2011-02-04 19:45:46 +08:00
|
|
|
#include "annotate.h"
|
2010-05-14 01:47:16 +08:00
|
|
|
#include "util.h"
|
2010-05-21 18:48:39 +08:00
|
|
|
#include "build-id.h"
|
2009-09-28 21:32:55 +08:00
|
|
|
#include "hist.h"
|
2009-12-14 23:10:39 +08:00
|
|
|
#include "session.h"
|
|
|
|
#include "sort.h"
|
2009-12-17 00:31:49 +08:00
|
|
|
#include <math.h>
|
2009-09-28 21:32:55 +08:00
|
|
|
|
2011-10-19 23:09:10 +08:00
|
|
|
static bool hists__filter_entry_by_dso(struct hists *hists,
|
|
|
|
struct hist_entry *he);
|
|
|
|
static bool hists__filter_entry_by_thread(struct hists *hists,
|
|
|
|
struct hist_entry *he);
|
2012-03-16 16:50:51 +08:00
|
|
|
static bool hists__filter_entry_by_symbol(struct hists *hists,
|
|
|
|
struct hist_entry *he);
|
2011-10-19 23:09:10 +08:00
|
|
|
|
2010-07-21 20:19:41 +08:00
|
|
|
enum hist_filter {
|
|
|
|
HIST_FILTER__DSO,
|
|
|
|
HIST_FILTER__THREAD,
|
|
|
|
HIST_FILTER__PARENT,
|
2012-03-16 16:50:51 +08:00
|
|
|
HIST_FILTER__SYMBOL,
|
2010-07-21 20:19:41 +08:00
|
|
|
};
|
|
|
|
|
2009-09-28 21:32:55 +08:00
|
|
|
struct callchain_param callchain_param = {
|
|
|
|
.mode = CHAIN_GRAPH_REL,
|
2011-06-07 23:49:46 +08:00
|
|
|
.min_percent = 0.5,
|
|
|
|
.order = ORDER_CALLEE
|
2009-09-28 21:32:55 +08:00
|
|
|
};
|
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
u16 hists__col_len(struct hists *hists, enum hist_column col)
|
2010-07-21 01:42:52 +08:00
|
|
|
{
|
2011-09-26 23:33:28 +08:00
|
|
|
return hists->col_len[col];
|
2010-07-21 01:42:52 +08:00
|
|
|
}
|
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
void hists__set_col_len(struct hists *hists, enum hist_column col, u16 len)
|
2010-07-21 01:42:52 +08:00
|
|
|
{
|
2011-09-26 23:33:28 +08:00
|
|
|
hists->col_len[col] = len;
|
2010-07-21 01:42:52 +08:00
|
|
|
}
|
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
bool hists__new_col_len(struct hists *hists, enum hist_column col, u16 len)
|
2010-07-21 01:42:52 +08:00
|
|
|
{
|
2011-09-26 23:33:28 +08:00
|
|
|
if (len > hists__col_len(hists, col)) {
|
|
|
|
hists__set_col_len(hists, col, len);
|
2010-07-21 01:42:52 +08:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2012-08-20 12:52:05 +08:00
|
|
|
void hists__reset_col_len(struct hists *hists)
|
2010-07-21 01:42:52 +08:00
|
|
|
{
|
|
|
|
enum hist_column col;
|
|
|
|
|
|
|
|
for (col = 0; col < HISTC_NR_COLS; ++col)
|
2011-09-26 23:33:28 +08:00
|
|
|
hists__set_col_len(hists, col, 0);
|
2010-07-21 01:42:52 +08:00
|
|
|
}
|
|
|
|
|
2012-02-10 06:21:01 +08:00
|
|
|
static void hists__set_unres_dso_col_len(struct hists *hists, int dso)
|
|
|
|
{
|
|
|
|
const unsigned int unresolved_col_width = BITS_PER_LONG / 4;
|
|
|
|
|
|
|
|
if (hists__col_len(hists, dso) < unresolved_col_width &&
|
|
|
|
!symbol_conf.col_width_list_str && !symbol_conf.field_sep &&
|
|
|
|
!symbol_conf.dso_list)
|
|
|
|
hists__set_col_len(hists, dso, unresolved_col_width);
|
|
|
|
}
|
|
|
|
|
2012-08-20 12:52:05 +08:00
|
|
|
void hists__calc_col_len(struct hists *hists, struct hist_entry *h)
|
2010-07-21 01:42:52 +08:00
|
|
|
{
|
2012-02-10 06:21:01 +08:00
|
|
|
const unsigned int unresolved_col_width = BITS_PER_LONG / 4;
|
2010-07-21 01:42:52 +08:00
|
|
|
u16 len;
|
|
|
|
|
|
|
|
if (h->ms.sym)
|
2012-02-10 06:21:01 +08:00
|
|
|
hists__new_col_len(hists, HISTC_SYMBOL, h->ms.sym->namelen + 4);
|
|
|
|
else
|
|
|
|
hists__set_unres_dso_col_len(hists, HISTC_DSO);
|
2010-07-21 01:42:52 +08:00
|
|
|
|
|
|
|
len = thread__comm_len(h->thread);
|
2011-09-26 23:33:28 +08:00
|
|
|
if (hists__new_col_len(hists, HISTC_COMM, len))
|
|
|
|
hists__set_col_len(hists, HISTC_THREAD, len + 6);
|
2010-07-21 01:42:52 +08:00
|
|
|
|
|
|
|
if (h->ms.map) {
|
|
|
|
len = dso__name_len(h->ms.map->dso);
|
2011-09-26 23:33:28 +08:00
|
|
|
hists__new_col_len(hists, HISTC_DSO, len);
|
2010-07-21 01:42:52 +08:00
|
|
|
}
|
2012-02-10 06:21:01 +08:00
|
|
|
|
|
|
|
if (h->branch_info) {
|
|
|
|
int symlen;
|
|
|
|
/*
|
|
|
|
* +4 accounts for '[x] ' priv level info
|
|
|
|
* +2 account of 0x prefix on raw addresses
|
|
|
|
*/
|
|
|
|
if (h->branch_info->from.sym) {
|
|
|
|
symlen = (int)h->branch_info->from.sym->namelen + 4;
|
|
|
|
hists__new_col_len(hists, HISTC_SYMBOL_FROM, symlen);
|
|
|
|
|
|
|
|
symlen = dso__name_len(h->branch_info->from.map->dso);
|
|
|
|
hists__new_col_len(hists, HISTC_DSO_FROM, symlen);
|
|
|
|
} else {
|
|
|
|
symlen = unresolved_col_width + 4 + 2;
|
|
|
|
hists__new_col_len(hists, HISTC_SYMBOL_FROM, symlen);
|
|
|
|
hists__set_unres_dso_col_len(hists, HISTC_DSO_FROM);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (h->branch_info->to.sym) {
|
|
|
|
symlen = (int)h->branch_info->to.sym->namelen + 4;
|
|
|
|
hists__new_col_len(hists, HISTC_SYMBOL_TO, symlen);
|
|
|
|
|
|
|
|
symlen = dso__name_len(h->branch_info->to.map->dso);
|
|
|
|
hists__new_col_len(hists, HISTC_DSO_TO, symlen);
|
|
|
|
} else {
|
|
|
|
symlen = unresolved_col_width + 4 + 2;
|
|
|
|
hists__new_col_len(hists, HISTC_SYMBOL_TO, symlen);
|
|
|
|
hists__set_unres_dso_col_len(hists, HISTC_DSO_TO);
|
|
|
|
}
|
|
|
|
}
|
2010-07-21 01:42:52 +08:00
|
|
|
}
|
|
|
|
|
2012-08-20 12:52:05 +08:00
|
|
|
void hists__output_recalc_col_len(struct hists *hists, int max_rows)
|
|
|
|
{
|
|
|
|
struct rb_node *next = rb_first(&hists->entries);
|
|
|
|
struct hist_entry *n;
|
|
|
|
int row = 0;
|
|
|
|
|
|
|
|
hists__reset_col_len(hists);
|
|
|
|
|
|
|
|
while (next && row++ < max_rows) {
|
|
|
|
n = rb_entry(next, struct hist_entry, rb_node);
|
|
|
|
if (!n->filtered)
|
|
|
|
hists__calc_col_len(hists, n);
|
|
|
|
next = rb_next(&n->rb_node);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-01-04 22:27:03 +08:00
|
|
|
static void hist_entry__add_cpumode_period(struct hist_entry *he,
|
2010-05-15 01:19:35 +08:00
|
|
|
unsigned int cpumode, u64 period)
|
2010-04-19 13:32:50 +08:00
|
|
|
{
|
2010-05-10 00:02:23 +08:00
|
|
|
switch (cpumode) {
|
2010-04-19 13:32:50 +08:00
|
|
|
case PERF_RECORD_MISC_KERNEL:
|
2012-01-04 22:27:03 +08:00
|
|
|
he->period_sys += period;
|
2010-04-19 13:32:50 +08:00
|
|
|
break;
|
|
|
|
case PERF_RECORD_MISC_USER:
|
2012-01-04 22:27:03 +08:00
|
|
|
he->period_us += period;
|
2010-04-19 13:32:50 +08:00
|
|
|
break;
|
|
|
|
case PERF_RECORD_MISC_GUEST_KERNEL:
|
2012-01-04 22:27:03 +08:00
|
|
|
he->period_guest_sys += period;
|
2010-04-19 13:32:50 +08:00
|
|
|
break;
|
|
|
|
case PERF_RECORD_MISC_GUEST_USER:
|
2012-01-04 22:27:03 +08:00
|
|
|
he->period_guest_us += period;
|
2010-04-19 13:32:50 +08:00
|
|
|
break;
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
perf top: Reuse the 'report' hist_entry/hists classes
This actually fixes several problems we had in the old 'perf top':
1. Unresolved symbols not show, limitation that came from the old
"KernelTop" codebase, to solve it we would need to do changes
that would make sym_entry have most of the hist_entry fields.
2. It was using the number of samples, not the sum of sample->period.
And brings the --sort code that allows us to have all the views in
'perf report', for instance:
[root@emilia ~]# perf top --sort dso
PerfTop: 5903 irqs/sec kernel:77.5% exact: 0.0% [1000Hz cycles], (all, 8 CPUs)
------------------------------------------------------------------------------
31.59% libcrypto.so.1.0.0
21.55% [kernel]
18.57% libpython2.6.so.1.0
7.04% libc-2.12.so
6.99% _backend_agg.so
4.72% sshd
1.48% multiarray.so
1.39% libfreetype.so.6.3.22
1.37% perf
0.71% libgobject-2.0.so.0.2200.5
0.53% [tg3]
0.48% libglib-2.0.so.0.2200.5
0.44% libstdc++.so.6.0.13
0.40% libcairo.so.2.10800.8
0.38% libm-2.12.so
0.34% umath.so
0.30% libgdk-x11-2.0.so.0.1800.9
0.22% libpthread-2.12.so
0.20% libgtk-x11-2.0.so.0.1800.9
0.20% librt-2.12.so
0.15% _path.so
0.13% libpango-1.0.so.0.2800.1
0.11% libatlas.so.3.0
0.09% ft2font.so
0.09% libpangoft2-1.0.so.0.2800.1
0.08% libX11.so.6.3.0
0.07% [vdso]
0.06% cyclictest
^C
All the filter lists can be used as well: --dsos, --comms, --symbols,
etc.
The 'perf report' TUI is also reused, being possible to apply all the
zoom operations, do annotation, etc.
This change will allow multiple simplifications in the symbol system as
well, that will be detailed in upcoming changesets.
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-xzaaldxq7zhqrrxdxjifk1mh@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-10-06 06:16:15 +08:00
|
|
|
static void hist_entry__decay(struct hist_entry *he)
|
|
|
|
{
|
|
|
|
he->period = (he->period * 7) / 8;
|
|
|
|
he->nr_events = (he->nr_events * 7) / 8;
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool hists__decay_entry(struct hists *hists, struct hist_entry *he)
|
|
|
|
{
|
2011-10-20 16:45:44 +08:00
|
|
|
u64 prev_period = he->period;
|
|
|
|
|
|
|
|
if (prev_period == 0)
|
2011-10-13 19:01:33 +08:00
|
|
|
return true;
|
2011-10-20 16:45:44 +08:00
|
|
|
|
perf top: Reuse the 'report' hist_entry/hists classes
This actually fixes several problems we had in the old 'perf top':
1. Unresolved symbols not show, limitation that came from the old
"KernelTop" codebase, to solve it we would need to do changes
that would make sym_entry have most of the hist_entry fields.
2. It was using the number of samples, not the sum of sample->period.
And brings the --sort code that allows us to have all the views in
'perf report', for instance:
[root@emilia ~]# perf top --sort dso
PerfTop: 5903 irqs/sec kernel:77.5% exact: 0.0% [1000Hz cycles], (all, 8 CPUs)
------------------------------------------------------------------------------
31.59% libcrypto.so.1.0.0
21.55% [kernel]
18.57% libpython2.6.so.1.0
7.04% libc-2.12.so
6.99% _backend_agg.so
4.72% sshd
1.48% multiarray.so
1.39% libfreetype.so.6.3.22
1.37% perf
0.71% libgobject-2.0.so.0.2200.5
0.53% [tg3]
0.48% libglib-2.0.so.0.2200.5
0.44% libstdc++.so.6.0.13
0.40% libcairo.so.2.10800.8
0.38% libm-2.12.so
0.34% umath.so
0.30% libgdk-x11-2.0.so.0.1800.9
0.22% libpthread-2.12.so
0.20% libgtk-x11-2.0.so.0.1800.9
0.20% librt-2.12.so
0.15% _path.so
0.13% libpango-1.0.so.0.2800.1
0.11% libatlas.so.3.0
0.09% ft2font.so
0.09% libpangoft2-1.0.so.0.2800.1
0.08% libX11.so.6.3.0
0.07% [vdso]
0.06% cyclictest
^C
All the filter lists can be used as well: --dsos, --comms, --symbols,
etc.
The 'perf report' TUI is also reused, being possible to apply all the
zoom operations, do annotation, etc.
This change will allow multiple simplifications in the symbol system as
well, that will be detailed in upcoming changesets.
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-xzaaldxq7zhqrrxdxjifk1mh@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-10-06 06:16:15 +08:00
|
|
|
hist_entry__decay(he);
|
2011-10-20 16:45:44 +08:00
|
|
|
|
|
|
|
if (!he->filtered)
|
|
|
|
hists->stats.total_period -= prev_period - he->period;
|
|
|
|
|
perf top: Reuse the 'report' hist_entry/hists classes
This actually fixes several problems we had in the old 'perf top':
1. Unresolved symbols not show, limitation that came from the old
"KernelTop" codebase, to solve it we would need to do changes
that would make sym_entry have most of the hist_entry fields.
2. It was using the number of samples, not the sum of sample->period.
And brings the --sort code that allows us to have all the views in
'perf report', for instance:
[root@emilia ~]# perf top --sort dso
PerfTop: 5903 irqs/sec kernel:77.5% exact: 0.0% [1000Hz cycles], (all, 8 CPUs)
------------------------------------------------------------------------------
31.59% libcrypto.so.1.0.0
21.55% [kernel]
18.57% libpython2.6.so.1.0
7.04% libc-2.12.so
6.99% _backend_agg.so
4.72% sshd
1.48% multiarray.so
1.39% libfreetype.so.6.3.22
1.37% perf
0.71% libgobject-2.0.so.0.2200.5
0.53% [tg3]
0.48% libglib-2.0.so.0.2200.5
0.44% libstdc++.so.6.0.13
0.40% libcairo.so.2.10800.8
0.38% libm-2.12.so
0.34% umath.so
0.30% libgdk-x11-2.0.so.0.1800.9
0.22% libpthread-2.12.so
0.20% libgtk-x11-2.0.so.0.1800.9
0.20% librt-2.12.so
0.15% _path.so
0.13% libpango-1.0.so.0.2800.1
0.11% libatlas.so.3.0
0.09% ft2font.so
0.09% libpangoft2-1.0.so.0.2800.1
0.08% libX11.so.6.3.0
0.07% [vdso]
0.06% cyclictest
^C
All the filter lists can be used as well: --dsos, --comms, --symbols,
etc.
The 'perf report' TUI is also reused, being possible to apply all the
zoom operations, do annotation, etc.
This change will allow multiple simplifications in the symbol system as
well, that will be detailed in upcoming changesets.
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-xzaaldxq7zhqrrxdxjifk1mh@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-10-06 06:16:15 +08:00
|
|
|
return he->period == 0;
|
|
|
|
}
|
|
|
|
|
2011-10-17 19:05:04 +08:00
|
|
|
static void __hists__decay_entries(struct hists *hists, bool zap_user,
|
|
|
|
bool zap_kernel, bool threaded)
|
perf top: Reuse the 'report' hist_entry/hists classes
This actually fixes several problems we had in the old 'perf top':
1. Unresolved symbols not show, limitation that came from the old
"KernelTop" codebase, to solve it we would need to do changes
that would make sym_entry have most of the hist_entry fields.
2. It was using the number of samples, not the sum of sample->period.
And brings the --sort code that allows us to have all the views in
'perf report', for instance:
[root@emilia ~]# perf top --sort dso
PerfTop: 5903 irqs/sec kernel:77.5% exact: 0.0% [1000Hz cycles], (all, 8 CPUs)
------------------------------------------------------------------------------
31.59% libcrypto.so.1.0.0
21.55% [kernel]
18.57% libpython2.6.so.1.0
7.04% libc-2.12.so
6.99% _backend_agg.so
4.72% sshd
1.48% multiarray.so
1.39% libfreetype.so.6.3.22
1.37% perf
0.71% libgobject-2.0.so.0.2200.5
0.53% [tg3]
0.48% libglib-2.0.so.0.2200.5
0.44% libstdc++.so.6.0.13
0.40% libcairo.so.2.10800.8
0.38% libm-2.12.so
0.34% umath.so
0.30% libgdk-x11-2.0.so.0.1800.9
0.22% libpthread-2.12.so
0.20% libgtk-x11-2.0.so.0.1800.9
0.20% librt-2.12.so
0.15% _path.so
0.13% libpango-1.0.so.0.2800.1
0.11% libatlas.so.3.0
0.09% ft2font.so
0.09% libpangoft2-1.0.so.0.2800.1
0.08% libX11.so.6.3.0
0.07% [vdso]
0.06% cyclictest
^C
All the filter lists can be used as well: --dsos, --comms, --symbols,
etc.
The 'perf report' TUI is also reused, being possible to apply all the
zoom operations, do annotation, etc.
This change will allow multiple simplifications in the symbol system as
well, that will be detailed in upcoming changesets.
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-xzaaldxq7zhqrrxdxjifk1mh@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-10-06 06:16:15 +08:00
|
|
|
{
|
|
|
|
struct rb_node *next = rb_first(&hists->entries);
|
|
|
|
struct hist_entry *n;
|
|
|
|
|
|
|
|
while (next) {
|
|
|
|
n = rb_entry(next, struct hist_entry, rb_node);
|
|
|
|
next = rb_next(&n->rb_node);
|
2011-10-13 19:01:33 +08:00
|
|
|
/*
|
|
|
|
* We may be annotating this, for instance, so keep it here in
|
|
|
|
* case some it gets new samples, we'll eventually free it when
|
|
|
|
* the user stops browsing and it agains gets fully decayed.
|
|
|
|
*/
|
2011-10-17 19:05:04 +08:00
|
|
|
if (((zap_user && n->level == '.') ||
|
|
|
|
(zap_kernel && n->level != '.') ||
|
|
|
|
hists__decay_entry(hists, n)) &&
|
|
|
|
!n->used) {
|
perf top: Reuse the 'report' hist_entry/hists classes
This actually fixes several problems we had in the old 'perf top':
1. Unresolved symbols not show, limitation that came from the old
"KernelTop" codebase, to solve it we would need to do changes
that would make sym_entry have most of the hist_entry fields.
2. It was using the number of samples, not the sum of sample->period.
And brings the --sort code that allows us to have all the views in
'perf report', for instance:
[root@emilia ~]# perf top --sort dso
PerfTop: 5903 irqs/sec kernel:77.5% exact: 0.0% [1000Hz cycles], (all, 8 CPUs)
------------------------------------------------------------------------------
31.59% libcrypto.so.1.0.0
21.55% [kernel]
18.57% libpython2.6.so.1.0
7.04% libc-2.12.so
6.99% _backend_agg.so
4.72% sshd
1.48% multiarray.so
1.39% libfreetype.so.6.3.22
1.37% perf
0.71% libgobject-2.0.so.0.2200.5
0.53% [tg3]
0.48% libglib-2.0.so.0.2200.5
0.44% libstdc++.so.6.0.13
0.40% libcairo.so.2.10800.8
0.38% libm-2.12.so
0.34% umath.so
0.30% libgdk-x11-2.0.so.0.1800.9
0.22% libpthread-2.12.so
0.20% libgtk-x11-2.0.so.0.1800.9
0.20% librt-2.12.so
0.15% _path.so
0.13% libpango-1.0.so.0.2800.1
0.11% libatlas.so.3.0
0.09% ft2font.so
0.09% libpangoft2-1.0.so.0.2800.1
0.08% libX11.so.6.3.0
0.07% [vdso]
0.06% cyclictest
^C
All the filter lists can be used as well: --dsos, --comms, --symbols,
etc.
The 'perf report' TUI is also reused, being possible to apply all the
zoom operations, do annotation, etc.
This change will allow multiple simplifications in the symbol system as
well, that will be detailed in upcoming changesets.
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-xzaaldxq7zhqrrxdxjifk1mh@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-10-06 06:16:15 +08:00
|
|
|
rb_erase(&n->rb_node, &hists->entries);
|
|
|
|
|
2011-10-13 20:06:54 +08:00
|
|
|
if (sort__need_collapse || threaded)
|
perf top: Reuse the 'report' hist_entry/hists classes
This actually fixes several problems we had in the old 'perf top':
1. Unresolved symbols not show, limitation that came from the old
"KernelTop" codebase, to solve it we would need to do changes
that would make sym_entry have most of the hist_entry fields.
2. It was using the number of samples, not the sum of sample->period.
And brings the --sort code that allows us to have all the views in
'perf report', for instance:
[root@emilia ~]# perf top --sort dso
PerfTop: 5903 irqs/sec kernel:77.5% exact: 0.0% [1000Hz cycles], (all, 8 CPUs)
------------------------------------------------------------------------------
31.59% libcrypto.so.1.0.0
21.55% [kernel]
18.57% libpython2.6.so.1.0
7.04% libc-2.12.so
6.99% _backend_agg.so
4.72% sshd
1.48% multiarray.so
1.39% libfreetype.so.6.3.22
1.37% perf
0.71% libgobject-2.0.so.0.2200.5
0.53% [tg3]
0.48% libglib-2.0.so.0.2200.5
0.44% libstdc++.so.6.0.13
0.40% libcairo.so.2.10800.8
0.38% libm-2.12.so
0.34% umath.so
0.30% libgdk-x11-2.0.so.0.1800.9
0.22% libpthread-2.12.so
0.20% libgtk-x11-2.0.so.0.1800.9
0.20% librt-2.12.so
0.15% _path.so
0.13% libpango-1.0.so.0.2800.1
0.11% libatlas.so.3.0
0.09% ft2font.so
0.09% libpangoft2-1.0.so.0.2800.1
0.08% libX11.so.6.3.0
0.07% [vdso]
0.06% cyclictest
^C
All the filter lists can be used as well: --dsos, --comms, --symbols,
etc.
The 'perf report' TUI is also reused, being possible to apply all the
zoom operations, do annotation, etc.
This change will allow multiple simplifications in the symbol system as
well, that will be detailed in upcoming changesets.
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-xzaaldxq7zhqrrxdxjifk1mh@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-10-06 06:16:15 +08:00
|
|
|
rb_erase(&n->rb_node_in, &hists->entries_collapsed);
|
|
|
|
|
|
|
|
hist_entry__free(n);
|
|
|
|
--hists->nr_entries;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-10-17 19:05:04 +08:00
|
|
|
void hists__decay_entries(struct hists *hists, bool zap_user, bool zap_kernel)
|
2011-10-13 20:06:54 +08:00
|
|
|
{
|
2011-10-17 19:05:04 +08:00
|
|
|
return __hists__decay_entries(hists, zap_user, zap_kernel, false);
|
2011-10-13 20:06:54 +08:00
|
|
|
}
|
|
|
|
|
2011-10-17 19:05:04 +08:00
|
|
|
void hists__decay_entries_threaded(struct hists *hists,
|
|
|
|
bool zap_user, bool zap_kernel)
|
2011-10-13 20:06:54 +08:00
|
|
|
{
|
2011-10-17 19:05:04 +08:00
|
|
|
return __hists__decay_entries(hists, zap_user, zap_kernel, true);
|
2011-10-13 20:06:54 +08:00
|
|
|
}
|
|
|
|
|
2009-09-28 21:32:55 +08:00
|
|
|
/*
|
2010-05-15 01:19:35 +08:00
|
|
|
* histogram, sorted on item, collects periods
|
2009-09-28 21:32:55 +08:00
|
|
|
*/
|
|
|
|
|
2010-05-10 00:02:23 +08:00
|
|
|
static struct hist_entry *hist_entry__new(struct hist_entry *template)
|
|
|
|
{
|
2010-08-23 02:05:22 +08:00
|
|
|
size_t callchain_size = symbol_conf.use_callchain ? sizeof(struct callchain_root) : 0;
|
2012-01-04 22:27:03 +08:00
|
|
|
struct hist_entry *he = malloc(sizeof(*he) + callchain_size);
|
2010-05-10 00:02:23 +08:00
|
|
|
|
2012-01-04 22:27:03 +08:00
|
|
|
if (he != NULL) {
|
|
|
|
*he = *template;
|
|
|
|
he->nr_events = 1;
|
|
|
|
if (he->ms.map)
|
|
|
|
he->ms.map->referenced = true;
|
2010-05-10 00:02:23 +08:00
|
|
|
if (symbol_conf.use_callchain)
|
2012-01-04 22:27:03 +08:00
|
|
|
callchain_init(he->callchain);
|
2010-05-10 00:02:23 +08:00
|
|
|
}
|
|
|
|
|
2012-01-04 22:27:03 +08:00
|
|
|
return he;
|
2010-05-10 00:02:23 +08:00
|
|
|
}
|
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
static void hists__inc_nr_entries(struct hists *hists, struct hist_entry *h)
|
2010-05-11 00:57:51 +08:00
|
|
|
{
|
2010-07-21 01:42:52 +08:00
|
|
|
if (!h->filtered) {
|
2011-09-26 23:33:28 +08:00
|
|
|
hists__calc_col_len(hists, h);
|
|
|
|
++hists->nr_entries;
|
2011-10-06 04:50:23 +08:00
|
|
|
hists->stats.total_period += h->period;
|
2010-07-21 01:42:52 +08:00
|
|
|
}
|
2010-05-11 00:57:51 +08:00
|
|
|
}
|
|
|
|
|
2010-07-21 20:19:41 +08:00
|
|
|
static u8 symbol__parent_filter(const struct symbol *parent)
|
|
|
|
{
|
|
|
|
if (symbol_conf.exclude_other && parent == NULL)
|
|
|
|
return 1 << HIST_FILTER__PARENT;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2012-02-10 06:21:01 +08:00
|
|
|
static struct hist_entry *add_hist_entry(struct hists *hists,
|
|
|
|
struct hist_entry *entry,
|
perf hist: Introduce hists class and move lots of methods to it
In cbbc79a we introduced support for multiple events by introducing a
new "event_stat_id" struct and then made several perf_session methods
receive a point to it instead of a pointer to perf_session, and kept the
event_stats and hists rb_tree in perf_session.
While working on the new newt based browser, I realised that it would be
better to introduce a new class, "hists" (short for "histograms"),
renaming the "event_stat_id" struct and the perf_session methods that
were really "hists" methods, as they manipulate only struct hists
members, not touching anything in the other perf_session members.
Other optimizations, such as calculating the maximum lenght of a symbol
name present in an hists instance will be possible as we add them,
avoiding a re-traversal just for finding that information.
The rationale for the name "hists" to replace "event_stat_id" is that we
may have multiple sets of hists for the same event_stat id, as, for
instance, the 'perf diff' tool has, so event stat id is not what
characterizes what this struct and the functions that manipulate it do.
Cc: Eric B Munson <ebmunson@us.ibm.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-11 00:04:11 +08:00
|
|
|
struct addr_location *al,
|
2012-02-10 06:21:01 +08:00
|
|
|
u64 period)
|
2009-10-03 21:42:45 +08:00
|
|
|
{
|
2011-10-06 04:50:23 +08:00
|
|
|
struct rb_node **p;
|
2009-10-03 21:42:45 +08:00
|
|
|
struct rb_node *parent = NULL;
|
|
|
|
struct hist_entry *he;
|
|
|
|
int cmp;
|
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
pthread_mutex_lock(&hists->lock);
|
|
|
|
|
|
|
|
p = &hists->entries_in->rb_node;
|
|
|
|
|
2009-10-03 21:42:45 +08:00
|
|
|
while (*p != NULL) {
|
|
|
|
parent = *p;
|
2011-10-06 04:50:23 +08:00
|
|
|
he = rb_entry(parent, struct hist_entry, rb_node_in);
|
2009-10-03 21:42:45 +08:00
|
|
|
|
2012-02-10 06:21:01 +08:00
|
|
|
cmp = hist_entry__cmp(entry, he);
|
2009-10-03 21:42:45 +08:00
|
|
|
|
|
|
|
if (!cmp) {
|
2010-05-15 01:19:35 +08:00
|
|
|
he->period += period;
|
|
|
|
++he->nr_events;
|
2012-03-27 15:14:18 +08:00
|
|
|
|
|
|
|
/* If the map of an existing hist_entry has
|
|
|
|
* become out-of-date due to an exec() or
|
|
|
|
* similar, update it. Otherwise we will
|
|
|
|
* mis-adjust symbol addresses when computing
|
|
|
|
* the history counter to increment.
|
|
|
|
*/
|
|
|
|
if (he->ms.map != entry->ms.map) {
|
|
|
|
he->ms.map = entry->ms.map;
|
|
|
|
if (he->ms.map)
|
|
|
|
he->ms.map->referenced = true;
|
|
|
|
}
|
2010-05-10 00:02:23 +08:00
|
|
|
goto out;
|
2009-10-03 21:42:45 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
if (cmp < 0)
|
|
|
|
p = &(*p)->rb_left;
|
|
|
|
else
|
|
|
|
p = &(*p)->rb_right;
|
|
|
|
}
|
|
|
|
|
2012-02-10 06:21:01 +08:00
|
|
|
he = hist_entry__new(entry);
|
2009-10-03 21:42:45 +08:00
|
|
|
if (!he)
|
2011-10-06 04:50:23 +08:00
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
rb_link_node(&he->rb_node_in, parent, p);
|
|
|
|
rb_insert_color(&he->rb_node_in, hists->entries_in);
|
2010-05-10 00:02:23 +08:00
|
|
|
out:
|
2010-05-15 01:19:35 +08:00
|
|
|
hist_entry__add_cpumode_period(he, al->cpumode, period);
|
2011-10-06 04:50:23 +08:00
|
|
|
out_unlock:
|
|
|
|
pthread_mutex_unlock(&hists->lock);
|
2009-10-03 21:42:45 +08:00
|
|
|
return he;
|
|
|
|
}
|
|
|
|
|
2012-02-10 06:21:01 +08:00
|
|
|
struct hist_entry *__hists__add_branch_entry(struct hists *self,
|
|
|
|
struct addr_location *al,
|
|
|
|
struct symbol *sym_parent,
|
|
|
|
struct branch_info *bi,
|
|
|
|
u64 period)
|
|
|
|
{
|
|
|
|
struct hist_entry entry = {
|
|
|
|
.thread = al->thread,
|
|
|
|
.ms = {
|
|
|
|
.map = bi->to.map,
|
|
|
|
.sym = bi->to.sym,
|
|
|
|
},
|
|
|
|
.cpu = al->cpu,
|
|
|
|
.ip = bi->to.addr,
|
|
|
|
.level = al->level,
|
|
|
|
.period = period,
|
|
|
|
.parent = sym_parent,
|
|
|
|
.filtered = symbol__parent_filter(sym_parent),
|
|
|
|
.branch_info = bi,
|
2012-10-04 20:49:35 +08:00
|
|
|
.hists = self,
|
2012-02-10 06:21:01 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
return add_hist_entry(self, &entry, al, period);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct hist_entry *__hists__add_entry(struct hists *self,
|
|
|
|
struct addr_location *al,
|
|
|
|
struct symbol *sym_parent, u64 period)
|
|
|
|
{
|
|
|
|
struct hist_entry entry = {
|
|
|
|
.thread = al->thread,
|
|
|
|
.ms = {
|
|
|
|
.map = al->map,
|
|
|
|
.sym = al->sym,
|
|
|
|
},
|
|
|
|
.cpu = al->cpu,
|
|
|
|
.ip = al->addr,
|
|
|
|
.level = al->level,
|
|
|
|
.period = period,
|
|
|
|
.parent = sym_parent,
|
|
|
|
.filtered = symbol__parent_filter(sym_parent),
|
2012-10-04 20:49:35 +08:00
|
|
|
.hists = self,
|
2012-02-10 06:21:01 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
return add_hist_entry(self, &entry, al, period);
|
|
|
|
}
|
|
|
|
|
2009-09-28 21:32:55 +08:00
|
|
|
int64_t
|
|
|
|
hist_entry__cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
struct sort_entry *se;
|
|
|
|
int64_t cmp = 0;
|
|
|
|
|
|
|
|
list_for_each_entry(se, &hist_entry__sort_list, list) {
|
2010-04-15 01:11:29 +08:00
|
|
|
cmp = se->se_cmp(left, right);
|
2009-09-28 21:32:55 +08:00
|
|
|
if (cmp)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return cmp;
|
|
|
|
}
|
|
|
|
|
|
|
|
int64_t
|
|
|
|
hist_entry__collapse(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
struct sort_entry *se;
|
|
|
|
int64_t cmp = 0;
|
|
|
|
|
|
|
|
list_for_each_entry(se, &hist_entry__sort_list, list) {
|
|
|
|
int64_t (*f)(struct hist_entry *, struct hist_entry *);
|
|
|
|
|
2010-04-15 01:11:29 +08:00
|
|
|
f = se->se_collapse ?: se->se_cmp;
|
2009-09-28 21:32:55 +08:00
|
|
|
|
|
|
|
cmp = f(left, right);
|
|
|
|
if (cmp)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return cmp;
|
|
|
|
}
|
|
|
|
|
|
|
|
void hist_entry__free(struct hist_entry *he)
|
|
|
|
{
|
|
|
|
free(he);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* collapse the histogram
|
|
|
|
*/
|
|
|
|
|
2012-09-11 06:15:03 +08:00
|
|
|
static bool hists__collapse_insert_entry(struct hists *hists __maybe_unused,
|
2011-01-14 11:51:58 +08:00
|
|
|
struct rb_root *root,
|
|
|
|
struct hist_entry *he)
|
2009-09-28 21:32:55 +08:00
|
|
|
{
|
2009-12-14 21:37:11 +08:00
|
|
|
struct rb_node **p = &root->rb_node;
|
2009-09-28 21:32:55 +08:00
|
|
|
struct rb_node *parent = NULL;
|
|
|
|
struct hist_entry *iter;
|
|
|
|
int64_t cmp;
|
|
|
|
|
|
|
|
while (*p != NULL) {
|
|
|
|
parent = *p;
|
2011-10-06 04:50:23 +08:00
|
|
|
iter = rb_entry(parent, struct hist_entry, rb_node_in);
|
2009-09-28 21:32:55 +08:00
|
|
|
|
|
|
|
cmp = hist_entry__collapse(iter, he);
|
|
|
|
|
|
|
|
if (!cmp) {
|
2012-09-26 15:47:28 +08:00
|
|
|
iter->period += he->period;
|
|
|
|
iter->period_sys += he->period_sys;
|
|
|
|
iter->period_us += he->period_us;
|
|
|
|
iter->period_guest_sys += he->period_guest_sys;
|
|
|
|
iter->period_guest_us += he->period_guest_us;
|
|
|
|
iter->nr_events += he->nr_events;
|
|
|
|
|
2011-01-14 11:51:58 +08:00
|
|
|
if (symbol_conf.use_callchain) {
|
2012-05-31 13:43:26 +08:00
|
|
|
callchain_cursor_reset(&callchain_cursor);
|
|
|
|
callchain_merge(&callchain_cursor,
|
|
|
|
iter->callchain,
|
2011-01-14 11:51:58 +08:00
|
|
|
he->callchain);
|
|
|
|
}
|
2009-09-28 21:32:55 +08:00
|
|
|
hist_entry__free(he);
|
2010-05-11 00:57:51 +08:00
|
|
|
return false;
|
2009-09-28 21:32:55 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
if (cmp < 0)
|
|
|
|
p = &(*p)->rb_left;
|
|
|
|
else
|
|
|
|
p = &(*p)->rb_right;
|
|
|
|
}
|
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
rb_link_node(&he->rb_node_in, parent, p);
|
|
|
|
rb_insert_color(&he->rb_node_in, root);
|
2010-05-11 00:57:51 +08:00
|
|
|
return true;
|
2009-09-28 21:32:55 +08:00
|
|
|
}
|
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
static struct rb_root *hists__get_rotate_entries_in(struct hists *hists)
|
2009-09-28 21:32:55 +08:00
|
|
|
{
|
2011-10-06 04:50:23 +08:00
|
|
|
struct rb_root *root;
|
|
|
|
|
|
|
|
pthread_mutex_lock(&hists->lock);
|
|
|
|
|
|
|
|
root = hists->entries_in;
|
|
|
|
if (++hists->entries_in > &hists->entries_in_array[1])
|
|
|
|
hists->entries_in = &hists->entries_in_array[0];
|
|
|
|
|
|
|
|
pthread_mutex_unlock(&hists->lock);
|
|
|
|
|
|
|
|
return root;
|
|
|
|
}
|
|
|
|
|
2011-10-19 23:09:10 +08:00
|
|
|
static void hists__apply_filters(struct hists *hists, struct hist_entry *he)
|
|
|
|
{
|
|
|
|
hists__filter_entry_by_dso(hists, he);
|
|
|
|
hists__filter_entry_by_thread(hists, he);
|
2012-03-16 16:50:51 +08:00
|
|
|
hists__filter_entry_by_symbol(hists, he);
|
2011-10-19 23:09:10 +08:00
|
|
|
}
|
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
static void __hists__collapse_resort(struct hists *hists, bool threaded)
|
|
|
|
{
|
|
|
|
struct rb_root *root;
|
2009-09-28 21:32:55 +08:00
|
|
|
struct rb_node *next;
|
|
|
|
struct hist_entry *n;
|
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
if (!sort__need_collapse && !threaded)
|
2009-09-28 21:32:55 +08:00
|
|
|
return;
|
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
root = hists__get_rotate_entries_in(hists);
|
|
|
|
next = rb_first(root);
|
2009-12-14 21:37:11 +08:00
|
|
|
|
2009-09-28 21:32:55 +08:00
|
|
|
while (next) {
|
2011-10-06 04:50:23 +08:00
|
|
|
n = rb_entry(next, struct hist_entry, rb_node_in);
|
|
|
|
next = rb_next(&n->rb_node_in);
|
2009-09-28 21:32:55 +08:00
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
rb_erase(&n->rb_node_in, root);
|
2011-10-19 23:09:10 +08:00
|
|
|
if (hists__collapse_insert_entry(hists, &hists->entries_collapsed, n)) {
|
|
|
|
/*
|
|
|
|
* If it wasn't combined with one of the entries already
|
|
|
|
* collapsed, we need to apply the filters that may have
|
|
|
|
* been set by, say, the hist_browser.
|
|
|
|
*/
|
|
|
|
hists__apply_filters(hists, n);
|
|
|
|
}
|
2009-09-28 21:32:55 +08:00
|
|
|
}
|
2011-10-06 04:50:23 +08:00
|
|
|
}
|
2009-12-14 21:37:11 +08:00
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
void hists__collapse_resort(struct hists *hists)
|
|
|
|
{
|
|
|
|
return __hists__collapse_resort(hists, false);
|
|
|
|
}
|
|
|
|
|
|
|
|
void hists__collapse_resort_threaded(struct hists *hists)
|
|
|
|
{
|
|
|
|
return __hists__collapse_resort(hists, true);
|
2009-09-28 21:32:55 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2010-05-15 01:19:35 +08:00
|
|
|
* reverse the map, sort on period.
|
2009-09-28 21:32:55 +08:00
|
|
|
*/
|
|
|
|
|
perf hist: Introduce hists class and move lots of methods to it
In cbbc79a we introduced support for multiple events by introducing a
new "event_stat_id" struct and then made several perf_session methods
receive a point to it instead of a pointer to perf_session, and kept the
event_stats and hists rb_tree in perf_session.
While working on the new newt based browser, I realised that it would be
better to introduce a new class, "hists" (short for "histograms"),
renaming the "event_stat_id" struct and the perf_session methods that
were really "hists" methods, as they manipulate only struct hists
members, not touching anything in the other perf_session members.
Other optimizations, such as calculating the maximum lenght of a symbol
name present in an hists instance will be possible as we add them,
avoiding a re-traversal just for finding that information.
The rationale for the name "hists" to replace "event_stat_id" is that we
may have multiple sets of hists for the same event_stat id, as, for
instance, the 'perf diff' tool has, so event stat id is not what
characterizes what this struct and the functions that manipulate it do.
Cc: Eric B Munson <ebmunson@us.ibm.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-11 00:04:11 +08:00
|
|
|
static void __hists__insert_output_entry(struct rb_root *entries,
|
|
|
|
struct hist_entry *he,
|
|
|
|
u64 min_callchain_hits)
|
2009-09-28 21:32:55 +08:00
|
|
|
{
|
perf hist: Introduce hists class and move lots of methods to it
In cbbc79a we introduced support for multiple events by introducing a
new "event_stat_id" struct and then made several perf_session methods
receive a point to it instead of a pointer to perf_session, and kept the
event_stats and hists rb_tree in perf_session.
While working on the new newt based browser, I realised that it would be
better to introduce a new class, "hists" (short for "histograms"),
renaming the "event_stat_id" struct and the perf_session methods that
were really "hists" methods, as they manipulate only struct hists
members, not touching anything in the other perf_session members.
Other optimizations, such as calculating the maximum lenght of a symbol
name present in an hists instance will be possible as we add them,
avoiding a re-traversal just for finding that information.
The rationale for the name "hists" to replace "event_stat_id" is that we
may have multiple sets of hists for the same event_stat id, as, for
instance, the 'perf diff' tool has, so event stat id is not what
characterizes what this struct and the functions that manipulate it do.
Cc: Eric B Munson <ebmunson@us.ibm.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-11 00:04:11 +08:00
|
|
|
struct rb_node **p = &entries->rb_node;
|
2009-09-28 21:32:55 +08:00
|
|
|
struct rb_node *parent = NULL;
|
|
|
|
struct hist_entry *iter;
|
|
|
|
|
2009-12-16 06:04:42 +08:00
|
|
|
if (symbol_conf.use_callchain)
|
2010-04-02 20:50:42 +08:00
|
|
|
callchain_param.sort(&he->sorted_chain, he->callchain,
|
2009-09-28 21:32:55 +08:00
|
|
|
min_callchain_hits, &callchain_param);
|
|
|
|
|
|
|
|
while (*p != NULL) {
|
|
|
|
parent = *p;
|
|
|
|
iter = rb_entry(parent, struct hist_entry, rb_node);
|
|
|
|
|
2010-05-15 01:19:35 +08:00
|
|
|
if (he->period > iter->period)
|
2009-09-28 21:32:55 +08:00
|
|
|
p = &(*p)->rb_left;
|
|
|
|
else
|
|
|
|
p = &(*p)->rb_right;
|
|
|
|
}
|
|
|
|
|
|
|
|
rb_link_node(&he->rb_node, parent, p);
|
perf hist: Introduce hists class and move lots of methods to it
In cbbc79a we introduced support for multiple events by introducing a
new "event_stat_id" struct and then made several perf_session methods
receive a point to it instead of a pointer to perf_session, and kept the
event_stats and hists rb_tree in perf_session.
While working on the new newt based browser, I realised that it would be
better to introduce a new class, "hists" (short for "histograms"),
renaming the "event_stat_id" struct and the perf_session methods that
were really "hists" methods, as they manipulate only struct hists
members, not touching anything in the other perf_session members.
Other optimizations, such as calculating the maximum lenght of a symbol
name present in an hists instance will be possible as we add them,
avoiding a re-traversal just for finding that information.
The rationale for the name "hists" to replace "event_stat_id" is that we
may have multiple sets of hists for the same event_stat id, as, for
instance, the 'perf diff' tool has, so event stat id is not what
characterizes what this struct and the functions that manipulate it do.
Cc: Eric B Munson <ebmunson@us.ibm.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-11 00:04:11 +08:00
|
|
|
rb_insert_color(&he->rb_node, entries);
|
2009-09-28 21:32:55 +08:00
|
|
|
}
|
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
static void __hists__output_resort(struct hists *hists, bool threaded)
|
2009-09-28 21:32:55 +08:00
|
|
|
{
|
2011-10-06 04:50:23 +08:00
|
|
|
struct rb_root *root;
|
2009-09-28 21:32:55 +08:00
|
|
|
struct rb_node *next;
|
|
|
|
struct hist_entry *n;
|
|
|
|
u64 min_callchain_hits;
|
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
min_callchain_hits = hists->stats.total_period * (callchain_param.min_percent / 100);
|
2009-09-28 21:32:55 +08:00
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
if (sort__need_collapse || threaded)
|
|
|
|
root = &hists->entries_collapsed;
|
|
|
|
else
|
|
|
|
root = hists->entries_in;
|
|
|
|
|
|
|
|
next = rb_first(root);
|
|
|
|
hists->entries = RB_ROOT;
|
2009-09-28 21:32:55 +08:00
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
hists->nr_entries = 0;
|
2011-10-27 19:19:48 +08:00
|
|
|
hists->stats.total_period = 0;
|
2011-09-26 23:33:28 +08:00
|
|
|
hists__reset_col_len(hists);
|
2010-05-11 00:57:51 +08:00
|
|
|
|
2009-09-28 21:32:55 +08:00
|
|
|
while (next) {
|
2011-10-06 04:50:23 +08:00
|
|
|
n = rb_entry(next, struct hist_entry, rb_node_in);
|
|
|
|
next = rb_next(&n->rb_node_in);
|
2009-09-28 21:32:55 +08:00
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
__hists__insert_output_entry(&hists->entries, n, min_callchain_hits);
|
2011-09-26 23:33:28 +08:00
|
|
|
hists__inc_nr_entries(hists, n);
|
2009-09-28 21:32:55 +08:00
|
|
|
}
|
2011-10-06 04:50:23 +08:00
|
|
|
}
|
2009-12-14 21:37:11 +08:00
|
|
|
|
2011-10-06 04:50:23 +08:00
|
|
|
void hists__output_resort(struct hists *hists)
|
|
|
|
{
|
|
|
|
return __hists__output_resort(hists, false);
|
|
|
|
}
|
|
|
|
|
|
|
|
void hists__output_resort_threaded(struct hists *hists)
|
|
|
|
{
|
|
|
|
return __hists__output_resort(hists, true);
|
2009-09-28 21:32:55 +08:00
|
|
|
}
|
2009-12-16 22:27:09 +08:00
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
static void hists__remove_entry_filter(struct hists *hists, struct hist_entry *h,
|
2010-07-16 23:35:07 +08:00
|
|
|
enum hist_filter filter)
|
|
|
|
{
|
|
|
|
h->filtered &= ~(1 << filter);
|
|
|
|
if (h->filtered)
|
|
|
|
return;
|
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
++hists->nr_entries;
|
2010-07-27 04:13:40 +08:00
|
|
|
if (h->ms.unfolded)
|
2011-09-26 23:33:28 +08:00
|
|
|
hists->nr_entries += h->nr_rows;
|
2010-07-27 04:13:40 +08:00
|
|
|
h->row_offset = 0;
|
2011-09-26 23:33:28 +08:00
|
|
|
hists->stats.total_period += h->period;
|
|
|
|
hists->stats.nr_events[PERF_RECORD_SAMPLE] += h->nr_events;
|
2010-07-16 23:35:07 +08:00
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
hists__calc_col_len(hists, h);
|
2010-07-16 23:35:07 +08:00
|
|
|
}
|
|
|
|
|
2011-10-19 23:09:10 +08:00
|
|
|
|
|
|
|
static bool hists__filter_entry_by_dso(struct hists *hists,
|
|
|
|
struct hist_entry *he)
|
|
|
|
{
|
|
|
|
if (hists->dso_filter != NULL &&
|
|
|
|
(he->ms.map == NULL || he->ms.map->dso != hists->dso_filter)) {
|
|
|
|
he->filtered |= (1 << HIST_FILTER__DSO);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2011-10-19 05:07:34 +08:00
|
|
|
void hists__filter_by_dso(struct hists *hists)
|
2010-05-11 22:10:15 +08:00
|
|
|
{
|
|
|
|
struct rb_node *nd;
|
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
hists->nr_entries = hists->stats.total_period = 0;
|
|
|
|
hists->stats.nr_events[PERF_RECORD_SAMPLE] = 0;
|
|
|
|
hists__reset_col_len(hists);
|
2010-05-11 22:10:15 +08:00
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) {
|
2010-05-11 22:10:15 +08:00
|
|
|
struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node);
|
|
|
|
|
|
|
|
if (symbol_conf.exclude_other && !h->parent)
|
|
|
|
continue;
|
|
|
|
|
2011-10-19 23:09:10 +08:00
|
|
|
if (hists__filter_entry_by_dso(hists, h))
|
2010-05-11 22:10:15 +08:00
|
|
|
continue;
|
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
hists__remove_entry_filter(hists, h, HIST_FILTER__DSO);
|
2010-05-11 22:10:15 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-10-19 23:09:10 +08:00
|
|
|
static bool hists__filter_entry_by_thread(struct hists *hists,
|
|
|
|
struct hist_entry *he)
|
|
|
|
{
|
|
|
|
if (hists->thread_filter != NULL &&
|
|
|
|
he->thread != hists->thread_filter) {
|
|
|
|
he->filtered |= (1 << HIST_FILTER__THREAD);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2011-10-19 05:07:34 +08:00
|
|
|
void hists__filter_by_thread(struct hists *hists)
|
2010-05-11 22:10:15 +08:00
|
|
|
{
|
|
|
|
struct rb_node *nd;
|
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
hists->nr_entries = hists->stats.total_period = 0;
|
|
|
|
hists->stats.nr_events[PERF_RECORD_SAMPLE] = 0;
|
|
|
|
hists__reset_col_len(hists);
|
2010-05-11 22:10:15 +08:00
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) {
|
2010-05-11 22:10:15 +08:00
|
|
|
struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node);
|
|
|
|
|
2011-10-19 23:09:10 +08:00
|
|
|
if (hists__filter_entry_by_thread(hists, h))
|
2010-05-11 22:10:15 +08:00
|
|
|
continue;
|
2010-07-16 23:35:07 +08:00
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
hists__remove_entry_filter(hists, h, HIST_FILTER__THREAD);
|
2010-05-11 22:10:15 +08:00
|
|
|
}
|
|
|
|
}
|
2010-05-12 10:18:06 +08:00
|
|
|
|
2012-03-16 16:50:51 +08:00
|
|
|
static bool hists__filter_entry_by_symbol(struct hists *hists,
|
|
|
|
struct hist_entry *he)
|
|
|
|
{
|
|
|
|
if (hists->symbol_filter_str != NULL &&
|
|
|
|
(!he->ms.sym || strstr(he->ms.sym->name,
|
|
|
|
hists->symbol_filter_str) == NULL)) {
|
|
|
|
he->filtered |= (1 << HIST_FILTER__SYMBOL);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
void hists__filter_by_symbol(struct hists *hists)
|
|
|
|
{
|
|
|
|
struct rb_node *nd;
|
|
|
|
|
|
|
|
hists->nr_entries = hists->stats.total_period = 0;
|
|
|
|
hists->stats.nr_events[PERF_RECORD_SAMPLE] = 0;
|
|
|
|
hists__reset_col_len(hists);
|
|
|
|
|
|
|
|
for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) {
|
|
|
|
struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node);
|
|
|
|
|
|
|
|
if (hists__filter_entry_by_symbol(hists, h))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
hists__remove_entry_filter(hists, h, HIST_FILTER__SYMBOL);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-02-04 23:43:24 +08:00
|
|
|
int hist_entry__inc_addr_samples(struct hist_entry *he, int evidx, u64 ip)
|
2010-05-12 10:18:06 +08:00
|
|
|
{
|
2011-02-04 23:43:24 +08:00
|
|
|
return symbol__inc_addr_samples(he->ms.sym, he->ms.map, evidx, ip);
|
2010-05-12 10:18:06 +08:00
|
|
|
}
|
|
|
|
|
2011-02-08 23:27:39 +08:00
|
|
|
int hist_entry__annotate(struct hist_entry *he, size_t privsize)
|
2010-05-12 10:18:06 +08:00
|
|
|
{
|
2011-02-08 23:27:39 +08:00
|
|
|
return symbol__annotate(he->ms.sym, he->ms.map, privsize);
|
2010-05-12 10:18:06 +08:00
|
|
|
}
|
2010-05-14 21:36:42 +08:00
|
|
|
|
2011-09-26 23:33:28 +08:00
|
|
|
void hists__inc_nr_events(struct hists *hists, u32 type)
|
2010-05-14 21:36:42 +08:00
|
|
|
{
|
2011-09-26 23:33:28 +08:00
|
|
|
++hists->stats.nr_events[0];
|
|
|
|
++hists->stats.nr_events[type];
|
2010-05-14 21:36:42 +08:00
|
|
|
}
|