On 10/16/2019 6:27 PM, Jiri Olsa wrote:
On Wed, Oct 16, 2019 at 06:22:06PM +0800, Jin, Yao wrote:
On 10/16/2019 3:27 PM, Jiri Olsa wrote:
On Wed, Oct 16, 2019 at 08:59:13AM +0800, Jin, Yao wrote:
On 10/15/2019 5:14 PM, Jiri Olsa wrote:
On Tue, Oct 15, 2019 at 10:53:57AM +0800, Jin Yao
On 10/16/2019 6:15 PM, Jiri Olsa wrote:
On Tue, Oct 15, 2019 at 10:53:18PM +0800, Jin, Yao wrote:
SNIP
+static struct block_header_column{
+ const char *name;
+ int width;
+} block_columns[PERF_HPP_REPORT__BLOCK_MAX_INDEX] = {
+ [PERF_HPP_REPORT__BLOCK_TOTAL_CYCLES_COV
On 10/16/2019 8:53 PM, Jiri Olsa wrote:
On Wed, Oct 16, 2019 at 06:51:07PM +0800, Jin, Yao wrote:
On 10/16/2019 6:15 PM, Jiri Olsa wrote:
On Tue, Oct 15, 2019 at 10:53:18PM +0800, Jin, Yao wrote:
SNIP
+static struct block_header_column{
+ const char *name;
+ int width
display the deprecated
events under the title "--- Following are deprecated events ---".
Signed-off-by: Jin Yao
---
tools/perf/pmu-events/jevents.c| 26 --
tools/perf/pmu-events/jevents.h| 3 +-
tools/perf/pmu-events/pmu-events.h | 1 +
g to back this up ;-)
if that's the case we can go with the original patch
jirka
I'm fine to go with the original patch. :)
Thanks
Jin Yao
So I think the only sane option is to hide it by default.
-Andi
Since now we go back to this version, can this patch be accepted?
Thanks
Jin Yao
On 10/15/2019 10:53 AM, Jin Yao wrote:
There are some deprecated events listed by perf list. But we can't remove
them from perf list with ease because some old scripts may use them.
Deprecated events ar
On 10/16/2019 8:53 PM, Jiri Olsa wrote:
On Wed, Oct 16, 2019 at 06:51:07PM +0800, Jin, Yao wrote:
On 10/16/2019 6:15 PM, Jiri Olsa wrote:
On Tue, Oct 15, 2019 at 10:53:18PM +0800, Jin, Yao wrote:
SNIP
+static struct block_header_column{
+ const char *name;
+ int width
On 10/22/2019 12:07 AM, Jiri Olsa wrote:
On Tue, Oct 15, 2019 at 01:33:48PM +0800, Jin Yao wrote:
SNIP
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -51,6 +51,7 @@
#include "util/util.h" // perf_tip()
#include "ui/ui.h"
#include "
On 10/22/2019 12:08 AM, Jiri Olsa wrote:
On Tue, Oct 15, 2019 at 01:33:48PM +0800, Jin Yao wrote:
SNIP
+ cycles += bi->cycles_aggr / bi->num_aggr;
+
+ he_block = hists__add_entry_block(&bh-&g
and tui. And also with the supporting
of --percent-limit.
v3:
---
1. Move common block info functions to block-info.h/block-info.c
2. Remove nasty hack for skipping calculation of column length.
3. Some minor cleanup.
v2:
---
Rebase to perf/core branch
Jin Yao (5):
perf util: C
more common part to block-info, such as
block_info__process_sym.
4. Remove the nasty hack for skipping calculation of column
length
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c| 118
tools/perf/util/Build| 1 +
tools/perf/util/block-i
116.0K0.42% 19
[random_r.c:357 -> random_r.c:394] libc-2.27.so
It only shows the entries which 'Sampled Cycles%' > 1%.
v3:
---
Minor change since the function name is changed:
block_total_cycles_percent -> block_info__
57 -> random_r.c:394] libc-2.27.so
v3:
---
Minor change since the function name is changed:
block_total_cycles_percent -> block_info__total_cycles_percent
Signed-off-by: Jin Yao
---
tools/perf/builtin-report.c| 30 +---
tools/perf/ui/browsers/hists.c | 62 +++
a new argument 'total_cycles' in hist__account_cycles(),
which will be added with the cycles of each sample.
Signed-off-by: Jin Yao
---
tools/perf/builtin-annotate.c | 2 +-
tools/perf/builtin-diff.c | 3 ++-
tools/perf/builtin-report.c | 2 +-
tools/perf/builtin-top.c | 3 ++-
1
[perf_event_nmi_handler+57 -> perf_event_nmi_handler+63] [kernel.kallsyms]
0.00% 10.02% 1
[perf_event_nmi_handler+68 -> perf_event_nmi_handler+74] [kernel.kallsyms]
v3:
---
1. Use common function block_info__pr
E_RUNNING
disabled 1
inherit 1
exclude_guest1
...
Signed-off-by: Jin Yao
---
tools/perf/pmu-events/jevents.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/tools/perf/pm
Hi Arnaldo,
Is this patch acceptable?
Thanks
Jin Yao
On 1/7/2019 5:05 PM, Jiri Olsa wrote:
On Mon, Jan 07, 2019 at 09:00:50AM +0800, Jin Yao wrote:
Following test shows the stat keeps running even if no longer
task to monitor (mgen exits at ~5s).
perf stat -e cycles -p `pgrep mgen` -I1000
In previous patch, we have already linked up the same basic blocks.
Now we compute the cycles diff value of basic blocks, in order to
sort by diff cycles later.
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 31 +++
tools/perf/util/sort.h| 2 ++
2 files
)
0 __random (3ab03:3ab0f)
0 __random (3ab14:3ab1b)
0 __random (3ab28:3ab2e)
0 __random (3ab4a:3ab53)
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 168 ++
tools/perf/util/hist.c| 2 +-
tools
Documentation the new option '--basic-block'.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 5 +
1 file changed, 5 insertions(+)
diff --git a/tools/perf/Documentation/perf-diff.txt
b/tools/perf/Documentation/perf-diff.txt
index da7809b..b242af8 100644
-
-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 66 +++
1 file changed, 66 insertions(+)
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index 09551fe..72c33ab 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/builtin-diff.c
, which will be
used in next patch for accounting cycles per basic block per
function.
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 81 ++-
1 file changed, 73 insertions(+), 8 deletions(-)
diff --git a/tools/perf/builtin-diff.c b/tools/perf
erf-diff option, which enables the displaying
of cycles difference of same program basic block amongst two or more
perf.data. The program basic block is the code block between two branches
in a function.
Jin Yao (9):
perf util: Create block_info structure
perf util: Add block_info in hist_entry
p
We will expand perf diff to support diff cycles of individual programs
blocks, so it requires all data files having branch stacks.
This patch checks HEADER_BRANCH_STACK in header, and only set the flag
has_br_stack when HEADER_BRANCH_STACK are set in all data files.
Signed-off-by: Jin Yao
field to hist_entry. In order not to
impact current interface, we creates a new function hists__add_entry_block.
Signed-off-by: Jin Yao
---
tools/perf/util/hist.c | 22 --
tools/perf/util/hist.h | 6 ++
tools/perf/util/sort.h | 1 +
3 files changed, 27 insertions(+), 2
(function), start/end addrress of this block, cycles. This patch
creates this structure and with some ops.
Signed-off-by: Jin Yao
---
tools/perf/util/symbol.c | 22 ++
tools/perf/util/symbol.h | 23 +++
2 files changed, 45 insertions(+)
diff --git a/tools/perf
compare, sort and print the basic blocks.
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 145 ++
1 file changed, 145 insertions(+)
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index e067ac6..09551fe 100644
--- a/tools/perf
On 5/22/2019 10:04 PM, Jiri Olsa wrote:
On Mon, May 20, 2019 at 09:27:55PM +0800, Jin Yao wrote:
Currently we only support sorting by diff cycles.
For example,
perf record -b ./div
perf record -b ./div
perf diff --basic-block
# Cycles diff Basic block (start:end
2.27.so [.] rand
2.14%[rand@plt+0 -> rand@plt+0]0 ±
0.0%div[.] rand@plt
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 5 ++
tools/perf/builtin-diff.c | 94 +++---
On 8/12/2019 4:35 PM, Jiri Olsa wrote:
On Sat, Aug 10, 2019 at 07:30:29AM +0800, Jin Yao wrote:
SNIP
static int process_block_per_sym(struct hist_entry *he)
@@ -684,6 +694,21 @@ static struct hist_entry *get_block_pair(struct hist_entry
*he,
return NULL;
}
+static void
On 8/12/2019 4:35 PM, Jiri Olsa wrote:
On Sat, Aug 10, 2019 at 07:30:29AM +0800, Jin Yao wrote:
SNIP
+
+ init_stats(&pair->diff.stats);
+ init_spark_values(pair->diff.svals, NUM_SPARKS);
+
+ for (int i = 0; i < pair->bl
On 8/12/2019 4:35 PM, Jiri Olsa wrote:
On Sat, Aug 10, 2019 at 07:30:29AM +0800, Jin Yao wrote:
SNIP
+ if (vals[i] != 0)
+ return 0;
+ return 1;
+}
+
+static int print_cycles_spark(char *bf, int size, unsigned long *svals, u64 n)
+{
+ int len
n function ‘compute_cycles_diff’:
builtin-diff.c:712:10: error: taking the absolute value of unsigned type
‘u64’ {aka ‘long unsigned int’} has no effect [-Werror=absolute-value]
712 | labs(pair->block_info->cycles_spark[i] -
| ^~~~
Because the result of u64 - u64 is s
iff.c: In function ‘compute_cycles_diff’:
builtin-diff.c:712:10: error: taking the absolute value of unsigned type ‘u64’
{aka ‘long unsigned int’} has no effect [-Werror=absolute-value]
712 | labs(pair->block_info->cycles_spark[i] -
| ^~~~
Because the re
diff.c:712:10: error: taking the absolute value of unsigned type ‘u64’
{aka ‘long unsigned int’} has no effect [-Werror=absolute-value]
712 | labs(pair->block_info->cycles_spark[i] -
| ^~~~
Because the result of u64 - u64 is still u64. No
Hi,
Any comments for this version?
Thanks
Jin Yao
On 7/25/2019 6:14 AM, Jin Yao wrote:
This patch prints the stddev and hist for the cycles diff of
program block. It can help us to understand if the cycles diff
is noisy or not.
This patch is inspired by Andi Kleen's patch
https://lw
On 7/30/2019 9:43 AM, Jin, Yao wrote:
On 7/30/2019 2:16 AM, Andi Kleen wrote:
diff --git a/tools/perf/pmu-events/jevents.c
b/tools/perf/pmu-events/jevents.c
index 1a91a197cafb..d413761621b0 100644
--- a/tools/perf/pmu-events/jevents.c
+++ b/tools/perf/pmu-events/jevents.c
@@ -453,6 +453,7
On 8/6/2019 4:34 PM, Jiri Olsa wrote:
On Thu, Jul 25, 2019 at 06:14:32AM +0800, Jin Yao wrote:
SNIP
static int cycles_printf(struct hist_entry *he, struct hist_entry *pair,
-struct perf_hpp *hpp, int width)
+struct perf_hpp *hpp, int width
On 8/6/2019 4:34 PM, Jiri Olsa wrote:
On Thu, Jul 25, 2019 at 06:14:32AM +0800, Jin Yao wrote:
SNIP
+}
+
double avg_stats(struct stats *stats)
{
return stats->mean;
diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
index 95b4de7a9d51..3448d319a220 100644
--- a/to
On 8/6/2019 4:34 PM, Jiri Olsa wrote:
On Thu, Jul 25, 2019 at 06:14:32AM +0800, Jin Yao wrote:
This patch prints the stddev and hist for the cycles diff of
program block. It can help us to understand if the cycles diff
is noisy or not.
This patch is inspired by Andi Kleen's patch
he fixed counter encodings between JSON and perf.
If big cores don't have this event, nothing will be generated in perf list.
In big cores pipeline.json, there is only "CPU_CLK_UNHALTED.THREAD", and
there is no "CPU_CLK_UNHALTED.CORE" defined. So at least for now,
CPU_CLK_UNHALTED.CORE will not be generated for big core.
Thanks
Jin Yao
On 8/8/2019 9:56 PM, Arnaldo Carvalho de Melo wrote:
Em Mon, Jul 29, 2019 at 03:27:55PM +0800, Jin Yao escreveu:
The events defined in pmu-events JSON are parsed and added into
perf tool. For fixed counters, we handle the encodings between
JSON and perf by using a static array fixed[].
But
On 6/24/2019 3:57 PM, Jiri Olsa wrote:
On Thu, Jun 20, 2019 at 10:36:39PM +0800, Jin Yao wrote:
SNIP
+
+static void *block_entry_zalloc(size_t size)
+{
+ return zalloc(size + sizeof(struct hist_entry));
+}
+
+static void block_entry_free(void *he)
+{
+ struct block_info *bi
On 6/25/2019 5:11 PM, Jiri Olsa wrote:
On Thu, Jun 20, 2019 at 10:36:39PM +0800, Jin Yao wrote:
SNIP
+
+static void *block_entry_zalloc(size_t size)
+{
+ return zalloc(size + sizeof(struct hist_entry));
+}
+
+static void block_entry_free(void *he)
+{
+ struct block_info *bi
check_file_brstack() from __cmd_diff() to cmd_diff().
Because later patch will check flag 'has_br_stack' before
ui_init().
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 29 +
1 file changed, 29 insertions(+)
diff --git a/tools/perf/builtin-diff.c b/tools/pe
Documentation the new computation selection 'cycles'.
v4:
---
Change the column 'Block cycles diff [start:end]' to
'[Program Block Range] Cycles Diff'
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 17 ++---
1 file changed,
field to hist_entry. In order not to
impact current interface, we creates a new function hists__add_entry_block.
Signed-off-by: Jin Yao
---
tools/perf/util/hist.c | 22 --
tools/perf/util/hist.h | 6 ++
tools/perf/util/sort.h | 1 +
3 files changed, 27 insertions(+), 2
the 'Baseline' and
'Shared Object'.
The output is sorted by "Baseline" and the basic blocks in the same
function are sorted by cycles diff.
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 80 ---
tools/perf/ui/stdi
data files.
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 90 +++
1 file changed, 90 insertions(+)
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index ff2c076..864423a 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/buil
pdate_blocked_averages (2bf980:2bf9d3)
48 update_blocked_averages (2bf934:2bf942)
-35 native_write_msr (267900:26790b)
26 native_irq_return_iret (c00a27:c00a27)
22 rcu_check_callbacks (2febb6:2febdc)
-21 __hrtimer_run_queues (30b220:30b2a3)
(function), start/end address of this block, cycles. This patch
creates this structure and with some ops.
Signed-off-by: Jin Yao
---
tools/perf/util/symbol.c | 22 ++
tools/perf/util/symbol.h | 23 +++
2 files changed, 45 insertions(+)
diff --git a/tools/perf
abs() -> labs(), in block_cycles_diff_cmp().
v2:
---
v1 adds the basic block entries to per data-file hists
but v2 adds the basic block entries to per symbol hists.
That is to keep current perf-diff format. Will show the
result in next patches.
Signed-off-by: Jin Yao
---
tools/perf/b
On 6/27/2019 3:27 PM, Jiri Olsa wrote:
On Thu, Jun 27, 2019 at 10:09:26PM +0800, Jin Yao wrote:
SNIP
+
+static int process_block_per_sym(struct hist_entry *he)
+{
+ struct annotation *notes;
+ struct cyc_hist *ch;
+ struct block_hist *bh;
+
+ if (!he->ms.map ||
On 6/8/2019 7:41 PM, Jin, Yao wrote:
On 6/5/2019 7:44 PM, Jiri Olsa wrote:
On Mon, Jun 03, 2019 at 10:36:14PM +0800, Jin Yao wrote:
SNIP
diff --git a/tools/perf/util/sort.h b/tools/perf/util/sort.h
index 43623fa..d1641da 100644
--- a/tools/perf/util/sort.h
+++ b/tools/perf/util/sort.h
On 7/3/2019 12:20 AM, Arnaldo Carvalho de Melo wrote:
Em Tue, Jul 02, 2019 at 01:17:39PM -0300, Arnaldo Carvalho de Melo escreveu:
Em Fri, Jun 28, 2019 at 05:23:02PM +0800, Jin Yao escreveu:
The target is to compare the performance difference (cycles
diff) for the same basic blocks in
On 6/5/2019 7:44 PM, Jiri Olsa wrote:
On Mon, Jun 03, 2019 at 10:36:14PM +0800, Jin Yao wrote:
SNIP
diff --git a/tools/perf/util/sort.h b/tools/perf/util/sort.h
index 43623fa..d1641da 100644
--- a/tools/perf/util/sort.h
+++ b/tools/perf/util/sort.h
@@ -79,6 +79,9 @@ struct hist_entry_diff
ymbol hists.
That is to keep current perf-diff format. Will show the
result in next patches.
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 190 +-
tools/perf/util/hist.c| 3 +
tools/perf/util/sort.h| 12 +++
3 files changed, 202 insertions
field to hist_entry. In order not to
impact current interface, we creates a new function hists__add_entry_block.
v6:
---
Remove the 'ops' argument in hists__add_entry_block
Signed-off-by: Jin Yao
---
tools/perf/util/hist.c | 20 ++--
tools/perf/util/hist.h | 5 ++
data files.
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 90 +++
1 file changed, 90 insertions(+)
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index 83b8c0f..823f162 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/buil
check_file_brstack() from __cmd_diff() to cmd_diff().
Because later patch will check flag 'has_br_stack' before
ui_init().
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 29 +
1 file changed, 29 insertions(+)
diff --git a/tools/perf/builtin-diff.c b/tools/pe
-35 native_write_msr (267900:26790b)
26 native_irq_return_iret (c00a27:c00a27)
22 rcu_check_callbacks (2febb6:2febdc)
-21 __hrtimer_run_queues (30b220:30b2a3)
19 pvclock_gtod_notify (14ba0:14c1b)
-18 task_tick_fair (2c5d29:2c5d41)
(function), start/end address of this block, cycles. This patch
creates this structure and with some ops.
Signed-off-by: Jin Yao
---
tools/perf/util/symbol.c | 22 ++
tools/perf/util/symbol.h | 23 +++
2 files changed, 45 insertions(+)
diff --git a/tools/perf
the 'Baseline' and
'Shared Object'.
The output is sorted by "Baseline" and the basic blocks in the same
function are sorted by cycles diff.
Signed-off-by: Jin Yao
---
tools/perf/builtin-diff.c | 80 ---
tools/perf/ui/stdi
Documentation the new computation selection 'cycles'.
v4:
---
Change the column 'Block cycles diff [start:end]' to
'[Program Block Range] Cycles Diff'
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 17 ++---
1 file changed,
x86_pmu_disable_event(event);
oops, I overlooed this, looks good
Acked-by: Jiri Olsa
thanks,
jirka
Hi,
Could this fix be accepted?
Thanks
Jin Yao
/*
* Needs to be called after x86_pmu_disable_event,
--
2.14.5
On 2/12/2019 9:07 PM, Peter Zijlstra wrote:
On Tue, Feb 12, 2019 at 10:42:38AM +0800, Jin, Yao wrote:
diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
index 4a9937076331..309ef5a64af5 100644
--- a/kernel/events/ring_buffer.c
+++ b/kernel/events/ring_buffer.c
@@ -734,6
For better filtering support for perf diff, it would be useful to
add --pid and --tid filter options.
For example,
perf diff --tid 13965
It only diff the samples for thread 13965.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 6 ++
tools/perf/builtin-diff.c
samples for CPU0 and CPU1.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 5 +
tools/perf/builtin-diff.c | 16
2 files changed, 21 insertions(+)
diff --git a/tools/perf/Documentation/perf-diff.txt
b/tools/perf/Documentation/perf-diff.txt
3971.150589 ...
perf diff --time 3946.361400,:3971.150589,
It analyzes the perf.data.old from the timestamp 3946.361400 to
the end of perf.data.old and analyzes the perf.data from the
timestamp 3971.150589 to the end of perf.data.
Signed-off-by: Jin Yao
---
tools/perf/Documentat
s for CPU0 and CPU1.
--pid:
Only diff samples for given process ID (comma separated list).
--tid:
Only diff samples for given thread ID (comma separated list).
For example,
perf diff --tid 13965
It only diff the samples for thread 13965.
Jin Yao (3):
perf diff: Support -
diff: Support --time filter option" according to
Jiri's comments. Others are no logical changes.
Jin Yao (3):
perf diff: Support --time filter option
perf diff: Support --cpu filter option
perf diff: Support --pid/--tid filter options
tools/perf/Documentation/pe
samples for CPU0 and CPU1.
v3:
---
No logical changes
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 5 +
tools/perf/builtin-diff.c | 16
2 files changed, 21 insertions(+)
diff --git a/tools/perf/Documentation/perf-diff.txt
b/tools/perf
For better filtering support for perf diff, it would be useful to
add --pid and --tid filter options.
For example,
perf diff --tid 13965
It only diff the samples for thread 13965.
v3:
---
No logical changes.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 6
is not set.
Refactor the code to make it simpler.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 41 ++
tools/perf/builtin-diff.c | 136 +
2 files changed, 164 insertions(+), 13 deletions(-)
diff --git a/tools/perf/Docu
On 3/4/2019 11:41 PM, Jiri Olsa wrote:
On Mon, Mar 04, 2019 at 08:52:38PM +0800, Jin Yao wrote:
For better support for perf diff, it would be useful to add --time filter
option to diff the samples within given time window.
It supports time percent with multipe time ranges. Time string is
&#
samples for CPU0 and CPU1.
v4:
---
No logical changes
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 5 +
tools/perf/builtin-diff.c | 16
2 files changed, 21 insertions(+)
diff --git a/tools/perf/Documentation/perf-diff.txt
b/tools/perf
it return error if strdup
is failed, and update __cmd_diff() accordingly.
Only the patch "perf diff: Support --time filter option" is impacted
Others are no logical changes.
Jin Yao (3):
perf diff: Support --time filter option
perf diff: Support --cpu filter option
perf di
dup
is failed, and update __cmd_diff() accordingly.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 41 +
tools/perf/builtin-diff.c | 148 +
2 files changed, 175 insertions(+), 14 deletions(-)
diff --git a/tools/perf/
For better filtering support for perf diff, it would be useful to
add --pid and --tid filter options.
For example,
perf diff --tid 13965
It only diff the samples for thread 13965.
v4:
---
No logical changes.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-diff.txt | 6
One week nearly passed, I guess there is no any objections for this
patch from community. :)
Thanks
Jin Yao
On 9/4/2018 3:45 PM, Jin, Yao wrote:
On 9/4/2018 3:13 PM, Peter Zijlstra wrote:
On Tue, Sep 04, 2018 at 06:58:17PM +0800, Jin Yao wrote:
root@skx /sys/devices# ls | grep uncore_iio
or 'system wide':
153.12 MiB uncore_iio_pcie1/bw_in_port0/
8.469790720 seconds time elapsed
v2:
---
Previously, it used braces around initialized string. For example,
{ "iio_cbdma" }.
The braces around scalar initializer would cause the build warning.
In v2, it remo
On 9/4/2018 3:13 PM, Peter Zijlstra wrote:
On Tue, Sep 04, 2018 at 06:58:17PM +0800, Jin Yao wrote:
root@skx /sys/devices# ls | grep uncore_iio
uncore_iio_0
uncore_iio_1
uncore_iio_2
uncore_iio_3
uncore_iio_4
uncore_iio_5
uncore_iio_free_running_0
uncore_iio_free_running_1
or%edi,%edi
:srand(s_randseed);
0.00 : 4004b3: mov$0x77359400,%ebx
:
:return i;
:}
Signed-off-by: Jin Yao
---
tools/perf/builtin-annotate.c | 7 +++
1 file changed, 7 insertions(+)
di
27;perf annotate: Support multiple events without group'
Jin Yao (3):
perf evlist: Create new function perf_evlist__force_leader
perf report: Use perf_evlist__force_leader to support '--group'
perf annotate: Support '--group' option
tools/perf/builtin-annotate.c
Since we have created a new function perf_evlist__force_leader,
so now remove the old code and use perf_evlist__force_leader
instead.
Signed-off-by: Jin Yao
---
tools/perf/builtin-report.c | 13 ++---
1 file changed, 2 insertions(+), 11 deletions(-)
diff --git a/tools/perf/builtin
For non-explicit group, perf report supports a option '--group'
which can enable group output. We also need to support perf annotate
with the same '--group'.
Create a new function perf_evlist__force_leader which contains
common code to force setting the group leader.
Si
On 5/22/2018 1:43 AM, Arnaldo Carvalho de Melo wrote:
Em Mon, May 21, 2018 at 10:57:46PM +0800, Jin Yao escreveu:
With the '--group' option, even for non-explicit group, perf annotate
will enable the group output.
For example,
perf record -e cycles,branches ./div
perf annotate ma
4407 samples)
..
The first line doesn't include the event 'branches'.
With this patch, it will show the correct group even string.
perf annotate --group --stdio
Percent | Source code & Disassembly of div for cycles, branches (44407
samples)
......
thers not changed
Jin Yao (5):
perf config: Add some new -DHAVE_XXX to CFLAGS
perf config: Rename to HAVE_DWARF_GETLOCATIONS_SUPPORT
perf version: Print the compiled-in status of libraries
perf: Support perf -vv
perf Documentation: Create perf-version.txt
Jiri Olsa (1):
tools include
Since a new option '--build-options' is created for 'perf version',
so we need to document it.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-version.txt | 24
1 file changed, 24 insertions(+)
create mode 100644 tools/perf/Documentation/pe
ld environment
2. Create '--build-options' option.
3. Use standard option parsing API 'parse_options'.
v2:
---
1. Use IS_BUILTIN macro to replace #ifdef/#endif block.
2. Print color for on/OFF.
Signed-off-by: Jin Yao
---
tools/perf/builtin-version.c | 82 +++
From: Jiri Olsa
Adding IS_BUILTIN macro and its dependencies into
tools world.
It's taken from kernel's include/linux/kconfig.h,
which can't be taken completely due to its kconfig
dependencies.
Signed-off-by: Jiri Olsa
---
tools/include/tools/config.h | 34 ++
In Makefile.config, to make all libraries flags have _SUPPORT suffix,
rename HAVE_DWARF_GETLOCATIONS to HAVE_DWARF_GETLOCATIONS_SUPPORT
Signed-off-by: Jin Yao
---
tools/perf/Makefile.config | 2 +-
tools/perf/util/dwarf-aux.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git
bpf: [ on ] # HAVE_LIBBPF_SUPPORT
v3:
---
One bug is found in v2. It didn't process the option like '-vabc'
correctly. Fix this bug.
v2:
---
Use a global variable version_verbose to record the number of 'v'.
Signed-off-by: Jin Yao
---
tools/perf/perf.c | 6 ++
too
On 4/3/2018 12:47 AM, Arnaldo Carvalho de Melo wrote:
Em Fri, Mar 30, 2018 at 05:27:14PM +0800, Jin Yao escreveu:
This patch checks the values passed by CFLAGS (-DHAVE_XXX) and then
print the status of libraries.
For example, if HAVE_DWARF_SUPPORT is defined, that means the
library "
04b1: xor%edi,%edi
:srand(s_randseed);
0.000.00 : 4004b3: mov$0x77359400,%ebx
:
:return i;
: }
Signed-off-by: Jin Yao
---
tools/perf/util/annotate.c | 3 +++
On 5/10/2018 12:42 AM, Arnaldo Carvalho de Melo wrote:
Em Wed, May 09, 2018 at 11:57:15PM +0800, Jin Yao escreveu:
When we perform following command lines:
perf record -e "{cycles,branches}" ./div
perf annotate main --stdio
Only shows one event "cycles" and the disp
On 5/10/2018 1:24 AM, Arnaldo Carvalho de Melo wrote:
Em Wed, May 09, 2018 at 01:42:51PM -0300, Arnaldo Carvalho de Melo escreveu:
Em Wed, May 09, 2018 at 11:57:15PM +0800, Jin Yao escreveu:
When we perform following command lines:
perf record -e "{cycles,branches}" ./div
perf ann
for (i = 0; i < 20; i++) {
flag = compute_flag();
4.85 5.83 38: xor%eax,%eax
0.01 0.01 → callq compute_flag
count++;
4.44 2.27movcount,%edx
0.00 0.00add
On 3/14/2018 9:54 PM, Arnaldo Carvalho de Melo wrote:
Em Wed, Mar 14, 2018 at 10:04:49AM +0800, Jin, Yao escreveu:
On 3/13/2018 11:20 PM, Arnaldo Carvalho de Melo wrote:
Em Tue, Mar 13, 2018 at 10:16:50PM +0800, Jin Yao escreveu:
There is a requirement to let perf annotate support
501 - 600 of 1316 matches
Mail list logo
