bpf: x86: add missing 'shift by register' instructions to x64 eBPF JIT

'shift by register' operations are supported by eBPF interpreter, but were
accidently left out of x64 JIT compiler. Fix it and add a testcase.

Reported-by: Brendan Gregg <brendan.d.gregg@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@plumgrid.com>
Fixes: 622582786c ("net: filter: x86: internal BPF JIT")
Signed-off-by: David S. Miller <davem@davemloft.net>

arch/x86/net/bpf_jit_comp.c

@@ -515,6 +515,48 @@ static int do_jit(struct bpf_prog *bpf_prog, int *addrs, u8 *image,
 			EMIT3(0xC1, add_1reg(b3, dst_reg), imm32);
 			break;
 
+		case BPF_ALU | BPF_LSH | BPF_X:
+		case BPF_ALU | BPF_RSH | BPF_X:
+		case BPF_ALU | BPF_ARSH | BPF_X:
+		case BPF_ALU64 | BPF_LSH | BPF_X:
+		case BPF_ALU64 | BPF_RSH | BPF_X:
+		case BPF_ALU64 | BPF_ARSH | BPF_X:
+
+			/* check for bad case when dst_reg == rcx */
+			if (dst_reg == BPF_REG_4) {
+				/* mov r11, dst_reg */
+				EMIT_mov(AUX_REG, dst_reg);
+				dst_reg = AUX_REG;
+			}
+
+			if (src_reg != BPF_REG_4) { /* common case */
+				EMIT1(0x51); /* push rcx */
+
+				/* mov rcx, src_reg */
+				EMIT_mov(BPF_REG_4, src_reg);
+			}
+
+			/* shl %rax, %cl | shr %rax, %cl | sar %rax, %cl */
+			if (BPF_CLASS(insn->code) == BPF_ALU64)
+				EMIT1(add_1mod(0x48, dst_reg));
+			else if (is_ereg(dst_reg))
+				EMIT1(add_1mod(0x40, dst_reg));
+
+			switch (BPF_OP(insn->code)) {
+			case BPF_LSH: b3 = 0xE0; break;
+			case BPF_RSH: b3 = 0xE8; break;
+			case BPF_ARSH: b3 = 0xF8; break;
+			}
+			EMIT2(0xD3, add_1reg(b3, dst_reg));
+
+			if (src_reg != BPF_REG_4)
+				EMIT1(0x59); /* pop rcx */
+
+			if (insn->dst_reg == BPF_REG_4)
+				/* mov dst_reg, r11 */
+				EMIT_mov(insn->dst_reg, AUX_REG);
+			break;
+
 		case BPF_ALU | BPF_END | BPF_FROM_BE:
 			switch (imm32) {
 			case 16:

lib/test_bpf.c

@@ -1341,6 +1341,44 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, -1 } }
 	},
+	{
+		"INT: shifts by register",
+		.u.insns_int = {
+			BPF_MOV64_IMM(R0, -1234),
+			BPF_MOV64_IMM(R1, 1),
+			BPF_ALU32_REG(BPF_RSH, R0, R1),
+			BPF_JMP_IMM(BPF_JEQ, R0, 0x7ffffd97, 1),
+			BPF_EXIT_INSN(),
+			BPF_MOV64_IMM(R2, 1),
+			BPF_ALU64_REG(BPF_LSH, R0, R2),
+			BPF_MOV32_IMM(R4, -1234),
+			BPF_JMP_REG(BPF_JEQ, R0, R4, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU64_IMM(BPF_AND, R4, 63),
+			BPF_ALU64_REG(BPF_LSH, R0, R4), /* R0 <= 46 */
+			BPF_MOV64_IMM(R3, 47),
+			BPF_ALU64_REG(BPF_ARSH, R0, R3),
+			BPF_JMP_IMM(BPF_JEQ, R0, -617, 1),
+			BPF_EXIT_INSN(),
+			BPF_MOV64_IMM(R2, 1),
+			BPF_ALU64_REG(BPF_LSH, R4, R2), /* R4 = 46 << 1 */
+			BPF_JMP_IMM(BPF_JEQ, R4, 92, 1),
+			BPF_EXIT_INSN(),
+			BPF_MOV64_IMM(R4, 4),
+			BPF_ALU64_REG(BPF_LSH, R4, R4), /* R4 = 4 << 4 */
+			BPF_JMP_IMM(BPF_JEQ, R4, 64, 1),
+			BPF_EXIT_INSN(),
+			BPF_MOV64_IMM(R4, 5),
+			BPF_ALU32_REG(BPF_LSH, R4, R4), /* R4 = 5 << 5 */
+			BPF_JMP_IMM(BPF_JEQ, R4, 160, 1),
+			BPF_EXIT_INSN(),
+			BPF_MOV64_IMM(R0, -1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, -1 } }
+	},
 	{
 		"INT: DIV + ABS",
 		.u.insns_int = {