On 6/26/13 11:23 PM, Namhyung Kim wrote:
On Wed, 26 Jun 2013 11:12:19 -0600, David Ahern wrote:
On 6/26/13 1:14 AM, Namhyung Kim wrote:
@@ -196,6 +222,8 @@ int cmd_ftrace(int argc, const char **argv, const char
*prefix __maybe_unused)
const struct option ftrace_options[] = {
OPT_STRING('t', "tracer", &ftrace.tracer, "tracer",
"tracer to use"),
+ OPT_STRING('p', "pid", &ftrace.target.tid, "pid",
+ "trace on existing process id"),
OPT_INCR('v', "verbose", &verbose,
"be more verbose"),
OPT_END()
You are calling it pid but assigning it as a tid which is inconsistent
with other perf commands. e.g., perf-record allows a list of pids (-p)
or tids (-t). Why not support that in perf-ftrace? And that leads to
the comment about consistency of options across perf commands: -t is
used here for tracer type to use.
Sorry, I should've mentioned it.
The pid filtering in ftrace is done via set_ftrace_pid file under the
tracing debugfs directory. IIRC It only supports process filtering not
thread filtering by iterating all threads in the kernel code. So --tid
option cannot be implemented as other perf commands.
So I chose not to and assigned -t option to --tracer.
Could that ever change? With the current code why call it pid in the
option but assign it to target.tid? Seems a like a source for confusion
later. Would be better to just assign to target.pid and let the
machinery do the right thing.
David
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
