x86/kprobes: Do not decode opcode in resume_execution()

Currently, kprobes decodes the opcode right after single-stepping in
resume_execution(). But the opcode was already decoded while preparing
arch_specific_insn in arch_copy_kprobe().

Decode the opcode in arch_copy_kprobe() instead of in resume_execution()
and set some flags which classify the opcode for the resuming process.

 [ bp: Massage commit message. ]

Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Link: https://lkml.kernel.org/r/160830072561.349576.3014979564448023213.stgit@devnote2

1 files changed, 7 insertions, 4 deletions

diff --git a/arch/x86/include/asm/kprobes.h b/arch/x86/include/asm/kprobes.h
index 991a7ad540c7..d20a3d6be36e 100644
--- a/arch/x86/include/asm/kprobes.h
+++ b/arch/x86/include/asm/kprobes.h
@@ -58,14 +58,17 @@ struct arch_specific_insn {
 	/* copy of the original instruction */
 	kprobe_opcode_t *insn;
 	/*
-	 * boostable = false: This instruction type is not boostable.
-	 * boostable = true: This instruction has been boosted: we have
+	 * boostable = 0: This instruction type is not boostable.
+	 * boostable = 1: This instruction has been boosted: we have
 	 * added a relative jump after the instruction copy in insn,
 	 * so no single-step and fixup are needed (unless there's
 	 * a post_handler).
 	 */
-	bool boostable;
-	bool if_modifier;
+	unsigned boostable:1;
+	unsigned if_modifier:1;
+	unsigned is_call:1;
+	unsigned is_pushf:1;
+	unsigned is_abs_ip:1;
 	/* Number of bytes of text poked */
 	int tp_len;
 };