b5a5276be..2366698df179 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 ec5b14763316..9a1e5f0a94e5 100644
--- a/arch/riscv/net/Makefile
+++ b/arch/riscv/net/Makefile
@@ -1,2 +1,9 @@
 # SPDX-License-Identifier: GPL-2.0-only
-obj-$(CONFIG_BPF_JIT) += bpf_jit_comp.o
+
+obj-$(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
new file mode 100644
index 000000000000..20e235d06f66
--- /dev/null
+++ b/arch/riscv/net/bpf_jit.h
@@ -0,0 +1,514 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Common functionality for RV32 and RV64 BPF JIT compilers
+ *
+ * Copyright (c) 2019 Björn Töpel <bjorn.topel@gmail.com>
+ *
+ */
+
+#ifndef _BPF_JIT_H
+#define _BPF_JIT_H
+
+#include <linux/bpf.h>
+#include <linux/filter.h>
+#include <asm/cacheflush.h>
+
+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);
+}
+
+/* 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);
+}
+
+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_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);
+}
+
+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_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);
+}
+
+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_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);
+}
+
+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);
+}
+
+/*
+ * 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);
+}
+
+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);
+}
+
+#endif /* __riscv_xlen == 64 */
+
+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_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 <luke.r.nels@gmail.com>
+ * Copyright (c) 2020 Xi Wang <xi.wang@gmail.com>
+ *
+ * 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 <linux/bpf.h>
+#include <linux/filter.h>
+#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);