perf annotate browser: Handle NULL jump targets
authorArnaldo Carvalho de Melo <acme@redhat.com>
Wed, 25 Apr 2012 10:48:42 +0000 (07:48 -0300)
committerArnaldo Carvalho de Melo <acme@redhat.com>
Wed, 25 Apr 2012 10:48:42 +0000 (07:48 -0300)
In annotate_browser__mark_jump_targets

702                     dlt = browser->offsets[dl->ops.target];
703                     bdlt = disasm_line__browser(dlt);
704                     bdlt->jump_target = true;
705             }
706
707     }

(gdb) p size
$5 = 2415
(gdb) p offset
$6 = 140
(gdb) p dl->ops.target
$7 = 143
(gdb) p browser->offsets[143]
$8 = (struct disasm_line *) 0x0
(gdb) p dl->name
$9 = 0x2363bd0 "je"
(gdb)

Really strange, the code assumed that at the jump target we would have
an assembly line, but only in the previous instruction offset we have a
'lock':

(gdb) p browser->offsets[144]
$10 = (struct disasm_line *) 0x0
(gdb) p browser->offsets[142]
$11 = (struct disasm_line *) 0x27bd620
(gdb) p browser->offsets[142]->name
$12 = 0x237a8a0 "lock"
(gdb)

I'll study this more, but for now I'll just check if there is a
disasm_line at dl->ops.target, i.e. a valid jump target.

Reported-by: Hagen Paul Pfeifer <hagen@jauu.net>
Reported-by: Ingo Molnar <mingo@kernel.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-inzjrzyqhkzyv78met2vula6@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
tools/perf/ui/browsers/annotate.c

index 9e3310cd02cd63c2a39595f411d2f0c08b123481..4c83fe3d7dad7ab57b0f6acaa964e52564a8ed24 100644 (file)
@@ -700,6 +700,13 @@ static void annotate_browser__mark_jump_targets(struct annotate_browser *browser
                }
 
                dlt = browser->offsets[dl->ops.target];
+               /*
+                * FIXME: Oops, no jump target? Buggy disassembler? Or do we
+                * have to adjust to the previous offset?
+                */
+               if (dlt == NULL)
+                       continue;
+
                bdlt = disasm_line__browser(dlt);
                bdlt->jump_target = true;
        }