[PATCH 5/5] perf/hw_breakpoint: Add fallback code for ptrace_set_breakpoint_addr

[Jiri Olsa]

Restoring the breakpoint after unsuccesfull address change,
so following user code no longer produces disabled breakpoint.

  ptrace(PTRACE_POKEUSER, child, offsetof(struct user, u_debugreg[0]), addr_1)
  ptrace(PTRACE_POKEUSER, child, offsetof(struct user, u_debugreg[7]), dr7)
  ptrace(PTRACE_POKEUSER, child, offsetof(struct user, u_debugreg[0]), -1)

The first 2 ptrace calls set breakpoint on addr_1. The 3rd ptrace
call tries to set it to bogus address (-1). This would normaly
end up with disabled breakpoint. This patch adds the code that
restores the breakpoint to its original state.

Link: http://lkml.kernel.org/n/tip-h9ut835vl297roen0v163zg6@xxxxxxxxxxxxxx
Signed-off-by: Jiri Olsa <jolsa@xxxxxxxxxx>
---
 arch/x86/kernel/ptrace.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index e2ee403865eb..22c06d0a38d1 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -693,9 +693,14 @@ static int ptrace_set_breakpoint_addr(struct task_struct *tsk, int nr,
 			t->ptrace_bps[nr] = bp;
 	} else {
 		struct perf_event_attr attr = bp->attr;
+		bool disabled = attr.disabled;
 
 		attr.bp_addr = addr;
 		err = modify_user_hw_breakpoint(bp, &attr);
+		if (err && !disabled) {
+			bp->attr.disabled = false;
+			WARN_ON(modify_user_hw_breakpoint(bp, &bp->attr));
+		}
 	}
 
 	return err;
-- 
2.17.1