2017-04-18 16:46:11 +03:00
|
|
|
#include <errno.h>
|
2017-04-17 21:23:08 +03:00
|
|
|
#include <inttypes.h>
|
2009-12-12 02:24:02 +03:00
|
|
|
#include <linux/kernel.h>
|
2013-06-11 19:29:18 +04:00
|
|
|
#include <traceevent/event-parse.h>
|
2017-03-16 19:41:59 +03:00
|
|
|
#include <api/fs/fs.h>
|
2009-12-12 02:24:02 +03:00
|
|
|
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
#include <byteswap.h>
|
2009-12-12 02:24:02 +03:00
|
|
|
#include <unistd.h>
|
|
|
|
#include <sys/types.h>
|
2010-05-19 01:29:23 +04:00
|
|
|
#include <sys/mman.h>
|
2009-12-12 02:24:02 +03:00
|
|
|
|
2011-03-06 03:40:06 +03:00
|
|
|
#include "evlist.h"
|
|
|
|
#include "evsel.h"
|
2017-04-25 21:45:35 +03:00
|
|
|
#include "memswap.h"
|
2009-12-12 02:24:02 +03:00
|
|
|
#include "session.h"
|
2011-11-28 14:30:20 +04:00
|
|
|
#include "tool.h"
|
2009-12-14 19:22:59 +03:00
|
|
|
#include "sort.h"
|
2009-12-12 02:24:02 +03:00
|
|
|
#include "util.h"
|
2011-07-04 15:57:50 +04:00
|
|
|
#include "cpumap.h"
|
2012-08-07 17:20:45 +04:00
|
|
|
#include "perf_regs.h"
|
2014-06-12 11:50:11 +04:00
|
|
|
#include "asm/bug.h"
|
2015-04-09 18:53:48 +03:00
|
|
|
#include "auxtrace.h"
|
2017-04-20 03:34:35 +03:00
|
|
|
#include "thread.h"
|
2015-05-29 16:33:30 +03:00
|
|
|
#include "thread-stack.h"
|
2015-10-25 17:51:42 +03:00
|
|
|
#include "stat.h"
|
2009-12-12 02:24:02 +03:00
|
|
|
|
2015-04-09 18:53:48 +03:00
|
|
|
static int perf_session__deliver_event(struct perf_session *session,
|
|
|
|
union perf_event *event,
|
|
|
|
struct perf_sample *sample,
|
|
|
|
struct perf_tool *tool,
|
|
|
|
u64 file_offset);
|
2015-03-03 18:20:38 +03:00
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
static int perf_session__open(struct perf_session *session)
|
2009-12-12 02:24:02 +03:00
|
|
|
{
|
2013-11-05 22:32:36 +04:00
|
|
|
struct perf_data_file *file = session->file;
|
2010-04-02 08:59:15 +04:00
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
if (perf_session__read_header(session) < 0) {
|
2015-11-09 23:12:03 +03:00
|
|
|
pr_err("incompatible file format (rerun with -v to learn more)\n");
|
2013-10-15 18:27:33 +04:00
|
|
|
return -1;
|
2009-12-12 02:24:02 +03:00
|
|
|
}
|
|
|
|
|
2013-10-15 18:27:34 +04:00
|
|
|
if (perf_data_file__is_pipe(file))
|
|
|
|
return 0;
|
|
|
|
|
2015-11-05 17:40:47 +03:00
|
|
|
if (perf_header__has_feat(&session->header, HEADER_STAT))
|
|
|
|
return 0;
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
if (!perf_evlist__valid_sample_type(session->evlist)) {
|
2015-11-09 23:12:03 +03:00
|
|
|
pr_err("non matching sample_type\n");
|
2013-10-15 18:27:33 +04:00
|
|
|
return -1;
|
2011-06-02 18:04:54 +04:00
|
|
|
}
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
if (!perf_evlist__valid_sample_id_all(session->evlist)) {
|
2015-11-09 23:12:03 +03:00
|
|
|
pr_err("non matching sample_id_all\n");
|
2013-10-15 18:27:33 +04:00
|
|
|
return -1;
|
2011-06-02 18:04:54 +04:00
|
|
|
}
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
if (!perf_evlist__valid_read_format(session->evlist)) {
|
2015-11-09 23:12:03 +03:00
|
|
|
pr_err("non matching read_format\n");
|
2013-10-15 18:27:33 +04:00
|
|
|
return -1;
|
2012-10-10 19:38:13 +04:00
|
|
|
}
|
|
|
|
|
2009-12-12 02:24:02 +03:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2012-08-02 02:31:00 +04:00
|
|
|
void perf_session__set_id_hdr_size(struct perf_session *session)
|
2010-12-02 15:25:28 +03:00
|
|
|
{
|
2012-08-02 02:31:00 +04:00
|
|
|
u16 id_hdr_size = perf_evlist__id_hdr_size(session->evlist);
|
|
|
|
|
|
|
|
machines__set_id_hdr_size(&session->machines, id_hdr_size);
|
2010-12-02 15:25:28 +03:00
|
|
|
}
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
int perf_session__create_kernel_maps(struct perf_session *session)
|
2010-04-19 09:32:50 +04:00
|
|
|
{
|
2013-11-05 22:32:36 +04:00
|
|
|
int ret = machine__create_kernel_maps(&session->machines.host);
|
2010-04-19 09:32:50 +04:00
|
|
|
|
|
|
|
if (ret >= 0)
|
2013-11-05 22:32:36 +04:00
|
|
|
ret = machines__create_guest_kernel_maps(&session->machines);
|
2010-04-19 09:32:50 +04:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
static void perf_session__destroy_kernel_maps(struct perf_session *session)
|
perf session: Free the ref_reloc_sym memory at the right place
Which is at perf_session__destroy_kernel_maps, counterpart to the
perf_session__create_kernel_maps where the kmap structure is located, just
after the vmlinux_maps.
Make it also check if the kernel maps were actually created, which may not
be the case if, for instance, perf_session__new can't complete due to
permission problems in, for instance, a 'perf report' case, when a
segfault will take place, that is how this was noticed.
The problem was introduced in d65a458, thus post .35.
This also adds code to release guest machines as them are also created
in perf_session__create_kernel_maps, so should be deleted on this newly
introduced counterpart, perf_session__destroy_kernel_maps.
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-08-03 01:18:28 +04:00
|
|
|
{
|
2013-11-05 22:32:36 +04:00
|
|
|
machines__destroy_kernel_maps(&session->machines);
|
perf session: Free the ref_reloc_sym memory at the right place
Which is at perf_session__destroy_kernel_maps, counterpart to the
perf_session__create_kernel_maps where the kmap structure is located, just
after the vmlinux_maps.
Make it also check if the kernel maps were actually created, which may not
be the case if, for instance, perf_session__new can't complete due to
permission problems in, for instance, a 'perf report' case, when a
segfault will take place, that is how this was noticed.
The problem was introduced in d65a458, thus post .35.
This also adds code to release guest machines as them are also created
in perf_session__create_kernel_maps, so should be deleted on this newly
introduced counterpart, perf_session__destroy_kernel_maps.
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-08-03 01:18:28 +04:00
|
|
|
}
|
|
|
|
|
2014-07-31 10:00:45 +04:00
|
|
|
static bool perf_session__has_comm_exec(struct perf_session *session)
|
|
|
|
{
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
|
2016-06-23 17:26:15 +03:00
|
|
|
evlist__for_each_entry(session->evlist, evsel) {
|
2014-07-31 10:00:45 +04:00
|
|
|
if (evsel->attr.comm_exec)
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void perf_session__set_comm_exec(struct perf_session *session)
|
|
|
|
{
|
|
|
|
bool comm_exec = perf_session__has_comm_exec(session);
|
|
|
|
|
|
|
|
machines__set_comm_exec(&session->machines, comm_exec);
|
|
|
|
}
|
|
|
|
|
2015-03-03 18:20:38 +03:00
|
|
|
static int ordered_events__deliver_event(struct ordered_events *oe,
|
2015-03-31 18:48:16 +03:00
|
|
|
struct ordered_event *event)
|
2015-03-03 18:20:38 +03:00
|
|
|
{
|
2015-03-31 18:48:16 +03:00
|
|
|
struct perf_sample sample;
|
|
|
|
struct perf_session *session = container_of(oe, struct perf_session,
|
|
|
|
ordered_events);
|
|
|
|
int ret = perf_evlist__parse_sample(session->evlist, event->event, &sample);
|
|
|
|
|
|
|
|
if (ret) {
|
|
|
|
pr_err("Can't parse sample, err = %d\n", ret);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2015-04-09 18:53:48 +03:00
|
|
|
return perf_session__deliver_event(session, event->event, &sample,
|
|
|
|
session->tool, event->file_offset);
|
2015-03-03 18:20:38 +03:00
|
|
|
}
|
|
|
|
|
2013-10-15 18:27:32 +04:00
|
|
|
struct perf_session *perf_session__new(struct perf_data_file *file,
|
|
|
|
bool repipe, struct perf_tool *tool)
|
2009-12-12 02:24:02 +03:00
|
|
|
{
|
2013-11-05 22:32:36 +04:00
|
|
|
struct perf_session *session = zalloc(sizeof(*session));
|
2011-12-07 13:02:54 +04:00
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
if (!session)
|
2009-12-12 02:24:02 +03:00
|
|
|
goto out;
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
session->repipe = repipe;
|
2015-03-31 18:48:16 +03:00
|
|
|
session->tool = tool;
|
2015-04-30 17:37:25 +03:00
|
|
|
INIT_LIST_HEAD(&session->auxtrace_index);
|
2013-11-05 22:32:36 +04:00
|
|
|
machines__init(&session->machines);
|
2015-03-31 18:48:16 +03:00
|
|
|
ordered_events__init(&session->ordered_events, ordered_events__deliver_event);
|
2009-12-12 02:24:02 +03:00
|
|
|
|
2013-10-15 18:27:33 +04:00
|
|
|
if (file) {
|
|
|
|
if (perf_data_file__open(file))
|
2010-01-28 02:05:52 +03:00
|
|
|
goto out_delete;
|
2013-10-15 18:27:33 +04:00
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
session->file = file;
|
2013-10-15 18:27:33 +04:00
|
|
|
|
|
|
|
if (perf_data_file__is_read(file)) {
|
2013-11-05 22:32:36 +04:00
|
|
|
if (perf_session__open(session) < 0)
|
2013-10-15 18:27:33 +04:00
|
|
|
goto out_close;
|
|
|
|
|
2017-04-10 23:14:30 +03:00
|
|
|
/*
|
|
|
|
* set session attributes that are present in perf.data
|
|
|
|
* but not in pipe-mode.
|
|
|
|
*/
|
|
|
|
if (!file->is_pipe) {
|
|
|
|
perf_session__set_id_hdr_size(session);
|
|
|
|
perf_session__set_comm_exec(session);
|
|
|
|
}
|
2013-10-15 18:27:33 +04:00
|
|
|
}
|
2015-09-09 18:25:00 +03:00
|
|
|
} else {
|
|
|
|
session->machines.host.env = &perf_env;
|
2013-10-15 18:27:33 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
if (!file || perf_data_file__is_write(file)) {
|
2010-01-28 02:05:52 +03:00
|
|
|
/*
|
|
|
|
* In O_RDONLY mode this will be performed when reading the
|
2011-01-29 19:01:45 +03:00
|
|
|
* kernel MMAP event, in perf_event__process_mmap().
|
2010-01-28 02:05:52 +03:00
|
|
|
*/
|
2013-11-05 22:32:36 +04:00
|
|
|
if (perf_session__create_kernel_maps(session) < 0)
|
2014-09-25 01:39:54 +04:00
|
|
|
pr_warning("Cannot read kernel map\n");
|
2010-01-28 02:05:52 +03:00
|
|
|
}
|
2009-12-28 02:37:02 +03:00
|
|
|
|
2017-04-10 23:14:30 +03:00
|
|
|
/*
|
|
|
|
* In pipe-mode, evlist is empty until PERF_RECORD_HEADER_ATTR is
|
|
|
|
* processed, so perf_evlist__sample_id_all is not meaningful here.
|
|
|
|
*/
|
|
|
|
if ((!file || !file->is_pipe) && tool && tool->ordering_requires_timestamps &&
|
2014-07-06 16:18:21 +04:00
|
|
|
tool->ordered_events && !perf_evlist__sample_id_all(session->evlist)) {
|
2010-12-10 06:09:16 +03:00
|
|
|
dump_printf("WARNING: No sample_id_all support, falling back to unordered processing\n");
|
2014-07-06 16:18:21 +04:00
|
|
|
tool->ordered_events = false;
|
2015-03-03 18:20:38 +03:00
|
|
|
}
|
2010-12-10 06:09:16 +03:00
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
return session;
|
2013-10-15 18:27:33 +04:00
|
|
|
|
|
|
|
out_close:
|
|
|
|
perf_data_file__close(file);
|
|
|
|
out_delete:
|
2013-11-05 22:32:36 +04:00
|
|
|
perf_session__delete(session);
|
2013-10-15 18:27:33 +04:00
|
|
|
out:
|
perf session: Move kmaps to perf_session
There is still some more work to do to disentangle map creation
from DSO loading, but this happens only for the kernel, and for
the early adopters of perf diff, where this disentanglement
matters most, we'll be testing different kernels, so no problem
here.
Further clarification: right now we create the kernel maps for
the various modules and discontiguous kernel text maps when
loading the DSO, we should do it as a two step process, first
creating the maps, for multiple mappings with the same DSO
store, then doing the dso load just once, for the first hit on
one of the maps sharing this DSO backing store.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1260741029-4430-6-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-14 00:50:29 +03:00
|
|
|
return NULL;
|
2009-12-12 02:24:02 +03:00
|
|
|
}
|
|
|
|
|
2011-11-09 19:24:25 +04:00
|
|
|
static void perf_session__delete_threads(struct perf_session *session)
|
|
|
|
{
|
2012-12-19 02:15:48 +04:00
|
|
|
machine__delete_threads(&session->machines.host);
|
2011-11-09 19:24:25 +04:00
|
|
|
}
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
void perf_session__delete(struct perf_session *session)
|
2009-12-12 02:24:02 +03:00
|
|
|
{
|
2016-06-22 16:02:16 +03:00
|
|
|
if (session == NULL)
|
|
|
|
return;
|
2015-04-09 18:53:48 +03:00
|
|
|
auxtrace__free(session);
|
2015-04-30 17:37:25 +03:00
|
|
|
auxtrace_index__free(&session->auxtrace_index);
|
2013-11-05 22:32:36 +04:00
|
|
|
perf_session__destroy_kernel_maps(session);
|
|
|
|
perf_session__delete_threads(session);
|
2015-09-08 19:30:00 +03:00
|
|
|
perf_env__exit(&session->header.env);
|
2013-11-05 22:32:36 +04:00
|
|
|
machines__exit(&session->machines);
|
|
|
|
if (session->file)
|
|
|
|
perf_data_file__close(session->file);
|
|
|
|
free(session);
|
2009-12-12 02:24:02 +03:00
|
|
|
}
|
2009-12-14 19:22:59 +03:00
|
|
|
|
2013-07-04 17:20:21 +04:00
|
|
|
static int process_event_synth_tracing_data_stub(struct perf_tool *tool
|
|
|
|
__maybe_unused,
|
|
|
|
union perf_event *event
|
2012-09-11 02:15:03 +04:00
|
|
|
__maybe_unused,
|
|
|
|
struct perf_session *session
|
|
|
|
__maybe_unused)
|
2011-11-25 14:19:45 +04:00
|
|
|
{
|
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-07-04 17:20:21 +04:00
|
|
|
static int process_event_synth_attr_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
2012-09-11 02:15:03 +04:00
|
|
|
struct perf_evlist **pevlist
|
|
|
|
__maybe_unused)
|
2011-11-12 04:45:41 +04:00
|
|
|
{
|
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:36 +03:00
|
|
|
static int process_event_synth_event_update_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
|
|
|
struct perf_evlist **pevlist
|
|
|
|
__maybe_unused)
|
|
|
|
{
|
2015-10-25 17:51:42 +03:00
|
|
|
if (dump_trace)
|
|
|
|
perf_event__fprintf_event_update(event, stdout);
|
|
|
|
|
2015-10-25 17:51:36 +03:00
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2012-09-11 02:15:03 +04:00
|
|
|
static int process_event_sample_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
|
|
|
struct perf_sample *sample __maybe_unused,
|
|
|
|
struct perf_evsel *evsel __maybe_unused,
|
|
|
|
struct machine *machine __maybe_unused)
|
2011-03-15 21:44:01 +03:00
|
|
|
{
|
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2012-09-11 02:15:03 +04:00
|
|
|
static int process_event_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
|
|
|
struct perf_sample *sample __maybe_unused,
|
|
|
|
struct machine *machine __maybe_unused)
|
2009-12-28 02:36:59 +03:00
|
|
|
{
|
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2012-09-11 02:15:03 +04:00
|
|
|
static int process_finished_round_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
2015-03-03 18:37:54 +03:00
|
|
|
struct ordered_events *oe __maybe_unused)
|
2011-11-28 13:56:39 +04:00
|
|
|
{
|
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2011-11-28 14:30:20 +04:00
|
|
|
static int process_finished_round(struct perf_tool *tool,
|
2011-11-25 14:19:45 +04:00
|
|
|
union perf_event *event,
|
2015-03-03 18:37:54 +03:00
|
|
|
struct ordered_events *oe);
|
2010-05-03 17:14:33 +04:00
|
|
|
|
2015-04-09 18:53:43 +03:00
|
|
|
static int skipn(int fd, off_t n)
|
|
|
|
{
|
|
|
|
char buf[4096];
|
|
|
|
ssize_t ret;
|
|
|
|
|
|
|
|
while (n > 0) {
|
|
|
|
ret = read(fd, buf, min(n, (off_t)sizeof(buf)));
|
|
|
|
if (ret <= 0)
|
|
|
|
return ret;
|
|
|
|
n -= ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static s64 process_event_auxtrace_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event,
|
|
|
|
struct perf_session *session
|
|
|
|
__maybe_unused)
|
|
|
|
{
|
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
if (perf_data_file__is_pipe(session->file))
|
|
|
|
skipn(perf_data_file__fd(session->file), event->auxtrace.size);
|
|
|
|
return event->auxtrace.size;
|
|
|
|
}
|
|
|
|
|
2016-03-07 22:44:39 +03:00
|
|
|
static int process_event_op2_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
|
|
|
struct perf_session *session __maybe_unused)
|
2015-04-09 18:53:47 +03:00
|
|
|
{
|
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:19 +03:00
|
|
|
|
|
|
|
static
|
|
|
|
int process_event_thread_map_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
|
|
|
struct perf_session *session __maybe_unused)
|
|
|
|
{
|
2015-10-25 17:51:42 +03:00
|
|
|
if (dump_trace)
|
|
|
|
perf_event__fprintf_thread_map(event, stdout);
|
|
|
|
|
2015-10-25 17:51:19 +03:00
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:23 +03:00
|
|
|
static
|
|
|
|
int process_event_cpu_map_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
|
|
|
struct perf_session *session __maybe_unused)
|
|
|
|
{
|
2015-10-25 17:51:42 +03:00
|
|
|
if (dump_trace)
|
|
|
|
perf_event__fprintf_cpu_map(event, stdout);
|
|
|
|
|
2015-10-25 17:51:23 +03:00
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:27 +03:00
|
|
|
static
|
|
|
|
int process_event_stat_config_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
|
|
|
struct perf_session *session __maybe_unused)
|
|
|
|
{
|
2015-10-25 17:51:42 +03:00
|
|
|
if (dump_trace)
|
|
|
|
perf_event__fprintf_stat_config(event, stdout);
|
|
|
|
|
2015-10-25 17:51:27 +03:00
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:30 +03:00
|
|
|
static int process_stat_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
|
|
|
struct perf_session *perf_session
|
|
|
|
__maybe_unused)
|
|
|
|
{
|
2015-10-25 17:51:42 +03:00
|
|
|
if (dump_trace)
|
|
|
|
perf_event__fprintf_stat(event, stdout);
|
|
|
|
|
2015-10-25 17:51:30 +03:00
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:33 +03:00
|
|
|
static int process_stat_round_stub(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event __maybe_unused,
|
|
|
|
struct perf_session *perf_session
|
|
|
|
__maybe_unused)
|
|
|
|
{
|
2015-10-25 17:51:42 +03:00
|
|
|
if (dump_trace)
|
|
|
|
perf_event__fprintf_stat_round(event, stdout);
|
|
|
|
|
2015-10-25 17:51:33 +03:00
|
|
|
dump_printf(": unhandled!\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-08-03 00:05:41 +04:00
|
|
|
void perf_tool__fill_defaults(struct perf_tool *tool)
|
2009-12-28 02:36:59 +03:00
|
|
|
{
|
2011-11-28 14:30:20 +04:00
|
|
|
if (tool->sample == NULL)
|
|
|
|
tool->sample = process_event_sample_stub;
|
|
|
|
if (tool->mmap == NULL)
|
|
|
|
tool->mmap = process_event_stub;
|
2013-09-23 05:44:59 +04:00
|
|
|
if (tool->mmap2 == NULL)
|
|
|
|
tool->mmap2 = process_event_stub;
|
2011-11-28 14:30:20 +04:00
|
|
|
if (tool->comm == NULL)
|
|
|
|
tool->comm = process_event_stub;
|
|
|
|
if (tool->fork == NULL)
|
|
|
|
tool->fork = process_event_stub;
|
|
|
|
if (tool->exit == NULL)
|
|
|
|
tool->exit = process_event_stub;
|
|
|
|
if (tool->lost == NULL)
|
|
|
|
tool->lost = perf_event__process_lost;
|
2015-05-10 22:13:15 +03:00
|
|
|
if (tool->lost_samples == NULL)
|
|
|
|
tool->lost_samples = perf_event__process_lost_samples;
|
2015-04-30 17:37:29 +03:00
|
|
|
if (tool->aux == NULL)
|
|
|
|
tool->aux = perf_event__process_aux;
|
2015-04-30 17:37:30 +03:00
|
|
|
if (tool->itrace_start == NULL)
|
|
|
|
tool->itrace_start = perf_event__process_itrace_start;
|
2015-07-21 12:44:03 +03:00
|
|
|
if (tool->context_switch == NULL)
|
|
|
|
tool->context_switch = perf_event__process_switch;
|
2011-11-28 14:30:20 +04:00
|
|
|
if (tool->read == NULL)
|
|
|
|
tool->read = process_event_sample_stub;
|
|
|
|
if (tool->throttle == NULL)
|
|
|
|
tool->throttle = process_event_stub;
|
|
|
|
if (tool->unthrottle == NULL)
|
|
|
|
tool->unthrottle = process_event_stub;
|
|
|
|
if (tool->attr == NULL)
|
|
|
|
tool->attr = process_event_synth_attr_stub;
|
2015-10-25 17:51:36 +03:00
|
|
|
if (tool->event_update == NULL)
|
|
|
|
tool->event_update = process_event_synth_event_update_stub;
|
2011-11-28 14:30:20 +04:00
|
|
|
if (tool->tracing_data == NULL)
|
|
|
|
tool->tracing_data = process_event_synth_tracing_data_stub;
|
|
|
|
if (tool->build_id == NULL)
|
2016-03-07 22:44:39 +03:00
|
|
|
tool->build_id = process_event_op2_stub;
|
2011-11-28 14:30:20 +04:00
|
|
|
if (tool->finished_round == NULL) {
|
2014-07-06 16:18:21 +04:00
|
|
|
if (tool->ordered_events)
|
2011-11-28 14:30:20 +04:00
|
|
|
tool->finished_round = process_finished_round;
|
2010-05-03 17:14:33 +04:00
|
|
|
else
|
2011-11-28 14:30:20 +04:00
|
|
|
tool->finished_round = process_finished_round_stub;
|
2010-05-03 17:14:33 +04:00
|
|
|
}
|
2014-10-27 16:49:22 +03:00
|
|
|
if (tool->id_index == NULL)
|
2016-03-07 22:44:39 +03:00
|
|
|
tool->id_index = process_event_op2_stub;
|
2015-04-09 18:53:43 +03:00
|
|
|
if (tool->auxtrace_info == NULL)
|
2016-03-07 22:44:39 +03:00
|
|
|
tool->auxtrace_info = process_event_op2_stub;
|
2015-04-09 18:53:43 +03:00
|
|
|
if (tool->auxtrace == NULL)
|
|
|
|
tool->auxtrace = process_event_auxtrace_stub;
|
2015-04-09 18:53:47 +03:00
|
|
|
if (tool->auxtrace_error == NULL)
|
2016-03-07 22:44:39 +03:00
|
|
|
tool->auxtrace_error = process_event_op2_stub;
|
2015-10-25 17:51:19 +03:00
|
|
|
if (tool->thread_map == NULL)
|
|
|
|
tool->thread_map = process_event_thread_map_stub;
|
2015-10-25 17:51:23 +03:00
|
|
|
if (tool->cpu_map == NULL)
|
|
|
|
tool->cpu_map = process_event_cpu_map_stub;
|
2015-10-25 17:51:27 +03:00
|
|
|
if (tool->stat_config == NULL)
|
|
|
|
tool->stat_config = process_event_stat_config_stub;
|
2015-10-25 17:51:30 +03:00
|
|
|
if (tool->stat == NULL)
|
|
|
|
tool->stat = process_stat_stub;
|
2015-10-25 17:51:33 +03:00
|
|
|
if (tool->stat_round == NULL)
|
|
|
|
tool->stat_round = process_stat_round_stub;
|
2016-03-08 11:38:44 +03:00
|
|
|
if (tool->time_conv == NULL)
|
|
|
|
tool->time_conv = process_event_op2_stub;
|
2009-12-28 02:36:59 +03:00
|
|
|
}
|
2014-12-17 23:24:45 +03:00
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
static void swap_sample_id_all(union perf_event *event, void *data)
|
|
|
|
{
|
|
|
|
void *end = (void *) event + event->header.size;
|
|
|
|
int size = end - data;
|
|
|
|
|
|
|
|
BUG_ON(size % sizeof(u64));
|
|
|
|
mem_bswap_64(data, size);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void perf_event__all64_swap(union perf_event *event,
|
2012-09-11 02:15:03 +04:00
|
|
|
bool sample_id_all __maybe_unused)
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
{
|
2011-01-29 19:01:45 +03:00
|
|
|
struct perf_event_header *hdr = &event->header;
|
|
|
|
mem_bswap_64(hdr + 1, event->header.size - sizeof(*hdr));
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
}
|
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
static void perf_event__comm_swap(union perf_event *event, bool sample_id_all)
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
{
|
2011-01-29 19:01:45 +03:00
|
|
|
event->comm.pid = bswap_32(event->comm.pid);
|
|
|
|
event->comm.tid = bswap_32(event->comm.tid);
|
2012-05-30 16:23:43 +04:00
|
|
|
|
|
|
|
if (sample_id_all) {
|
|
|
|
void *data = &event->comm.comm;
|
|
|
|
|
2012-09-11 02:15:01 +04:00
|
|
|
data += PERF_ALIGN(strlen(data) + 1, sizeof(u64));
|
2012-05-30 16:23:43 +04:00
|
|
|
swap_sample_id_all(event, data);
|
|
|
|
}
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
}
|
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
static void perf_event__mmap_swap(union perf_event *event,
|
|
|
|
bool sample_id_all)
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
{
|
2011-01-29 19:01:45 +03:00
|
|
|
event->mmap.pid = bswap_32(event->mmap.pid);
|
|
|
|
event->mmap.tid = bswap_32(event->mmap.tid);
|
|
|
|
event->mmap.start = bswap_64(event->mmap.start);
|
|
|
|
event->mmap.len = bswap_64(event->mmap.len);
|
|
|
|
event->mmap.pgoff = bswap_64(event->mmap.pgoff);
|
2012-05-30 16:23:43 +04:00
|
|
|
|
|
|
|
if (sample_id_all) {
|
|
|
|
void *data = &event->mmap.filename;
|
|
|
|
|
2012-09-11 02:15:01 +04:00
|
|
|
data += PERF_ALIGN(strlen(data) + 1, sizeof(u64));
|
2012-05-30 16:23:43 +04:00
|
|
|
swap_sample_id_all(event, data);
|
|
|
|
}
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
}
|
|
|
|
|
2013-08-21 14:10:25 +04:00
|
|
|
static void perf_event__mmap2_swap(union perf_event *event,
|
|
|
|
bool sample_id_all)
|
|
|
|
{
|
|
|
|
event->mmap2.pid = bswap_32(event->mmap2.pid);
|
|
|
|
event->mmap2.tid = bswap_32(event->mmap2.tid);
|
|
|
|
event->mmap2.start = bswap_64(event->mmap2.start);
|
|
|
|
event->mmap2.len = bswap_64(event->mmap2.len);
|
|
|
|
event->mmap2.pgoff = bswap_64(event->mmap2.pgoff);
|
|
|
|
event->mmap2.maj = bswap_32(event->mmap2.maj);
|
|
|
|
event->mmap2.min = bswap_32(event->mmap2.min);
|
|
|
|
event->mmap2.ino = bswap_64(event->mmap2.ino);
|
|
|
|
|
|
|
|
if (sample_id_all) {
|
|
|
|
void *data = &event->mmap2.filename;
|
|
|
|
|
|
|
|
data += PERF_ALIGN(strlen(data) + 1, sizeof(u64));
|
|
|
|
swap_sample_id_all(event, data);
|
|
|
|
}
|
|
|
|
}
|
2012-05-30 16:23:43 +04:00
|
|
|
static void perf_event__task_swap(union perf_event *event, bool sample_id_all)
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
{
|
2011-01-29 19:01:45 +03:00
|
|
|
event->fork.pid = bswap_32(event->fork.pid);
|
|
|
|
event->fork.tid = bswap_32(event->fork.tid);
|
|
|
|
event->fork.ppid = bswap_32(event->fork.ppid);
|
|
|
|
event->fork.ptid = bswap_32(event->fork.ptid);
|
|
|
|
event->fork.time = bswap_64(event->fork.time);
|
2012-05-30 16:23:43 +04:00
|
|
|
|
|
|
|
if (sample_id_all)
|
|
|
|
swap_sample_id_all(event, &event->fork + 1);
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
}
|
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
static void perf_event__read_swap(union perf_event *event, bool sample_id_all)
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
{
|
2011-01-29 19:01:45 +03:00
|
|
|
event->read.pid = bswap_32(event->read.pid);
|
|
|
|
event->read.tid = bswap_32(event->read.tid);
|
|
|
|
event->read.value = bswap_64(event->read.value);
|
|
|
|
event->read.time_enabled = bswap_64(event->read.time_enabled);
|
|
|
|
event->read.time_running = bswap_64(event->read.time_running);
|
|
|
|
event->read.id = bswap_64(event->read.id);
|
2012-05-30 16:23:43 +04:00
|
|
|
|
|
|
|
if (sample_id_all)
|
|
|
|
swap_sample_id_all(event, &event->read + 1);
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
}
|
|
|
|
|
2015-04-30 17:37:29 +03:00
|
|
|
static void perf_event__aux_swap(union perf_event *event, bool sample_id_all)
|
|
|
|
{
|
|
|
|
event->aux.aux_offset = bswap_64(event->aux.aux_offset);
|
|
|
|
event->aux.aux_size = bswap_64(event->aux.aux_size);
|
|
|
|
event->aux.flags = bswap_64(event->aux.flags);
|
|
|
|
|
|
|
|
if (sample_id_all)
|
|
|
|
swap_sample_id_all(event, &event->aux + 1);
|
|
|
|
}
|
|
|
|
|
2015-04-30 17:37:30 +03:00
|
|
|
static void perf_event__itrace_start_swap(union perf_event *event,
|
|
|
|
bool sample_id_all)
|
|
|
|
{
|
|
|
|
event->itrace_start.pid = bswap_32(event->itrace_start.pid);
|
|
|
|
event->itrace_start.tid = bswap_32(event->itrace_start.tid);
|
|
|
|
|
|
|
|
if (sample_id_all)
|
|
|
|
swap_sample_id_all(event, &event->itrace_start + 1);
|
|
|
|
}
|
|
|
|
|
2015-07-21 12:44:03 +03:00
|
|
|
static void perf_event__switch_swap(union perf_event *event, bool sample_id_all)
|
|
|
|
{
|
|
|
|
if (event->header.type == PERF_RECORD_SWITCH_CPU_WIDE) {
|
|
|
|
event->context_switch.next_prev_pid =
|
|
|
|
bswap_32(event->context_switch.next_prev_pid);
|
|
|
|
event->context_switch.next_prev_tid =
|
|
|
|
bswap_32(event->context_switch.next_prev_tid);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (sample_id_all)
|
|
|
|
swap_sample_id_all(event, &event->context_switch + 1);
|
|
|
|
}
|
|
|
|
|
2013-09-01 14:36:15 +04:00
|
|
|
static void perf_event__throttle_swap(union perf_event *event,
|
|
|
|
bool sample_id_all)
|
|
|
|
{
|
|
|
|
event->throttle.time = bswap_64(event->throttle.time);
|
|
|
|
event->throttle.id = bswap_64(event->throttle.id);
|
|
|
|
event->throttle.stream_id = bswap_64(event->throttle.stream_id);
|
|
|
|
|
|
|
|
if (sample_id_all)
|
|
|
|
swap_sample_id_all(event, &event->throttle + 1);
|
|
|
|
}
|
|
|
|
|
2012-05-16 10:59:03 +04:00
|
|
|
static u8 revbyte(u8 b)
|
|
|
|
{
|
|
|
|
int rev = (b >> 4) | ((b & 0xf) << 4);
|
|
|
|
rev = ((rev & 0xcc) >> 2) | ((rev & 0x33) << 2);
|
|
|
|
rev = ((rev & 0xaa) >> 1) | ((rev & 0x55) << 1);
|
|
|
|
return (u8) rev;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* XXX this is hack in attempt to carry flags bitfield
|
2016-02-24 21:02:25 +03:00
|
|
|
* through endian village. ABI says:
|
2012-05-16 10:59:03 +04:00
|
|
|
*
|
|
|
|
* Bit-fields are allocated from right to left (least to most significant)
|
|
|
|
* on little-endian implementations and from left to right (most to least
|
|
|
|
* significant) on big-endian implementations.
|
|
|
|
*
|
|
|
|
* The above seems to be byte specific, so we need to reverse each
|
|
|
|
* byte of the bitfield. 'Internet' also says this might be implementation
|
|
|
|
* specific and we probably need proper fix and carry perf_event_attr
|
|
|
|
* bitfield flags in separate data file FEAT_ section. Thought this seems
|
|
|
|
* to work for now.
|
|
|
|
*/
|
|
|
|
static void swap_bitfield(u8 *p, unsigned len)
|
|
|
|
{
|
|
|
|
unsigned i;
|
|
|
|
|
|
|
|
for (i = 0; i < len; i++) {
|
|
|
|
*p = revbyte(*p);
|
|
|
|
p++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-07-15 22:34:09 +04:00
|
|
|
/* exported for swapping attributes in file header */
|
|
|
|
void perf_event__attr_swap(struct perf_event_attr *attr)
|
|
|
|
{
|
|
|
|
attr->type = bswap_32(attr->type);
|
|
|
|
attr->size = bswap_32(attr->size);
|
2015-06-17 12:56:39 +03:00
|
|
|
|
|
|
|
#define bswap_safe(f, n) \
|
|
|
|
(attr->size > (offsetof(struct perf_event_attr, f) + \
|
|
|
|
sizeof(attr->f) * (n)))
|
|
|
|
#define bswap_field(f, sz) \
|
|
|
|
do { \
|
|
|
|
if (bswap_safe(f, 0)) \
|
|
|
|
attr->f = bswap_##sz(attr->f); \
|
|
|
|
} while(0)
|
perf tools: Per event max-stack settings
The tooling counterpart, now it is possible to do:
# perf record -e sched:sched_switch/max-stack=10/ -e cycles/call-graph=dwarf,max-stack=4/ -e cpu-cycles/call-graph=dwarf,max-stack=1024/ usleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.052 MB perf.data (5 samples) ]
# perf evlist -v
sched:sched_switch: type: 2, size: 112, config: 0x110, { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|CALLCHAIN|CPU|PERIOD|RAW|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, sample_max_stack: 10
cycles/call-graph=dwarf,max-stack=4/: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|PERIOD|REGS_USER|STACK_USER|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, freq: 1, enable_on_exec: 1, sample_id_all: 1, exclude_guest: 1, exclude_callchain_user: 1, sample_regs_user: 0xff0fff, sample_stack_user: 8192, sample_max_stack: 4
cpu-cycles/call-graph=dwarf,max-stack=1024/: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|PERIOD|REGS_USER|STACK_USER|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, freq: 1, enable_on_exec: 1, sample_id_all: 1, exclude_guest: 1, exclude_callchain_user: 1, sample_regs_user: 0xff0fff, sample_stack_user: 8192, sample_max_stack: 1024
# Tip: use 'perf evlist --trace-fields' to show fields for tracepoint events
Using just /max-stack=N/ means /call-graph=fp,max-stack=N/, that should
be further configurable by means of some .perfconfig knob.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Milian Wolff <milian.wolff@kdab.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Zefan Li <lizefan@huawei.com>
Link: http://lkml.kernel.org/n/tip-kolmn1yo40p7jhswxwrc7rrd@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2016-04-29 01:03:42 +03:00
|
|
|
#define bswap_field_16(f) bswap_field(f, 16)
|
2015-06-17 12:56:39 +03:00
|
|
|
#define bswap_field_32(f) bswap_field(f, 32)
|
|
|
|
#define bswap_field_64(f) bswap_field(f, 64)
|
|
|
|
|
|
|
|
bswap_field_64(config);
|
|
|
|
bswap_field_64(sample_period);
|
|
|
|
bswap_field_64(sample_type);
|
|
|
|
bswap_field_64(read_format);
|
|
|
|
bswap_field_32(wakeup_events);
|
|
|
|
bswap_field_32(bp_type);
|
|
|
|
bswap_field_64(bp_addr);
|
|
|
|
bswap_field_64(bp_len);
|
|
|
|
bswap_field_64(branch_sample_type);
|
|
|
|
bswap_field_64(sample_regs_user);
|
|
|
|
bswap_field_32(sample_stack_user);
|
|
|
|
bswap_field_32(aux_watermark);
|
perf tools: Per event max-stack settings
The tooling counterpart, now it is possible to do:
# perf record -e sched:sched_switch/max-stack=10/ -e cycles/call-graph=dwarf,max-stack=4/ -e cpu-cycles/call-graph=dwarf,max-stack=1024/ usleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.052 MB perf.data (5 samples) ]
# perf evlist -v
sched:sched_switch: type: 2, size: 112, config: 0x110, { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|CALLCHAIN|CPU|PERIOD|RAW|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, sample_max_stack: 10
cycles/call-graph=dwarf,max-stack=4/: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|PERIOD|REGS_USER|STACK_USER|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, freq: 1, enable_on_exec: 1, sample_id_all: 1, exclude_guest: 1, exclude_callchain_user: 1, sample_regs_user: 0xff0fff, sample_stack_user: 8192, sample_max_stack: 4
cpu-cycles/call-graph=dwarf,max-stack=1024/: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|PERIOD|REGS_USER|STACK_USER|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, freq: 1, enable_on_exec: 1, sample_id_all: 1, exclude_guest: 1, exclude_callchain_user: 1, sample_regs_user: 0xff0fff, sample_stack_user: 8192, sample_max_stack: 1024
# Tip: use 'perf evlist --trace-fields' to show fields for tracepoint events
Using just /max-stack=N/ means /call-graph=fp,max-stack=N/, that should
be further configurable by means of some .perfconfig knob.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Milian Wolff <milian.wolff@kdab.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Zefan Li <lizefan@huawei.com>
Link: http://lkml.kernel.org/n/tip-kolmn1yo40p7jhswxwrc7rrd@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2016-04-29 01:03:42 +03:00
|
|
|
bswap_field_16(sample_max_stack);
|
2015-06-17 12:56:39 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* After read_format are bitfields. Check read_format because
|
|
|
|
* we are unable to use offsetof on bitfield.
|
|
|
|
*/
|
|
|
|
if (bswap_safe(read_format, 1))
|
|
|
|
swap_bitfield((u8 *) (&attr->read_format + 1),
|
|
|
|
sizeof(u64));
|
|
|
|
#undef bswap_field_64
|
|
|
|
#undef bswap_field_32
|
|
|
|
#undef bswap_field
|
|
|
|
#undef bswap_safe
|
2011-07-15 22:34:09 +04:00
|
|
|
}
|
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
static void perf_event__hdr_attr_swap(union perf_event *event,
|
2012-09-11 02:15:03 +04:00
|
|
|
bool sample_id_all __maybe_unused)
|
2010-04-02 08:59:19 +04:00
|
|
|
{
|
|
|
|
size_t size;
|
|
|
|
|
2011-07-15 22:34:09 +04:00
|
|
|
perf_event__attr_swap(&event->attr.attr);
|
2010-04-02 08:59:19 +04:00
|
|
|
|
2011-01-29 19:01:45 +03:00
|
|
|
size = event->header.size;
|
|
|
|
size -= (void *)&event->attr.id - (void *)event;
|
|
|
|
mem_bswap_64(event->attr.id, size);
|
2010-04-02 08:59:19 +04:00
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:36 +03:00
|
|
|
static void perf_event__event_update_swap(union perf_event *event,
|
|
|
|
bool sample_id_all __maybe_unused)
|
|
|
|
{
|
|
|
|
event->event_update.type = bswap_64(event->event_update.type);
|
|
|
|
event->event_update.id = bswap_64(event->event_update.id);
|
|
|
|
}
|
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
static void perf_event__event_type_swap(union perf_event *event,
|
2012-09-11 02:15:03 +04:00
|
|
|
bool sample_id_all __maybe_unused)
|
2010-04-02 08:59:20 +04:00
|
|
|
{
|
2011-01-29 19:01:45 +03:00
|
|
|
event->event_type.event_type.event_id =
|
|
|
|
bswap_64(event->event_type.event_type.event_id);
|
2010-04-02 08:59:20 +04:00
|
|
|
}
|
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
static void perf_event__tracing_data_swap(union perf_event *event,
|
2012-09-11 02:15:03 +04:00
|
|
|
bool sample_id_all __maybe_unused)
|
2010-04-02 08:59:21 +04:00
|
|
|
{
|
2011-01-29 19:01:45 +03:00
|
|
|
event->tracing_data.size = bswap_32(event->tracing_data.size);
|
2010-04-02 08:59:21 +04:00
|
|
|
}
|
|
|
|
|
2015-04-09 18:53:43 +03:00
|
|
|
static void perf_event__auxtrace_info_swap(union perf_event *event,
|
|
|
|
bool sample_id_all __maybe_unused)
|
|
|
|
{
|
|
|
|
size_t size;
|
|
|
|
|
|
|
|
event->auxtrace_info.type = bswap_32(event->auxtrace_info.type);
|
|
|
|
|
|
|
|
size = event->header.size;
|
|
|
|
size -= (void *)&event->auxtrace_info.priv - (void *)event;
|
|
|
|
mem_bswap_64(event->auxtrace_info.priv, size);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void perf_event__auxtrace_swap(union perf_event *event,
|
|
|
|
bool sample_id_all __maybe_unused)
|
|
|
|
{
|
|
|
|
event->auxtrace.size = bswap_64(event->auxtrace.size);
|
|
|
|
event->auxtrace.offset = bswap_64(event->auxtrace.offset);
|
|
|
|
event->auxtrace.reference = bswap_64(event->auxtrace.reference);
|
|
|
|
event->auxtrace.idx = bswap_32(event->auxtrace.idx);
|
|
|
|
event->auxtrace.tid = bswap_32(event->auxtrace.tid);
|
|
|
|
event->auxtrace.cpu = bswap_32(event->auxtrace.cpu);
|
|
|
|
}
|
|
|
|
|
2015-04-09 18:53:47 +03:00
|
|
|
static void perf_event__auxtrace_error_swap(union perf_event *event,
|
|
|
|
bool sample_id_all __maybe_unused)
|
|
|
|
{
|
|
|
|
event->auxtrace_error.type = bswap_32(event->auxtrace_error.type);
|
|
|
|
event->auxtrace_error.code = bswap_32(event->auxtrace_error.code);
|
|
|
|
event->auxtrace_error.cpu = bswap_32(event->auxtrace_error.cpu);
|
|
|
|
event->auxtrace_error.pid = bswap_32(event->auxtrace_error.pid);
|
|
|
|
event->auxtrace_error.tid = bswap_32(event->auxtrace_error.tid);
|
|
|
|
event->auxtrace_error.ip = bswap_64(event->auxtrace_error.ip);
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:19 +03:00
|
|
|
static void perf_event__thread_map_swap(union perf_event *event,
|
|
|
|
bool sample_id_all __maybe_unused)
|
|
|
|
{
|
|
|
|
unsigned i;
|
|
|
|
|
|
|
|
event->thread_map.nr = bswap_64(event->thread_map.nr);
|
|
|
|
|
|
|
|
for (i = 0; i < event->thread_map.nr; i++)
|
|
|
|
event->thread_map.entries[i].pid = bswap_64(event->thread_map.entries[i].pid);
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:23 +03:00
|
|
|
static void perf_event__cpu_map_swap(union perf_event *event,
|
|
|
|
bool sample_id_all __maybe_unused)
|
|
|
|
{
|
|
|
|
struct cpu_map_data *data = &event->cpu_map.data;
|
|
|
|
struct cpu_map_entries *cpus;
|
|
|
|
struct cpu_map_mask *mask;
|
|
|
|
unsigned i;
|
|
|
|
|
|
|
|
data->type = bswap_64(data->type);
|
|
|
|
|
|
|
|
switch (data->type) {
|
|
|
|
case PERF_CPU_MAP__CPUS:
|
|
|
|
cpus = (struct cpu_map_entries *)data->data;
|
|
|
|
|
|
|
|
cpus->nr = bswap_16(cpus->nr);
|
|
|
|
|
|
|
|
for (i = 0; i < cpus->nr; i++)
|
|
|
|
cpus->cpu[i] = bswap_16(cpus->cpu[i]);
|
|
|
|
break;
|
|
|
|
case PERF_CPU_MAP__MASK:
|
|
|
|
mask = (struct cpu_map_mask *) data->data;
|
|
|
|
|
|
|
|
mask->nr = bswap_16(mask->nr);
|
|
|
|
mask->long_size = bswap_16(mask->long_size);
|
|
|
|
|
|
|
|
switch (mask->long_size) {
|
|
|
|
case 4: mem_bswap_32(&mask->mask, mask->nr); break;
|
|
|
|
case 8: mem_bswap_64(&mask->mask, mask->nr); break;
|
|
|
|
default:
|
|
|
|
pr_err("cpu_map swap: unsupported long size\n");
|
|
|
|
}
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:27 +03:00
|
|
|
static void perf_event__stat_config_swap(union perf_event *event,
|
|
|
|
bool sample_id_all __maybe_unused)
|
|
|
|
{
|
|
|
|
u64 size;
|
|
|
|
|
|
|
|
size = event->stat_config.nr * sizeof(event->stat_config.data[0]);
|
|
|
|
size += 1; /* nr item itself */
|
|
|
|
mem_bswap_64(&event->stat_config.nr, size);
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:30 +03:00
|
|
|
static void perf_event__stat_swap(union perf_event *event,
|
|
|
|
bool sample_id_all __maybe_unused)
|
|
|
|
{
|
|
|
|
event->stat.id = bswap_64(event->stat.id);
|
|
|
|
event->stat.thread = bswap_32(event->stat.thread);
|
|
|
|
event->stat.cpu = bswap_32(event->stat.cpu);
|
|
|
|
event->stat.val = bswap_64(event->stat.val);
|
|
|
|
event->stat.ena = bswap_64(event->stat.ena);
|
|
|
|
event->stat.run = bswap_64(event->stat.run);
|
|
|
|
}
|
|
|
|
|
2015-10-25 17:51:33 +03:00
|
|
|
static void perf_event__stat_round_swap(union perf_event *event,
|
|
|
|
bool sample_id_all __maybe_unused)
|
|
|
|
{
|
|
|
|
event->stat_round.type = bswap_64(event->stat_round.type);
|
|
|
|
event->stat_round.time = bswap_64(event->stat_round.time);
|
|
|
|
}
|
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
typedef void (*perf_event__swap_op)(union perf_event *event,
|
|
|
|
bool sample_id_all);
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
|
2011-01-29 19:01:45 +03:00
|
|
|
static perf_event__swap_op perf_event__swap_ops[] = {
|
|
|
|
[PERF_RECORD_MMAP] = perf_event__mmap_swap,
|
2013-08-21 14:10:25 +04:00
|
|
|
[PERF_RECORD_MMAP2] = perf_event__mmap2_swap,
|
2011-01-29 19:01:45 +03:00
|
|
|
[PERF_RECORD_COMM] = perf_event__comm_swap,
|
|
|
|
[PERF_RECORD_FORK] = perf_event__task_swap,
|
|
|
|
[PERF_RECORD_EXIT] = perf_event__task_swap,
|
|
|
|
[PERF_RECORD_LOST] = perf_event__all64_swap,
|
|
|
|
[PERF_RECORD_READ] = perf_event__read_swap,
|
2013-09-01 14:36:15 +04:00
|
|
|
[PERF_RECORD_THROTTLE] = perf_event__throttle_swap,
|
|
|
|
[PERF_RECORD_UNTHROTTLE] = perf_event__throttle_swap,
|
2011-01-29 19:01:45 +03:00
|
|
|
[PERF_RECORD_SAMPLE] = perf_event__all64_swap,
|
2015-04-30 17:37:29 +03:00
|
|
|
[PERF_RECORD_AUX] = perf_event__aux_swap,
|
2015-04-30 17:37:30 +03:00
|
|
|
[PERF_RECORD_ITRACE_START] = perf_event__itrace_start_swap,
|
2015-05-10 22:13:15 +03:00
|
|
|
[PERF_RECORD_LOST_SAMPLES] = perf_event__all64_swap,
|
2015-07-21 12:44:03 +03:00
|
|
|
[PERF_RECORD_SWITCH] = perf_event__switch_swap,
|
|
|
|
[PERF_RECORD_SWITCH_CPU_WIDE] = perf_event__switch_swap,
|
2011-07-15 22:34:09 +04:00
|
|
|
[PERF_RECORD_HEADER_ATTR] = perf_event__hdr_attr_swap,
|
2011-01-29 19:01:45 +03:00
|
|
|
[PERF_RECORD_HEADER_EVENT_TYPE] = perf_event__event_type_swap,
|
|
|
|
[PERF_RECORD_HEADER_TRACING_DATA] = perf_event__tracing_data_swap,
|
|
|
|
[PERF_RECORD_HEADER_BUILD_ID] = NULL,
|
2014-10-27 16:49:22 +03:00
|
|
|
[PERF_RECORD_ID_INDEX] = perf_event__all64_swap,
|
2015-04-09 18:53:43 +03:00
|
|
|
[PERF_RECORD_AUXTRACE_INFO] = perf_event__auxtrace_info_swap,
|
|
|
|
[PERF_RECORD_AUXTRACE] = perf_event__auxtrace_swap,
|
2015-04-09 18:53:47 +03:00
|
|
|
[PERF_RECORD_AUXTRACE_ERROR] = perf_event__auxtrace_error_swap,
|
2015-10-25 17:51:19 +03:00
|
|
|
[PERF_RECORD_THREAD_MAP] = perf_event__thread_map_swap,
|
2015-10-25 17:51:23 +03:00
|
|
|
[PERF_RECORD_CPU_MAP] = perf_event__cpu_map_swap,
|
2015-10-25 17:51:27 +03:00
|
|
|
[PERF_RECORD_STAT_CONFIG] = perf_event__stat_config_swap,
|
2015-10-25 17:51:30 +03:00
|
|
|
[PERF_RECORD_STAT] = perf_event__stat_swap,
|
2015-10-25 17:51:33 +03:00
|
|
|
[PERF_RECORD_STAT_ROUND] = perf_event__stat_round_swap,
|
2015-10-25 17:51:36 +03:00
|
|
|
[PERF_RECORD_EVENT_UPDATE] = perf_event__event_update_swap,
|
2016-03-08 11:38:44 +03:00
|
|
|
[PERF_RECORD_TIME_CONV] = perf_event__all64_swap,
|
2011-01-29 19:01:45 +03:00
|
|
|
[PERF_RECORD_HEADER_MAX] = NULL,
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
};
|
|
|
|
|
2010-05-03 17:14:33 +04:00
|
|
|
/*
|
|
|
|
* When perf record finishes a pass on every buffers, it records this pseudo
|
|
|
|
* event.
|
|
|
|
* We record the max timestamp t found in the pass n.
|
|
|
|
* Assuming these timestamps are monotonic across cpus, we know that if
|
|
|
|
* a buffer still has events with timestamps below t, they will be all
|
|
|
|
* available and then read in the pass n + 1.
|
|
|
|
* Hence when we start to read the pass n + 2, we can safely flush every
|
|
|
|
* events with timestamps below t.
|
|
|
|
*
|
|
|
|
* ============ PASS n =================
|
|
|
|
* CPU 0 | CPU 1
|
|
|
|
* |
|
|
|
|
* cnt1 timestamps | cnt2 timestamps
|
|
|
|
* 1 | 2
|
|
|
|
* 2 | 3
|
|
|
|
* - | 4 <--- max recorded
|
|
|
|
*
|
|
|
|
* ============ PASS n + 1 ==============
|
|
|
|
* CPU 0 | CPU 1
|
|
|
|
* |
|
|
|
|
* cnt1 timestamps | cnt2 timestamps
|
|
|
|
* 3 | 5
|
|
|
|
* 4 | 6
|
|
|
|
* 5 | 7 <---- max recorded
|
|
|
|
*
|
|
|
|
* Flush every events below timestamp 4
|
|
|
|
*
|
|
|
|
* ============ PASS n + 2 ==============
|
|
|
|
* CPU 0 | CPU 1
|
|
|
|
* |
|
|
|
|
* cnt1 timestamps | cnt2 timestamps
|
|
|
|
* 6 | 8
|
|
|
|
* 7 | 9
|
|
|
|
* - | 10
|
|
|
|
*
|
|
|
|
* Flush every events below timestamp 7
|
|
|
|
* etc...
|
|
|
|
*/
|
2015-03-03 17:58:45 +03:00
|
|
|
static int process_finished_round(struct perf_tool *tool __maybe_unused,
|
2012-09-11 02:15:03 +04:00
|
|
|
union perf_event *event __maybe_unused,
|
2015-03-03 18:37:54 +03:00
|
|
|
struct ordered_events *oe)
|
2010-05-03 17:14:33 +04:00
|
|
|
{
|
2015-06-23 10:52:48 +03:00
|
|
|
if (dump_trace)
|
|
|
|
fprintf(stdout, "\n");
|
2015-03-03 17:58:45 +03:00
|
|
|
return ordered_events__flush(oe, OE_FLUSH__ROUND);
|
2010-05-03 17:14:33 +04:00
|
|
|
}
|
|
|
|
|
2015-03-03 17:58:45 +03:00
|
|
|
int perf_session__queue_event(struct perf_session *s, union perf_event *event,
|
|
|
|
struct perf_sample *sample, u64 file_offset)
|
2010-04-24 02:04:12 +04:00
|
|
|
{
|
2015-03-03 19:02:24 +03:00
|
|
|
return ordered_events__queue(&s->ordered_events, event, sample, file_offset);
|
perf session: Parse sample earlier
At perf_session__process_event, so that we reduce the number of lines in eache
tool sample processing routine that now receives a sample_data pointer already
parsed.
This will also be useful in the next patch, where we'll allow sample the
identity fields in MMAP, FORK, EXIT, etc, when it will be possible to see (cpu,
timestamp) just after before every event.
Also validate callchains in perf_session__process_event, i.e. as early as
possible, and keep a counter of the number of events discarded due to invalid
callchains, warning the user about it if it happens.
There is an assumption that was kept that all events have the same sample_type,
that will be dealt with in the future, when this preexisting limitation will be
removed.
Tested-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ian Munsie <imunsie@au1.ibm.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Ian Munsie <imunsie@au1.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <1291318772-30880-4-git-send-email-acme@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-12-02 19:10:21 +03:00
|
|
|
}
|
2010-04-24 02:04:12 +04:00
|
|
|
|
2015-01-05 21:23:05 +03:00
|
|
|
static void callchain__lbr_callstack_printf(struct perf_sample *sample)
|
perf session: Parse sample earlier
At perf_session__process_event, so that we reduce the number of lines in eache
tool sample processing routine that now receives a sample_data pointer already
parsed.
This will also be useful in the next patch, where we'll allow sample the
identity fields in MMAP, FORK, EXIT, etc, when it will be possible to see (cpu,
timestamp) just after before every event.
Also validate callchains in perf_session__process_event, i.e. as early as
possible, and keep a counter of the number of events discarded due to invalid
callchains, warning the user about it if it happens.
There is an assumption that was kept that all events have the same sample_type,
that will be dealt with in the future, when this preexisting limitation will be
removed.
Tested-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ian Munsie <imunsie@au1.ibm.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Ian Munsie <imunsie@au1.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <1291318772-30880-4-git-send-email-acme@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-12-02 19:10:21 +03:00
|
|
|
{
|
2015-01-05 21:23:05 +03:00
|
|
|
struct ip_callchain *callchain = sample->callchain;
|
|
|
|
struct branch_stack *lbr_stack = sample->branch_stack;
|
|
|
|
u64 kernel_callchain_nr = callchain->nr;
|
perf session: Parse sample earlier
At perf_session__process_event, so that we reduce the number of lines in eache
tool sample processing routine that now receives a sample_data pointer already
parsed.
This will also be useful in the next patch, where we'll allow sample the
identity fields in MMAP, FORK, EXIT, etc, when it will be possible to see (cpu,
timestamp) just after before every event.
Also validate callchains in perf_session__process_event, i.e. as early as
possible, and keep a counter of the number of events discarded due to invalid
callchains, warning the user about it if it happens.
There is an assumption that was kept that all events have the same sample_type,
that will be dealt with in the future, when this preexisting limitation will be
removed.
Tested-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ian Munsie <imunsie@au1.ibm.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Ian Munsie <imunsie@au1.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <1291318772-30880-4-git-send-email-acme@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-12-02 19:10:21 +03:00
|
|
|
unsigned int i;
|
2010-04-24 02:04:12 +04:00
|
|
|
|
2015-01-05 21:23:05 +03:00
|
|
|
for (i = 0; i < kernel_callchain_nr; i++) {
|
|
|
|
if (callchain->ips[i] == PERF_CONTEXT_USER)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((i != kernel_callchain_nr) && lbr_stack->nr) {
|
|
|
|
u64 total_nr;
|
|
|
|
/*
|
|
|
|
* LBR callstack can only get user call chain,
|
|
|
|
* i is kernel call chain number,
|
|
|
|
* 1 is PERF_CONTEXT_USER.
|
|
|
|
*
|
|
|
|
* The user call chain is stored in LBR registers.
|
|
|
|
* LBR are pair registers. The caller is stored
|
|
|
|
* in "from" register, while the callee is stored
|
|
|
|
* in "to" register.
|
|
|
|
* For example, there is a call stack
|
|
|
|
* "A"->"B"->"C"->"D".
|
|
|
|
* The LBR registers will recorde like
|
|
|
|
* "C"->"D", "B"->"C", "A"->"B".
|
|
|
|
* So only the first "to" register and all "from"
|
|
|
|
* registers are needed to construct the whole stack.
|
|
|
|
*/
|
|
|
|
total_nr = i + 1 + lbr_stack->nr + 1;
|
|
|
|
kernel_callchain_nr = i + 1;
|
|
|
|
|
|
|
|
printf("... LBR call chain: nr:%" PRIu64 "\n", total_nr);
|
|
|
|
|
|
|
|
for (i = 0; i < kernel_callchain_nr; i++)
|
|
|
|
printf("..... %2d: %016" PRIx64 "\n",
|
|
|
|
i, callchain->ips[i]);
|
|
|
|
|
|
|
|
printf("..... %2d: %016" PRIx64 "\n",
|
|
|
|
(int)(kernel_callchain_nr), lbr_stack->entries[0].to);
|
|
|
|
for (i = 0; i < lbr_stack->nr; i++)
|
|
|
|
printf("..... %2d: %016" PRIx64 "\n",
|
|
|
|
(int)(i + kernel_callchain_nr + 1), lbr_stack->entries[i].from);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void callchain__printf(struct perf_evsel *evsel,
|
|
|
|
struct perf_sample *sample)
|
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
struct ip_callchain *callchain = sample->callchain;
|
|
|
|
|
2016-04-18 16:35:03 +03:00
|
|
|
if (perf_evsel__has_branch_callstack(evsel))
|
2015-01-05 21:23:05 +03:00
|
|
|
callchain__lbr_callstack_printf(sample);
|
|
|
|
|
|
|
|
printf("... FP chain: nr:%" PRIu64 "\n", callchain->nr);
|
perf session: Parse sample earlier
At perf_session__process_event, so that we reduce the number of lines in eache
tool sample processing routine that now receives a sample_data pointer already
parsed.
This will also be useful in the next patch, where we'll allow sample the
identity fields in MMAP, FORK, EXIT, etc, when it will be possible to see (cpu,
timestamp) just after before every event.
Also validate callchains in perf_session__process_event, i.e. as early as
possible, and keep a counter of the number of events discarded due to invalid
callchains, warning the user about it if it happens.
There is an assumption that was kept that all events have the same sample_type,
that will be dealt with in the future, when this preexisting limitation will be
removed.
Tested-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ian Munsie <imunsie@au1.ibm.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Ian Munsie <imunsie@au1.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <1291318772-30880-4-git-send-email-acme@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-12-02 19:10:21 +03:00
|
|
|
|
2015-01-05 21:23:05 +03:00
|
|
|
for (i = 0; i < callchain->nr; i++)
|
2011-01-23 01:37:02 +03:00
|
|
|
printf("..... %2d: %016" PRIx64 "\n",
|
2015-01-05 21:23:05 +03:00
|
|
|
i, callchain->ips[i]);
|
2010-04-24 02:04:12 +04:00
|
|
|
}
|
|
|
|
|
2012-02-10 02:21:01 +04:00
|
|
|
static void branch_stack__printf(struct perf_sample *sample)
|
|
|
|
{
|
|
|
|
uint64_t i;
|
|
|
|
|
|
|
|
printf("... branch stack: nr:%" PRIu64 "\n", sample->branch_stack->nr);
|
|
|
|
|
2015-07-18 18:24:46 +03:00
|
|
|
for (i = 0; i < sample->branch_stack->nr; i++) {
|
|
|
|
struct branch_entry *e = &sample->branch_stack->entries[i];
|
|
|
|
|
|
|
|
printf("..... %2"PRIu64": %016" PRIx64 " -> %016" PRIx64 " %hu cycles %s%s%s%s %x\n",
|
|
|
|
i, e->from, e->to,
|
2017-02-17 18:27:26 +03:00
|
|
|
(unsigned short)e->flags.cycles,
|
2015-07-18 18:24:46 +03:00
|
|
|
e->flags.mispred ? "M" : " ",
|
|
|
|
e->flags.predicted ? "P" : " ",
|
|
|
|
e->flags.abort ? "A" : " ",
|
|
|
|
e->flags.in_tx ? "T" : " ",
|
|
|
|
(unsigned)e->flags.reserved);
|
|
|
|
}
|
2012-02-10 02:21:01 +04:00
|
|
|
}
|
|
|
|
|
2012-08-07 17:20:45 +04:00
|
|
|
static void regs_dump__printf(u64 mask, u64 *regs)
|
|
|
|
{
|
|
|
|
unsigned rid, i = 0;
|
|
|
|
|
|
|
|
for_each_set_bit(rid, (unsigned long *) &mask, sizeof(mask) * 8) {
|
|
|
|
u64 val = regs[i++];
|
|
|
|
|
|
|
|
printf(".... %-5s 0x%" PRIx64 "\n",
|
|
|
|
perf_reg_name(rid), val);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-09-24 15:48:39 +04:00
|
|
|
static const char *regs_abi[] = {
|
|
|
|
[PERF_SAMPLE_REGS_ABI_NONE] = "none",
|
|
|
|
[PERF_SAMPLE_REGS_ABI_32] = "32-bit",
|
|
|
|
[PERF_SAMPLE_REGS_ABI_64] = "64-bit",
|
|
|
|
};
|
|
|
|
|
|
|
|
static inline const char *regs_dump_abi(struct regs_dump *d)
|
|
|
|
{
|
|
|
|
if (d->abi > PERF_SAMPLE_REGS_ABI_64)
|
|
|
|
return "unknown";
|
|
|
|
|
|
|
|
return regs_abi[d->abi];
|
|
|
|
}
|
|
|
|
|
|
|
|
static void regs__printf(const char *type, struct regs_dump *regs)
|
|
|
|
{
|
|
|
|
u64 mask = regs->mask;
|
|
|
|
|
|
|
|
printf("... %s regs: mask 0x%" PRIx64 " ABI %s\n",
|
|
|
|
type,
|
|
|
|
mask,
|
|
|
|
regs_dump_abi(regs));
|
|
|
|
|
|
|
|
regs_dump__printf(mask, regs->regs);
|
|
|
|
}
|
|
|
|
|
2014-01-07 16:47:25 +04:00
|
|
|
static void regs_user__printf(struct perf_sample *sample)
|
2012-08-07 17:20:45 +04:00
|
|
|
{
|
|
|
|
struct regs_dump *user_regs = &sample->user_regs;
|
|
|
|
|
2014-09-24 15:48:39 +04:00
|
|
|
if (user_regs->regs)
|
|
|
|
regs__printf("user", user_regs);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void regs_intr__printf(struct perf_sample *sample)
|
|
|
|
{
|
|
|
|
struct regs_dump *intr_regs = &sample->intr_regs;
|
|
|
|
|
|
|
|
if (intr_regs->regs)
|
|
|
|
regs__printf("intr", intr_regs);
|
2012-08-07 17:20:45 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
static void stack_user__printf(struct stack_dump *dump)
|
|
|
|
{
|
|
|
|
printf("... ustack: size %" PRIu64 ", offset 0x%x\n",
|
|
|
|
dump->size, dump->offset);
|
|
|
|
}
|
|
|
|
|
2015-02-14 21:08:51 +03:00
|
|
|
static void perf_evlist__print_tstamp(struct perf_evlist *evlist,
|
2011-01-29 19:01:45 +03:00
|
|
|
union perf_event *event,
|
2011-01-29 18:02:00 +03:00
|
|
|
struct perf_sample *sample)
|
2010-12-02 15:25:28 +03:00
|
|
|
{
|
2015-02-14 21:08:51 +03:00
|
|
|
u64 sample_type = __perf_evlist__combined_sample_type(evlist);
|
2012-08-02 02:15:52 +04:00
|
|
|
|
2010-12-02 15:25:28 +03:00
|
|
|
if (event->header.type != PERF_RECORD_SAMPLE &&
|
2015-02-14 21:08:51 +03:00
|
|
|
!perf_evlist__sample_id_all(evlist)) {
|
2010-12-02 15:25:28 +03:00
|
|
|
fputs("-1 -1 ", stdout);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2012-08-02 02:15:52 +04:00
|
|
|
if ((sample_type & PERF_SAMPLE_CPU))
|
2010-12-02 15:25:28 +03:00
|
|
|
printf("%u ", sample->cpu);
|
|
|
|
|
2012-08-02 02:15:52 +04:00
|
|
|
if (sample_type & PERF_SAMPLE_TIME)
|
2011-01-23 01:37:02 +03:00
|
|
|
printf("%" PRIu64 " ", sample->time);
|
2010-12-02 15:25:28 +03:00
|
|
|
}
|
|
|
|
|
2012-10-10 19:38:13 +04:00
|
|
|
static void sample_read__printf(struct perf_sample *sample, u64 read_format)
|
|
|
|
{
|
|
|
|
printf("... sample_read:\n");
|
|
|
|
|
|
|
|
if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
|
|
|
|
printf("...... time enabled %016" PRIx64 "\n",
|
|
|
|
sample->read.time_enabled);
|
|
|
|
|
|
|
|
if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
|
|
|
|
printf("...... time running %016" PRIx64 "\n",
|
|
|
|
sample->read.time_running);
|
|
|
|
|
|
|
|
if (read_format & PERF_FORMAT_GROUP) {
|
|
|
|
u64 i;
|
|
|
|
|
|
|
|
printf(".... group nr %" PRIu64 "\n", sample->read.group.nr);
|
|
|
|
|
|
|
|
for (i = 0; i < sample->read.group.nr; i++) {
|
|
|
|
struct sample_read_value *value;
|
|
|
|
|
|
|
|
value = &sample->read.group.values[i];
|
|
|
|
printf("..... id %016" PRIx64
|
|
|
|
", value %016" PRIx64 "\n",
|
|
|
|
value->id, value->value);
|
|
|
|
}
|
|
|
|
} else
|
|
|
|
printf("..... id %016" PRIx64 ", value %016" PRIx64 "\n",
|
|
|
|
sample->read.one.id, sample->read.one.value);
|
|
|
|
}
|
|
|
|
|
2015-02-14 21:08:51 +03:00
|
|
|
static void dump_event(struct perf_evlist *evlist, union perf_event *event,
|
2011-01-29 18:02:00 +03:00
|
|
|
u64 file_offset, struct perf_sample *sample)
|
2010-12-07 15:48:47 +03:00
|
|
|
{
|
|
|
|
if (!dump_trace)
|
|
|
|
return;
|
|
|
|
|
2011-01-23 01:37:02 +03:00
|
|
|
printf("\n%#" PRIx64 " [%#x]: event: %d\n",
|
|
|
|
file_offset, event->header.size, event->header.type);
|
2010-12-07 15:48:47 +03:00
|
|
|
|
|
|
|
trace_event(event);
|
|
|
|
|
|
|
|
if (sample)
|
2015-02-14 21:08:51 +03:00
|
|
|
perf_evlist__print_tstamp(evlist, event, sample);
|
2010-12-07 15:48:47 +03:00
|
|
|
|
2011-01-23 01:37:02 +03:00
|
|
|
printf("%#" PRIx64 " [%#x]: PERF_RECORD_%s", file_offset,
|
2011-01-29 19:01:45 +03:00
|
|
|
event->header.size, perf_event__name(event->header.type));
|
2010-12-07 15:48:47 +03:00
|
|
|
}
|
|
|
|
|
2012-08-07 17:20:45 +04:00
|
|
|
static void dump_sample(struct perf_evsel *evsel, union perf_event *event,
|
2011-01-29 18:02:00 +03:00
|
|
|
struct perf_sample *sample)
|
2010-12-07 15:48:47 +03:00
|
|
|
{
|
2012-08-02 02:15:52 +04:00
|
|
|
u64 sample_type;
|
|
|
|
|
2010-12-09 17:20:20 +03:00
|
|
|
if (!dump_trace)
|
|
|
|
return;
|
|
|
|
|
2014-02-26 07:43:46 +04:00
|
|
|
printf("(IP, 0x%x): %d/%d: %#" PRIx64 " period: %" PRIu64 " addr: %#" PRIx64 "\n",
|
2011-01-23 01:37:02 +03:00
|
|
|
event->header.misc, sample->pid, sample->tid, sample->ip,
|
2011-05-30 23:08:23 +04:00
|
|
|
sample->period, sample->addr);
|
2010-12-07 15:48:47 +03:00
|
|
|
|
2012-08-07 17:20:45 +04:00
|
|
|
sample_type = evsel->attr.sample_type;
|
2012-08-02 02:15:52 +04:00
|
|
|
|
|
|
|
if (sample_type & PERF_SAMPLE_CALLCHAIN)
|
2015-01-05 21:23:05 +03:00
|
|
|
callchain__printf(evsel, sample);
|
2012-02-10 02:21:01 +04:00
|
|
|
|
2016-04-18 16:35:03 +03:00
|
|
|
if ((sample_type & PERF_SAMPLE_BRANCH_STACK) && !perf_evsel__has_branch_callstack(evsel))
|
2012-02-10 02:21:01 +04:00
|
|
|
branch_stack__printf(sample);
|
2012-08-07 17:20:45 +04:00
|
|
|
|
|
|
|
if (sample_type & PERF_SAMPLE_REGS_USER)
|
2014-01-07 16:47:25 +04:00
|
|
|
regs_user__printf(sample);
|
2012-08-07 17:20:45 +04:00
|
|
|
|
2014-09-24 15:48:39 +04:00
|
|
|
if (sample_type & PERF_SAMPLE_REGS_INTR)
|
|
|
|
regs_intr__printf(sample);
|
|
|
|
|
2012-08-07 17:20:45 +04:00
|
|
|
if (sample_type & PERF_SAMPLE_STACK_USER)
|
|
|
|
stack_user__printf(&sample->user_stack);
|
2013-01-24 19:10:29 +04:00
|
|
|
|
|
|
|
if (sample_type & PERF_SAMPLE_WEIGHT)
|
|
|
|
printf("... weight: %" PRIu64 "\n", sample->weight);
|
2013-01-24 19:10:35 +04:00
|
|
|
|
|
|
|
if (sample_type & PERF_SAMPLE_DATA_SRC)
|
|
|
|
printf(" . data_src: 0x%"PRIx64"\n", sample->data_src);
|
2012-10-10 19:38:13 +04:00
|
|
|
|
2013-09-20 18:40:43 +04:00
|
|
|
if (sample_type & PERF_SAMPLE_TRANSACTION)
|
|
|
|
printf("... transaction: %" PRIx64 "\n", sample->transaction);
|
|
|
|
|
2012-10-10 19:38:13 +04:00
|
|
|
if (sample_type & PERF_SAMPLE_READ)
|
|
|
|
sample_read__printf(sample, evsel->attr.read_format);
|
2010-12-07 15:48:47 +03:00
|
|
|
}
|
|
|
|
|
2015-02-14 20:26:15 +03:00
|
|
|
static struct machine *machines__find_for_cpumode(struct machines *machines,
|
2013-08-27 12:23:06 +04:00
|
|
|
union perf_event *event,
|
|
|
|
struct perf_sample *sample)
|
2011-11-28 13:56:39 +04:00
|
|
|
{
|
2013-12-20 22:41:47 +04:00
|
|
|
struct machine *machine;
|
2011-11-28 13:56:39 +04:00
|
|
|
|
2012-07-21 03:25:48 +04:00
|
|
|
if (perf_guest &&
|
2016-03-23 00:23:43 +03:00
|
|
|
((sample->cpumode == PERF_RECORD_MISC_GUEST_KERNEL) ||
|
|
|
|
(sample->cpumode == PERF_RECORD_MISC_GUEST_USER))) {
|
2012-04-09 12:22:23 +04:00
|
|
|
u32 pid;
|
|
|
|
|
2013-08-21 14:10:25 +04:00
|
|
|
if (event->header.type == PERF_RECORD_MMAP
|
|
|
|
|| event->header.type == PERF_RECORD_MMAP2)
|
2012-04-09 12:22:23 +04:00
|
|
|
pid = event->mmap.pid;
|
|
|
|
else
|
2013-08-27 12:23:06 +04:00
|
|
|
pid = sample->pid;
|
2012-04-09 12:22:23 +04:00
|
|
|
|
2015-02-14 20:26:15 +03:00
|
|
|
machine = machines__find(machines, pid);
|
2013-12-20 22:41:47 +04:00
|
|
|
if (!machine)
|
perf kvm record/report: 'unprocessable sample' error while recording/reporting guest data
While recording guest samples in host using perf kvm record, it will
populate unprocessable sample error, though samples will be recorded
properly. While generating report using perf kvm report, no samples will
be processed and same error will populate. We have seen this behaviour
with upstream perf(4.4-rc3) on x86 and ppc64 hardware.
Reason behind this failure is, when it tries to fetch machine from
rb_tree of machines, it fails. As a part of tracing a bug, we figured
out that this code was incorrectly refactored in commit 54245fdc3576
("perf session: Remove wrappers to machines__find").
This patch will change the functionality such that if it can't fetch
machine in first trial, it will create one node of machine and add that to
rb_tree. So next time when it tries to fetch same machine from rb_tree,
it won't fail. Actually it was the case before refactoring of code in
aforementioned commit.
This patch is generated from acme perf/core branch.
Below I've mention an example that demonstrate the behaviour before and
after applying patch.
Before applying patch:
[Note: One needs to run guest before recording data in host]
ravi@ravi-bangoria:~$ ./perf kvm record -a
Warning:
5903 unprocessable samples recorded.
Do you have a KVM guest running and not using 'perf kvm'?
[ perf record: Captured and wrote 1.409 MB perf.data.guest (285 samples) ]
ravi@ravi-bangoria:~$ ./perf kvm report --stdio
Warning:
5903 unprocessable samples recorded.
Do you have a KVM guest running and not using 'perf kvm'?
# To display the perf.data header info, please use --header/--header-only options.
#
# Total Lost Samples: 0
#
# Samples: 285 of event 'cycles'
# Event count (approx.): 88715406
#
# Overhead Command Shared Object Symbol
# ........ ....... ............. ......
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
After applying patch:
ravi@ravi-bangoria:~$ ./perf kvm record -a
[ perf record: Captured and wrote 1.188 MB perf.data.guest (17 samples) ]
ravi@ravi-bangoria:~$ ./perf kvm report --stdio
# To display the perf.data header info, please use --header/--header-only options.
#
# Total Lost Samples: 0
#
# Samples: 17 of event 'cycles'
# Event count (approx.): 700746
#
# Overhead Command Shared Object Symbol
# ........ ....... ................ ......................
#
34.19% :5758 [unknown] [g] 0xffffffff818682ab
22.79% :5758 [unknown] [g] 0xffffffff812dc7f8
22.79% :5758 [unknown] [g] 0xffffffff818650d0
14.83% :5758 [unknown] [g] 0xffffffff8161a1b6
2.49% :5758 [unknown] [g] 0xffffffff818692bf
0.48% :5758 [unknown] [g] 0xffffffff81869253
0.05% :5758 [unknown] [g] 0xffffffff81869250
Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.vnet.ibm.com>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: stable@vger.kernel.org # v3.19+
Fixes: 54245fdc3576 ("perf session: Remove wrappers to machines__find")
Link: http://lkml.kernel.org/r/1449471302-11283-1-git-send-email-ravi.bangoria@linux.vnet.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-07 09:55:02 +03:00
|
|
|
machine = machines__findnew(machines, DEFAULT_GUEST_KERNEL_ID);
|
2013-12-20 22:41:47 +04:00
|
|
|
return machine;
|
2012-04-09 12:22:23 +04:00
|
|
|
}
|
2011-11-28 13:56:39 +04:00
|
|
|
|
2015-02-14 20:26:15 +03:00
|
|
|
return &machines->host;
|
2011-11-28 13:56:39 +04:00
|
|
|
}
|
|
|
|
|
2015-02-14 21:05:28 +03:00
|
|
|
static int deliver_sample_value(struct perf_evlist *evlist,
|
2012-10-10 20:52:24 +04:00
|
|
|
struct perf_tool *tool,
|
|
|
|
union perf_event *event,
|
|
|
|
struct perf_sample *sample,
|
|
|
|
struct sample_read_value *v,
|
|
|
|
struct machine *machine)
|
|
|
|
{
|
2015-02-14 21:05:28 +03:00
|
|
|
struct perf_sample_id *sid = perf_evlist__id2sid(evlist, v->id);
|
2012-10-10 20:52:24 +04:00
|
|
|
|
|
|
|
if (sid) {
|
|
|
|
sample->id = v->id;
|
|
|
|
sample->period = v->value - sid->period;
|
|
|
|
sid->period = v->value;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!sid || sid->evsel == NULL) {
|
2015-02-14 21:05:28 +03:00
|
|
|
++evlist->stats.nr_unknown_id;
|
2012-10-10 20:52:24 +04:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return tool->sample(tool, event, sample, sid->evsel, machine);
|
|
|
|
}
|
|
|
|
|
2015-02-14 21:05:28 +03:00
|
|
|
static int deliver_sample_group(struct perf_evlist *evlist,
|
2012-10-10 20:52:24 +04:00
|
|
|
struct perf_tool *tool,
|
|
|
|
union perf_event *event,
|
|
|
|
struct perf_sample *sample,
|
|
|
|
struct machine *machine)
|
|
|
|
{
|
|
|
|
int ret = -EINVAL;
|
|
|
|
u64 i;
|
|
|
|
|
|
|
|
for (i = 0; i < sample->read.group.nr; i++) {
|
2015-02-14 21:05:28 +03:00
|
|
|
ret = deliver_sample_value(evlist, tool, event, sample,
|
2012-10-10 20:52:24 +04:00
|
|
|
&sample->read.group.values[i],
|
|
|
|
machine);
|
|
|
|
if (ret)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
2015-02-14 21:05:28 +03:00
|
|
|
perf_evlist__deliver_sample(struct perf_evlist *evlist,
|
2012-10-10 20:52:24 +04:00
|
|
|
struct perf_tool *tool,
|
|
|
|
union perf_event *event,
|
|
|
|
struct perf_sample *sample,
|
|
|
|
struct perf_evsel *evsel,
|
|
|
|
struct machine *machine)
|
|
|
|
{
|
|
|
|
/* We know evsel != NULL. */
|
|
|
|
u64 sample_type = evsel->attr.sample_type;
|
|
|
|
u64 read_format = evsel->attr.read_format;
|
|
|
|
|
2017-01-17 16:22:33 +03:00
|
|
|
/* Standard sample delivery. */
|
2012-10-10 20:52:24 +04:00
|
|
|
if (!(sample_type & PERF_SAMPLE_READ))
|
|
|
|
return tool->sample(tool, event, sample, evsel, machine);
|
|
|
|
|
|
|
|
/* For PERF_SAMPLE_READ we have either single or group mode. */
|
|
|
|
if (read_format & PERF_FORMAT_GROUP)
|
2015-02-14 21:05:28 +03:00
|
|
|
return deliver_sample_group(evlist, tool, event, sample,
|
2012-10-10 20:52:24 +04:00
|
|
|
machine);
|
|
|
|
else
|
2015-02-14 21:05:28 +03:00
|
|
|
return deliver_sample_value(evlist, tool, event, sample,
|
2012-10-10 20:52:24 +04:00
|
|
|
&sample->read.one, machine);
|
|
|
|
}
|
|
|
|
|
2015-03-03 18:20:38 +03:00
|
|
|
static int machines__deliver_event(struct machines *machines,
|
|
|
|
struct perf_evlist *evlist,
|
|
|
|
union perf_event *event,
|
|
|
|
struct perf_sample *sample,
|
|
|
|
struct perf_tool *tool, u64 file_offset)
|
2010-12-05 16:32:55 +03:00
|
|
|
{
|
2011-03-15 21:44:01 +03:00
|
|
|
struct perf_evsel *evsel;
|
2011-11-28 13:56:39 +04:00
|
|
|
struct machine *machine;
|
2011-03-15 21:44:01 +03:00
|
|
|
|
2015-02-14 21:08:51 +03:00
|
|
|
dump_event(evlist, event, file_offset, sample);
|
2010-12-07 15:48:55 +03:00
|
|
|
|
2015-02-14 21:05:28 +03:00
|
|
|
evsel = perf_evlist__id2evsel(evlist, sample->id);
|
2011-10-29 18:15:04 +04:00
|
|
|
|
2015-03-03 17:48:12 +03:00
|
|
|
machine = machines__find_for_cpumode(machines, event, sample);
|
2011-11-28 13:56:39 +04:00
|
|
|
|
2010-12-05 16:32:55 +03:00
|
|
|
switch (event->header.type) {
|
|
|
|
case PERF_RECORD_SAMPLE:
|
2011-03-15 21:44:01 +03:00
|
|
|
if (evsel == NULL) {
|
2015-02-14 21:05:28 +03:00
|
|
|
++evlist->stats.nr_unknown_id;
|
2012-04-12 16:21:01 +04:00
|
|
|
return 0;
|
2011-03-15 21:44:01 +03:00
|
|
|
}
|
2015-09-03 15:31:00 +03:00
|
|
|
dump_sample(evsel, event, sample);
|
2012-02-10 21:05:04 +04:00
|
|
|
if (machine == NULL) {
|
2015-02-14 21:05:28 +03:00
|
|
|
++evlist->stats.nr_unprocessable_samples;
|
2012-04-12 16:21:01 +04:00
|
|
|
return 0;
|
2012-02-10 21:05:04 +04:00
|
|
|
}
|
2015-02-14 21:05:28 +03:00
|
|
|
return perf_evlist__deliver_sample(evlist, tool, event, sample, evsel, machine);
|
2010-12-05 16:32:55 +03:00
|
|
|
case PERF_RECORD_MMAP:
|
2011-11-28 14:30:20 +04:00
|
|
|
return tool->mmap(tool, event, sample, machine);
|
2013-08-21 14:10:25 +04:00
|
|
|
case PERF_RECORD_MMAP2:
|
2015-06-17 16:51:10 +03:00
|
|
|
if (event->header.misc & PERF_RECORD_MISC_PROC_MAP_PARSE_TIMEOUT)
|
|
|
|
++evlist->stats.nr_proc_map_timeout;
|
2013-08-21 14:10:25 +04:00
|
|
|
return tool->mmap2(tool, event, sample, machine);
|
2010-12-05 16:32:55 +03:00
|
|
|
case PERF_RECORD_COMM:
|
2011-11-28 14:30:20 +04:00
|
|
|
return tool->comm(tool, event, sample, machine);
|
perf tools: Add PERF_RECORD_NAMESPACES to include namespaces related info
Introduce a new option to record PERF_RECORD_NAMESPACES events emitted
by the kernel when fork, clone, setns or unshare are invoked. And update
perf-record documentation with the new option to record namespace
events.
Committer notes:
Combined it with a later patch to allow printing it via 'perf report -D'
and be able to test the feature introduced in this patch. Had to move
here also perf_ns__name(), that was introduced in another later patch.
Also used PRIu64 and PRIx64 to fix the build in some enfironments wrt:
util/event.c:1129:39: error: format '%lx' expects argument of type 'long unsigned int', but argument 6 has type 'long long unsigned int' [-Werror=format=]
ret += fprintf(fp, "%u/%s: %lu/0x%lx%s", idx
^
Testing it:
# perf record --namespaces -a
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 1.083 MB perf.data (423 samples) ]
#
# perf report -D
<SNIP>
3 2028902078892 0x115140 [0xa0]: PERF_RECORD_NAMESPACES 14783/14783 - nr_namespaces: 7
[0/net: 3/0xf0000081, 1/uts: 3/0xeffffffe, 2/ipc: 3/0xefffffff, 3/pid: 3/0xeffffffc,
4/user: 3/0xeffffffd, 5/mnt: 3/0xf0000000, 6/cgroup: 3/0xeffffffb]
0x1151e0 [0x30]: event: 9
.
. ... raw event: size 48 bytes
. 0000: 09 00 00 00 02 00 30 00 c4 71 82 68 0c 7f 00 00 ......0..q.h....
. 0010: a9 39 00 00 a9 39 00 00 94 28 fe 63 d8 01 00 00 .9...9...(.c....
. 0020: 03 00 00 00 00 00 00 00 ce c4 02 00 00 00 00 00 ................
<SNIP>
NAMESPACES events: 1
<SNIP>
#
Signed-off-by: Hari Bathini <hbathini@linux.vnet.ibm.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Starovoitov <ast@fb.com>
Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>
Cc: Aravinda Prasad <aravinda@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sargun Dhillon <sargun@sargun.me>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/148891930386.25309.18412039920746995488.stgit@hbathini.in.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-03-07 23:41:43 +03:00
|
|
|
case PERF_RECORD_NAMESPACES:
|
|
|
|
return tool->namespaces(tool, event, sample, machine);
|
2010-12-05 16:32:55 +03:00
|
|
|
case PERF_RECORD_FORK:
|
2011-11-28 14:30:20 +04:00
|
|
|
return tool->fork(tool, event, sample, machine);
|
2010-12-05 16:32:55 +03:00
|
|
|
case PERF_RECORD_EXIT:
|
2011-11-28 14:30:20 +04:00
|
|
|
return tool->exit(tool, event, sample, machine);
|
2010-12-05 16:32:55 +03:00
|
|
|
case PERF_RECORD_LOST:
|
2011-11-28 14:30:20 +04:00
|
|
|
if (tool->lost == perf_event__process_lost)
|
2015-02-14 21:05:28 +03:00
|
|
|
evlist->stats.total_lost += event->lost.lost;
|
2011-11-28 14:30:20 +04:00
|
|
|
return tool->lost(tool, event, sample, machine);
|
2015-05-10 22:13:15 +03:00
|
|
|
case PERF_RECORD_LOST_SAMPLES:
|
|
|
|
if (tool->lost_samples == perf_event__process_lost_samples)
|
|
|
|
evlist->stats.total_lost_samples += event->lost_samples.lost;
|
|
|
|
return tool->lost_samples(tool, event, sample, machine);
|
2010-12-05 16:32:55 +03:00
|
|
|
case PERF_RECORD_READ:
|
2011-11-28 14:30:20 +04:00
|
|
|
return tool->read(tool, event, sample, evsel, machine);
|
2010-12-05 16:32:55 +03:00
|
|
|
case PERF_RECORD_THROTTLE:
|
2011-11-28 14:30:20 +04:00
|
|
|
return tool->throttle(tool, event, sample, machine);
|
2010-12-05 16:32:55 +03:00
|
|
|
case PERF_RECORD_UNTHROTTLE:
|
2011-11-28 14:30:20 +04:00
|
|
|
return tool->unthrottle(tool, event, sample, machine);
|
2015-04-30 17:37:29 +03:00
|
|
|
case PERF_RECORD_AUX:
|
2017-03-16 19:41:59 +03:00
|
|
|
if (tool->aux == perf_event__process_aux) {
|
|
|
|
if (event->aux.flags & PERF_AUX_FLAG_TRUNCATED)
|
|
|
|
evlist->stats.total_aux_lost += 1;
|
|
|
|
if (event->aux.flags & PERF_AUX_FLAG_PARTIAL)
|
|
|
|
evlist->stats.total_aux_partial += 1;
|
|
|
|
}
|
2015-04-30 17:37:29 +03:00
|
|
|
return tool->aux(tool, event, sample, machine);
|
2015-04-30 17:37:30 +03:00
|
|
|
case PERF_RECORD_ITRACE_START:
|
|
|
|
return tool->itrace_start(tool, event, sample, machine);
|
2015-07-21 12:44:03 +03:00
|
|
|
case PERF_RECORD_SWITCH:
|
|
|
|
case PERF_RECORD_SWITCH_CPU_WIDE:
|
|
|
|
return tool->context_switch(tool, event, sample, machine);
|
2010-12-05 16:32:55 +03:00
|
|
|
default:
|
2015-02-14 21:05:28 +03:00
|
|
|
++evlist->stats.nr_unknown_events;
|
2010-12-05 16:32:55 +03:00
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-04-09 18:53:48 +03:00
|
|
|
static int perf_session__deliver_event(struct perf_session *session,
|
|
|
|
union perf_event *event,
|
|
|
|
struct perf_sample *sample,
|
|
|
|
struct perf_tool *tool,
|
|
|
|
u64 file_offset)
|
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
ret = auxtrace__process_event(session, event, sample, tool);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
if (ret > 0)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
return machines__deliver_event(&session->machines, session->evlist,
|
|
|
|
event, sample, tool, file_offset);
|
|
|
|
}
|
|
|
|
|
2014-07-23 23:19:58 +04:00
|
|
|
static s64 perf_session__process_user_event(struct perf_session *session,
|
|
|
|
union perf_event *event,
|
|
|
|
u64 file_offset)
|
2009-12-28 02:36:59 +03:00
|
|
|
{
|
2015-03-03 18:37:54 +03:00
|
|
|
struct ordered_events *oe = &session->ordered_events;
|
2015-03-31 18:48:16 +03:00
|
|
|
struct perf_tool *tool = session->tool;
|
2013-10-15 18:27:34 +04:00
|
|
|
int fd = perf_data_file__fd(session->file);
|
2011-11-12 04:45:41 +04:00
|
|
|
int err;
|
|
|
|
|
2015-02-14 21:08:51 +03:00
|
|
|
dump_event(session->evlist, event, file_offset, NULL);
|
2009-12-28 02:36:59 +03:00
|
|
|
|
2010-12-05 16:32:55 +03:00
|
|
|
/* These events are processed right away */
|
2009-12-28 02:36:59 +03:00
|
|
|
switch (event->header.type) {
|
2010-04-02 08:59:19 +04:00
|
|
|
case PERF_RECORD_HEADER_ATTR:
|
2013-07-04 17:20:21 +04:00
|
|
|
err = tool->attr(tool, event, &session->evlist);
|
2014-07-31 10:00:45 +04:00
|
|
|
if (err == 0) {
|
2012-08-02 02:31:00 +04:00
|
|
|
perf_session__set_id_hdr_size(session);
|
2014-07-31 10:00:45 +04:00
|
|
|
perf_session__set_comm_exec(session);
|
|
|
|
}
|
2011-11-12 04:45:41 +04:00
|
|
|
return err;
|
2015-10-25 17:51:36 +03:00
|
|
|
case PERF_RECORD_EVENT_UPDATE:
|
|
|
|
return tool->event_update(tool, event, &session->evlist);
|
2014-02-04 18:37:48 +04:00
|
|
|
case PERF_RECORD_HEADER_EVENT_TYPE:
|
|
|
|
/*
|
|
|
|
* Depreceated, but we need to handle it for sake
|
|
|
|
* of old data files create in pipe mode.
|
|
|
|
*/
|
|
|
|
return 0;
|
2010-04-02 08:59:21 +04:00
|
|
|
case PERF_RECORD_HEADER_TRACING_DATA:
|
|
|
|
/* setup for reading amidst mmap */
|
2013-10-15 18:27:34 +04:00
|
|
|
lseek(fd, file_offset, SEEK_SET);
|
2013-07-04 17:20:21 +04:00
|
|
|
return tool->tracing_data(tool, event, session);
|
2010-04-02 08:59:22 +04:00
|
|
|
case PERF_RECORD_HEADER_BUILD_ID:
|
2011-11-28 14:30:20 +04:00
|
|
|
return tool->build_id(tool, event, session);
|
2010-05-03 17:14:33 +04:00
|
|
|
case PERF_RECORD_FINISHED_ROUND:
|
2015-03-03 18:37:54 +03:00
|
|
|
return tool->finished_round(tool, event, oe);
|
2014-10-27 16:49:22 +03:00
|
|
|
case PERF_RECORD_ID_INDEX:
|
|
|
|
return tool->id_index(tool, event, session);
|
2015-04-09 18:53:43 +03:00
|
|
|
case PERF_RECORD_AUXTRACE_INFO:
|
|
|
|
return tool->auxtrace_info(tool, event, session);
|
|
|
|
case PERF_RECORD_AUXTRACE:
|
|
|
|
/* setup for reading amidst mmap */
|
|
|
|
lseek(fd, file_offset + event->header.size, SEEK_SET);
|
|
|
|
return tool->auxtrace(tool, event, session);
|
2015-04-09 18:53:47 +03:00
|
|
|
case PERF_RECORD_AUXTRACE_ERROR:
|
2015-04-09 18:53:50 +03:00
|
|
|
perf_session__auxtrace_error_inc(session, event);
|
2015-04-09 18:53:47 +03:00
|
|
|
return tool->auxtrace_error(tool, event, session);
|
2015-10-25 17:51:19 +03:00
|
|
|
case PERF_RECORD_THREAD_MAP:
|
|
|
|
return tool->thread_map(tool, event, session);
|
2015-10-25 17:51:23 +03:00
|
|
|
case PERF_RECORD_CPU_MAP:
|
|
|
|
return tool->cpu_map(tool, event, session);
|
2015-10-25 17:51:27 +03:00
|
|
|
case PERF_RECORD_STAT_CONFIG:
|
|
|
|
return tool->stat_config(tool, event, session);
|
2015-10-25 17:51:30 +03:00
|
|
|
case PERF_RECORD_STAT:
|
|
|
|
return tool->stat(tool, event, session);
|
2015-10-25 17:51:33 +03:00
|
|
|
case PERF_RECORD_STAT_ROUND:
|
|
|
|
return tool->stat_round(tool, event, session);
|
2016-03-08 11:38:44 +03:00
|
|
|
case PERF_RECORD_TIME_CONV:
|
|
|
|
session->time_conv = event->time_conv;
|
|
|
|
return tool->time_conv(tool, event, session);
|
2009-12-28 02:36:59 +03:00
|
|
|
default:
|
2010-12-07 15:49:01 +03:00
|
|
|
return -EINVAL;
|
2009-12-28 02:36:59 +03:00
|
|
|
}
|
2010-12-07 15:49:01 +03:00
|
|
|
}
|
|
|
|
|
2014-10-27 16:49:23 +03:00
|
|
|
int perf_session__deliver_synth_event(struct perf_session *session,
|
|
|
|
union perf_event *event,
|
2015-03-03 17:58:45 +03:00
|
|
|
struct perf_sample *sample)
|
2014-10-27 16:49:23 +03:00
|
|
|
{
|
2015-03-03 17:48:12 +03:00
|
|
|
struct perf_evlist *evlist = session->evlist;
|
2015-03-31 18:48:16 +03:00
|
|
|
struct perf_tool *tool = session->tool;
|
2015-03-03 17:48:12 +03:00
|
|
|
|
|
|
|
events_stats__inc(&evlist->stats, event->header.type);
|
2014-10-27 16:49:23 +03:00
|
|
|
|
|
|
|
if (event->header.type >= PERF_RECORD_USER_TYPE_START)
|
2015-03-03 17:58:45 +03:00
|
|
|
return perf_session__process_user_event(session, event, 0);
|
2014-10-27 16:49:23 +03:00
|
|
|
|
2015-03-03 17:48:12 +03:00
|
|
|
return machines__deliver_event(&session->machines, evlist, event, sample, tool, 0);
|
2014-10-27 16:49:23 +03:00
|
|
|
}
|
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
static void event_swap(union perf_event *event, bool sample_id_all)
|
|
|
|
{
|
|
|
|
perf_event__swap_op swap;
|
|
|
|
|
|
|
|
swap = perf_event__swap_ops[event->header.type];
|
|
|
|
if (swap)
|
|
|
|
swap(event, sample_id_all);
|
|
|
|
}
|
|
|
|
|
2014-07-31 10:00:57 +04:00
|
|
|
int perf_session__peek_event(struct perf_session *session, off_t file_offset,
|
|
|
|
void *buf, size_t buf_sz,
|
|
|
|
union perf_event **event_ptr,
|
|
|
|
struct perf_sample *sample)
|
|
|
|
{
|
|
|
|
union perf_event *event;
|
|
|
|
size_t hdr_sz, rest;
|
|
|
|
int fd;
|
|
|
|
|
|
|
|
if (session->one_mmap && !session->header.needs_swap) {
|
|
|
|
event = file_offset - session->one_mmap_offset +
|
|
|
|
session->one_mmap_addr;
|
|
|
|
goto out_parse_sample;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (perf_data_file__is_pipe(session->file))
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
fd = perf_data_file__fd(session->file);
|
|
|
|
hdr_sz = sizeof(struct perf_event_header);
|
|
|
|
|
|
|
|
if (buf_sz < hdr_sz)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
if (lseek(fd, file_offset, SEEK_SET) == (off_t)-1 ||
|
2015-05-19 16:05:45 +03:00
|
|
|
readn(fd, buf, hdr_sz) != (ssize_t)hdr_sz)
|
2014-07-31 10:00:57 +04:00
|
|
|
return -1;
|
|
|
|
|
|
|
|
event = (union perf_event *)buf;
|
|
|
|
|
|
|
|
if (session->header.needs_swap)
|
|
|
|
perf_event_header__bswap(&event->header);
|
|
|
|
|
2015-05-19 16:05:45 +03:00
|
|
|
if (event->header.size < hdr_sz || event->header.size > buf_sz)
|
2014-07-31 10:00:57 +04:00
|
|
|
return -1;
|
|
|
|
|
|
|
|
rest = event->header.size - hdr_sz;
|
|
|
|
|
2015-05-19 16:05:45 +03:00
|
|
|
if (readn(fd, buf, rest) != (ssize_t)rest)
|
2014-07-31 10:00:57 +04:00
|
|
|
return -1;
|
|
|
|
|
|
|
|
if (session->header.needs_swap)
|
|
|
|
event_swap(event, perf_evlist__sample_id_all(session->evlist));
|
|
|
|
|
|
|
|
out_parse_sample:
|
|
|
|
|
|
|
|
if (sample && event->header.type < PERF_RECORD_USER_TYPE_START &&
|
|
|
|
perf_evlist__parse_sample(session->evlist, event, sample))
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
*event_ptr = event;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2014-07-23 23:19:58 +04:00
|
|
|
static s64 perf_session__process_event(struct perf_session *session,
|
2015-03-03 17:58:45 +03:00
|
|
|
union perf_event *event, u64 file_offset)
|
2010-12-07 15:49:01 +03:00
|
|
|
{
|
2015-02-14 21:05:28 +03:00
|
|
|
struct perf_evlist *evlist = session->evlist;
|
2015-03-31 18:48:16 +03:00
|
|
|
struct perf_tool *tool = session->tool;
|
2011-01-29 18:02:00 +03:00
|
|
|
struct perf_sample sample;
|
2010-12-07 15:49:01 +03:00
|
|
|
int ret;
|
|
|
|
|
2012-05-30 16:23:43 +04:00
|
|
|
if (session->header.needs_swap)
|
2015-02-14 21:05:28 +03:00
|
|
|
event_swap(event, perf_evlist__sample_id_all(evlist));
|
2010-12-07 15:49:01 +03:00
|
|
|
|
|
|
|
if (event->header.type >= PERF_RECORD_HEADER_MAX)
|
|
|
|
return -EINVAL;
|
|
|
|
|
2015-02-14 21:05:28 +03:00
|
|
|
events_stats__inc(&evlist->stats, event->header.type);
|
2010-12-07 15:49:01 +03:00
|
|
|
|
|
|
|
if (event->header.type >= PERF_RECORD_USER_TYPE_START)
|
2015-03-03 17:58:45 +03:00
|
|
|
return perf_session__process_user_event(session, event, file_offset);
|
2010-12-05 16:32:55 +03:00
|
|
|
|
2010-12-07 15:48:58 +03:00
|
|
|
/*
|
|
|
|
* For all kernel events we get the sample data
|
|
|
|
*/
|
2015-02-14 21:05:28 +03:00
|
|
|
ret = perf_evlist__parse_sample(evlist, event, &sample);
|
2011-05-22 04:17:22 +04:00
|
|
|
if (ret)
|
|
|
|
return ret;
|
2010-12-07 15:48:58 +03:00
|
|
|
|
2014-07-06 16:18:21 +04:00
|
|
|
if (tool->ordered_events) {
|
2015-03-03 17:58:45 +03:00
|
|
|
ret = perf_session__queue_event(session, event, &sample, file_offset);
|
2010-12-05 16:32:55 +03:00
|
|
|
if (ret != -ETIME)
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2015-04-09 18:53:48 +03:00
|
|
|
return perf_session__deliver_event(session, event, &sample, tool,
|
|
|
|
file_offset);
|
2009-12-28 02:36:59 +03:00
|
|
|
}
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
void perf_event_header__bswap(struct perf_event_header *hdr)
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
{
|
2013-11-05 22:32:36 +04:00
|
|
|
hdr->type = bswap_32(hdr->type);
|
|
|
|
hdr->misc = bswap_16(hdr->misc);
|
|
|
|
hdr->size = bswap_16(hdr->size);
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
}
|
|
|
|
|
2011-11-09 19:24:25 +04:00
|
|
|
struct thread *perf_session__findnew(struct perf_session *session, pid_t pid)
|
|
|
|
{
|
2014-07-14 14:02:25 +04:00
|
|
|
return machine__findnew_thread(&session->machines.host, -1, pid);
|
2011-11-09 19:24:25 +04:00
|
|
|
}
|
|
|
|
|
2015-12-09 05:11:23 +03:00
|
|
|
int perf_session__register_idle_thread(struct perf_session *session)
|
2009-12-28 02:36:59 +03:00
|
|
|
{
|
2014-07-14 14:02:25 +04:00
|
|
|
struct thread *thread;
|
2015-12-09 05:11:23 +03:00
|
|
|
int err = 0;
|
2009-12-28 02:36:59 +03:00
|
|
|
|
2014-07-14 14:02:25 +04:00
|
|
|
thread = machine__findnew_thread(&session->machines.host, 0, 0);
|
2013-09-11 18:18:24 +04:00
|
|
|
if (thread == NULL || thread__set_comm(thread, "swapper", 0)) {
|
2009-12-28 02:36:59 +03:00
|
|
|
pr_err("problem inserting idle task.\n");
|
2015-12-09 05:11:23 +03:00
|
|
|
err = -1;
|
2009-12-28 02:36:59 +03:00
|
|
|
}
|
|
|
|
|
perf tools: Add PERF_RECORD_NAMESPACES to include namespaces related info
Introduce a new option to record PERF_RECORD_NAMESPACES events emitted
by the kernel when fork, clone, setns or unshare are invoked. And update
perf-record documentation with the new option to record namespace
events.
Committer notes:
Combined it with a later patch to allow printing it via 'perf report -D'
and be able to test the feature introduced in this patch. Had to move
here also perf_ns__name(), that was introduced in another later patch.
Also used PRIu64 and PRIx64 to fix the build in some enfironments wrt:
util/event.c:1129:39: error: format '%lx' expects argument of type 'long unsigned int', but argument 6 has type 'long long unsigned int' [-Werror=format=]
ret += fprintf(fp, "%u/%s: %lu/0x%lx%s", idx
^
Testing it:
# perf record --namespaces -a
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 1.083 MB perf.data (423 samples) ]
#
# perf report -D
<SNIP>
3 2028902078892 0x115140 [0xa0]: PERF_RECORD_NAMESPACES 14783/14783 - nr_namespaces: 7
[0/net: 3/0xf0000081, 1/uts: 3/0xeffffffe, 2/ipc: 3/0xefffffff, 3/pid: 3/0xeffffffc,
4/user: 3/0xeffffffd, 5/mnt: 3/0xf0000000, 6/cgroup: 3/0xeffffffb]
0x1151e0 [0x30]: event: 9
.
. ... raw event: size 48 bytes
. 0000: 09 00 00 00 02 00 30 00 c4 71 82 68 0c 7f 00 00 ......0..q.h....
. 0010: a9 39 00 00 a9 39 00 00 94 28 fe 63 d8 01 00 00 .9...9...(.c....
. 0020: 03 00 00 00 00 00 00 00 ce c4 02 00 00 00 00 00 ................
<SNIP>
NAMESPACES events: 1
<SNIP>
#
Signed-off-by: Hari Bathini <hbathini@linux.vnet.ibm.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Starovoitov <ast@fb.com>
Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>
Cc: Aravinda Prasad <aravinda@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sargun Dhillon <sargun@sargun.me>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/148891930386.25309.18412039920746995488.stgit@hbathini.in.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-03-07 23:41:43 +03:00
|
|
|
if (thread == NULL || thread__set_namespaces(thread, 0, NULL)) {
|
|
|
|
pr_err("problem inserting idle task.\n");
|
|
|
|
err = -1;
|
|
|
|
}
|
|
|
|
|
2015-12-09 05:11:23 +03:00
|
|
|
/* machine__findnew_thread() got the thread, so put it */
|
|
|
|
thread__put(thread);
|
|
|
|
return err;
|
2009-12-28 02:36:59 +03:00
|
|
|
}
|
|
|
|
|
2016-07-14 11:34:46 +03:00
|
|
|
static void
|
|
|
|
perf_session__warn_order(const struct perf_session *session)
|
|
|
|
{
|
|
|
|
const struct ordered_events *oe = &session->ordered_events;
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
bool should_warn = true;
|
|
|
|
|
|
|
|
evlist__for_each_entry(session->evlist, evsel) {
|
|
|
|
if (evsel->attr.write_backward)
|
|
|
|
should_warn = false;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!should_warn)
|
|
|
|
return;
|
|
|
|
if (oe->nr_unordered_events != 0)
|
|
|
|
ui__warning("%u out of order events recorded.\n", oe->nr_unordered_events);
|
|
|
|
}
|
|
|
|
|
2015-03-31 18:48:16 +03:00
|
|
|
static void perf_session__warn_about_errors(const struct perf_session *session)
|
2011-01-04 21:25:15 +03:00
|
|
|
{
|
2015-03-31 18:48:16 +03:00
|
|
|
const struct events_stats *stats = &session->evlist->stats;
|
|
|
|
|
|
|
|
if (session->tool->lost == perf_event__process_lost &&
|
2015-02-14 20:57:13 +03:00
|
|
|
stats->nr_events[PERF_RECORD_LOST] != 0) {
|
2011-10-29 18:15:04 +04:00
|
|
|
ui__warning("Processed %d events and lost %d chunks!\n\n"
|
|
|
|
"Check IO/CPU overload!\n\n",
|
2015-02-14 20:57:13 +03:00
|
|
|
stats->nr_events[0],
|
|
|
|
stats->nr_events[PERF_RECORD_LOST]);
|
2011-01-04 21:25:15 +03:00
|
|
|
}
|
|
|
|
|
2015-05-10 22:13:15 +03:00
|
|
|
if (session->tool->lost_samples == perf_event__process_lost_samples) {
|
|
|
|
double drop_rate;
|
|
|
|
|
|
|
|
drop_rate = (double)stats->total_lost_samples /
|
|
|
|
(double) (stats->nr_events[PERF_RECORD_SAMPLE] + stats->total_lost_samples);
|
|
|
|
if (drop_rate > 0.05) {
|
2015-06-11 15:44:24 +03:00
|
|
|
ui__warning("Processed %" PRIu64 " samples and lost %3.2f%% samples!\n\n",
|
2015-05-10 22:13:15 +03:00
|
|
|
stats->nr_events[PERF_RECORD_SAMPLE] + stats->total_lost_samples,
|
|
|
|
drop_rate * 100.0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-09-25 16:15:37 +03:00
|
|
|
if (session->tool->aux == perf_event__process_aux &&
|
|
|
|
stats->total_aux_lost != 0) {
|
|
|
|
ui__warning("AUX data lost %" PRIu64 " times out of %u!\n\n",
|
|
|
|
stats->total_aux_lost,
|
|
|
|
stats->nr_events[PERF_RECORD_AUX]);
|
|
|
|
}
|
|
|
|
|
2017-03-16 19:41:59 +03:00
|
|
|
if (session->tool->aux == perf_event__process_aux &&
|
|
|
|
stats->total_aux_partial != 0) {
|
|
|
|
bool vmm_exclusive = false;
|
|
|
|
|
|
|
|
(void)sysfs__read_bool("module/kvm_intel/parameters/vmm_exclusive",
|
|
|
|
&vmm_exclusive);
|
|
|
|
|
|
|
|
ui__warning("AUX data had gaps in it %" PRIu64 " times out of %u!\n\n"
|
|
|
|
"Are you running a KVM guest in the background?%s\n\n",
|
|
|
|
stats->total_aux_partial,
|
|
|
|
stats->nr_events[PERF_RECORD_AUX],
|
|
|
|
vmm_exclusive ?
|
|
|
|
"\nReloading kvm_intel module with vmm_exclusive=0\n"
|
|
|
|
"will reduce the gaps to only guest's timeslices." :
|
|
|
|
"");
|
|
|
|
}
|
|
|
|
|
2015-02-14 20:57:13 +03:00
|
|
|
if (stats->nr_unknown_events != 0) {
|
2011-01-04 21:25:15 +03:00
|
|
|
ui__warning("Found %u unknown events!\n\n"
|
|
|
|
"Is this an older tool processing a perf.data "
|
|
|
|
"file generated by a more recent tool?\n\n"
|
|
|
|
"If that is not the case, consider "
|
|
|
|
"reporting to linux-kernel@vger.kernel.org.\n\n",
|
2015-02-14 20:57:13 +03:00
|
|
|
stats->nr_unknown_events);
|
2011-01-04 21:25:15 +03:00
|
|
|
}
|
|
|
|
|
2015-02-14 20:57:13 +03:00
|
|
|
if (stats->nr_unknown_id != 0) {
|
2011-03-15 21:44:01 +03:00
|
|
|
ui__warning("%u samples with id not present in the header\n",
|
2015-02-14 20:57:13 +03:00
|
|
|
stats->nr_unknown_id);
|
2011-03-15 21:44:01 +03:00
|
|
|
}
|
|
|
|
|
2015-02-14 20:57:13 +03:00
|
|
|
if (stats->nr_invalid_chains != 0) {
|
2015-02-14 20:50:11 +03:00
|
|
|
ui__warning("Found invalid callchains!\n\n"
|
|
|
|
"%u out of %u events were discarded for this reason.\n\n"
|
|
|
|
"Consider reporting to linux-kernel@vger.kernel.org.\n\n",
|
2015-02-14 20:57:13 +03:00
|
|
|
stats->nr_invalid_chains,
|
|
|
|
stats->nr_events[PERF_RECORD_SAMPLE]);
|
2015-02-14 20:50:11 +03:00
|
|
|
}
|
2012-02-10 21:05:04 +04:00
|
|
|
|
2015-02-14 20:57:13 +03:00
|
|
|
if (stats->nr_unprocessable_samples != 0) {
|
2012-02-10 21:05:04 +04:00
|
|
|
ui__warning("%u unprocessable samples recorded.\n"
|
|
|
|
"Do you have a KVM guest running and not using 'perf kvm'?\n",
|
2015-02-14 20:57:13 +03:00
|
|
|
stats->nr_unprocessable_samples);
|
2012-02-10 21:05:04 +04:00
|
|
|
}
|
2014-11-26 18:39:31 +03:00
|
|
|
|
2016-07-14 11:34:46 +03:00
|
|
|
perf_session__warn_order(session);
|
2015-04-09 18:53:50 +03:00
|
|
|
|
|
|
|
events_stats__auxtrace_error_warn(stats);
|
2015-06-17 16:51:10 +03:00
|
|
|
|
|
|
|
if (stats->nr_proc_map_timeout != 0) {
|
|
|
|
ui__warning("%d map information files for pre-existing threads were\n"
|
|
|
|
"not processed, if there are samples for addresses they\n"
|
|
|
|
"will not be resolved, you may find out which are these\n"
|
|
|
|
"threads by running with -v and redirecting the output\n"
|
2015-06-17 16:51:11 +03:00
|
|
|
"to a file.\n"
|
|
|
|
"The time limit to process proc map is too short?\n"
|
|
|
|
"Increase it by --proc-map-timeout\n",
|
2015-06-17 16:51:10 +03:00
|
|
|
stats->nr_proc_map_timeout);
|
|
|
|
}
|
2011-01-04 21:25:15 +03:00
|
|
|
}
|
|
|
|
|
2015-05-29 16:33:30 +03:00
|
|
|
static int perf_session__flush_thread_stack(struct thread *thread,
|
|
|
|
void *p __maybe_unused)
|
|
|
|
{
|
|
|
|
return thread_stack__flush(thread);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int perf_session__flush_thread_stacks(struct perf_session *session)
|
|
|
|
{
|
|
|
|
return machines__for_each_thread(&session->machines,
|
|
|
|
perf_session__flush_thread_stack,
|
|
|
|
NULL);
|
|
|
|
}
|
|
|
|
|
2010-04-02 08:59:15 +04:00
|
|
|
volatile int session_done;
|
|
|
|
|
2015-03-03 17:58:45 +03:00
|
|
|
static int __perf_session__process_pipe_events(struct perf_session *session)
|
2010-04-02 08:59:15 +04:00
|
|
|
{
|
2015-03-03 17:48:12 +03:00
|
|
|
struct ordered_events *oe = &session->ordered_events;
|
2015-03-31 18:48:16 +03:00
|
|
|
struct perf_tool *tool = session->tool;
|
2013-11-05 22:32:36 +04:00
|
|
|
int fd = perf_data_file__fd(session->file);
|
perf tools: Fix piped mode read code
In __perf_session__process_pipe_events(), there was a risk we would read
more than what a union perf_event struct can hold. this could happen in
case, perf is reading a file which contains new record types it does not
know about and which are larger than anything it knows about.
In general, perf is supposed to skip records it does not understand, but
in pipe mode, those have to be read and ignored. The fixed size header
contains the size of the record, but that size may be larger than union
perf_event, yet it was used as the backing to the read in:
union perf_event event;
void *p;
size = event->header.size;
p = &event;
p += sizeof(struct perf_event_header);
if (size - sizeof(struct perf_event_header)) {
err = readn(self->fd, p, size - sizeof(struct perf_event_header));
We fix this by allocating a buffer based on the size reported in the
header. We reuse the buffer as much as we can. We realloc in case it
becomes too small. In the common case, the performance impact is
negligible.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1337081295-10303-3-git-send-email-eranian@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-15 15:28:12 +04:00
|
|
|
union perf_event *event;
|
|
|
|
uint32_t size, cur_size = 0;
|
|
|
|
void *buf = NULL;
|
2014-07-23 23:19:58 +04:00
|
|
|
s64 skip = 0;
|
2010-04-02 08:59:15 +04:00
|
|
|
u64 head;
|
2013-11-28 14:30:14 +04:00
|
|
|
ssize_t err;
|
2010-04-02 08:59:15 +04:00
|
|
|
void *p;
|
|
|
|
|
2011-11-28 14:30:20 +04:00
|
|
|
perf_tool__fill_defaults(tool);
|
2010-04-02 08:59:15 +04:00
|
|
|
|
|
|
|
head = 0;
|
perf tools: Fix piped mode read code
In __perf_session__process_pipe_events(), there was a risk we would read
more than what a union perf_event struct can hold. this could happen in
case, perf is reading a file which contains new record types it does not
know about and which are larger than anything it knows about.
In general, perf is supposed to skip records it does not understand, but
in pipe mode, those have to be read and ignored. The fixed size header
contains the size of the record, but that size may be larger than union
perf_event, yet it was used as the backing to the read in:
union perf_event event;
void *p;
size = event->header.size;
p = &event;
p += sizeof(struct perf_event_header);
if (size - sizeof(struct perf_event_header)) {
err = readn(self->fd, p, size - sizeof(struct perf_event_header));
We fix this by allocating a buffer based on the size reported in the
header. We reuse the buffer as much as we can. We realloc in case it
becomes too small. In the common case, the performance impact is
negligible.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1337081295-10303-3-git-send-email-eranian@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-15 15:28:12 +04:00
|
|
|
cur_size = sizeof(union perf_event);
|
|
|
|
|
|
|
|
buf = malloc(cur_size);
|
|
|
|
if (!buf)
|
|
|
|
return -errno;
|
perf inject: Copy events when reordering events in pipe mode
__perf_session__process_pipe_events reuses the same memory buffer to
process all events in the pipe.
When reordering is needed (e.g. -b option), events are not immediately
flushed, but kept around until reordering is possible, causing
memory corruption.
The problem is usually observed by a "Unknown sample error" output. It
can easily be reproduced by:
perf record -o - noploop | perf inject -b > output
Committer testing:
Before:
$ perf record -o - stress -t 2 -c 2 | perf inject -b > /dev/null
stress: info: [8297] dispatching hogs: 2 cpu, 0 io, 0 vm, 0 hdd
stress: info: [8297] successful run completed in 2s
[ perf record: Woken up 3 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
Warning:
Found 1 unknown events!
Is this an older tool processing a perf.data file generated by a more recent tool?
If that is not the case, consider reporting to linux-kernel@vger.kernel.org.
$
After:
$ perf record -o - stress -t 2 -c 2 | perf inject -b > /dev/null
stress: info: [9027] dispatching hogs: 2 cpu, 0 io, 0 vm, 0 hdd
stress: info: [9027] successful run completed in 2s
[ perf record: Woken up 3 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
no symbols found in /usr/bin/stress, maybe install a debug package?
no symbols found in /usr/bin/stress, maybe install a debug package?
$
Signed-off-by: David Carrillo-Cisneros <davidcc@google.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Paul Turner <pjt@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Simon Que <sque@chromium.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/20170410201432.24807-3-davidcc@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-04-10 23:14:27 +03:00
|
|
|
ordered_events__set_copy_on_queue(oe, true);
|
2010-04-02 08:59:15 +04:00
|
|
|
more:
|
perf tools: Fix piped mode read code
In __perf_session__process_pipe_events(), there was a risk we would read
more than what a union perf_event struct can hold. this could happen in
case, perf is reading a file which contains new record types it does not
know about and which are larger than anything it knows about.
In general, perf is supposed to skip records it does not understand, but
in pipe mode, those have to be read and ignored. The fixed size header
contains the size of the record, but that size may be larger than union
perf_event, yet it was used as the backing to the read in:
union perf_event event;
void *p;
size = event->header.size;
p = &event;
p += sizeof(struct perf_event_header);
if (size - sizeof(struct perf_event_header)) {
err = readn(self->fd, p, size - sizeof(struct perf_event_header));
We fix this by allocating a buffer based on the size reported in the
header. We reuse the buffer as much as we can. We realloc in case it
becomes too small. In the common case, the performance impact is
negligible.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1337081295-10303-3-git-send-email-eranian@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-15 15:28:12 +04:00
|
|
|
event = buf;
|
2013-10-15 18:27:34 +04:00
|
|
|
err = readn(fd, event, sizeof(struct perf_event_header));
|
2010-04-02 08:59:15 +04:00
|
|
|
if (err <= 0) {
|
|
|
|
if (err == 0)
|
|
|
|
goto done;
|
|
|
|
|
|
|
|
pr_err("failed to read event header\n");
|
|
|
|
goto out_err;
|
|
|
|
}
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
if (session->header.needs_swap)
|
perf tools: Fix piped mode read code
In __perf_session__process_pipe_events(), there was a risk we would read
more than what a union perf_event struct can hold. this could happen in
case, perf is reading a file which contains new record types it does not
know about and which are larger than anything it knows about.
In general, perf is supposed to skip records it does not understand, but
in pipe mode, those have to be read and ignored. The fixed size header
contains the size of the record, but that size may be larger than union
perf_event, yet it was used as the backing to the read in:
union perf_event event;
void *p;
size = event->header.size;
p = &event;
p += sizeof(struct perf_event_header);
if (size - sizeof(struct perf_event_header)) {
err = readn(self->fd, p, size - sizeof(struct perf_event_header));
We fix this by allocating a buffer based on the size reported in the
header. We reuse the buffer as much as we can. We realloc in case it
becomes too small. In the common case, the performance impact is
negligible.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1337081295-10303-3-git-send-email-eranian@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-15 15:28:12 +04:00
|
|
|
perf_event_header__bswap(&event->header);
|
2010-04-02 08:59:15 +04:00
|
|
|
|
perf tools: Fix piped mode read code
In __perf_session__process_pipe_events(), there was a risk we would read
more than what a union perf_event struct can hold. this could happen in
case, perf is reading a file which contains new record types it does not
know about and which are larger than anything it knows about.
In general, perf is supposed to skip records it does not understand, but
in pipe mode, those have to be read and ignored. The fixed size header
contains the size of the record, but that size may be larger than union
perf_event, yet it was used as the backing to the read in:
union perf_event event;
void *p;
size = event->header.size;
p = &event;
p += sizeof(struct perf_event_header);
if (size - sizeof(struct perf_event_header)) {
err = readn(self->fd, p, size - sizeof(struct perf_event_header));
We fix this by allocating a buffer based on the size reported in the
header. We reuse the buffer as much as we can. We realloc in case it
becomes too small. In the common case, the performance impact is
negligible.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1337081295-10303-3-git-send-email-eranian@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-15 15:28:12 +04:00
|
|
|
size = event->header.size;
|
2013-07-04 17:20:27 +04:00
|
|
|
if (size < sizeof(struct perf_event_header)) {
|
|
|
|
pr_err("bad event header size\n");
|
|
|
|
goto out_err;
|
|
|
|
}
|
2010-04-02 08:59:15 +04:00
|
|
|
|
perf tools: Fix piped mode read code
In __perf_session__process_pipe_events(), there was a risk we would read
more than what a union perf_event struct can hold. this could happen in
case, perf is reading a file which contains new record types it does not
know about and which are larger than anything it knows about.
In general, perf is supposed to skip records it does not understand, but
in pipe mode, those have to be read and ignored. The fixed size header
contains the size of the record, but that size may be larger than union
perf_event, yet it was used as the backing to the read in:
union perf_event event;
void *p;
size = event->header.size;
p = &event;
p += sizeof(struct perf_event_header);
if (size - sizeof(struct perf_event_header)) {
err = readn(self->fd, p, size - sizeof(struct perf_event_header));
We fix this by allocating a buffer based on the size reported in the
header. We reuse the buffer as much as we can. We realloc in case it
becomes too small. In the common case, the performance impact is
negligible.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1337081295-10303-3-git-send-email-eranian@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-15 15:28:12 +04:00
|
|
|
if (size > cur_size) {
|
|
|
|
void *new = realloc(buf, size);
|
|
|
|
if (!new) {
|
|
|
|
pr_err("failed to allocate memory to read event\n");
|
|
|
|
goto out_err;
|
|
|
|
}
|
|
|
|
buf = new;
|
|
|
|
cur_size = size;
|
|
|
|
event = buf;
|
|
|
|
}
|
|
|
|
p = event;
|
2010-04-02 08:59:15 +04:00
|
|
|
p += sizeof(struct perf_event_header);
|
|
|
|
|
2010-05-05 09:27:40 +04:00
|
|
|
if (size - sizeof(struct perf_event_header)) {
|
2013-10-15 18:27:34 +04:00
|
|
|
err = readn(fd, p, size - sizeof(struct perf_event_header));
|
2010-05-05 09:27:40 +04:00
|
|
|
if (err <= 0) {
|
|
|
|
if (err == 0) {
|
|
|
|
pr_err("unexpected end of event stream\n");
|
|
|
|
goto done;
|
|
|
|
}
|
2010-04-02 08:59:15 +04:00
|
|
|
|
2010-05-05 09:27:40 +04:00
|
|
|
pr_err("failed to read event data\n");
|
|
|
|
goto out_err;
|
|
|
|
}
|
2010-04-02 08:59:15 +04:00
|
|
|
}
|
|
|
|
|
2015-03-03 17:58:45 +03:00
|
|
|
if ((skip = perf_session__process_event(session, event, head)) < 0) {
|
2012-04-16 22:42:51 +04:00
|
|
|
pr_err("%#" PRIx64 " [%#x]: failed to process type: %d\n",
|
perf tools: Fix piped mode read code
In __perf_session__process_pipe_events(), there was a risk we would read
more than what a union perf_event struct can hold. this could happen in
case, perf is reading a file which contains new record types it does not
know about and which are larger than anything it knows about.
In general, perf is supposed to skip records it does not understand, but
in pipe mode, those have to be read and ignored. The fixed size header
contains the size of the record, but that size may be larger than union
perf_event, yet it was used as the backing to the read in:
union perf_event event;
void *p;
size = event->header.size;
p = &event;
p += sizeof(struct perf_event_header);
if (size - sizeof(struct perf_event_header)) {
err = readn(self->fd, p, size - sizeof(struct perf_event_header));
We fix this by allocating a buffer based on the size reported in the
header. We reuse the buffer as much as we can. We realloc in case it
becomes too small. In the common case, the performance impact is
negligible.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1337081295-10303-3-git-send-email-eranian@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-15 15:28:12 +04:00
|
|
|
head, event->header.size, event->header.type);
|
2012-04-16 22:42:51 +04:00
|
|
|
err = -EINVAL;
|
|
|
|
goto out_err;
|
2010-04-02 08:59:15 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
head += size;
|
|
|
|
|
|
|
|
if (skip > 0)
|
|
|
|
head += skip;
|
|
|
|
|
|
|
|
if (!session_done())
|
|
|
|
goto more;
|
|
|
|
done:
|
2013-10-18 16:29:02 +04:00
|
|
|
/* do the final flush for ordered samples */
|
2015-03-03 17:58:45 +03:00
|
|
|
err = ordered_events__flush(oe, OE_FLUSH__FINAL);
|
2015-04-09 18:53:48 +03:00
|
|
|
if (err)
|
|
|
|
goto out_err;
|
|
|
|
err = auxtrace__flush_events(session, tool);
|
2015-05-29 16:33:30 +03:00
|
|
|
if (err)
|
|
|
|
goto out_err;
|
|
|
|
err = perf_session__flush_thread_stacks(session);
|
2010-04-02 08:59:15 +04:00
|
|
|
out_err:
|
perf tools: Fix piped mode read code
In __perf_session__process_pipe_events(), there was a risk we would read
more than what a union perf_event struct can hold. this could happen in
case, perf is reading a file which contains new record types it does not
know about and which are larger than anything it knows about.
In general, perf is supposed to skip records it does not understand, but
in pipe mode, those have to be read and ignored. The fixed size header
contains the size of the record, but that size may be larger than union
perf_event, yet it was used as the backing to the read in:
union perf_event event;
void *p;
size = event->header.size;
p = &event;
p += sizeof(struct perf_event_header);
if (size - sizeof(struct perf_event_header)) {
err = readn(self->fd, p, size - sizeof(struct perf_event_header));
We fix this by allocating a buffer based on the size reported in the
header. We reuse the buffer as much as we can. We realloc in case it
becomes too small. In the common case, the performance impact is
negligible.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1337081295-10303-3-git-send-email-eranian@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-15 15:28:12 +04:00
|
|
|
free(buf);
|
2015-03-31 18:48:16 +03:00
|
|
|
perf_session__warn_about_errors(session);
|
2014-06-11 00:50:03 +04:00
|
|
|
ordered_events__free(&session->ordered_events);
|
2015-04-09 18:53:48 +03:00
|
|
|
auxtrace__free_events(session);
|
2010-04-02 08:59:15 +04:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2011-05-23 15:06:28 +04:00
|
|
|
static union perf_event *
|
|
|
|
fetch_mmaped_event(struct perf_session *session,
|
|
|
|
u64 head, size_t mmap_size, char *buf)
|
|
|
|
{
|
|
|
|
union perf_event *event;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Ensure we have enough space remaining to read
|
|
|
|
* the size of the event in the headers.
|
|
|
|
*/
|
|
|
|
if (head + sizeof(event->header) > mmap_size)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
event = (union perf_event *)(buf + head);
|
|
|
|
|
|
|
|
if (session->header.needs_swap)
|
|
|
|
perf_event_header__bswap(&event->header);
|
|
|
|
|
2013-07-04 17:20:27 +04:00
|
|
|
if (head + event->header.size > mmap_size) {
|
|
|
|
/* We're not fetching the event so swap back again */
|
|
|
|
if (session->header.needs_swap)
|
|
|
|
perf_event_header__bswap(&event->header);
|
2011-05-23 15:06:28 +04:00
|
|
|
return NULL;
|
2013-07-04 17:20:27 +04:00
|
|
|
}
|
2011-05-23 15:06:28 +04:00
|
|
|
|
|
|
|
return event;
|
|
|
|
}
|
|
|
|
|
2012-11-10 23:12:19 +04:00
|
|
|
/*
|
|
|
|
* On 64bit we can mmap the data file in one go. No need for tiny mmap
|
|
|
|
* slices. On 32bit we use 32MB.
|
|
|
|
*/
|
|
|
|
#if BITS_PER_LONG == 64
|
|
|
|
#define MMAP_SIZE ULLONG_MAX
|
|
|
|
#define NUM_MMAPS 1
|
|
|
|
#else
|
|
|
|
#define MMAP_SIZE (32 * 1024 * 1024ULL)
|
|
|
|
#define NUM_MMAPS 128
|
|
|
|
#endif
|
|
|
|
|
2015-01-29 11:06:43 +03:00
|
|
|
static int __perf_session__process_events(struct perf_session *session,
|
|
|
|
u64 data_offset, u64 data_size,
|
2015-03-03 17:58:45 +03:00
|
|
|
u64 file_size)
|
2009-12-28 02:36:59 +03:00
|
|
|
{
|
2015-03-03 17:48:12 +03:00
|
|
|
struct ordered_events *oe = &session->ordered_events;
|
2015-03-31 18:48:16 +03:00
|
|
|
struct perf_tool *tool = session->tool;
|
2013-10-15 18:27:34 +04:00
|
|
|
int fd = perf_data_file__fd(session->file);
|
2014-07-23 23:19:58 +04:00
|
|
|
u64 head, page_offset, file_offset, file_pos, size;
|
2010-11-30 20:49:49 +03:00
|
|
|
int err, mmap_prot, mmap_flags, map_idx = 0;
|
2012-10-06 21:57:10 +04:00
|
|
|
size_t mmap_size;
|
2012-11-10 23:12:19 +04:00
|
|
|
char *buf, *mmaps[NUM_MMAPS];
|
2011-01-29 19:01:45 +03:00
|
|
|
union perf_event *event;
|
2013-10-23 22:40:38 +04:00
|
|
|
struct ui_progress prog;
|
2014-07-23 23:19:58 +04:00
|
|
|
s64 skip;
|
2010-11-30 20:49:38 +03:00
|
|
|
|
2011-11-28 14:30:20 +04:00
|
|
|
perf_tool__fill_defaults(tool);
|
2009-12-28 02:36:59 +03:00
|
|
|
|
2010-11-30 20:49:38 +03:00
|
|
|
page_offset = page_size * (data_offset / page_size);
|
|
|
|
file_offset = page_offset;
|
|
|
|
head = data_offset - page_offset;
|
2009-12-28 02:36:59 +03:00
|
|
|
|
perf record: Avoid infinite loop at buildid processing with no samples
If a session contains no events, we can get stuck in an infinite loop in
__perf_session__process_events, with a non-zero file_size and data_offset, but
a zero data_size.
In this case, we can mmap the entirety of the file (consisting of the file and
attribute headers), and fetch_mmaped_event will correctly refuse to read any
(unmapped and non-existent) event headers. This causes
__perf_session__process_events to unmap the file and retry with the exact same
parameters, getting stuck in an infinite loop.
This has been observed to result in an exit-time hang when counting
rare/unschedulable events with perf record, and can be triggered artificially
with the script below:
----
#!/bin/sh
printf "REPRO: launching perf\n";
./perf record -e software/config=9/ sleep 1 &
PERF_PID=$!;
sleep 0.002;
kill -2 $PERF_PID;
printf "REPRO: waiting for perf (%d) to exit...\n" "$PERF_PID";
wait $PERF_PID;
printf "REPRO: perf exited\n";
----
To avoid this, have __perf_session__process_events bail out early when
the file has no data (i.e. it has no events).
Commiter note:
I only managed to reproduce this when setting
/proc/sys/kernel/kptr_restrict to '1' and changing the code to
purposefully not process any samples and no synthesized samples, i.e.
kptr_restrict prevents 'record' from synthesizing the kernel mmaps for
vmlinux + modules and since it is a workload started from perf, we don't
synthesize mmap/comm records for existing threads.
Adrian Hunter managed to reproduce it in his environment tho.
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@kernel.org>
Tested-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1442423929-12253-1-git-send-email-mark.rutland@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-09-16 20:18:49 +03:00
|
|
|
if (data_size == 0)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
if (data_offset + data_size < file_size)
|
2010-11-30 20:49:44 +03:00
|
|
|
file_size = data_offset + data_size;
|
|
|
|
|
2013-10-23 22:40:38 +04:00
|
|
|
ui_progress__init(&prog, file_size, "Processing events...");
|
2010-11-30 20:49:46 +03:00
|
|
|
|
2012-11-10 23:12:19 +04:00
|
|
|
mmap_size = MMAP_SIZE;
|
2014-07-14 14:02:51 +04:00
|
|
|
if (mmap_size > file_size) {
|
2010-11-30 20:49:46 +03:00
|
|
|
mmap_size = file_size;
|
2014-07-14 14:02:51 +04:00
|
|
|
session->one_mmap = true;
|
|
|
|
}
|
2010-11-30 20:49:46 +03:00
|
|
|
|
2010-11-30 20:49:49 +03:00
|
|
|
memset(mmaps, 0, sizeof(mmaps));
|
|
|
|
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
mmap_prot = PROT_READ;
|
|
|
|
mmap_flags = MAP_SHARED;
|
|
|
|
|
2010-11-30 20:49:38 +03:00
|
|
|
if (session->header.needs_swap) {
|
perf tools: Cross platform perf.data analysis support
There are still some problems related to loading vmlinux files,
but those are unrelated to the feature implemented in this
patch, so will get fixed in the next patches, but here are some
results:
1. collect perf.data file on a Fedora 12 machine, x86_64, 64-bit
userland
2. transfer it to a Debian Testing machine, PARISC64, 32-bit
userland
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | head -5
74f9930ee94475b6b3238caf3725a50d59cb994b [kernel.kallsyms]
55fdd56670453ea66c011158c4b9d30179c1d049 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/ipt_MASQUERADE.ko
41adff63c730890480980d5d8ba513f1c216a858 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/iptable_nat.ko
90a33def1077bb8e97b8a78546dc96c2de62df46 /lib/modules/2.6.33-rc4-tip+/kernel/net/ipv4/netfilter/nf_nat.ko
984c7bea90ce1376d5c8e7ef43a781801286e62d /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/tun.ko
acme@parisc:~/git/linux-2.6-tip$ perf buildid-list | tail -5
22492f3753c6a67de5c7ccbd6b863390c92c0723 /usr/lib64/libXt.so.6.0.0
353802bb7e1b895ba43507cc678f951e778e4c6f /usr/lib64/libMagickCore.so.2.0.0
d10c2897558595efe7be8b0584cf7e6398bc776c /usr/lib64/libfprint.so.0.0.0
a83ecfb519a788774a84d5ddde633c9ba56c03ab /home/acme/bin/perf
d3ca765a8ecf257d263801d7ad8c49c189082317 /usr/lib64/libdwarf.so.0.0
acme@parisc:~/git/linux-2.6-tip$
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm
The file [kernel.kallsyms] cannot be used, trying to use /proc/kallsyms...
^^^^ The problem related to vmlinux handling, it shouldn't be trying this
^^^^ rather alien /proc/kallsyms at all...
/lib64/libpthread-2.10.2.so with build id 5c68f7afeb33309c78037e374b0deee84dd441f6 not found, continuing without symbols
/lib64/libc-2.10.2.so with build id eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 not found, continuing without symbols
/home/acme/bin/perf with build id a83ecfb519a788774a84d5ddde633c9ba56c03ab not found, continuing without symbols
/usr/sbin/openvpn with build id f2037a091ef36b591187a858d75e203690ea9409 not found, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/e1000e/e1000e.ko, continuing without symbols
Failed to open /lib/modules/2.6.33-rc4-tip+/kernel/drivers/net/wireless/iwlwifi/iwlcore.ko, continuing without symbols
<SNIP more complaints about not finding the right build-ids,
those will have to wait for 'perf archive' or plain
copying what was collected by 'perf record' on the x86_64,
source machine, see further below for an example of this >
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get when running the same command for the
same perf.data file on the F12, x86_64, source machine:
[root@doppio linux-2.6-tip]# perf report --sort comm
# Samples: 293085637
#
# Overhead Command
# ........ ...............
#
61.70% find
23.50% perf
5.86% swapper
3.12% sshd
2.39% init
0.87% bash
0.86% sleep
0.59% dbus-daemon
0.25% hald
0.24% NetworkManager
0.19% hald-addon-rfki
0.15% openvpn
0.07% phy0
0.07% events/0
0.05% iwl3945
0.05% events/1
0.03% kondemand/0
[root@doppio linux-2.6-tip]#
The other modes work as well, modulo the problem with vmlinux:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object
# ........ ............... .................................
#
35.11% find ffffffff81002b5a
18.25% perf ffffffff8102235f
16.17% find libc-2.10.2.so
9.07% find find
5.80% swapper ffffffff8102235f
3.95% perf libc-2.10.2.so
2.33% init ffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k
1.35% find [e1000e]
0.68% sleep libc-2.10.2.so
acme@parisc:~/git/linux-2.6-tip$
And the lack of the right buildids:
acme@parisc:~/git/linux-2.6-tip$ perf report --sort comm,dso,symbol 2> /dev/null | head -15
# Samples: 293085637
#
# Overhead Command Shared Object Symbol
# ........ ............... ................................. ......
#
35.11% find ffffffff81002b5a [k] 0xffffffff81002b5a
18.25% perf ffffffff8102235f [k] 0xffffffff8102235f
16.17% find libc-2.10.2.so [.] 0x00000000045782
9.07% find find [.] 0x0000000000fb0e
5.80% swapper ffffffff8102235f [k] 0xffffffff8102235f
3.95% perf libc-2.10.2.so [.] 0x0000000007f398
2.33% init ffffffff810091b9 [k] 0xffffffff810091b9
1.65% sshd libcrypto.so.0.9.8k [.] 0x00000000105440
1.35% find [e1000e] [k] 0x00000000010948
0.68% sleep libc-2.10.2.so [.] 0x0000000011ad5b
acme@parisc:~/git/linux-2.6-tip$
But if we:
acme@parisc:~/git/linux-2.6-tip$ ls ~/.debug
ls: cannot access /home/acme/.debug: No such file or directory
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/lib64/libc-2.10.2.so/
acme@parisc:~/git/linux-2.6-tip$ scp doppio:.debug/lib64/libc-2.10.2.so/* ~/.debug/lib64/libc-2.10.2.so/
acme@doppio's password:
eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 100% 1783KB 714.7KB/s 00:02
acme@parisc:~/git/linux-2.6-tip$ mkdir -p ~/.debug/.build-id/eb
acme@parisc:~/git/linux-2.6-tip$ ln -s ../../lib64/libc-2.10.2.so/eb4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1 ~/.debug/.build-id/eb/4ec8fa8b2a5eb18cad173c92f27ed8887ed1c1
acme@parisc:~/git/linux-2.6-tip$ perf report --dsos libc-2.10.2.so 2> /dev/null
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
acme@parisc:~/git/linux-2.6-tip$
Which matches what we get on the source, F12, x86_64 machine:
[root@doppio linux-2.6-tip]# perf report --dsos libc-2.10.2.so
# dso: libc-2.10.2.so
# Samples: 64281170
#
# Overhead Command Symbol
# ........ ............... ......
#
14.98% perf [.] __GI_strcmp
12.30% find [.] __GI_memmove
9.25% find [.] _int_malloc
7.60% find [.] _IO_vfprintf_internal
6.10% find [.] _IO_new_file_xsputn
6.02% find [.] __GI_close
3.08% find [.] _IO_file_overflow_internal
3.08% find [.] malloc_consolidate
3.08% find [.] _int_free
3.08% find [.] __strchrnul
3.08% find [.] __getdents64
3.08% find [.] __write_nocancel
3.08% sleep [.] __GI__dl_addr
3.08% sshd [.] __libc_select
3.08% find [.] _IO_new_file_write
3.07% find [.] _IO_new_do_write
3.06% find [.] __GI___errno_location
3.05% find [.] __GI___libc_malloc
3.04% perf [.] __GI_memcpy
1.71% find [.] __fprintf_chk
1.29% bash [.] __gconv_transform_utf8_internal
0.79% dbus-daemon [.] __GI_strlen
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@doppio linux-2.6-tip]#
So I think this is really, really nice in that it demonstrates
the portability of perf.data files and the use of build-ids
accross such aliens worlds :-)
There are some things to fix tho, like the bitmap on the header,
but things are looking good.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263478990-8200-2-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-14 17:23:10 +03:00
|
|
|
mmap_prot |= PROT_WRITE;
|
|
|
|
mmap_flags = MAP_PRIVATE;
|
|
|
|
}
|
2009-12-28 02:36:59 +03:00
|
|
|
remap:
|
2013-10-15 18:27:34 +04:00
|
|
|
buf = mmap(NULL, mmap_size, mmap_prot, mmap_flags, fd,
|
2010-11-30 20:49:46 +03:00
|
|
|
file_offset);
|
2009-12-28 02:36:59 +03:00
|
|
|
if (buf == MAP_FAILED) {
|
|
|
|
pr_err("failed to mmap file\n");
|
|
|
|
err = -errno;
|
|
|
|
goto out_err;
|
|
|
|
}
|
2010-11-30 20:49:49 +03:00
|
|
|
mmaps[map_idx] = buf;
|
|
|
|
map_idx = (map_idx + 1) & (ARRAY_SIZE(mmaps) - 1);
|
2010-11-30 20:49:44 +03:00
|
|
|
file_pos = file_offset + head;
|
2014-07-14 14:02:51 +04:00
|
|
|
if (session->one_mmap) {
|
|
|
|
session->one_mmap_addr = buf;
|
|
|
|
session->one_mmap_offset = file_offset;
|
|
|
|
}
|
2009-12-28 02:36:59 +03:00
|
|
|
|
|
|
|
more:
|
2011-05-23 15:06:28 +04:00
|
|
|
event = fetch_mmaped_event(session, head, mmap_size, buf);
|
|
|
|
if (!event) {
|
2010-11-30 20:49:49 +03:00
|
|
|
if (mmaps[map_idx]) {
|
|
|
|
munmap(mmaps[map_idx], mmap_size);
|
|
|
|
mmaps[map_idx] = NULL;
|
|
|
|
}
|
2009-12-28 02:36:59 +03:00
|
|
|
|
2010-11-30 20:49:38 +03:00
|
|
|
page_offset = page_size * (head / page_size);
|
|
|
|
file_offset += page_offset;
|
|
|
|
head -= page_offset;
|
2009-12-28 02:36:59 +03:00
|
|
|
goto remap;
|
|
|
|
}
|
|
|
|
|
|
|
|
size = event->header.size;
|
|
|
|
|
2013-07-04 17:20:27 +04:00
|
|
|
if (size < sizeof(struct perf_event_header) ||
|
2015-03-03 17:58:45 +03:00
|
|
|
(skip = perf_session__process_event(session, event, file_pos)) < 0) {
|
2012-04-16 22:42:51 +04:00
|
|
|
pr_err("%#" PRIx64 " [%#x]: failed to process type: %d\n",
|
|
|
|
file_offset + head, event->header.size,
|
|
|
|
event->header.type);
|
|
|
|
err = -EINVAL;
|
|
|
|
goto out_err;
|
2009-12-28 02:36:59 +03:00
|
|
|
}
|
|
|
|
|
2014-07-23 23:19:57 +04:00
|
|
|
if (skip)
|
|
|
|
size += skip;
|
|
|
|
|
2009-12-28 02:36:59 +03:00
|
|
|
head += size;
|
2010-11-30 20:49:44 +03:00
|
|
|
file_pos += size;
|
2009-12-28 02:36:59 +03:00
|
|
|
|
2013-10-23 22:40:38 +04:00
|
|
|
ui_progress__update(&prog, size);
|
2010-11-30 20:49:46 +03:00
|
|
|
|
2013-09-17 23:34:28 +04:00
|
|
|
if (session_done())
|
2013-10-18 16:29:02 +04:00
|
|
|
goto out;
|
2013-09-17 23:34:28 +04:00
|
|
|
|
2010-11-30 20:49:44 +03:00
|
|
|
if (file_pos < file_size)
|
2009-12-28 02:36:59 +03:00
|
|
|
goto more;
|
2010-11-30 20:49:44 +03:00
|
|
|
|
2013-10-18 16:29:02 +04:00
|
|
|
out:
|
2010-04-24 02:04:12 +04:00
|
|
|
/* do the final flush for ordered samples */
|
2015-03-03 17:58:45 +03:00
|
|
|
err = ordered_events__flush(oe, OE_FLUSH__FINAL);
|
2015-04-09 18:53:48 +03:00
|
|
|
if (err)
|
|
|
|
goto out_err;
|
|
|
|
err = auxtrace__flush_events(session, tool);
|
2015-05-29 16:33:30 +03:00
|
|
|
if (err)
|
|
|
|
goto out_err;
|
|
|
|
err = perf_session__flush_thread_stacks(session);
|
2009-12-28 02:36:59 +03:00
|
|
|
out_err:
|
2012-11-13 17:30:34 +04:00
|
|
|
ui_progress__finish();
|
2015-03-31 18:48:16 +03:00
|
|
|
perf_session__warn_about_errors(session);
|
2016-04-13 11:21:04 +03:00
|
|
|
/*
|
|
|
|
* We may switching perf.data output, make ordered_events
|
|
|
|
* reusable.
|
|
|
|
*/
|
|
|
|
ordered_events__reinit(&session->ordered_events);
|
2015-04-09 18:53:48 +03:00
|
|
|
auxtrace__free_events(session);
|
2014-07-14 14:02:51 +04:00
|
|
|
session->one_mmap = false;
|
2009-12-28 02:36:59 +03:00
|
|
|
return err;
|
|
|
|
}
|
2009-12-28 02:37:01 +03:00
|
|
|
|
2015-03-03 17:58:45 +03:00
|
|
|
int perf_session__process_events(struct perf_session *session)
|
2010-02-03 21:52:05 +03:00
|
|
|
{
|
2013-11-05 22:32:36 +04:00
|
|
|
u64 size = perf_data_file__size(session->file);
|
2010-02-03 21:52:05 +03:00
|
|
|
int err;
|
|
|
|
|
2015-12-09 05:11:23 +03:00
|
|
|
if (perf_session__register_idle_thread(session) < 0)
|
2010-02-03 21:52:05 +03:00
|
|
|
return -ENOMEM;
|
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
if (!perf_data_file__is_pipe(session->file))
|
|
|
|
err = __perf_session__process_events(session,
|
|
|
|
session->header.data_offset,
|
2015-03-03 17:58:45 +03:00
|
|
|
session->header.data_size, size);
|
2010-04-02 08:59:15 +04:00
|
|
|
else
|
2015-03-03 17:58:45 +03:00
|
|
|
err = __perf_session__process_pipe_events(session);
|
2010-07-27 18:46:12 +04:00
|
|
|
|
2010-02-03 21:52:05 +03:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2012-08-02 02:15:52 +04:00
|
|
|
bool perf_session__has_traces(struct perf_session *session, const char *msg)
|
2009-12-28 02:37:01 +03:00
|
|
|
{
|
2013-08-08 06:50:58 +04:00
|
|
|
struct perf_evsel *evsel;
|
|
|
|
|
2016-06-23 17:26:15 +03:00
|
|
|
evlist__for_each_entry(session->evlist, evsel) {
|
2013-08-08 06:50:58 +04:00
|
|
|
if (evsel->attr.type == PERF_TYPE_TRACEPOINT)
|
|
|
|
return true;
|
2009-12-28 02:37:01 +03:00
|
|
|
}
|
|
|
|
|
2013-08-08 06:50:58 +04:00
|
|
|
pr_err("No trace sample to read. Did you call 'perf %s'?\n", msg);
|
|
|
|
return false;
|
2009-12-28 02:37:01 +03:00
|
|
|
}
|
2010-01-05 21:50:31 +03:00
|
|
|
|
2011-11-28 13:56:39 +04:00
|
|
|
int maps__set_kallsyms_ref_reloc_sym(struct map **maps,
|
|
|
|
const char *symbol_name, u64 addr)
|
2010-01-05 21:50:31 +03:00
|
|
|
{
|
|
|
|
char *bracket;
|
2017-02-14 22:19:56 +03:00
|
|
|
int i;
|
2010-04-19 09:32:50 +04:00
|
|
|
struct ref_reloc_sym *ref;
|
|
|
|
|
|
|
|
ref = zalloc(sizeof(struct ref_reloc_sym));
|
|
|
|
if (ref == NULL)
|
|
|
|
return -ENOMEM;
|
2010-01-05 21:50:31 +03:00
|
|
|
|
2010-04-19 09:32:50 +04:00
|
|
|
ref->name = strdup(symbol_name);
|
|
|
|
if (ref->name == NULL) {
|
|
|
|
free(ref);
|
2010-01-05 21:50:31 +03:00
|
|
|
return -ENOMEM;
|
2010-04-19 09:32:50 +04:00
|
|
|
}
|
2010-01-05 21:50:31 +03:00
|
|
|
|
2010-04-19 09:32:50 +04:00
|
|
|
bracket = strchr(ref->name, ']');
|
2010-01-05 21:50:31 +03:00
|
|
|
if (bracket)
|
|
|
|
*bracket = '\0';
|
|
|
|
|
2010-04-19 09:32:50 +04:00
|
|
|
ref->addr = addr;
|
2010-02-03 21:52:00 +03:00
|
|
|
|
|
|
|
for (i = 0; i < MAP__NR_TYPES; ++i) {
|
2010-04-19 09:32:50 +04:00
|
|
|
struct kmap *kmap = map__kmap(maps[i]);
|
2015-04-07 11:22:45 +03:00
|
|
|
|
|
|
|
if (!kmap)
|
|
|
|
continue;
|
2010-04-19 09:32:50 +04:00
|
|
|
kmap->ref_reloc_sym = ref;
|
2010-02-03 21:52:00 +03:00
|
|
|
}
|
|
|
|
|
2010-01-05 21:50:31 +03:00
|
|
|
return 0;
|
|
|
|
}
|
2010-05-10 02:57:08 +04:00
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
size_t perf_session__fprintf_dsos(struct perf_session *session, FILE *fp)
|
2010-05-10 02:57:08 +04:00
|
|
|
{
|
2013-11-05 22:32:36 +04:00
|
|
|
return machines__fprintf_dsos(&session->machines, fp);
|
2010-05-10 02:57:08 +04:00
|
|
|
}
|
2010-05-19 20:41:23 +04:00
|
|
|
|
2013-11-05 22:32:36 +04:00
|
|
|
size_t perf_session__fprintf_dsos_buildid(struct perf_session *session, FILE *fp,
|
2012-12-07 16:53:58 +04:00
|
|
|
bool (skip)(struct dso *dso, int parm), int parm)
|
2010-05-19 20:41:23 +04:00
|
|
|
{
|
2013-11-05 22:32:36 +04:00
|
|
|
return machines__fprintf_dsos_buildid(&session->machines, fp, skip, parm);
|
2010-05-19 20:41:23 +04:00
|
|
|
}
|
2011-03-06 03:40:06 +03:00
|
|
|
|
|
|
|
size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp)
|
|
|
|
{
|
2015-04-09 18:53:48 +03:00
|
|
|
size_t ret;
|
|
|
|
const char *msg = "";
|
|
|
|
|
|
|
|
if (perf_header__has_feat(&session->header, HEADER_AUXTRACE))
|
|
|
|
msg = " (excludes AUX area (e.g. instruction trace) decoded / synthesized events)";
|
|
|
|
|
2015-06-23 10:52:49 +03:00
|
|
|
ret = fprintf(fp, "\nAggregated stats:%s\n", msg);
|
2011-03-06 03:40:06 +03:00
|
|
|
|
2015-02-14 20:50:11 +03:00
|
|
|
ret += events_stats__fprintf(&session->evlist->stats, fp);
|
2011-03-06 03:40:06 +03:00
|
|
|
return ret;
|
|
|
|
}
|
2011-03-10 08:23:27 +03:00
|
|
|
|
2011-11-09 19:24:25 +04:00
|
|
|
size_t perf_session__fprintf(struct perf_session *session, FILE *fp)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* FIXME: Here we have to actually print all the machines in this
|
|
|
|
* session, not just the host...
|
|
|
|
*/
|
2012-12-19 02:15:48 +04:00
|
|
|
return machine__fprintf(&session->machines.host, fp);
|
2011-11-09 19:24:25 +04:00
|
|
|
}
|
|
|
|
|
2011-04-07 07:54:20 +04:00
|
|
|
struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session,
|
|
|
|
unsigned int type)
|
|
|
|
{
|
|
|
|
struct perf_evsel *pos;
|
|
|
|
|
2016-06-23 17:26:15 +03:00
|
|
|
evlist__for_each_entry(session->evlist, pos) {
|
2011-04-07 07:54:20 +04:00
|
|
|
if (pos->attr.type == type)
|
|
|
|
return pos;
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2011-07-04 15:57:50 +04:00
|
|
|
int perf_session__cpu_bitmap(struct perf_session *session,
|
|
|
|
const char *cpu_list, unsigned long *cpu_bitmap)
|
|
|
|
{
|
2014-01-20 15:39:39 +04:00
|
|
|
int i, err = -1;
|
2011-07-04 15:57:50 +04:00
|
|
|
struct cpu_map *map;
|
|
|
|
|
|
|
|
for (i = 0; i < PERF_TYPE_MAX; ++i) {
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
|
|
|
|
evsel = perf_session__find_first_evtype(session, i);
|
|
|
|
if (!evsel)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
if (!(evsel->attr.sample_type & PERF_SAMPLE_CPU)) {
|
|
|
|
pr_err("File does not contain CPU events. "
|
|
|
|
"Remove -c option to proceed.\n");
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
map = cpu_map__new(cpu_list);
|
2011-11-13 21:45:27 +04:00
|
|
|
if (map == NULL) {
|
|
|
|
pr_err("Invalid cpu_list\n");
|
|
|
|
return -1;
|
|
|
|
}
|
2011-07-04 15:57:50 +04:00
|
|
|
|
|
|
|
for (i = 0; i < map->nr; i++) {
|
|
|
|
int cpu = map->map[i];
|
|
|
|
|
|
|
|
if (cpu >= MAX_NR_CPUS) {
|
|
|
|
pr_err("Requested CPU %d too large. "
|
|
|
|
"Consider raising MAX_NR_CPUS\n", cpu);
|
2014-01-20 15:39:39 +04:00
|
|
|
goto out_delete_map;
|
2011-07-04 15:57:50 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
set_bit(cpu, cpu_bitmap);
|
|
|
|
}
|
|
|
|
|
2014-01-20 15:39:39 +04:00
|
|
|
err = 0;
|
|
|
|
|
|
|
|
out_delete_map:
|
2015-06-23 01:36:04 +03:00
|
|
|
cpu_map__put(map);
|
2014-01-20 15:39:39 +04:00
|
|
|
return err;
|
2011-07-04 15:57:50 +04:00
|
|
|
}
|
perf tools: Make perf.data more self-descriptive (v8)
The goal of this patch is to include more information about the host
environment into the perf.data so it is more self-descriptive. Overtime,
profiles are captured on various machines and it becomes hard to track
what was recorded, on what machine and when.
This patch provides a way to solve this by extending the perf.data file
with basic information about the host machine. To add those extensions,
we leverage the feature bits capabilities of the perf.data format. The
change is backward compatible with existing perf.data files.
We define the following useful new extensions:
- HEADER_HOSTNAME: the hostname
- HEADER_OSRELEASE: the kernel release number
- HEADER_ARCH: the hw architecture
- HEADER_CPUDESC: generic CPU description
- HEADER_NRCPUS: number of online/avail cpus
- HEADER_CMDLINE: perf command line
- HEADER_VERSION: perf version
- HEADER_TOPOLOGY: cpu topology
- HEADER_EVENT_DESC: full event description (attrs)
- HEADER_CPUID: easy-to-parse low level CPU identication
The small granularity for the entries is to make it easier to extend
without breaking backward compatiblity. Many entries are provided as
ASCII strings.
Perf report/script have been modified to print the basic information as
easy-to-parse ASCII strings. Extended information about CPU and NUMA
topology may be requested with the -I option.
Thanks to David Ahern for reviewing and testing the many versions of
this patch.
$ perf report --stdio
# ========
# captured on : Mon Sep 26 15:22:14 2011
# hostname : quad
# os release : 3.1.0-rc4-tip
# perf version : 3.1.0-rc4
# arch : x86_64
# nrcpus online : 4
# nrcpus avail : 4
# cpudesc : Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
# cpuid : GenuineIntel,6,15,11
# total memory : 8105360 kB
# cmdline : /home/eranian/perfmon/official/tip/build/tools/perf/perf record date
# event : name = cycles, type = 0, config = 0x0, config1 = 0x0, config2 = 0x0, excl_usr = 0, excl_kern = 0, id = { 29, 30, 31,
# HEADER_CPU_TOPOLOGY info available, use -I to display
# HEADER_NUMA_TOPOLOGY info available, use -I to display
# ========
#
...
$ perf report --stdio -I
# ========
# captured on : Mon Sep 26 15:22:14 2011
# hostname : quad
# os release : 3.1.0-rc4-tip
# perf version : 3.1.0-rc4
# arch : x86_64
# nrcpus online : 4
# nrcpus avail : 4
# cpudesc : Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
# cpuid : GenuineIntel,6,15,11
# total memory : 8105360 kB
# cmdline : /home/eranian/perfmon/official/tip/build/tools/perf/perf record date
# event : name = cycles, type = 0, config = 0x0, config1 = 0x0, config2 = 0x0, excl_usr = 0, excl_kern = 0, id = { 29, 30, 31,
# sibling cores : 0-3
# sibling threads : 0
# sibling threads : 1
# sibling threads : 2
# sibling threads : 3
# node0 meminfo : total = 8320608 kB, free = 7571024 kB
# node0 cpu list : 0-3
# ========
#
...
Reviewed-by: David Ahern <dsahern@gmail.com>
Tested-by: David Ahern <dsahern@gmail.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Robert Richter <robert.richter@amd.com>
Cc: Andi Kleen <ak@linux.intel.com>
Link: http://lkml.kernel.org/r/20110930134040.GA5575@quad
Signed-off-by: Stephane Eranian <eranian@google.com>
[ committer notes: Use --show-info in the tools as was in the docs, rename
perf_header_fprintf_info to perf_file_section__fprintf_info, fixup
conflict with f69b64f7 "perf: Support setting the disassembler style" ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-30 17:40:40 +04:00
|
|
|
|
|
|
|
void perf_session__fprintf_info(struct perf_session *session, FILE *fp,
|
|
|
|
bool full)
|
|
|
|
{
|
|
|
|
if (session == NULL || fp == NULL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
fprintf(fp, "# ========\n");
|
|
|
|
perf_header__fprintf_info(session, fp, full);
|
|
|
|
fprintf(fp, "# ========\n#\n");
|
|
|
|
}
|
2012-06-27 20:08:42 +04:00
|
|
|
|
|
|
|
|
|
|
|
int __perf_session__set_tracepoints_handlers(struct perf_session *session,
|
|
|
|
const struct perf_evsel_str_handler *assocs,
|
|
|
|
size_t nr_assocs)
|
|
|
|
{
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
size_t i;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
for (i = 0; i < nr_assocs; i++) {
|
2013-09-06 22:19:01 +04:00
|
|
|
/*
|
|
|
|
* Adding a handler for an event not in the session,
|
|
|
|
* just ignore it.
|
|
|
|
*/
|
|
|
|
evsel = perf_evlist__find_tracepoint_by_name(session->evlist, assocs[i].name);
|
2012-06-27 20:08:42 +04:00
|
|
|
if (evsel == NULL)
|
2013-09-06 22:19:01 +04:00
|
|
|
continue;
|
2012-06-27 20:08:42 +04:00
|
|
|
|
|
|
|
err = -EEXIST;
|
2013-11-06 17:17:38 +04:00
|
|
|
if (evsel->handler != NULL)
|
2013-09-06 22:19:01 +04:00
|
|
|
goto out;
|
2013-11-06 17:17:38 +04:00
|
|
|
evsel->handler = assocs[i].handler;
|
2012-06-27 20:08:42 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
err = 0;
|
|
|
|
out:
|
|
|
|
return err;
|
|
|
|
}
|
2014-10-27 16:49:22 +03:00
|
|
|
|
|
|
|
int perf_event__process_id_index(struct perf_tool *tool __maybe_unused,
|
|
|
|
union perf_event *event,
|
|
|
|
struct perf_session *session)
|
|
|
|
{
|
|
|
|
struct perf_evlist *evlist = session->evlist;
|
|
|
|
struct id_index_event *ie = &event->id_index;
|
|
|
|
size_t i, nr, max_nr;
|
|
|
|
|
|
|
|
max_nr = (ie->header.size - sizeof(struct id_index_event)) /
|
|
|
|
sizeof(struct id_index_entry);
|
|
|
|
nr = ie->nr;
|
|
|
|
if (nr > max_nr)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
if (dump_trace)
|
|
|
|
fprintf(stdout, " nr: %zu\n", nr);
|
|
|
|
|
|
|
|
for (i = 0; i < nr; i++) {
|
|
|
|
struct id_index_entry *e = &ie->entries[i];
|
|
|
|
struct perf_sample_id *sid;
|
|
|
|
|
|
|
|
if (dump_trace) {
|
|
|
|
fprintf(stdout, " ... id: %"PRIu64, e->id);
|
|
|
|
fprintf(stdout, " idx: %"PRIu64, e->idx);
|
|
|
|
fprintf(stdout, " cpu: %"PRId64, e->cpu);
|
|
|
|
fprintf(stdout, " tid: %"PRId64"\n", e->tid);
|
|
|
|
}
|
|
|
|
|
|
|
|
sid = perf_evlist__id2sid(evlist, e->id);
|
|
|
|
if (!sid)
|
|
|
|
return -ENOENT;
|
|
|
|
sid->idx = e->idx;
|
|
|
|
sid->cpu = e->cpu;
|
|
|
|
sid->tid = e->tid;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
int perf_event__synthesize_id_index(struct perf_tool *tool,
|
|
|
|
perf_event__handler_t process,
|
|
|
|
struct perf_evlist *evlist,
|
|
|
|
struct machine *machine)
|
|
|
|
{
|
|
|
|
union perf_event *ev;
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
size_t nr = 0, i = 0, sz, max_nr, n;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
pr_debug2("Synthesizing id index\n");
|
|
|
|
|
|
|
|
max_nr = (UINT16_MAX - sizeof(struct id_index_event)) /
|
|
|
|
sizeof(struct id_index_entry);
|
|
|
|
|
2016-06-23 17:26:15 +03:00
|
|
|
evlist__for_each_entry(evlist, evsel)
|
2014-10-27 16:49:22 +03:00
|
|
|
nr += evsel->ids;
|
|
|
|
|
|
|
|
n = nr > max_nr ? max_nr : nr;
|
|
|
|
sz = sizeof(struct id_index_event) + n * sizeof(struct id_index_entry);
|
|
|
|
ev = zalloc(sz);
|
|
|
|
if (!ev)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
ev->id_index.header.type = PERF_RECORD_ID_INDEX;
|
|
|
|
ev->id_index.header.size = sz;
|
|
|
|
ev->id_index.nr = n;
|
|
|
|
|
2016-06-23 17:26:15 +03:00
|
|
|
evlist__for_each_entry(evlist, evsel) {
|
2014-10-27 16:49:22 +03:00
|
|
|
u32 j;
|
|
|
|
|
|
|
|
for (j = 0; j < evsel->ids; j++) {
|
|
|
|
struct id_index_entry *e;
|
|
|
|
struct perf_sample_id *sid;
|
|
|
|
|
|
|
|
if (i >= n) {
|
|
|
|
err = process(tool, ev, NULL, machine);
|
|
|
|
if (err)
|
|
|
|
goto out_err;
|
|
|
|
nr -= n;
|
|
|
|
i = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
e = &ev->id_index.entries[i++];
|
|
|
|
|
|
|
|
e->id = evsel->id[j];
|
|
|
|
|
|
|
|
sid = perf_evlist__id2sid(evlist, e->id);
|
|
|
|
if (!sid) {
|
|
|
|
free(ev);
|
|
|
|
return -ENOENT;
|
|
|
|
}
|
|
|
|
|
|
|
|
e->idx = sid->idx;
|
|
|
|
e->cpu = sid->cpu;
|
|
|
|
e->tid = sid->tid;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
sz = sizeof(struct id_index_event) + nr * sizeof(struct id_index_entry);
|
|
|
|
ev->id_index.header.size = sz;
|
|
|
|
ev->id_index.nr = nr;
|
|
|
|
|
|
|
|
err = process(tool, ev, NULL, machine);
|
|
|
|
out_err:
|
|
|
|
free(ev);
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|