powerpc/ebpf/jit: Implement JIT compiler for extended BPF

PPC64 eBPF JIT compiler.

Enable with:
  echo 1 > /proc/sys/net/core/bpf_jit_enable
or
  echo 2 > /proc/sys/net/core/bpf_jit_enable

... to see the generated JIT code. This can further be processed with
tools/net/bpf_jit_disasm.

With CONFIG_TEST_BPF=m and 'modprobe test_bpf':

 test_bpf: Summary: 305 PASSED, 0 FAILED, [297/297 JIT'ed]

... on both ppc64 BE and LE.

The details of the approach are documented through various comments in
the code.

Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
diff --git a/arch/powerpc/net/bpf_jit_comp64.c b/arch/powerpc/net/bpf_jit_comp64.c
new file mode 100644
index 0000000..6073b78
--- /dev/null
+++ b/arch/powerpc/net/bpf_jit_comp64.c
@@ -0,0 +1,954 @@
+/*
+ * bpf_jit_comp64.c: eBPF JIT compiler
+ *
+ * Copyright 2016 Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
+ *		  IBM Corporation
+ *
+ * Based on the powerpc classic BPF JIT compiler by Matt Evans
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2
+ * of the License.
+ */
+#include <linux/moduleloader.h>
+#include <asm/cacheflush.h>
+#include <linux/netdevice.h>
+#include <linux/filter.h>
+#include <linux/if_vlan.h>
+#include <asm/kprobes.h>
+
+#include "bpf_jit64.h"
+
+int bpf_jit_enable __read_mostly;
+
+static void bpf_jit_fill_ill_insns(void *area, unsigned int size)
+{
+	int *p = area;
+
+	/* Fill whole space with trap instructions */
+	while (p < (int *)((char *)area + size))
+		*p++ = BREAKPOINT_INSTRUCTION;
+}
+
+static inline void bpf_flush_icache(void *start, void *end)
+{
+	smp_wmb();
+	flush_icache_range((unsigned long)start, (unsigned long)end);
+}
+
+static inline bool bpf_is_seen_register(struct codegen_context *ctx, int i)
+{
+	return (ctx->seen & (1 << (31 - b2p[i])));
+}
+
+static inline void bpf_set_seen_register(struct codegen_context *ctx, int i)
+{
+	ctx->seen |= (1 << (31 - b2p[i]));
+}
+
+static inline bool bpf_has_stack_frame(struct codegen_context *ctx)
+{
+	/*
+	 * We only need a stack frame if:
+	 * - we call other functions (kernel helpers), or
+	 * - the bpf program uses its stack area
+	 * The latter condition is deduced from the usage of BPF_REG_FP
+	 */
+	return ctx->seen & SEEN_FUNC || bpf_is_seen_register(ctx, BPF_REG_FP);
+}
+
+static void bpf_jit_emit_skb_loads(u32 *image, struct codegen_context *ctx)
+{
+	/*
+	 * Load skb->len and skb->data_len
+	 * r3 points to skb
+	 */
+	PPC_LWZ(b2p[SKB_HLEN_REG], 3, offsetof(struct sk_buff, len));
+	PPC_LWZ(b2p[TMP_REG_1], 3, offsetof(struct sk_buff, data_len));
+	/* header_len = len - data_len */
+	PPC_SUB(b2p[SKB_HLEN_REG], b2p[SKB_HLEN_REG], b2p[TMP_REG_1]);
+
+	/* skb->data pointer */
+	PPC_BPF_LL(b2p[SKB_DATA_REG], 3, offsetof(struct sk_buff, data));
+}
+
+static void bpf_jit_emit_func_call(u32 *image, struct codegen_context *ctx, u64 func)
+{
+#ifdef PPC64_ELF_ABI_v1
+	/* func points to the function descriptor */
+	PPC_LI64(b2p[TMP_REG_2], func);
+	/* Load actual entry point from function descriptor */
+	PPC_BPF_LL(b2p[TMP_REG_1], b2p[TMP_REG_2], 0);
+	/* ... and move it to LR */
+	PPC_MTLR(b2p[TMP_REG_1]);
+	/*
+	 * Load TOC from function descriptor at offset 8.
+	 * We can clobber r2 since we get called through a
+	 * function pointer (so caller will save/restore r2)
+	 * and since we don't use a TOC ourself.
+	 */
+	PPC_BPF_LL(2, b2p[TMP_REG_2], 8);
+#else
+	/* We can clobber r12 */
+	PPC_FUNC_ADDR(12, func);
+	PPC_MTLR(12);
+#endif
+	PPC_BLRL();
+}
+
+static void bpf_jit_build_prologue(u32 *image, struct codegen_context *ctx)
+{
+	int i;
+	bool new_stack_frame = bpf_has_stack_frame(ctx);
+
+	if (new_stack_frame) {
+		/*
+		 * We need a stack frame, but we don't necessarily need to
+		 * save/restore LR unless we call other functions
+		 */
+		if (ctx->seen & SEEN_FUNC) {
+			EMIT(PPC_INST_MFLR | __PPC_RT(R0));
+			PPC_BPF_STL(0, 1, PPC_LR_STKOFF);
+		}
+
+		PPC_BPF_STLU(1, 1, -BPF_PPC_STACKFRAME);
+	}
+
+	/*
+	 * Back up non-volatile regs -- BPF registers 6-10
+	 * If we haven't created our own stack frame, we save these
+	 * in the protected zone below the previous stack frame
+	 */
+	for (i = BPF_REG_6; i <= BPF_REG_10; i++)
+		if (bpf_is_seen_register(ctx, i))
+			PPC_BPF_STL(b2p[i], 1,
+				(new_stack_frame ? BPF_PPC_STACKFRAME : 0) -
+					(8 * (32 - b2p[i])));
+
+	/*
+	 * Save additional non-volatile regs if we cache skb
+	 * Also, setup skb data
+	 */
+	if (ctx->seen & SEEN_SKB) {
+		PPC_BPF_STL(b2p[SKB_HLEN_REG], 1,
+			BPF_PPC_STACKFRAME - (8 * (32 - b2p[SKB_HLEN_REG])));
+		PPC_BPF_STL(b2p[SKB_DATA_REG], 1,
+			BPF_PPC_STACKFRAME - (8 * (32 - b2p[SKB_DATA_REG])));
+		bpf_jit_emit_skb_loads(image, ctx);
+	}
+
+	/* Setup frame pointer to point to the bpf stack area */
+	if (bpf_is_seen_register(ctx, BPF_REG_FP))
+		PPC_ADDI(b2p[BPF_REG_FP], 1,
+				BPF_PPC_STACKFRAME - BPF_PPC_STACK_SAVE);
+}
+
+static void bpf_jit_build_epilogue(u32 *image, struct codegen_context *ctx)
+{
+	int i;
+	bool new_stack_frame = bpf_has_stack_frame(ctx);
+
+	/* Move result to r3 */
+	PPC_MR(3, b2p[BPF_REG_0]);
+
+	/* Restore NVRs */
+	for (i = BPF_REG_6; i <= BPF_REG_10; i++)
+		if (bpf_is_seen_register(ctx, i))
+			PPC_BPF_LL(b2p[i], 1,
+				(new_stack_frame ? BPF_PPC_STACKFRAME : 0) -
+					(8 * (32 - b2p[i])));
+
+	/* Restore non-volatile registers used for skb cache */
+	if (ctx->seen & SEEN_SKB) {
+		PPC_BPF_LL(b2p[SKB_HLEN_REG], 1,
+			BPF_PPC_STACKFRAME - (8 * (32 - b2p[SKB_HLEN_REG])));
+		PPC_BPF_LL(b2p[SKB_DATA_REG], 1,
+			BPF_PPC_STACKFRAME - (8 * (32 - b2p[SKB_DATA_REG])));
+	}
+
+	/* Tear down our stack frame */
+	if (new_stack_frame) {
+		PPC_ADDI(1, 1, BPF_PPC_STACKFRAME);
+		if (ctx->seen & SEEN_FUNC) {
+			PPC_BPF_LL(0, 1, PPC_LR_STKOFF);
+			PPC_MTLR(0);
+		}
+	}
+
+	PPC_BLR();
+}
+
+/* Assemble the body code between the prologue & epilogue */
+static int bpf_jit_build_body(struct bpf_prog *fp, u32 *image,
+			      struct codegen_context *ctx,
+			      u32 *addrs)
+{
+	const struct bpf_insn *insn = fp->insnsi;
+	int flen = fp->len;
+	int i;
+
+	/* Start of epilogue code - will only be valid 2nd pass onwards */
+	u32 exit_addr = addrs[flen];
+
+	for (i = 0; i < flen; i++) {
+		u32 code = insn[i].code;
+		u32 dst_reg = b2p[insn[i].dst_reg];
+		u32 src_reg = b2p[insn[i].src_reg];
+		s16 off = insn[i].off;
+		s32 imm = insn[i].imm;
+		u64 imm64;
+		u8 *func;
+		u32 true_cond;
+		int stack_local_off;
+
+		/*
+		 * addrs[] maps a BPF bytecode address into a real offset from
+		 * the start of the body code.
+		 */
+		addrs[i] = ctx->idx * 4;
+
+		/*
+		 * As an optimization, we note down which non-volatile registers
+		 * are used so that we can only save/restore those in our
+		 * prologue and epilogue. We do this here regardless of whether
+		 * the actual BPF instruction uses src/dst registers or not
+		 * (for instance, BPF_CALL does not use them). The expectation
+		 * is that those instructions will have src_reg/dst_reg set to
+		 * 0. Even otherwise, we just lose some prologue/epilogue
+		 * optimization but everything else should work without
+		 * any issues.
+		 */
+		if (dst_reg >= 24 && dst_reg <= 31)
+			bpf_set_seen_register(ctx, insn[i].dst_reg);
+		if (src_reg >= 24 && src_reg <= 31)
+			bpf_set_seen_register(ctx, insn[i].src_reg);
+
+		switch (code) {
+		/*
+		 * Arithmetic operations: ADD/SUB/MUL/DIV/MOD/NEG
+		 */
+		case BPF_ALU | BPF_ADD | BPF_X: /* (u32) dst += (u32) src */
+		case BPF_ALU64 | BPF_ADD | BPF_X: /* dst += src */
+			PPC_ADD(dst_reg, dst_reg, src_reg);
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_SUB | BPF_X: /* (u32) dst -= (u32) src */
+		case BPF_ALU64 | BPF_SUB | BPF_X: /* dst -= src */
+			PPC_SUB(dst_reg, dst_reg, src_reg);
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_ADD | BPF_K: /* (u32) dst += (u32) imm */
+		case BPF_ALU | BPF_SUB | BPF_K: /* (u32) dst -= (u32) imm */
+		case BPF_ALU64 | BPF_ADD | BPF_K: /* dst += imm */
+		case BPF_ALU64 | BPF_SUB | BPF_K: /* dst -= imm */
+			if (BPF_OP(code) == BPF_SUB)
+				imm = -imm;
+			if (imm) {
+				if (imm >= -32768 && imm < 32768)
+					PPC_ADDI(dst_reg, dst_reg, IMM_L(imm));
+				else {
+					PPC_LI32(b2p[TMP_REG_1], imm);
+					PPC_ADD(dst_reg, dst_reg, b2p[TMP_REG_1]);
+				}
+			}
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_MUL | BPF_X: /* (u32) dst *= (u32) src */
+		case BPF_ALU64 | BPF_MUL | BPF_X: /* dst *= src */
+			if (BPF_CLASS(code) == BPF_ALU)
+				PPC_MULW(dst_reg, dst_reg, src_reg);
+			else
+				PPC_MULD(dst_reg, dst_reg, src_reg);
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_MUL | BPF_K: /* (u32) dst *= (u32) imm */
+		case BPF_ALU64 | BPF_MUL | BPF_K: /* dst *= imm */
+			if (imm >= -32768 && imm < 32768)
+				PPC_MULI(dst_reg, dst_reg, IMM_L(imm));
+			else {
+				PPC_LI32(b2p[TMP_REG_1], imm);
+				if (BPF_CLASS(code) == BPF_ALU)
+					PPC_MULW(dst_reg, dst_reg,
+							b2p[TMP_REG_1]);
+				else
+					PPC_MULD(dst_reg, dst_reg,
+							b2p[TMP_REG_1]);
+			}
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_DIV | BPF_X: /* (u32) dst /= (u32) src */
+		case BPF_ALU | BPF_MOD | BPF_X: /* (u32) dst %= (u32) src */
+			PPC_CMPWI(src_reg, 0);
+			PPC_BCC_SHORT(COND_NE, (ctx->idx * 4) + 12);
+			PPC_LI(b2p[BPF_REG_0], 0);
+			PPC_JMP(exit_addr);
+			if (BPF_OP(code) == BPF_MOD) {
+				PPC_DIVWU(b2p[TMP_REG_1], dst_reg, src_reg);
+				PPC_MULW(b2p[TMP_REG_1], src_reg,
+						b2p[TMP_REG_1]);
+				PPC_SUB(dst_reg, dst_reg, b2p[TMP_REG_1]);
+			} else
+				PPC_DIVWU(dst_reg, dst_reg, src_reg);
+			goto bpf_alu32_trunc;
+		case BPF_ALU64 | BPF_DIV | BPF_X: /* dst /= src */
+		case BPF_ALU64 | BPF_MOD | BPF_X: /* dst %= src */
+			PPC_CMPDI(src_reg, 0);
+			PPC_BCC_SHORT(COND_NE, (ctx->idx * 4) + 12);
+			PPC_LI(b2p[BPF_REG_0], 0);
+			PPC_JMP(exit_addr);
+			if (BPF_OP(code) == BPF_MOD) {
+				PPC_DIVD(b2p[TMP_REG_1], dst_reg, src_reg);
+				PPC_MULD(b2p[TMP_REG_1], src_reg,
+						b2p[TMP_REG_1]);
+				PPC_SUB(dst_reg, dst_reg, b2p[TMP_REG_1]);
+			} else
+				PPC_DIVD(dst_reg, dst_reg, src_reg);
+			break;
+		case BPF_ALU | BPF_MOD | BPF_K: /* (u32) dst %= (u32) imm */
+		case BPF_ALU | BPF_DIV | BPF_K: /* (u32) dst /= (u32) imm */
+		case BPF_ALU64 | BPF_MOD | BPF_K: /* dst %= imm */
+		case BPF_ALU64 | BPF_DIV | BPF_K: /* dst /= imm */
+			if (imm == 0)
+				return -EINVAL;
+			else if (imm == 1)
+				goto bpf_alu32_trunc;
+
+			PPC_LI32(b2p[TMP_REG_1], imm);
+			switch (BPF_CLASS(code)) {
+			case BPF_ALU:
+				if (BPF_OP(code) == BPF_MOD) {
+					PPC_DIVWU(b2p[TMP_REG_2], dst_reg,
+							b2p[TMP_REG_1]);
+					PPC_MULW(b2p[TMP_REG_1],
+							b2p[TMP_REG_1],
+							b2p[TMP_REG_2]);
+					PPC_SUB(dst_reg, dst_reg,
+							b2p[TMP_REG_1]);
+				} else
+					PPC_DIVWU(dst_reg, dst_reg,
+							b2p[TMP_REG_1]);
+				break;
+			case BPF_ALU64:
+				if (BPF_OP(code) == BPF_MOD) {
+					PPC_DIVD(b2p[TMP_REG_2], dst_reg,
+							b2p[TMP_REG_1]);
+					PPC_MULD(b2p[TMP_REG_1],
+							b2p[TMP_REG_1],
+							b2p[TMP_REG_2]);
+					PPC_SUB(dst_reg, dst_reg,
+							b2p[TMP_REG_1]);
+				} else
+					PPC_DIVD(dst_reg, dst_reg,
+							b2p[TMP_REG_1]);
+				break;
+			}
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_NEG: /* (u32) dst = -dst */
+		case BPF_ALU64 | BPF_NEG: /* dst = -dst */
+			PPC_NEG(dst_reg, dst_reg);
+			goto bpf_alu32_trunc;
+
+		/*
+		 * Logical operations: AND/OR/XOR/[A]LSH/[A]RSH
+		 */
+		case BPF_ALU | BPF_AND | BPF_X: /* (u32) dst = dst & src */
+		case BPF_ALU64 | BPF_AND | BPF_X: /* dst = dst & src */
+			PPC_AND(dst_reg, dst_reg, src_reg);
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_AND | BPF_K: /* (u32) dst = dst & imm */
+		case BPF_ALU64 | BPF_AND | BPF_K: /* dst = dst & imm */
+			if (!IMM_H(imm))
+				PPC_ANDI(dst_reg, dst_reg, IMM_L(imm));
+			else {
+				/* Sign-extended */
+				PPC_LI32(b2p[TMP_REG_1], imm);
+				PPC_AND(dst_reg, dst_reg, b2p[TMP_REG_1]);
+			}
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_OR | BPF_X: /* dst = (u32) dst | (u32) src */
+		case BPF_ALU64 | BPF_OR | BPF_X: /* dst = dst | src */
+			PPC_OR(dst_reg, dst_reg, src_reg);
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_OR | BPF_K:/* dst = (u32) dst | (u32) imm */
+		case BPF_ALU64 | BPF_OR | BPF_K:/* dst = dst | imm */
+			if (imm < 0 && BPF_CLASS(code) == BPF_ALU64) {
+				/* Sign-extended */
+				PPC_LI32(b2p[TMP_REG_1], imm);
+				PPC_OR(dst_reg, dst_reg, b2p[TMP_REG_1]);
+			} else {
+				if (IMM_L(imm))
+					PPC_ORI(dst_reg, dst_reg, IMM_L(imm));
+				if (IMM_H(imm))
+					PPC_ORIS(dst_reg, dst_reg, IMM_H(imm));
+			}
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_XOR | BPF_X: /* (u32) dst ^= src */
+		case BPF_ALU64 | BPF_XOR | BPF_X: /* dst ^= src */
+			PPC_XOR(dst_reg, dst_reg, src_reg);
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_XOR | BPF_K: /* (u32) dst ^= (u32) imm */
+		case BPF_ALU64 | BPF_XOR | BPF_K: /* dst ^= imm */
+			if (imm < 0 && BPF_CLASS(code) == BPF_ALU64) {
+				/* Sign-extended */
+				PPC_LI32(b2p[TMP_REG_1], imm);
+				PPC_XOR(dst_reg, dst_reg, b2p[TMP_REG_1]);
+			} else {
+				if (IMM_L(imm))
+					PPC_XORI(dst_reg, dst_reg, IMM_L(imm));
+				if (IMM_H(imm))
+					PPC_XORIS(dst_reg, dst_reg, IMM_H(imm));
+			}
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_LSH | BPF_X: /* (u32) dst <<= (u32) src */
+			/* slw clears top 32 bits */
+			PPC_SLW(dst_reg, dst_reg, src_reg);
+			break;
+		case BPF_ALU64 | BPF_LSH | BPF_X: /* dst <<= src; */
+			PPC_SLD(dst_reg, dst_reg, src_reg);
+			break;
+		case BPF_ALU | BPF_LSH | BPF_K: /* (u32) dst <<== (u32) imm */
+			/* with imm 0, we still need to clear top 32 bits */
+			PPC_SLWI(dst_reg, dst_reg, imm);
+			break;
+		case BPF_ALU64 | BPF_LSH | BPF_K: /* dst <<== imm */
+			if (imm != 0)
+				PPC_SLDI(dst_reg, dst_reg, imm);
+			break;
+		case BPF_ALU | BPF_RSH | BPF_X: /* (u32) dst >>= (u32) src */
+			PPC_SRW(dst_reg, dst_reg, src_reg);
+			break;
+		case BPF_ALU64 | BPF_RSH | BPF_X: /* dst >>= src */
+			PPC_SRD(dst_reg, dst_reg, src_reg);
+			break;
+		case BPF_ALU | BPF_RSH | BPF_K: /* (u32) dst >>= (u32) imm */
+			PPC_SRWI(dst_reg, dst_reg, imm);
+			break;
+		case BPF_ALU64 | BPF_RSH | BPF_K: /* dst >>= imm */
+			if (imm != 0)
+				PPC_SRDI(dst_reg, dst_reg, imm);
+			break;
+		case BPF_ALU64 | BPF_ARSH | BPF_X: /* (s64) dst >>= src */
+			PPC_SRAD(dst_reg, dst_reg, src_reg);
+			break;
+		case BPF_ALU64 | BPF_ARSH | BPF_K: /* (s64) dst >>= imm */
+			if (imm != 0)
+				PPC_SRADI(dst_reg, dst_reg, imm);
+			break;
+
+		/*
+		 * MOV
+		 */
+		case BPF_ALU | BPF_MOV | BPF_X: /* (u32) dst = src */
+		case BPF_ALU64 | BPF_MOV | BPF_X: /* dst = src */
+			PPC_MR(dst_reg, src_reg);
+			goto bpf_alu32_trunc;
+		case BPF_ALU | BPF_MOV | BPF_K: /* (u32) dst = imm */
+		case BPF_ALU64 | BPF_MOV | BPF_K: /* dst = (s64) imm */
+			PPC_LI32(dst_reg, imm);
+			if (imm < 0)
+				goto bpf_alu32_trunc;
+			break;
+
+bpf_alu32_trunc:
+		/* Truncate to 32-bits */
+		if (BPF_CLASS(code) == BPF_ALU)
+			PPC_RLWINM(dst_reg, dst_reg, 0, 0, 31);
+		break;
+
+		/*
+		 * BPF_FROM_BE/LE
+		 */
+		case BPF_ALU | BPF_END | BPF_FROM_LE:
+		case BPF_ALU | BPF_END | BPF_FROM_BE:
+#ifdef __BIG_ENDIAN__
+			if (BPF_SRC(code) == BPF_FROM_BE)
+				goto emit_clear;
+#else /* !__BIG_ENDIAN__ */
+			if (BPF_SRC(code) == BPF_FROM_LE)
+				goto emit_clear;
+#endif
+			switch (imm) {
+			case 16:
+				/* Rotate 8 bits left & mask with 0x0000ff00 */
+				PPC_RLWINM(b2p[TMP_REG_1], dst_reg, 8, 16, 23);
+				/* Rotate 8 bits right & insert LSB to reg */
+				PPC_RLWIMI(b2p[TMP_REG_1], dst_reg, 24, 24, 31);
+				/* Move result back to dst_reg */
+				PPC_MR(dst_reg, b2p[TMP_REG_1]);
+				break;
+			case 32:
+				/*
+				 * Rotate word left by 8 bits:
+				 * 2 bytes are already in their final position
+				 * -- byte 2 and 4 (of bytes 1, 2, 3 and 4)
+				 */
+				PPC_RLWINM(b2p[TMP_REG_1], dst_reg, 8, 0, 31);
+				/* Rotate 24 bits and insert byte 1 */
+				PPC_RLWIMI(b2p[TMP_REG_1], dst_reg, 24, 0, 7);
+				/* Rotate 24 bits and insert byte 3 */
+				PPC_RLWIMI(b2p[TMP_REG_1], dst_reg, 24, 16, 23);
+				PPC_MR(dst_reg, b2p[TMP_REG_1]);
+				break;
+			case 64:
+				/*
+				 * Way easier and faster(?) to store the value
+				 * into stack and then use ldbrx
+				 *
+				 * First, determine where in stack we can store
+				 * this:
+				 * - if we have allotted a stack frame, then we
+				 *   will utilize the area set aside by
+				 *   BPF_PPC_STACK_LOCALS
+				 * - else, we use the area beneath the NV GPR
+				 *   save area
+				 *
+				 * ctx->seen will be reliable in pass2, but
+				 * the instructions generated will remain the
+				 * same across all passes
+				 */
+				if (bpf_has_stack_frame(ctx))
+					stack_local_off = STACK_FRAME_MIN_SIZE;
+				else
+					stack_local_off = -(BPF_PPC_STACK_SAVE + 8);
+
+				PPC_STD(dst_reg, 1, stack_local_off);
+				PPC_ADDI(b2p[TMP_REG_1], 1, stack_local_off);
+				PPC_LDBRX(dst_reg, 0, b2p[TMP_REG_1]);
+				break;
+			}
+			break;
+
+emit_clear:
+			switch (imm) {
+			case 16:
+				/* zero-extend 16 bits into 64 bits */
+				PPC_RLDICL(dst_reg, dst_reg, 0, 48);
+				break;
+			case 32:
+				/* zero-extend 32 bits into 64 bits */
+				PPC_RLDICL(dst_reg, dst_reg, 0, 32);
+				break;
+			case 64:
+				/* nop */
+				break;
+			}
+			break;
+
+		/*
+		 * BPF_ST(X)
+		 */
+		case BPF_STX | BPF_MEM | BPF_B: /* *(u8 *)(dst + off) = src */
+		case BPF_ST | BPF_MEM | BPF_B: /* *(u8 *)(dst + off) = imm */
+			if (BPF_CLASS(code) == BPF_ST) {
+				PPC_LI(b2p[TMP_REG_1], imm);
+				src_reg = b2p[TMP_REG_1];
+			}
+			PPC_STB(src_reg, dst_reg, off);
+			break;
+		case BPF_STX | BPF_MEM | BPF_H: /* (u16 *)(dst + off) = src */
+		case BPF_ST | BPF_MEM | BPF_H: /* (u16 *)(dst + off) = imm */
+			if (BPF_CLASS(code) == BPF_ST) {
+				PPC_LI(b2p[TMP_REG_1], imm);
+				src_reg = b2p[TMP_REG_1];
+			}
+			PPC_STH(src_reg, dst_reg, off);
+			break;
+		case BPF_STX | BPF_MEM | BPF_W: /* *(u32 *)(dst + off) = src */
+		case BPF_ST | BPF_MEM | BPF_W: /* *(u32 *)(dst + off) = imm */
+			if (BPF_CLASS(code) == BPF_ST) {
+				PPC_LI32(b2p[TMP_REG_1], imm);
+				src_reg = b2p[TMP_REG_1];
+			}
+			PPC_STW(src_reg, dst_reg, off);
+			break;
+		case BPF_STX | BPF_MEM | BPF_DW: /* (u64 *)(dst + off) = src */
+		case BPF_ST | BPF_MEM | BPF_DW: /* *(u64 *)(dst + off) = imm */
+			if (BPF_CLASS(code) == BPF_ST) {
+				PPC_LI32(b2p[TMP_REG_1], imm);
+				src_reg = b2p[TMP_REG_1];
+			}
+			PPC_STD(src_reg, dst_reg, off);
+			break;
+
+		/*
+		 * BPF_STX XADD (atomic_add)
+		 */
+		/* *(u32 *)(dst + off) += src */
+		case BPF_STX | BPF_XADD | BPF_W:
+			/* Get EA into TMP_REG_1 */
+			PPC_ADDI(b2p[TMP_REG_1], dst_reg, off);
+			/* error if EA is not word-aligned */
+			PPC_ANDI(b2p[TMP_REG_2], b2p[TMP_REG_1], 0x03);
+			PPC_BCC_SHORT(COND_EQ, (ctx->idx * 4) + 12);
+			PPC_LI(b2p[BPF_REG_0], 0);
+			PPC_JMP(exit_addr);
+			/* load value from memory into TMP_REG_2 */
+			PPC_BPF_LWARX(b2p[TMP_REG_2], 0, b2p[TMP_REG_1], 0);
+			/* add value from src_reg into this */
+			PPC_ADD(b2p[TMP_REG_2], b2p[TMP_REG_2], src_reg);
+			/* store result back */
+			PPC_BPF_STWCX(b2p[TMP_REG_2], 0, b2p[TMP_REG_1]);
+			/* we're done if this succeeded */
+			PPC_BCC_SHORT(COND_EQ, (ctx->idx * 4) + (7*4));
+			/* otherwise, let's try once more */
+			PPC_BPF_LWARX(b2p[TMP_REG_2], 0, b2p[TMP_REG_1], 0);
+			PPC_ADD(b2p[TMP_REG_2], b2p[TMP_REG_2], src_reg);
+			PPC_BPF_STWCX(b2p[TMP_REG_2], 0, b2p[TMP_REG_1]);
+			/* exit if the store was not successful */
+			PPC_LI(b2p[BPF_REG_0], 0);
+			PPC_BCC(COND_NE, exit_addr);
+			break;
+		/* *(u64 *)(dst + off) += src */
+		case BPF_STX | BPF_XADD | BPF_DW:
+			PPC_ADDI(b2p[TMP_REG_1], dst_reg, off);
+			/* error if EA is not doubleword-aligned */
+			PPC_ANDI(b2p[TMP_REG_2], b2p[TMP_REG_1], 0x07);
+			PPC_BCC_SHORT(COND_EQ, (ctx->idx * 4) + (3*4));
+			PPC_LI(b2p[BPF_REG_0], 0);
+			PPC_JMP(exit_addr);
+			PPC_BPF_LDARX(b2p[TMP_REG_2], 0, b2p[TMP_REG_1], 0);
+			PPC_ADD(b2p[TMP_REG_2], b2p[TMP_REG_2], src_reg);
+			PPC_BPF_STDCX(b2p[TMP_REG_2], 0, b2p[TMP_REG_1]);
+			PPC_BCC_SHORT(COND_EQ, (ctx->idx * 4) + (7*4));
+			PPC_BPF_LDARX(b2p[TMP_REG_2], 0, b2p[TMP_REG_1], 0);
+			PPC_ADD(b2p[TMP_REG_2], b2p[TMP_REG_2], src_reg);
+			PPC_BPF_STDCX(b2p[TMP_REG_2], 0, b2p[TMP_REG_1]);
+			PPC_LI(b2p[BPF_REG_0], 0);
+			PPC_BCC(COND_NE, exit_addr);
+			break;
+
+		/*
+		 * BPF_LDX
+		 */
+		/* dst = *(u8 *)(ul) (src + off) */
+		case BPF_LDX | BPF_MEM | BPF_B:
+			PPC_LBZ(dst_reg, src_reg, off);
+			break;
+		/* dst = *(u16 *)(ul) (src + off) */
+		case BPF_LDX | BPF_MEM | BPF_H:
+			PPC_LHZ(dst_reg, src_reg, off);
+			break;
+		/* dst = *(u32 *)(ul) (src + off) */
+		case BPF_LDX | BPF_MEM | BPF_W:
+			PPC_LWZ(dst_reg, src_reg, off);
+			break;
+		/* dst = *(u64 *)(ul) (src + off) */
+		case BPF_LDX | BPF_MEM | BPF_DW:
+			PPC_LD(dst_reg, src_reg, off);
+			break;
+
+		/*
+		 * Doubleword load
+		 * 16 byte instruction that uses two 'struct bpf_insn'
+		 */
+		case BPF_LD | BPF_IMM | BPF_DW: /* dst = (u64) imm */
+			imm64 = ((u64)(u32) insn[i].imm) |
+				    (((u64)(u32) insn[i+1].imm) << 32);
+			/* Adjust for two bpf instructions */
+			addrs[++i] = ctx->idx * 4;
+			PPC_LI64(dst_reg, imm64);
+			break;
+
+		/*
+		 * Return/Exit
+		 */
+		case BPF_JMP | BPF_EXIT:
+			/*
+			 * If this isn't the very last instruction, branch to
+			 * the epilogue. If we _are_ the last instruction,
+			 * we'll just fall through to the epilogue.
+			 */
+			if (i != flen - 1)
+				PPC_JMP(exit_addr);
+			/* else fall through to the epilogue */
+			break;
+
+		/*
+		 * Call kernel helper
+		 */
+		case BPF_JMP | BPF_CALL:
+			ctx->seen |= SEEN_FUNC;
+			func = (u8 *) __bpf_call_base + imm;
+
+			/* Save skb pointer if we need to re-cache skb data */
+			if (bpf_helper_changes_skb_data(func))
+				PPC_BPF_STL(3, 1, STACK_FRAME_MIN_SIZE);
+
+			bpf_jit_emit_func_call(image, ctx, (u64)func);
+
+			/* move return value from r3 to BPF_REG_0 */
+			PPC_MR(b2p[BPF_REG_0], 3);
+
+			/* refresh skb cache */
+			if (bpf_helper_changes_skb_data(func)) {
+				/* reload skb pointer to r3 */
+				PPC_BPF_LL(3, 1, STACK_FRAME_MIN_SIZE);
+				bpf_jit_emit_skb_loads(image, ctx);
+			}
+			break;
+
+		/*
+		 * Jumps and branches
+		 */
+		case BPF_JMP | BPF_JA:
+			PPC_JMP(addrs[i + 1 + off]);
+			break;
+
+		case BPF_JMP | BPF_JGT | BPF_K:
+		case BPF_JMP | BPF_JGT | BPF_X:
+		case BPF_JMP | BPF_JSGT | BPF_K:
+		case BPF_JMP | BPF_JSGT | BPF_X:
+			true_cond = COND_GT;
+			goto cond_branch;
+		case BPF_JMP | BPF_JGE | BPF_K:
+		case BPF_JMP | BPF_JGE | BPF_X:
+		case BPF_JMP | BPF_JSGE | BPF_K:
+		case BPF_JMP | BPF_JSGE | BPF_X:
+			true_cond = COND_GE;
+			goto cond_branch;
+		case BPF_JMP | BPF_JEQ | BPF_K:
+		case BPF_JMP | BPF_JEQ | BPF_X:
+			true_cond = COND_EQ;
+			goto cond_branch;
+		case BPF_JMP | BPF_JNE | BPF_K:
+		case BPF_JMP | BPF_JNE | BPF_X:
+			true_cond = COND_NE;
+			goto cond_branch;
+		case BPF_JMP | BPF_JSET | BPF_K:
+		case BPF_JMP | BPF_JSET | BPF_X:
+			true_cond = COND_NE;
+			/* Fall through */
+
+cond_branch:
+			switch (code) {
+			case BPF_JMP | BPF_JGT | BPF_X:
+			case BPF_JMP | BPF_JGE | BPF_X:
+			case BPF_JMP | BPF_JEQ | BPF_X:
+			case BPF_JMP | BPF_JNE | BPF_X:
+				/* unsigned comparison */
+				PPC_CMPLD(dst_reg, src_reg);
+				break;
+			case BPF_JMP | BPF_JSGT | BPF_X:
+			case BPF_JMP | BPF_JSGE | BPF_X:
+				/* signed comparison */
+				PPC_CMPD(dst_reg, src_reg);
+				break;
+			case BPF_JMP | BPF_JSET | BPF_X:
+				PPC_AND_DOT(b2p[TMP_REG_1], dst_reg, src_reg);
+				break;
+			case BPF_JMP | BPF_JNE | BPF_K:
+			case BPF_JMP | BPF_JEQ | BPF_K:
+			case BPF_JMP | BPF_JGT | BPF_K:
+			case BPF_JMP | BPF_JGE | BPF_K:
+				/*
+				 * Need sign-extended load, so only positive
+				 * values can be used as imm in cmpldi
+				 */
+				if (imm >= 0 && imm < 32768)
+					PPC_CMPLDI(dst_reg, imm);
+				else {
+					/* sign-extending load */
+					PPC_LI32(b2p[TMP_REG_1], imm);
+					/* ... but unsigned comparison */
+					PPC_CMPLD(dst_reg, b2p[TMP_REG_1]);
+				}
+				break;
+			case BPF_JMP | BPF_JSGT | BPF_K:
+			case BPF_JMP | BPF_JSGE | BPF_K:
+				/*
+				 * signed comparison, so any 16-bit value
+				 * can be used in cmpdi
+				 */
+				if (imm >= -32768 && imm < 32768)
+					PPC_CMPDI(dst_reg, imm);
+				else {
+					PPC_LI32(b2p[TMP_REG_1], imm);
+					PPC_CMPD(dst_reg, b2p[TMP_REG_1]);
+				}
+				break;
+			case BPF_JMP | BPF_JSET | BPF_K:
+				/* andi does not sign-extend the immediate */
+				if (imm >= 0 && imm < 32768)
+					/* PPC_ANDI is _only/always_ dot-form */
+					PPC_ANDI(b2p[TMP_REG_1], dst_reg, imm);
+				else {
+					PPC_LI32(b2p[TMP_REG_1], imm);
+					PPC_AND_DOT(b2p[TMP_REG_1], dst_reg,
+						    b2p[TMP_REG_1]);
+				}
+				break;
+			}
+			PPC_BCC(true_cond, addrs[i + 1 + off]);
+			break;
+
+		/*
+		 * Loads from packet header/data
+		 * Assume 32-bit input value in imm and X (src_reg)
+		 */
+
+		/* Absolute loads */
+		case BPF_LD | BPF_W | BPF_ABS:
+			func = (u8 *)CHOOSE_LOAD_FUNC(imm, sk_load_word);
+			goto common_load_abs;
+		case BPF_LD | BPF_H | BPF_ABS:
+			func = (u8 *)CHOOSE_LOAD_FUNC(imm, sk_load_half);
+			goto common_load_abs;
+		case BPF_LD | BPF_B | BPF_ABS:
+			func = (u8 *)CHOOSE_LOAD_FUNC(imm, sk_load_byte);
+common_load_abs:
+			/*
+			 * Load from [imm]
+			 * Load into r4, which can just be passed onto
+			 *  skb load helpers as the second parameter
+			 */
+			PPC_LI32(4, imm);
+			goto common_load;
+
+		/* Indirect loads */
+		case BPF_LD | BPF_W | BPF_IND:
+			func = (u8 *)sk_load_word;
+			goto common_load_ind;
+		case BPF_LD | BPF_H | BPF_IND:
+			func = (u8 *)sk_load_half;
+			goto common_load_ind;
+		case BPF_LD | BPF_B | BPF_IND:
+			func = (u8 *)sk_load_byte;
+common_load_ind:
+			/*
+			 * Load from [src_reg + imm]
+			 * Treat src_reg as a 32-bit value
+			 */
+			PPC_EXTSW(4, src_reg);
+			if (imm) {
+				if (imm >= -32768 && imm < 32768)
+					PPC_ADDI(4, 4, IMM_L(imm));
+				else {
+					PPC_LI32(b2p[TMP_REG_1], imm);
+					PPC_ADD(4, 4, b2p[TMP_REG_1]);
+				}
+			}
+
+common_load:
+			ctx->seen |= SEEN_SKB;
+			ctx->seen |= SEEN_FUNC;
+			bpf_jit_emit_func_call(image, ctx, (u64)func);
+
+			/*
+			 * Helper returns 'lt' condition on error, and an
+			 * appropriate return value in BPF_REG_0
+			 */
+			PPC_BCC(COND_LT, exit_addr);
+			break;
+
+		/*
+		 * TODO: Tail call
+		 */
+		case BPF_JMP | BPF_CALL | BPF_X:
+
+		default:
+			/*
+			 * The filter contains something cruel & unusual.
+			 * We don't handle it, but also there shouldn't be
+			 * anything missing from our list.
+			 */
+			pr_err_ratelimited("eBPF filter opcode %04x (@%d) unsupported\n",
+					code, i);
+			return -ENOTSUPP;
+		}
+	}
+
+	/* Set end-of-body-code address for exit. */
+	addrs[i] = ctx->idx * 4;
+
+	return 0;
+}
+
+void bpf_jit_compile(struct bpf_prog *fp) { }
+
+struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
+{
+	u32 proglen;
+	u32 alloclen;
+	u8 *image = NULL;
+	u32 *code_base;
+	u32 *addrs;
+	struct codegen_context cgctx;
+	int pass;
+	int flen;
+	struct bpf_binary_header *bpf_hdr;
+
+	if (!bpf_jit_enable)
+		return fp;
+
+	flen = fp->len;
+	addrs = kzalloc((flen+1) * sizeof(*addrs), GFP_KERNEL);
+	if (addrs == NULL)
+		return fp;
+
+	cgctx.idx = 0;
+	cgctx.seen = 0;
+	/* Scouting faux-generate pass 0 */
+	if (bpf_jit_build_body(fp, 0, &cgctx, addrs))
+		/* We hit something illegal or unsupported. */
+		goto out;
+
+	/*
+	 * Pretend to build prologue, given the features we've seen.  This will
+	 * update ctgtx.idx as it pretends to output instructions, then we can
+	 * calculate total size from idx.
+	 */
+	bpf_jit_build_prologue(0, &cgctx);
+	bpf_jit_build_epilogue(0, &cgctx);
+
+	proglen = cgctx.idx * 4;
+	alloclen = proglen + FUNCTION_DESCR_SIZE;
+
+	bpf_hdr = bpf_jit_binary_alloc(alloclen, &image, 4,
+			bpf_jit_fill_ill_insns);
+	if (!bpf_hdr)
+		goto out;
+
+	code_base = (u32 *)(image + FUNCTION_DESCR_SIZE);
+
+	/* Code generation passes 1-2 */
+	for (pass = 1; pass < 3; pass++) {
+		/* Now build the prologue, body code & epilogue for real. */
+		cgctx.idx = 0;
+		bpf_jit_build_prologue(code_base, &cgctx);
+		bpf_jit_build_body(fp, code_base, &cgctx, addrs);
+		bpf_jit_build_epilogue(code_base, &cgctx);
+
+		if (bpf_jit_enable > 1)
+			pr_info("Pass %d: shrink = %d, seen = 0x%x\n", pass,
+				proglen - (cgctx.idx * 4), cgctx.seen);
+	}
+
+	if (bpf_jit_enable > 1)
+		/*
+		 * Note that we output the base address of the code_base
+		 * rather than image, since opcodes are in code_base.
+		 */
+		bpf_jit_dump(flen, proglen, pass, code_base);
+
+	if (image) {
+		bpf_flush_icache(bpf_hdr, image + alloclen);
+#ifdef PPC64_ELF_ABI_v1
+		/* Function descriptor nastiness: Address + TOC */
+		((u64 *)image)[0] = (u64)code_base;
+		((u64 *)image)[1] = local_paca->kernel_toc;
+#endif
+		fp->bpf_func = (void *)image;
+		fp->jited = 1;
+	}
+
+out:
+	kfree(addrs);
+	return fp;
+}
+
+void bpf_jit_free(struct bpf_prog *fp)
+{
+	unsigned long addr = (unsigned long)fp->bpf_func & PAGE_MASK;
+	struct bpf_binary_header *bpf_hdr = (void *)addr;
+
+	if (fp->jited)
+		bpf_jit_binary_free(bpf_hdr);
+
+	bpf_prog_unlock_free(fp);
+}