uprobes/x86: Remove incorrect WARN_ON() in uprobe_init_insn()
authorOleg Nesterov <oleg@redhat.com>
Fri, 18 May 2018 16:27:39 +0000 (18:27 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Tue, 17 Jul 2018 09:37:54 +0000 (11:37 +0200)
commit 90718e32e1dcc2479acfa208ccfc6442850b594c upstream.

insn_get_length() has the side-effect of processing the entire instruction
but only if it was decoded successfully, otherwise insn_complete() can fail
and in this case we need to just return an error without warning.

Reported-by: syzbot+30d675e3ca03c1c351e7@syzkaller.appspotmail.com
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: syzkaller-bugs@googlegroups.com
Link: https://lkml.kernel.org/lkml/20180518162739.GA5559@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/x86/kernel/uprobes.c

index 495c776..e78a6b1 100644 (file)
@@ -290,7 +290,7 @@ static int uprobe_init_insn(struct arch_uprobe *auprobe, struct insn *insn, bool
        insn_init(insn, auprobe->insn, sizeof(auprobe->insn), x86_64);
        /* has the side-effect of processing the entire instruction */
        insn_get_length(insn);
-       if (WARN_ON_ONCE(!insn_complete(insn)))
+       if (!insn_complete(insn))
                return -ENOEXEC;
 
        if (is_prefix_bad(insn))