diff options
| -rw-r--r-- | tools/perf/Documentation/perf-intel-pt.txt | 12 | ||||
| -rw-r--r-- | tools/perf/scripts/python/intel-pt-events.py | 65 |
2 files changed, 74 insertions, 3 deletions
diff --git a/tools/perf/Documentation/perf-intel-pt.txt b/tools/perf/Documentation/perf-intel-pt.txt index 92464a5d7eaf..7b6ccd2fa3bf 100644 --- a/tools/perf/Documentation/perf-intel-pt.txt +++ b/tools/perf/Documentation/perf-intel-pt.txt @@ -189,8 +189,16 @@ There is also script intel-pt-events.py which provides an example of how to unpack the raw data for power events and PTWRITE. The script also displays branches, and supports 2 additional modes selected by option: - --insn-trace - instruction trace - --src-trace - source trace + - --insn-trace - instruction trace + - --src-trace - source trace + +The intel-pt-events.py script also has options: + + - --all-switch-events - display all switch events, not only the last consecutive. + - --interleave [<n>] - interleave sample output for the same timestamp so that + no more than n samples for a CPU are displayed in a row. 'n' defaults to 4. + Note this only affects the order of output, and only when the timestamp is the + same. As mentioned above, it is easy to capture too much data. One way to limit the data captured is to use 'snapshot' mode which is explained further below. diff --git a/tools/perf/scripts/python/intel-pt-events.py b/tools/perf/scripts/python/intel-pt-events.py index 6be7fd8fd615..08862a2582f4 100644 --- a/tools/perf/scripts/python/intel-pt-events.py +++ b/tools/perf/scripts/python/intel-pt-events.py @@ -13,10 +13,12 @@ from __future__ import print_function +import io import os import sys import struct import argparse +import contextlib from libxed import LibXED from ctypes import create_string_buffer, addressof @@ -39,6 +41,11 @@ glb_src = False glb_source_file_name = None glb_line_number = None glb_dso = None +glb_stash_dict = {} +glb_output = None +glb_output_pos = 0 +glb_cpu = -1 +glb_time = 0 def get_optional_null(perf_dict, field): if field in perf_dict: @@ -70,6 +77,7 @@ def trace_begin(): ap.add_argument("--insn-trace", action='store_true') ap.add_argument("--src-trace", action='store_true') ap.add_argument("--all-switch-events", action='store_true') + ap.add_argument("--interleave", type=int, nargs='?', const=4, default=0) global glb_args global glb_insn global glb_src @@ -94,11 +102,39 @@ def trace_begin(): perf_set_itrace_options(perf_script_context, itrace) def trace_end(): + if glb_args.interleave: + flush_stashed_output() print("End") def trace_unhandled(event_name, context, event_fields_dict): print(' '.join(['%s=%s'%(k,str(v))for k,v in sorted(event_fields_dict.items())])) +def stash_output(): + global glb_stash_dict + global glb_output_pos + output_str = glb_output.getvalue()[glb_output_pos:] + n = len(output_str) + if n: + glb_output_pos += n + if glb_cpu not in glb_stash_dict: + glb_stash_dict[glb_cpu] = [] + glb_stash_dict[glb_cpu].append(output_str) + +def flush_stashed_output(): + global glb_stash_dict + while glb_stash_dict: + cpus = list(glb_stash_dict.keys()) + # Output at most glb_args.interleave output strings per cpu + for cpu in cpus: + items = glb_stash_dict[cpu] + countdown = glb_args.interleave + while len(items) and countdown: + sys.stdout.write(items[0]) + del items[0] + countdown -= 1 + if not items: + del glb_stash_dict[cpu] + def print_ptwrite(raw_buf): data = struct.unpack_from("<IQ", raw_buf) flags = data[0] @@ -375,15 +411,40 @@ def do_process_event(param_dict): print_common_start(comm, sample, name) print_common_ip(param_dict, sample, symbol, dso) +def interleave_events(param_dict): + global glb_cpu + global glb_time + global glb_output + global glb_output_pos + + sample = param_dict["sample"] + glb_cpu = sample["cpu"] + ts = sample["time"] + + if glb_time != ts: + glb_time = ts + flush_stashed_output() + + glb_output_pos = 0 + with contextlib.redirect_stdout(io.StringIO()) as glb_output: + do_process_event(param_dict) + + stash_output() + def process_event(param_dict): try: - do_process_event(param_dict) + if glb_args.interleave: + interleave_events(param_dict) + else: + do_process_event(param_dict) except broken_pipe_exception: # Stop python printing broken pipe errors and traceback sys.stdout = open(os.devnull, 'w') sys.exit(1) def auxtrace_error(typ, code, cpu, pid, tid, ip, ts, msg, cpumode, *x): + if glb_args.interleave: + flush_stashed_output() if len(x) >= 2 and x[0]: machine_pid = x[0] vcpu = x[1] @@ -403,6 +464,8 @@ def auxtrace_error(typ, code, cpu, pid, tid, ip, ts, msg, cpumode, *x): sys.exit(1) def context_switch(ts, cpu, pid, tid, np_pid, np_tid, machine_pid, out, out_preempt, *x): + if glb_args.interleave: + flush_stashed_output() if out: out_str = "Switch out " else: |