add --state to offcputime

man/man8/offcputime.8

@@ -2,7 +2,7 @@
 .SH NAME
 offcputime \- Summarize off-CPU time by kernel stack trace. Uses Linux eBPF/bcc.
 .SH SYNOPSIS
-.B offcputime [\-h] [\-u] [\-p PID] [\-v] [\-f] [duration]
+.B offcputime [\-h] [\-p PID | \-t TID | \-u | \-k] [\-U | \-K] [\-d] [\-f] [\-\-stack\-storage\-size STACK_STORAGE_SIZE] [\-m MIN_BLOCK_TIME] [\-M MAX_BLOCK_TIME] [\-\-state STATE] [duration]
 .SH DESCRIPTION
 This program shows stack traces and task names that were blocked and "off-CPU",
 and the total duration they were not running: their "off-CPU time".
@@ -64,6 +64,16 @@ Output folded format.
 \-\-stack-storage-size STACK_STORAGE_SIZE
 Change the number of unique stack traces that can be stored and displayed.
 .TP
+\-m MIN_BLOCK_TIME
+The minimum time in microseconds over which we store traces (default 1)
+.TP
+\-M MAX_BLOCK_TIME
+The maximum time in microseconds under which we store traces (default U64_MAX)
+.TP
+\-\-state
+Filter on this thread state bitmask (eg, 2 == TASK_UNINTERRUPTIBLE).
+See include/linux/sched.h for states.
+.TP
 duration
 Duration to trace, in seconds.
 .SH EXAMPLES

tools/offcputime.py

@@ -88,11 +88,16 @@ parser.add_argument("-M", "--max-block-time", default=(1 << 64) - 1,
     type=positive_nonzero_int,
     help="the amount of time in microseconds under which we " +
          "store traces (default U64_MAX)")
+parser.add_argument("--state", default=999,
+    type=positive_int,
+    help="filter on this thread state bitmask (eg, 2 == TASK_UNINTERRUPTIBLE" +
+         ") see include/linux/sched.h")
 args = parser.parse_args()
 if args.pid and args.tgid:
     parser.error("specify only one of -p and -t")
 folded = args.folded
 duration = int(args.duration)
+debug = 0
 
 # signal handler
 def signal_ignore(signal, frame):
@@ -123,7 +128,7 @@ int oncpu(struct pt_regs *ctx, struct task_struct *prev) {
     u64 ts, *tsp;
 
     // record previous thread sleep time
-    if (THREAD_FILTER) {
+    if ((THREAD_FILTER) && (STATE_FILTER)) {
         ts = bpf_ktime_get_ns();
         start.update(&pid, &ts);
     }
@@ -177,7 +182,15 @@ elif args.kernel_threads_only:
 else:
     thread_context = "all threads"
     thread_filter = '1'
+if args.state == 0:
+    state_filter = 'prev->state == 0'
+elif args.state != 999:
+    # these states are sometimes bitmask checked
+    state_filter = 'prev->state & %d' % args.state
+else:
+    state_filter = '1'
 bpf_text = bpf_text.replace('THREAD_FILTER', thread_filter)
+bpf_text = bpf_text.replace('STATE_FILTER', state_filter)
 
 # set stack storage size
 bpf_text = bpf_text.replace('STACK_STORAGE_SIZE', str(args.stack_storage_size))
@@ -210,6 +223,9 @@ if args.kernel_threads_only and args.user_stacks_only:
           "doesn't make sense.", file=stderr)
     exit(1)
 
+if (debug):
+    print(bpf_text)
+
 # initialize BPF
 b = BPF(text=bpf_text)
 b.attach_kprobe(event="finish_task_switch", fn_name="oncpu")

tools/offcputime_example.txt

@@ -718,9 +718,10 @@ creating your "off-CPU time flame graphs".
 
 USAGE message:
 
-# ./offcputime -h
+# ./offcputime.py -h
 usage: offcputime.py [-h] [-p PID | -t TID | -u | -k] [-U | -K] [-d] [-f]
                      [--stack-storage-size STACK_STORAGE_SIZE]
+                     [-m MIN_BLOCK_TIME] [-M MAX_BLOCK_TIME] [--state STATE]
                      [duration]
 
 Summarize off-CPU time by stack trace
@@ -747,11 +748,21 @@ optional arguments:
   --stack-storage-size STACK_STORAGE_SIZE
                         the number of unique stack traces that can be stored
                         and displayed (default 1024)
+  -m MIN_BLOCK_TIME, --min-block-time MIN_BLOCK_TIME
+                        the amount of time in microseconds over which we store
+                        traces (default 1)
+  -M MAX_BLOCK_TIME, --max-block-time MAX_BLOCK_TIME
+                        the amount of time in microseconds under which we
+                        store traces (default U64_MAX)
+  --state STATE         filter on this thread state bitmask (eg, 2 ==
+                        TASK_UNINTERRUPTIBLE) see include/linux/sched.h
 
 examples:
     ./offcputime             # trace off-CPU stack time until Ctrl-C
     ./offcputime 5           # trace for 5 seconds only
     ./offcputime -f 5        # 5 seconds, and output in folded format
+    ./offcputime -m 1000     # trace only events that last more than 1000 usec
+    ./offcputime -M 10000    # trace only events that last less than 10000 usec
     ./offcputime -p 185      # only trace threads for PID 185
     ./offcputime -t 188      # only trace thread 188
     ./offcputime -u          # only trace user threads (no kernel)