| /linux/tools/perf/Documentation/ |
| H A D | tips.txt | 1 For a higher level overview, try: perf report --sort comm,dso
2 Sample related events with: perf record -e '{cycles,instructions}:S'
3 Compare performance results with: perf diff [<old file> <new file>]
4 Boolean options have negative forms, e.g.: perf report --no-children
6 Customize output of perf script with: perf script -F event,ip,sym
7 Generate a script for your data: perf script -g <lang>
8 Save output of perf stat using: perf stat record <target workload>
9 Create an archive with symtabs to analyse on other machine: perf archiv
[all...] |
| H A D | perf.txt | 1 perf(1)
6 perf - Performance analysis tools for Linux
11 'perf' [--version] [--help] [OPTIONS] COMMAND [ARGS]
17 Run perf help command.
21 Display perf version.
44 List the most commonly used perf commands.
47 List available perf options.
64 perf-event-open - Print perf_event_open() arguments and
66 kmaps - Print kernel and module maps (perf script
67 and perf report without browser)
[all …]
|
| H A D | perf-daemon.txt | 1 perf-daemon(1)
7 perf-daemon - Run record sessions on background
13 'perf daemon'
14 'perf daemon' [<options>]
15 'perf daemon start' [<options>]
16 'perf daemon stop' [<options>]
17 'perf daemon signal' [<options>]
18 'perf daemon ping' [<options>]
26 You can imagine 'perf daemon' of background process with several
27 'perf record' child tasks, like:
[all …]
|
| H A D | perf-kvm.txt | 1 perf-kvm(1)
6 perf-kvm - Tool to trace/measure kvm guest os
11 'perf kvm' [--host] [--guest] [--guestmount=<path>
14 'perf kvm' [--host] [--guest] [--guestkallsyms=<path> --guestmodules=<path>
16 'perf kvm stat [record|report|live] [<options>]
20 There are a couple of variants of perf kvm:
22 'perf kvm [options] top <command>' to generates and displays
26 'perf kvm record <command>' to record the performance counter profile
27 of an arbitrary workload and save it into a perf data file. We set the
28 default behavior of perf kvm as --guest, so if neither --host nor --guest
[all …]
|
| H A D | Build.txt | 2 1) perf build
4 The perf build process consists of several separated building blocks,
5 which are linked together to form the perf binary:
7 - perf builtin commands
11 Several makefiles govern the perf build:
15 Makefile.perf with a -j option to do parallel builds.
17 - Makefile.perf
18 main makefile that triggers build of all perf objects including
34 2) perf build
36 The Makefile.perf triggers the build framework for build objects:
[all …]
|
| H A D | perf.data-directory-format.txt | 1 perf.data directory format
4 in following versions of perf. We will remove this
8 This document describes the on-disk perf.data directory format.
20 - there is a single perf.data file named 'data' within the directory.
23 $ tree -ps perf.data
24 perf.data
30 Currently the only 'perf record' option to output to a directory is
34 $ sudo perf record --kcore uname
36 [ perf record: Woken up 1 times to write data ]
37 [ perf record: Captured and wrote 0.015 MB perf.data (9 samples) ]
[all …]
|
| H A D | perf-diff.txt | 1 perf-diff(1)
6 perf-diff - Read perf.data files and display the differential profile
11 'perf diff' [baseline file] [data file1] [[data file2] ... ]
15 This command displays the performance difference amongst two or more perf.data
16 files captured via perf record.
18 If no parameters are passed it will assume perf.data.old and perf.data.
21 specified perf.data files.
24 As the perf.data files could come from different binaries, the symbols addresses
25 could vary. So perf diff is based on the comparison of the files and
62 Please see description of --sort in the perf-report man page.
[all …]
|
| H A D | perf-amd-ibs.txt | 1 perf-amd-ibs(1)
6 perf-amd-ibs - Support for AMD Instruction-Based Sampling (IBS) with perf tool
11 'perf record' -e ibs_op//
12 'perf record' -e ibs_fetch//
27 using the Linux perf utility. The following files will be created at boot time
44 (sample was generated at IP X but perf would record it at IP X+n). Hence,
66 # perf record -e ibs_op// -c 100000 -a
70 # perf record -e ibs_op// -c 100000 -C 10
74 # perf record -e ibs_op// -F 1000 -C 10
78 # perf record -e ibs_op/cnt_ctl=1/ -c 100000 -a
[all …]
|
| H A D | perf-script.txt | 1 perf-script(1)
6 perf-script - Read perf.data (created by perf record) and display trace output
11 'perf script' [<options>]
12 'perf script' [<options>] record <script> [<record-options>] <command>
13 'perf script' [<options>] report <script> [script-args]
14 'perf script' [<options>] <script> <required-script-args> [<record-options>] <command>
15 'perf script' [<options>] <top-script> [script-args]
21 There are several variants of perf scrip
[all...] |
| H A D | perf-script-perl.txt | 1 perf-script-perl(1)
6 perf-script-perl - Process trace data with a Perl script
11 'perf script' [-s [Perl]:script[.pl] ]
16 This perf script option is used to process perf script data using perf's
24 You can avoid reading the rest of this document by running 'perf script
25 -g perl' in the same directory as an existing perf.data trace file.
31 ~/libexec/perf-core/scripts/perl for typical examples showing how to
33 the check-perf-script.pl script, while not interesting for its results,
39 When perf script is invoked using a trace script, a user-defined
47 available as calls back into the perf executable (see below).
[all …]
|
| /linux/tools/perf/ |
| H A D | command-list.txt | 2 # List of known perf commands.
5 perf-annotate mainporcelain common
6 perf-archive mainporcelain common
7 perf-bench mainporcelain common
8 perf-buildid-cache mainporcelain common
9 perf-buildid-list mainporcelain common
10 perf-data mainporcelain common
11 perf-diff mainporcelain common
12 perf-c2c mainporcelain common
13 perf-config mainporcelain common
[all …]
|
| H A D | Build | 1 perf-bench-y += builtin-bench.o
2 perf-y += builtin-annotate.o
3 perf-y += builtin-check.o
4 perf-y += builtin-config.o
5 perf-y += builtin-diff.o
6 perf-y += builtin-evlist.o
7 perf-y += builtin-ftrace.o
8 perf-y += builtin-help.o
9 perf-y += builtin-buildid-list.o
10 perf
[all...] |
| /linux/drivers/ntb/test/ |
| H A D | ntb_perf.c | 45 * PCIe NTB Perf Linux driver
126 * Perf driver data definition
144 struct perf_ctx *perf; member
169 struct perf_ctx *perf; member
205 int (*cmd_recv)(struct perf_ctx *perf, int *pidx, enum perf_cmd *cmd,
247 static void perf_terminate_test(struct perf_ctx *perf);
253 link = ntb_link_is_up(peer->perf->ntb, NULL, NULL); in perf_link_is_up()
260 struct perf_ctx *perf = peer->perf; in perf_spad_cmd_send() local
264 dev_dbg(&perf->ntb->dev, "CMD send: %d 0x%llx\n", cmd, data); in perf_spad_cmd_send()
277 sts = ntb_peer_spad_read(perf->ntb, peer->pidx, in perf_spad_cmd_send()
[all …]
|
| /linux/drivers/gpu/drm/msm/ |
| H A D | msm_perf.c | 41 static int wait_sample(struct msm_perf_state *perf) in wait_sample() argument
45 if (time_after(perf->next_jiffies, start_jiffies)) { in wait_sample()
47 perf->next_jiffies - start_jiffies; in wait_sample()
54 perf->next_jiffies += SAMPLE_TIME; in wait_sample()
58 static int refill_buf(struct msm_perf_state *perf) in refill_buf() argument
60 struct msm_drm_private *priv = perf->dev->dev_private; in refill_buf()
62 char *ptr = perf->buf; in refill_buf()
63 int rem = sizeof(perf->buf); in refill_buf()
66 if ((perf->cnt++ % 32) == 0) { in refill_buf()
86 ret = wait_sample(perf); in refill_buf()
[all …]
|
| /linux/drivers/gpu/drm/nouveau/nvkm/subdev/bios/ |
| H A D | perf.c | 26 #include <subdev/bios/perf.h>
34 u32 perf = 0; in nvbios_perf_table() local
38 perf = nvbios_rd32(bios, bit_P.offset + 0); in nvbios_perf_table()
39 if (perf) { in nvbios_perf_table()
40 *ver = nvbios_rd08(bios, perf + 0); in nvbios_perf_table()
41 *hdr = nvbios_rd08(bios, perf + 1); in nvbios_perf_table()
43 *cnt = nvbios_rd08(bios, perf + 5); in nvbios_perf_table()
44 *len = nvbios_rd08(bios, perf + 2); in nvbios_perf_table()
45 *snr = nvbios_rd08(bios, perf + 4); in nvbios_perf_table()
46 *ssz = nvbios_rd08(bios, perf + 3); in nvbios_perf_table()
[all …]
|
| /linux/tools/perf/tests/ |
| H A D | hists_cumulate.c | 27 /* perf [kernel] schedule() */
29 /* perf [perf] main() */
31 /* perf [perf] cmd_record() */
33 /* perf [libc] malloc() */
35 /* perf [libc] free() */
37 /* perf [perf] main() */
39 /* perf [kerne
[all...] |
| H A D | hists_common.h | 43 * perf: 100 perf main
44 * perf: 100 perf run_command
45 * perf: 100 perf cmd_record
46 * perf: 100 libc malloc
47 * perf: 100 libc free
48 * perf: 100 libc realloc
49 * perf: 100 [kernel] schedule
50 * perf: 100 [kernel] page_fault
51 * perf: 100 [kernel] sys_perf_event_open
52 * perf: 200 perf main
[all …]
|
| H A D | hists_output.c | 28 /* perf [kernel] schedule() */
30 /* perf [perf] main() */
32 /* perf [perf] cmd_record() */
34 /* perf [libc] malloc() */
36 /* perf [libc] free() */
38 /* perf [perf] main() */
40 /* perf [kerne
[all...] |
| /linux/tools/perf/bench/ |
| H A D | Build | 1 perf-bench-y += sched-messaging.o
2 perf-bench-y += sched-pipe.o
3 perf-bench-y += sched-seccomp-notify.o
4 perf-bench-y += syscall.o
5 perf-bench-y += mem-functions.o
6 perf-bench-y += futex.o
7 perf-bench-y += futex-hash.o
8 perf-bench-y += futex-wake.o
9 perf-bench-y += futex-wake-parallel.o
10 perf
[all...] |
| /linux/tools/perf/trace/beauty/ |
| H A D | Build | 1 perf-y += clone.o
2 perf-y += fcntl.o
3 perf-y += flock.o
4 perf-y += fs_at_flags.o
5 perf-y += fsmount.o
6 perf-y += fspick.o
8 perf-y += ioctl.o
10 perf-y += kcmp.o
11 perf-y += mount_flags.o
12 perf
[all...] |
| /linux/tools/perf/tests/shell/common/ |
| H A D | patterns.sh | 161 export RE_LINE_RECORD1="^\[\s+perf\s+record:\s+Woken up $RE_NUMBER times? to write data\s+\].*$"
162 # The first line of perf-record "OK" output
164 # [ perf record: Woken up 1 times to write data ]
167 export RE_LINE_RECORD2="^\[\s+perf\s+record:\s+Captured and wrote $RE_NUMBER\s*MB\s+(?:[\w\+\.-]*(?…
168 # The second line of perf-record "OK" output
170 # [ perf record: Captured and wrote 0.405 MB perf.data (109 samples) ]
171 # [ perf record: Captured and wrote 0.405 MB perf.data (~109 samples) ]
172 # [ perf record: Captured and wrote 0.405 MB /some/temp/dir/perf.data (109 samples) ]
173 # [ perf record: Captured and wrote 0.405 MB ./perf.data (109 samples) ]
174 # [ perf record: Captured and wrote 0.405 MB ./perf.data.3 (109 samples) ]
[all …]
|
| /linux/drivers/cpufreq/ |
| H A D | amd-pstate.c | 139 static inline u8 freq_to_perf(union perf_cached perf, u32 nominal_freq, unsigned int freq_val) in freq_to_perf() argument
141 u32 perf_val = DIV_ROUND_UP_ULL((u64)freq_val * perf.nominal_perf, nominal_freq); in freq_to_perf()
143 return (u8)clamp(perf_val, perf.lowest_perf, perf.highest_perf); in freq_to_perf()
146 static inline u32 perf_to_freq(union perf_cached perf, u32 nominal_freq, u8 perf_val) in perf_to_freq() argument
149 perf.nominal_perf); in perf_to_freq()
202 pr_debug("Could not retrieve energy perf value (%d)\n", ret); in msr_get_epp()
223 pr_debug("Could not retrieve energy perf value (%d)\n", ret); in shmem_get_epp()
246 union perf_cached perf = READ_ONCE(cpudata->perf); in msr_update_perf() local
249 perf.highest_perf, in msr_update_perf()
297 union perf_cached perf = cpudata->perf; in msr_set_epp() local
[all …]
|
| /linux/Documentation/trace/coresight/ |
| H A D | coresight-perf.rst | 4 CoreSight - Perf
10 Perf is able to locally access CoreSight trace data and store it to the
11 output perf data files. This data can then be later decoded to give the
13 can log such data with a perf record command like::
15 perf record -e cs_etm//u testbinary
18 a perf.data trace file. That file would have AUX sections if CoreSight
22 perf report --stdio --dump -i perf.data
43 To compile perf with CoreSight support in the tools/perf directory do::
53 For complete information on building perf with CoreSight support and
86 profiling result. To mitigate the issue of excessive trace data, Perf
[all …]
|
| /linux/drivers/gpu/drm/i915/ |
| H A D | i915_perf.c | 29 * DOC: i915 Perf Overview
35 * This i915 perf interface enables userspace to configure and open a file
51 * DOC: i915 Perf History and Comparison with Core Perf
53 * The interface was initially inspired by the core Perf infrastructure but
56 * i915 perf file descriptors represent a "stream" instead of an "event"; where
57 * a perf event primarily corresponds to a single 64bit value, while a stream
61 * of related counters. Samples for an i915 perf stream capturing OA metrics
64 * selected by the user opening the stream. Perf has support for grouping
68 * i915 perf stream configurations are provided as an array of u64 (key,value)
72 * i915 perf doesn't support exposing metrics via an mmap'd circular buffer.
[all …]
|
| /linux/drivers/gpu/drm/i915/selftests/ |
| H A D | i915_perf.c | 20 alloc_empty_config(struct i915_perf *perf) in alloc_empty_config() argument
28 oa_config->perf = perf; in alloc_empty_config()
33 mutex_lock(&perf->metrics_lock); in alloc_empty_config()
35 oa_config->id = idr_alloc(&perf->metrics_idr, oa_config, 2, 0, GFP_KERNEL); in alloc_empty_config()
37 mutex_unlock(&perf->metrics_lock); in alloc_empty_config()
42 mutex_unlock(&perf->metrics_lock); in alloc_empty_config()
48 destroy_empty_config(struct i915_perf *perf) in destroy_empty_config() argument
53 mutex_lock(&perf->metrics_lock); in destroy_empty_config()
55 idr_for_each_entry(&perf->metrics_idr, tmp, id) { in destroy_empty_config()
63 idr_remove(&perf->metrics_idr, oa_config->id); in destroy_empty_config()
[all …]
|