tracing: Fix user stack trace "??" output
authorEiichi Tsukata <devel@etsukata.com>
Sun, 30 Jun 2019 08:54:38 +0000 (17:54 +0900)
committerSteven Rostedt (VMware) <rostedt@goodmis.org>
Fri, 19 Jul 2019 16:12:39 +0000 (12:12 -0400)
Commit c5c27a0a5838 ("x86/stacktrace: Remove the pointless ULONG_MAX
marker") removes ULONG_MAX marker from user stack trace entries but
trace_user_stack_print() still uses the marker and it outputs unnecessary
"??".

For example:

            less-1911  [001] d..2    34.758944: <user stack trace>
   =>  <00007f16f2295910>
   => ??
   => ??
   => ??
   => ??
   => ??
   => ??
   => ??

The user stack trace code zeroes the storage before saving the stack, so if
the trace is shorter than the maximum number of entries it can terminate
the print loop if a zero entry is detected.

Link: http://lkml.kernel.org/r/20190630085438.25545-1-devel@etsukata.com
Cc: stable@vger.kernel.org
Fixes: 4285f2fcef80 ("tracing: Remove the ULONG_MAX stack trace hackery")
Signed-off-by: Eiichi Tsukata <devel@etsukata.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
kernel/trace/trace_output.c

index 54373d93e251a1c01aa85ec918024cd3d5e9bca8..1d6178a188f471cec3ed49995866d912bf0c60e5 100644 (file)
@@ -1109,17 +1109,10 @@ static enum print_line_t trace_user_stack_print(struct trace_iterator *iter,
        for (i = 0; i < FTRACE_STACK_ENTRIES; i++) {
                unsigned long ip = field->caller[i];
 
-               if (ip == ULONG_MAX || trace_seq_has_overflowed(s))
+               if (!ip || trace_seq_has_overflowed(s))
                        break;
 
                trace_seq_puts(s, " => ");
-
-               if (!ip) {
-                       trace_seq_puts(s, "??");
-                       trace_seq_putc(s, '\n');
-                       continue;
-               }
-
                seq_print_user_ip(s, mm, ip, flags);
                trace_seq_putc(s, '\n');
        }