x86/kprobes: Identify far indirect JMP correctly
authorMasami Hiramatsu <mhiramat@kernel.org>
Tue, 2 Mar 2021 15:25:34 +0000 (00:25 +0900)
committerPeter Zijlstra <peterz@infradead.org>
Tue, 23 Mar 2021 15:07:56 +0000 (16:07 +0100)
Since Grp5 far indirect JMP is FF "mod 101 r/m", it should be
(modrm & 0x38) == 0x28, and near indirect JMP is also 0x38 == 0x20.
So we can mask modrm with 0x30 and check 0x20.
This is actually what the original code does, it also doesn't care
the last bit. So the result code is same.

Thus, I think this is just a cosmetic cleanup.

Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/161469873475.49483.13257083019966335137.stgit@devnote2
arch/x86/kernel/kprobes/core.c

index 9b31790..f6ec57f 100644 (file)
@@ -467,8 +467,7 @@ static void set_resume_flags(struct kprobe *p, struct insn *insn)
                        p->ainsn.is_call = 1;
                        p->ainsn.is_abs_ip = 1;
                        break;
-               } else if (((opcode & 0x31) == 0x20) ||
-                          ((opcode & 0x31) == 0x21)) {
+               } else if ((opcode & 0x30) == 0x20) {
                        /*
                         * jmp near and far, absolute indirect
                         * ip is correct.