oProfile: oops when profile_pc() returns ~0LU

Instruction pointer returned by profile_pc() can be a random value.  This
break the assumption than we can safely set struct op_sample.eip field to a
magic value to signal to the per-cpu buffer reader side special event like
task switch ending up in a segfault in get_task_mm() when profile_pc()
return ~0UL.  Fixed by sanitizing the sampled eip and reject/log invalid
eip.

Problem reported by Sami Farin, patch tested by him.

Signed-off-by: Philippe Elie <phil.el@wanadoo.fr>
Tested-by: Sami Farin <safari-kernel@safari.iki.fi>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

1 files changed, 7 insertions, 0 deletions

diff --git a/drivers/oprofile/cpu_buffer.c b/drivers/oprofile/cpu_buffer.c
index a83c3db7d18f..c93d3d2640ab 100644
--- a/drivers/oprofile/cpu_buffer.c
+++ b/drivers/oprofile/cpu_buffer.c
@@ -64,6 +64,8 @@ int alloc_cpu_buffers(void)
                 b->head_pos = 0;
                 b->sample_received = 0;
                 b->sample_lost_overflow = 0;
+                b->backtrace_aborted = 0;
+                b->sample_invalid_eip = 0;
                 b->cpu = i;
                 INIT_DELAYED_WORK(&b->work, wq_sync_buffer);
         }
@@ -175,6 +177,11 @@ static int log_sample(struct oprofile_cpu_buffer * cpu_buf, unsigned long pc,
 
         cpu_buf->sample_received++;
 
+        if (pc == ESCAPE_CODE) {
+                cpu_buf->sample_invalid_eip++;
+                return 0;
+        }
+
         if (nr_available_slots(cpu_buf) < 3) {
                 cpu_buf->sample_lost_overflow++;
                 return 0;