[PATCH v2 5/6] x86/alternative: Relax text_poke_bp() constraint

[Peter Zijlstra <peterz@infradead.org>]

Currently text_poke_bp() is very strict to only allow patching a
single instruction; however with straight-line-speculation it will be
required to patch: ret; int3, which is two instructions.

As such, relax the constraints a little to allow int3 padding for all
instructions that do not imply the execution of the next instruction,
ie: RET, JMP.d8 and JMP.d32.

While there, rename the text_poke_loc::rel32 field to ::disp.

Note: this fills up the text_poke_loc structure which is now a round
  16 bytes big.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
 arch/x86/kernel/alternative.c |   44 ++++++++++++++++++++++++++++--------------
 1 file changed, 30 insertions(+), 14 deletions(-)

--- a/arch/x86/kernel/alternative.c
+++ b/arch/x86/kernel/alternative.c
@@ -1114,10 +1114,11 @@ void text_poke_sync(void)
 
 struct text_poke_loc {
 	s32 rel_addr; /* addr := _stext + rel_addr */
-	s32 rel32;
+	s32 disp;
+	u8 len;
 	u8 opcode;
 	const u8 text[POKE_MAX_OPCODE_SIZE];
-	u8 old;
+	u8 old; /* see text_poke_bp_batch() */
 };
 
 struct bp_patching_desc {
@@ -1165,7 +1166,7 @@ noinstr int poke_int3_handler(struct pt_
 {
 	struct bp_patching_desc *desc;
 	struct text_poke_loc *tp;
-	int len, ret = 0;
+	int ret = 0;
 	void *ip;
 
 	if (user_mode(regs))
@@ -1205,8 +1206,7 @@ noinstr int poke_int3_handler(struct pt_
 			goto out_put;
 	}
 
-	len = text_opcode_size(tp->opcode);
-	ip += len;
+	ip += tp->len;
 
 	switch (tp->opcode) {
 	case INT3_INSN_OPCODE:
@@ -1221,12 +1221,12 @@ noinstr int poke_int3_handler(struct pt_
 		break;
 
 	case CALL_INSN_OPCODE:
-		int3_emulate_call(regs, (long)ip + tp->rel32);
+		int3_emulate_call(regs, (long)ip + tp->disp);
 		break;
 
 	case JMP32_INSN_OPCODE:
 	case JMP8_INSN_OPCODE:
-		int3_emulate_jmp(regs, (long)ip + tp->rel32);
+		int3_emulate_jmp(regs, (long)ip + tp->disp);
 		break;
 
 	default:
@@ -1301,7 +1301,7 @@ static void text_poke_bp_batch(struct te
 	 */
 	for (do_sync = 0, i = 0; i < nr_entries; i++) {
 		u8 old[POKE_MAX_OPCODE_SIZE] = { tp[i].old, };
-		int len = text_opcode_size(tp[i].opcode);
+		int len = tp[i].len;
 
 		if (len - INT3_INSN_SIZE > 0) {
 			memcpy(old + INT3_INSN_SIZE,
@@ -1378,20 +1378,36 @@ static void text_poke_loc_init(struct te
 			       const void *opcode, size_t len, const void *emulate)
 {
 	struct insn insn;
-	int ret;
+	int ret, i;
 
 	memcpy((void *)tp->text, opcode, len);
 	if (!emulate)
 		emulate = opcode;
 
 	ret = insn_decode_kernel(&insn, emulate);
-
 	BUG_ON(ret < 0);
-	BUG_ON(len != insn.length);
 
 	tp->rel_addr = addr - (void *)_stext;
+	tp->len = len;
 	tp->opcode = insn.opcode.bytes[0];
 
+	switch(tp->opcode) {
+	case RET_INSN_OPCODE:
+	case JMP32_INSN_OPCODE:
+	case JMP8_INSN_OPCODE:
+		/*
+		 * Control flow instructions without implied execution of the
+		 * next instruction can be padded with INT3.
+		 */
+		for (i = insn.length; i < len; i++)
+			BUG_ON(tp->text[i] != INT3_INSN_OPCODE);
+		break;
+
+	default:
+		BUG_ON(len != insn.length);
+	};
+
+
 	switch (tp->opcode) {
 	case INT3_INSN_OPCODE:
 	case RET_INSN_OPCODE:
@@ -1400,7 +1416,7 @@ static void text_poke_loc_init(struct te
 	case CALL_INSN_OPCODE:
 	case JMP32_INSN_OPCODE:
 	case JMP8_INSN_OPCODE:
-		tp->rel32 = insn.immediate.value;
+		tp->disp = insn.immediate.value;
 		break;
 
 	default: /* assume NOP */
@@ -1408,13 +1424,13 @@ static void text_poke_loc_init(struct te
 		case 2: /* NOP2 -- emulate as JMP8+0 */
 			BUG_ON(memcmp(emulate, x86_nops[len], len));
 			tp->opcode = JMP8_INSN_OPCODE;
-			tp->rel32 = 0;
+			tp->disp = 0;
 			break;
 
 		case 5: /* NOP5 -- emulate as JMP32+0 */
 			BUG_ON(memcmp(emulate, x86_nops[len], len));
 			tp->opcode = JMP32_INSN_OPCODE;
-			tp->rel32 = 0;
+			tp->disp = 0;
 			break;
 
 		default: /* unknown instruction */