diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2020-06-01 13:23:59 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2020-06-01 13:23:59 -0700 |
commit | a7092c82042b4ba3000cf7b369d1032161c5d4c9 (patch) | |
tree | 240c3b73ac4b25fcca218871f109383ed95fd52a /tools/perf/scripts/python | |
parent | 69fc06f70f4569c9969f99fe25bdc9a6bb537b43 (diff) | |
parent | 5cde265384cad739b162cf08afba6da8857778bd (diff) |
Merge tag 'perf-core-2020-06-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf updates from Ingo Molnar:
"Kernel side changes:
- Add AMD Fam17h RAPL support
- Introduce CAP_PERFMON to kernel and user space
- Add Zhaoxin CPU support
- Misc fixes and cleanups
Tooling changes:
- perf record:
Introduce '--switch-output-event' to use arbitrary events to be
setup and read from a side band thread and, when they take place a
signal be sent to the main 'perf record' thread, reusing the core
for '--switch-output' to take perf.data snapshots from the ring
buffer used for '--overwrite', e.g.:
# perf record --overwrite -e sched:* \
--switch-output-event syscalls:*connect* \
workload
will take perf.data.YYYYMMDDHHMMSS snapshots up to around the
connect syscalls.
Add '--num-synthesize-threads' option to control degree of
parallelism of the synthesize_mmap() code which is scanning
/proc/PID/task/PID/maps and can be time consuming. This mimics
pre-existing behaviour in 'perf top'.
- perf bench:
Add a multi-threaded synthesize benchmark and kallsyms parsing
benchmark.
- Intel PT support:
Stitch LBR records from multiple samples to get deeper backtraces,
there are caveats, see the csets for details.
Allow using Intel PT to synthesize callchains for regular events.
Add support for synthesizing branch stacks for regular events
(cycles, instructions, etc) from Intel PT data.
Misc changes:
- Updated perf vendor events for power9 and Coresight.
- Add flamegraph.py script via 'perf flamegraph'
- Misc other changes, fixes and cleanups - see the Git log for details
Also, since over the last couple of years perf tooling has matured and
decoupled from the kernel perf changes to a large degree, going
forward Arnaldo is going to send perf tooling changes via direct pull
requests"
* tag 'perf-core-2020-06-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (163 commits)
perf/x86/rapl: Add AMD Fam17h RAPL support
perf/x86/rapl: Make perf_probe_msr() more robust and flexible
perf/x86/rapl: Flip logic on default events visibility
perf/x86/rapl: Refactor to share the RAPL code between Intel and AMD CPUs
perf/x86/rapl: Move RAPL support to common x86 code
perf/core: Replace zero-length array with flexible-array
perf/x86: Replace zero-length array with flexible-array
perf/x86/intel: Add more available bits for OFFCORE_RESPONSE of Intel Tremont
perf/x86/rapl: Add Ice Lake RAPL support
perf flamegraph: Use /bin/bash for report and record scripts
perf cs-etm: Move definition of 'traceid_list' global variable from header file
libsymbols kallsyms: Move hex2u64 out of header
libsymbols kallsyms: Parse using io api
perf bench: Add kallsyms parsing
perf: cs-etm: Update to build with latest opencsd version.
perf symbol: Fix kernel symbol address display
perf inject: Rename perf_evsel__*() operating on 'struct evsel *' to evsel__*()
perf annotate: Rename perf_evsel__*() operating on 'struct evsel *' to evsel__*()
perf trace: Rename perf_evsel__*() operating on 'struct evsel *' to evsel__*()
perf script: Rename perf_evsel__*() operating on 'struct evsel *' to evsel__*()
...
Diffstat (limited to 'tools/perf/scripts/python')
-rwxr-xr-x | tools/perf/scripts/python/bin/flamegraph-record | 2 | ||||
-rwxr-xr-x | tools/perf/scripts/python/bin/flamegraph-report | 3 | ||||
-rwxr-xr-x | tools/perf/scripts/python/flamegraph.py | 124 |
3 files changed, 129 insertions, 0 deletions
diff --git a/tools/perf/scripts/python/bin/flamegraph-record b/tools/perf/scripts/python/bin/flamegraph-record new file mode 100755 index 000000000000..7df5a19c0163 --- /dev/null +++ b/tools/perf/scripts/python/bin/flamegraph-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -g "$@" diff --git a/tools/perf/scripts/python/bin/flamegraph-report b/tools/perf/scripts/python/bin/flamegraph-report new file mode 100755 index 000000000000..53c5dc90c87e --- /dev/null +++ b/tools/perf/scripts/python/bin/flamegraph-report @@ -0,0 +1,3 @@ +#!/bin/bash +# description: create flame graphs +perf script -s "$PERF_EXEC_PATH"/scripts/python/flamegraph.py -- "$@" diff --git a/tools/perf/scripts/python/flamegraph.py b/tools/perf/scripts/python/flamegraph.py new file mode 100755 index 000000000000..61f3be9add6b --- /dev/null +++ b/tools/perf/scripts/python/flamegraph.py @@ -0,0 +1,124 @@ +# flamegraph.py - create flame graphs from perf samples +# SPDX-License-Identifier: GPL-2.0 +# +# Usage: +# +# perf record -a -g -F 99 sleep 60 +# perf script report flamegraph +# +# Combined: +# +# perf script flamegraph -a -F 99 sleep 60 +# +# Written by Andreas Gerstmayr <agerstmayr@redhat.com> +# Flame Graphs invented by Brendan Gregg <bgregg@netflix.com> +# Works in tandem with d3-flame-graph by Martin Spier <mspier@netflix.com> + +from __future__ import print_function +import sys +import os +import argparse +import json + + +class Node: + def __init__(self, name, libtype=""): + self.name = name + self.libtype = libtype + self.value = 0 + self.children = [] + + def toJSON(self): + return { + "n": self.name, + "l": self.libtype, + "v": self.value, + "c": self.children + } + + +class FlameGraphCLI: + def __init__(self, args): + self.args = args + self.stack = Node("root") + + if self.args.format == "html" and \ + not os.path.isfile(self.args.template): + print("Flame Graph template {} does not exist. Please install " + "the js-d3-flame-graph (RPM) or libjs-d3-flame-graph (deb) " + "package, specify an existing flame graph template " + "(--template PATH) or another output format " + "(--format FORMAT).".format(self.args.template), + file=sys.stderr) + sys.exit(1) + + def find_or_create_node(self, node, name, dso): + libtype = "kernel" if dso == "[kernel.kallsyms]" else "" + if name is None: + name = "[unknown]" + + for child in node.children: + if child.name == name and child.libtype == libtype: + return child + + child = Node(name, libtype) + node.children.append(child) + return child + + def process_event(self, event): + node = self.find_or_create_node(self.stack, event["comm"], None) + if "callchain" in event: + for entry in reversed(event['callchain']): + node = self.find_or_create_node( + node, entry.get("sym", {}).get("name"), event.get("dso")) + else: + node = self.find_or_create_node( + node, entry.get("symbol"), event.get("dso")) + node.value += 1 + + def trace_end(self): + json_str = json.dumps(self.stack, default=lambda x: x.toJSON()) + + if self.args.format == "html": + try: + with open(self.args.template) as f: + output_str = f.read().replace("/** @flamegraph_json **/", + json_str) + except IOError as e: + print("Error reading template file: {}".format(e), file=sys.stderr) + sys.exit(1) + output_fn = self.args.output or "flamegraph.html" + else: + output_str = json_str + output_fn = self.args.output or "stacks.json" + + if output_fn == "-": + sys.stdout.write(output_str) + else: + print("dumping data to {}".format(output_fn)) + try: + with open(output_fn, "w") as out: + out.write(output_str) + except IOError as e: + print("Error writing output file: {}".format(e), file=sys.stderr) + sys.exit(1) + + +if __name__ == "__main__": + parser = argparse.ArgumentParser(description="Create flame graphs.") + parser.add_argument("-f", "--format", + default="html", choices=["json", "html"], + help="output file format") + parser.add_argument("-o", "--output", + help="output file name") + parser.add_argument("--template", + default="/usr/share/d3-flame-graph/d3-flamegraph-base.html", + help="path to flamegraph HTML template") + parser.add_argument("-i", "--input", + help=argparse.SUPPRESS) + + args = parser.parse_args() + cli = FlameGraphCLI(args) + + process_event = cli.process_event + trace_end = cli.trace_end |