Merge pull request iovisor#1699 from palmtenor/profile-sample

Add -c option for profile.py

man/man8/profile.8

@@ -2,7 +2,7 @@
 .SH NAME
 profile \- Profile CPU usage by sampling stack traces. Uses Linux eBPF/bcc.
 .SH SYNOPSIS
-.B profile [\-adfh] [\-p PID] [\-U | \-k] [\-F FREQUENCY]
+.B profile [\-adfh] [\-p PID] [\-U | \-K] [\-F FREQUENCY | \-c COUNT]
 .B [\-\-stack\-storage\-size COUNT] [duration]
 .SH DESCRIPTION
 This is a CPU profiler. It works by taking samples of stack traces at timed
@@ -32,7 +32,10 @@ Trace this process ID only (filtered in-kernel). Without this, all CPUs are
 profiled.
 .TP
 \-F frequency
-Frequency to sample stacks (default 49).
+Frequency to sample stacks.
+.TP
+\-c count
+Sample stacks every one in this many events.
 .TP
 \-f
 Print output in folded stack format.
@@ -67,6 +70,10 @@ Profile at 99 Hertz for 5 seconds only:
 #
 .B profile -F 99 5
 .TP
+Profile 1 in a million events for 5 seconds only:
+#
+.B profile -c 1000000 5
+.TP
 Profile PID 181 only:
 #
 .B profile -p 181

tools/profile.py

@@ -63,6 +63,7 @@ def positive_nonzero_int(val):
 examples = """examples:
  ./profile # profile stack traces at 49 Hertz until Ctrl-C
  ./profile -F 99 # profile stack traces at 99 Hertz
+ ./profile -c 1000000 # profile stack traces every 1 in a million events
  ./profile 5 # profile at 49 Hertz for 5 seconds only
  ./profile -f 5 # output in folded format for flame graphs
  ./profile -p 185 # only profile threads for PID 185
@@ -82,8 +83,11 @@ def positive_nonzero_int(val):
  help="show stacks from user space only (no kernel space stacks)")
 stack_group.add_argument("-K", "--kernel-stacks-only", action="store_true",
  help="show stacks from kernel space only (no user space stacks)")
-parser.add_argument("-F", "--frequency", type=positive_int, default=49,
- help="sample frequency, Hertz (default 49)")
+sample_group = parser.add_mutually_exclusive_group()
+sample_group.add_argument("-F", "--frequency", type=positive_int,
+ help="sample frequency, Hertz")
+sample_group.add_argument("-c", "--count", type=positive_int,
+ help="sample period, number of events")
 parser.add_argument("-d", "--delimited", action="store_true",
  help="insert delimiter between kernel/user stacks")
 parser.add_argument("-a", "--annotations", action="store_true",
@@ -214,10 +218,22 @@ def positive_nonzero_int(val):
 bpf_text = bpf_text.replace('USER_STACK_GET', user_stack_get)
 bpf_text = bpf_text.replace('KERNEL_STACK_GET', kernel_stack_get)
 
+sample_freq = 0
+sample_period = 0
+if args.frequency:
+ sample_freq = args.frequency
+elif args.count:
+ sample_period = args.count
+else:
+ # If user didn't specify anything, use default 49Hz sampling
+ sample_freq = 49
+sample_context = "%s%d %s" % (("", sample_freq, "Hertz") if sample_freq
+ else ("every ", sample_period, "events"))
+
 # header
 if not args.folded:
- print("Sampling at %d Hertz of %s by %s stack" %
- (args.frequency, thread_context, stack_context), end="")
+ print("Sampling at %s of %s by %s stack" %
+ (sample_context, thread_context, stack_context), end="")
  if duration < 99999999:
  print(" for %d secs." % duration)
  else:
@@ -232,7 +248,7 @@ def positive_nonzero_int(val):
 b = BPF(text=bpf_text)
 b.attach_perf_event(ev_type=PerfType.SOFTWARE,
  ev_config=PerfSWConfig.CPU_CLOCK, fn_name="do_perf_event",
- sample_period=0, sample_freq=args.frequency)
+ sample_period=sample_period, sample_freq=sample_freq)
 
 # signal handler
 def signal_ignore(signal, frame):

tools/profile_example.txt

@@ -704,8 +704,8 @@ Run ./profile -h to see the default.
 USAGE message:
 
 # ./profile -h
-usage: profile [-h] [-p PID] [-U | -K] [-F FREQUENCY] [-d] [-a] [-f]
- [--stack-storage-size STACK_STORAGE_SIZE]
+usage: profile [-h] [-p PID] [-U | -K] [-F FREQUENCY | -c COUNT] [-d] [-a]
+ [-f] [--stack-storage-size STACK_STORAGE_SIZE]
  [duration]
 
 Profile CPU stack traces at a timed interval
@@ -723,7 +723,9 @@ optional arguments:
  show stacks from kernel space only (no user space
  stacks)
  -F FREQUENCY, --frequency FREQUENCY
- sample frequency, Hertz (default 49)
+ sample frequency, Hertz
+ -c COUNT, --count COUNT
+ sample period, number of events
  -d, --delimited insert delimiter between kernel/user stacks
  -a, --annotations add _[k] annotations to kernel frames
  -f, --folded output folded format, one line per stack (for flame
@@ -735,6 +737,7 @@ optional arguments:
 examples:
  ./profile # profile stack traces at 49 Hertz until Ctrl-C
  ./profile -F 99 # profile stack traces at 99 Hertz
+ ./profile -c 1000000 # profile stack traces every 1 in a million events
  ./profile 5 # profile at 49 Hertz for 5 seconds only
  ./profile -f 5 # output in folded format for flame graphs
  ./profile -p 185 # only profile threads for PID 185