Hi,
https://01.org/numatop/ is slightly faster to access.
Or please access https://github.com/01org/numatop to get the source directly.
Thanks
Jin Yao
-Original Message-
From: Jin, Yao
Sent: Wednesday, April 17, 2013 10:30 AM
To: 'l...@lwn.net'; 'linux-kernel@
changed.
Please let me know what do you think for this patch and if I should go ahead.
Thanks
Jin Yao
-Original Message-
From: Arnaldo Carvalho de Melo [mailto:a...@kernel.org]
Sent: Friday, September 9, 2016 1:11 AM
To: Andi Kleen
Cc: Stephane Eranian ; Peter Zijlstra
; Ingo Moln
set is not correct, so finally objdump uses wrong
start address to disassemble /lib/x86_64-linux-gnu/libc-2.23.so.
This patch lets dso__load not try ubuntu debuginfo
(/usr/lib/debug/lib/x86_64-linux-gnu/libc-2.23.so in this example), because
1: the image doesn't have valid assembly code (check
nts on the
left side of annotate view for stdio, tui and gtk modes.
Jin Yao (4):
perf annotate: create a new hists to manage multiple events samples
perf annotate: Display multiple events for stdio mode
perf annotate: Display multiple events for tui mode
perf annotate: Display multiple even
ed-off-by: Jin Yao
---
tools/perf/ui/browsers/annotate.c | 49 ---
tools/perf/ui/browsers/hists.c| 2 +-
tools/perf/util/annotate.h| 6 +++--
tools/perf/util/hist.h| 8 ---
4 files changed, 45 insertions(+), 20 deletions(-)
diff -
e hists is per event. So we need a new
hists to manage the samples for multiple events and use a new
hist_event data structure to save the map/symbol information
for per event.
Signed-off-by: Jin Yao
---
tools/perf/builtin-annotate.c | 60 +++
tools
For example:
perf record -e cycles,branches ./div
perf annotate main --gtk
Both the cycles and branches are displayed at the left column in
gtk window.
Signed-off-by: Jin Yao
---
tools/perf/ui/gtk/annotate.c | 35 ++-
1 file changed, 22 insertions(+), 13
00b4c(%rip)
:
: if (flag)
3.994.32 :400500: je 400532
Signed-off-by: Jin Yao
---
tools/perf/builtin-annotate.c | 2 +-
tools/perf/builtin-top.c | 3 +-
tools/perf/util/annotate.c| 179 ---
stead to check fused instruction
pair.
v1: Inital post
Jin Yao (2):
perf util: Check for fused instruction
perf report: Implement visual marker for macro fusion in annotate
tools/perf/arch/x86/annotate/instructions.c | 37 +
tools/perf/ui/browser.c
Arnaldo's fix to let the display be better.
To get the evsel->evlist->env->cpuid, save the evsel in
annotate_browser.
v2: No more changes, just uses a new function "ins__is_fused"
to check if the instructions are fused.
v1: Initial post
Signed-off-by: Jin
t as a weak function that will be overridden
by the host arch doesn't work. So now it's implemented
as an arch-specific function.
v1: Initial post
Signed-off-by: Jin Yao
---
tools/perf/arch/x86/annotate/instructions.c | 37 +
tools/perf/util/an
On 10/24/2017 3:16 PM, Jiri Olsa wrote:
On Tue, Oct 24, 2017 at 10:03:05AM +0800, Jin, Yao wrote:
SNIP
hum, could you still unset the sample if there's no time given?
and keep the speed in this case..
jirka
Hi Jiri,
I check this question again. The '--time' option is
On 11/4/2017 12:29 AM, Arnaldo Carvalho de Melo wrote:
Em Tue, Oct 24, 2017 at 09:16:59AM +0200, Jiri Olsa escreveu:
On Tue, Oct 24, 2017 at 10:03:05AM +0800, Jin, Yao wrote:
SNIP
hum, could you still unset the sample if there's no time given?
and keep the speed in this case..
jirka
On 11/4/2017 6:24 PM, Jiri Olsa wrote:
On Fri, Nov 03, 2017 at 01:29:42PM -0300, Arnaldo Carvalho de Melo wrote:
Em Tue, Oct 24, 2017 at 09:16:59AM +0200, Jiri Olsa escreveu:
On Tue, Oct 24, 2017 at 10:03:05AM +0800, Jin, Yao wrote:
SNIP
hum, could you still unset the sample if there'
. Following passes.
perf script --time 10%/10x12321xsdfdasfdsafdsafdsa
Now it uses strtol to replace atoi.
Signed-off-by: Jin Yao
---
tools/perf/util/time-utils.c | 205 ---
tools/perf/util/time-utils.h | 3 +
2 files changed, 196 insertions(+), 12
urns error message to user.
v4: Remove perf_time__skip_sample, only uses perf_time__ranges_skip_sample
v3: Since the definitions of first_sample_time/last_sample_time
are moved from perf_session to perf_evlist so change the
related code.
Signed-off-by: Jin Yao
---
tools/perf/Document
ssion. Just define them in perf_evlist
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf.data-file-format.txt | 4 ++
tools/perf/util/evlist.h | 2 +
tools/perf/util/header.c | 60 ++
tools/perf/util/hea
d, returns error message to user.
v4: Remove perf_time__skip_sample, only uses perf_time__ranges_skip_sample
v3: Since the definitions of first_sample_time/last_sample_time
are moved from perf_session to perf_evlist so change the
related code.
Signed-off-by: Jin Yao
---
tools/perf/Doc
ltiple time ranges.
For example:
1. Select the second 10% time slice
perf report --time 10%/2
2. Select from 0% to 10% time slice
perf report --time 0%-10%
It also support the multiple time ranges.
3. Select the first and second 10% time slices
perf report --time 10%/1,10%/2
4. Select
oundaries".
While after consideration, I think a new option is not very
necessary.
v3: Remove the definitions of first_sample_time and last_sample_time
from struct record and directly save them in perf_evlist.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-record.txt | 3 +++
t
ned-off-by: Jin Yao
---
tools/perf/util/time-utils.c | 28
tools/perf/util/time-utils.h | 3 +++
2 files changed, 31 insertions(+)
diff --git a/tools/perf/util/time-utils.c b/tools/perf/util/time-utils.c
index 79e4281..b380356 100644
--- a/tools/perf/util/time-uti
FD(evsel, cpu, thread) = -1;
}
It's not correct!
This change is introduced by "475fb533fb7d"
("perf evsel: Fix buffer overflow while freeing events")
This fix is to let xyarray__max_x() return max_x (ncpus) and
let xyarry__max_y() return max_y (nthreads)
S
On 10/5/2017 4:50 PM, Jiri Olsa wrote:
On Tue, Oct 03, 2017 at 10:22:32PM +0800, Jin Yao wrote:
v4:
---
1. Use perf script time style for timestamp printing. Also add with
the printing of sample duration. For example:
perf report --header
time of first sample : 5276531.323099
On 10/20/2017 4:25 AM, Arnaldo Carvalho de Melo wrote:
Em Thu, Oct 19, 2017 at 05:21:27PM -0300, Arnaldo Carvalho de Melo escreveu:
Em Tue, Oct 03, 2017 at 10:22:34PM +0800, Jin Yao escreveu:
In perf record, it's walked on all samples yet. So it's very easy to get
You're s
or 'system wide':
153.12 MiB uncore_iio_pcie1/bw_in_port0/
8.469790720 seconds time elapsed
Signed-off-by: Jin Yao
---
arch/x86/events/intel/uncore.c | 9 +++--
arch/x86/events/intel/uncore.h | 1 +
arch/x86/events/intel/uncore_snbep.c | 10 ++
3 fil
On 1/23/2018 10:40 PM, Jiri Olsa wrote:
On Tue, Jan 23, 2018 at 07:02:44AM +0800, Jin, Yao wrote:
SNIP
threads = thread_map__new_str(target->pid, target->tid, target->uid,
- target->per_thread);
+
es 'target->per_thread &&
target->system_wide' as a condition to check for all threads case.
v2:
---
According to Jiri's comment, add description to explain why the patch
uses 'target->per_thread && target->system_wide'.
v2 doesn't hav
7;t work.
This patch displays warnings that these libraries are missing in build
when perf report / perf annotate are used.
Signed-off-by: Jin Yao
---
tools/perf/builtin-annotate.c | 2 ++
tools/perf/builtin-report.c | 2 ++
tools/perf/util/symbol.c | 21 +
tools/per
On 1/11/2018 11:30 PM, Jiri Olsa wrote:
On Thu, Jan 11, 2018 at 07:03:06PM +0800, Jin Yao wrote:
We keep having bug reports that when users build perf on their own,
we already have same warnings during the build
Yes, there will be warnings displayed during the build if some libraries
On 1/16/2018 9:17 PM, Jiri Olsa wrote:
On Tue, Jan 16, 2018 at 09:06:09PM +0800, Jin, Yao wrote:
Just tested. But looks it's not OK for '--per-thread' case.
yea, I haven't tested much.. might need soem tweaking,
but my point was that it could be doable on one place
in
e
classes.
- Arnaldo
Hi,
Very sorry, I just see this mail today.
Please let me look at this issue first.
Thanks
Jin Yao
threads in a system, resulting in new kernel events being created for all
threads rather than just the thread of interest.
This patch keeps the newly introduced func
se.
I just think for your perf record case, the target->system_wide will not
be set. Instead, if target->per-thread and target->system_wide are both
set, that means we needs to trace on all threads, right?
Thanks
Jin Yao
On 1/13/2018 7:12 AM, Mathieu Poirier wrote:
Commit ("73c
es 'target->per_thread &&
target->system_wide' as a condition to check for all threads case.
Signed-off-by: Jin Yao
---
tools/perf/util/evlist.c | 2 +-
tools/perf/util/thread_map.c | 4 ++--
tools/perf/util/thread_map.h | 2 +-
3 files changed, 4 insertions(+), 4 dele
On 1/23/2018 5:10 AM, Mathieu Poirier wrote:
On 22 January 2018 at 15:15, Jin Yao wrote:
Mathieu Poirier reports issue in commit ("73c0ca1eee3d perf thread_map:
Enumerate all threads from /proc") that it has negative impact on
'perf record --per-thread'. It has the
On 1/23/2018 7:56 AM, Mathieu Poirier wrote:
On 22 January 2018 at 15:15, Jin Yao wrote:
Mathieu Poirier reports issue in commit ("73c0ca1eee3d perf thread_map:
Enumerate all threads from /proc") that it has negative impact on
'perf record --per-thread'. It has the
On 1/5/2018 3:09 AM, Arnaldo Carvalho de Melo wrote:
Em Fri, Dec 08, 2017 at 09:13:42PM +0800, Jin Yao escreveu:
In the default 'perf record' configuration, all samples are processed,
to create the HEADER_BUILD_ID table. So it's very easy to get the
first/last samples and s
On 1/5/2018 8:53 PM, Arnaldo Carvalho de Melo wrote:
Em Fri, Jan 05, 2018 at 09:15:03AM +0800, Jin, Yao escreveu:
On 1/5/2018 3:09 AM, Arnaldo Carvalho de Melo wrote:
Em Fri, Dec 08, 2017 at 09:13:42PM +0800, Jin Yao escreveu:
In the default 'perf record' configuration, all s
2, y = 121212;
│ │
│ │s_randseed = time(0);
│ │srand(s_randseed);
│ │
│ │for (i = 0; i < 20; i++) {
1.37 │0.4882:└─→sub$0x1,%ebx
28.21 │0.4817 ↑ jne38
Signed-off-by: Jin Yao
---
tools/perf/ui/bro
On 1/16/2018 10:48 PM, Arnaldo Carvalho de Melo wrote:
Em Tue, Jan 16, 2018 at 12:55:19PM +0100, Jiri Olsa escreveu:
On Wed, Jan 10, 2018 at 11:00:25PM +0800, Jin Yao wrote:
It's follow-up patches to improve the perf time slice feature
(perf report/script --time xxx)
1. Improve the
On 1/8/2018 10:38 PM, Arnaldo Carvalho de Melo wrote:
Em Mon, Jan 08, 2018 at 11:31:49AM -0300, Arnaldo Carvalho de Melo escreveu:
Em Fri, Dec 08, 2017 at 09:13:43PM +0800, Jin Yao escreveu:
Current perf report/script/... have a --time option to limit the time
range of output. But right now
cify which one is wanted" when
you notice a % in the --time string, etc.
4. Add an indication of what percentage ranges are being used. For example,
Samples: 128 of event 'cycles:ppp', Event count (approx.): 21386169 (10%/1)
I will develop a follow-up patchset to improve this feature.
Thanks
Jin Yao
umber limitation
perf script: Remove the time slices number limitation
Jin Yao (8):
perf report: Improve error msg when no first/last sample time found
perf script: Improve error msg when no first/last sample time found
perf util: Improve error checking for time percent input
perf util: Su
--buildid-all' is enabled, needs to set '--timestamp-boundary')."
Signed-off-by: Jin Yao
---
tools/perf/builtin-report.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index dd4df9a..a6c5cf2 10
Add a time slices indication to the perf report header.
For example,
# perf report --stdio --time 10%
# Total Lost Samples: 0
#
# Samples: 9K of event 'cycles:ppp' (time slices: 10%)
# Event count (approx.): 8951288803
Signed-off-by: Jin Yao
---
tools/perf/builtin-re
Previously it was only allowed to use at most 10 time slices
in 'perf report --time'.
This patch removes this limitation.
For example, following command line is OK (12 time slices)
perf report --stdio --time
1%/1,1%/2,1%/3,1%/4,1%/5,1%/6,1%/7,1%/8,1%/9,1%/10,1%/11,1%/12
Signed-off-b
root@skl:/tmp# perf report --stdio --time 1abc%/1
Invalid time string
Signed-off-by: Jin Yao
---
tools/perf/util/time-utils.c | 8 ++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/tools/perf/util/time-utils.c b/tools/perf/util/time-utils.c
index 3f7f18f..88510ab 100644
--- a/
Previously it was only allowed to use at most 10 time slices
in 'perf script --time'.
This patch removes this limitation.
For example, following command line is OK (12 time slices)
perf script --time
1%/1,1%/2,1%/3,1%/4,1%/5,1%/6,1%/7,1%/8,1%/9,1%/10,1%/11,1%/12
Signed-off-b
Previously, the time percent slice needs an index to specify
which one the user wants.
While it may be easy for using if the index can be omitted.
So with this patch, for example,
perf report --stdio --time 10%/1 should be equivalent to
perf report --stdio --time 10%
Signed-off-by: Jin Yao
--buildid-all' is enabled, needs to set '--timestamp-boundary')."
Signed-off-by: Jin Yao
---
tools/perf/builtin-script.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index c1cce47..4f691af 10
e one entry even if no comma is found.
Signed-off-by: Jin Yao
---
tools/perf/util/time-utils.c | 28
tools/perf/util/time-utils.h | 2 ++
2 files changed, 30 insertions(+)
diff --git a/tools/perf/util/time-utils.c b/tools/perf/util/time-utils.c
index 5769f97..6193b46 1
.651675247 seconds time elapsed
Signed-off-by: Jin Yao
---
tools/perf/builtin-stat.c| 14 +-
tools/perf/util/evsel.c | 3 +++
tools/perf/util/thread_map.c | 1 +
tools/perf/util/thread_map.h | 1 +
4 files changed, 18 insertions(+), 1 deletion(-)
diff --git a/tools/p
On 1/16/2018 7:55 PM, Jiri Olsa wrote:
On Wed, Jan 10, 2018 at 11:00:25PM +0800, Jin Yao wrote:
It's follow-up patches to improve the perf time slice feature
(perf report/script --time xxx)
1. Improve the error message
perf report: Improve error msg when no first/last sample time
_ADMIN
>= 2: Disallow kernel profiling by users without CAP_SYS_ADMIN
To make this setting permanent, edit /etc/sysctl.conf too, e.g.:
kernel.perf_event_paranoid = -1
Thanks
Jin Yao
On 1/16/2018 8:51 PM, Jiri Olsa wrote:
On Tue, Jan 16, 2018 at 11:43:08PM +0800, Jin Yao wrote:
If we
On 1/16/2018 9:17 PM, Jiri Olsa wrote:
On Tue, Jan 16, 2018 at 09:06:09PM +0800, Jin, Yao wrote:
Just tested. But looks it's not OK for '--per-thread' case.
yea, I haven't tested much.. might need soem tweaking,
but my point was that it could be doable on one place
in
The tui_dump is created as a parameter and it will be finally passed to
symbol__tui_annotate() and be saved to browser.dump.
It's a switch for TUI routines to indicate if TUI output needs to be
dumped to stdio.
Signed-off-by: Jin Yao
---
tools/perf/builtin-annotate.c | 2 +-
tools/pe
%edx,%eax
│}
26.93 │3.38 2 ← retq
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-annotate.txt | 3 +++
tools/perf/builtin-annotate.c | 12 ++--
2 files changed, 13 insertions(+), 2 deletions(-)
diff --git a/tools/perf/Documentation/perf-annotate.tx
TUI routines and drop the lagacy stdio code. But right
now we'd better keep it until the '--tui-dump' option is good enough.
Jin Yao (4):
perf browser: Add a new 'dump' flag
perf browser: bypass ui_init if in tui dump mode
perf annotate: Process the new switch flag
We have a new requirement to provide the TUI output option for
non-interactive mode. For example, write the TUI output to stdio directly.
This patch creates a new flag 'dump' in struct ui_browser. Once it's on,
for the formatted buffer, we just print it on stdio.
Signed-
We create a tui dump mode in previous patch. In tui dump mode, the output
is written to stdio. We need to bypass ui_init() in setup_browser().
Signed-off-by: Jin Yao
---
tools/perf/builtin-annotate.c | 2 +-
tools/perf/builtin-c2c.c | 2 +-
tools/perf/builtin-report.c | 2 +-
tools/perf
libslang: [ on ]
libcrypto: [ on ]
libunwind: [ on ]
libdw-dwarf-unwind: [ on ]
zlib: [ on ]
lzma: [ on ]
get_cpuid: [ on ]
bpf: [ on ]
Jin Yao (3):
perf config: Add -DNO_GLIBC t
-off-by: Jin Yao
---
tools/perf/Makefile.config | 2 ++
1 file changed, 2 insertions(+)
diff --git a/tools/perf/Makefile.config b/tools/perf/Makefile.config
index 98ff736..5883dd6 100644
--- a/tools/perf/Makefile.config
+++ b/tools/perf/Makefile.config
@@ -324,6 +324,8 @@ else
NO_LIBBPF
lzma: [ on ]
get_cpuid: [ on ]
bpf: [ on ]
Signed-off-by: Jin Yao
---
tools/perf/perf.c | 6 ++
1 file changed, 6 insertions(+)
diff --git a/tools/perf/perf.c b/tools/perf/perf.c
index 1b3fc8e..300c83d 100644
--- a/tools/perf/perf.c
+++ b/to
This patch checks the values passed by CFLAGS (-DXXX) and then
print the status of libraries.
For example, if HAVE_DWARF_SUPPORT is defined, that means the
library "dwarf" is compiled-in. The patch will print the status
"on" for this library.
Signed-off-by: Jin Yao
---
On 3/26/2018 5:07 PM, Jiri Olsa wrote:
On Mon, Mar 26, 2018 at 02:00:31AM -0700, Andi Kleen wrote:
On Tue, Mar 27, 2018 at 12:07:01AM +0800, Jin Yao wrote:
We keep having bug reports that when users build perf on their own,
but they don't install some needed libraries such as libelf,
l
On 3/26/2018 5:39 PM, Jiri Olsa wrote:
On Tue, Mar 27, 2018 at 12:07:03AM +0800, Jin Yao wrote:
This patch checks the values passed by CFLAGS (-DXXX) and then
print the status of libraries.
For example, if HAVE_DWARF_SUPPORT is defined, that means the
library "dwarf" is compil
On 3/26/2018 5:39 PM, Jiri Olsa wrote:
On Tue, Mar 27, 2018 at 12:07:03AM +0800, Jin Yao wrote:
This patch checks the values passed by CFLAGS (-DXXX) and then
print the status of libraries.
For example, if HAVE_DWARF_SUPPORT is defined, that means the
library "dwarf" is compil
On 3/26/2018 9:51 PM, Jin, Yao wrote:
On 3/26/2018 5:39 PM, Jiri Olsa wrote:
On Tue, Mar 27, 2018 at 12:07:03AM +0800, Jin Yao wrote:
This patch checks the values passed by CFLAGS (-DXXX) and then
print the status of libraries.
For example, if HAVE_DWARF_SUPPORT is defined, that means the
On 3/27/2018 1:58 PM, Ingo Molnar wrote:
* Jin Yao wrote:
+#ifdef HAVE_DWARF_SUPPORT
+#ifdef HAVE_DWARF_GETLOCATIONS
+#ifdef NO_GLIBC
+#ifdef HAVE_GTK2_SUPPORT
+#ifdef HAVE_LIBAUDIT_SUPPORT
+#ifdef HAVE_LIBBFD_SUPPORT
+#ifdef HAVE_LIBELF_SUPPORT
+#ifdef HAVE_LIBNUMA_SUPPORT
+#ifdef
On 3/27/2018 2:03 PM, Ingo Molnar wrote:
* Jin Yao wrote:
+++ b/tools/perf/perf.c
@@ -64,6 +64,7 @@ static struct cmd_struct commands[] = {
{ "top", cmd_top,0 },
{ "annotate", cmd_annotate, 0 },
{ "version", cmd_versio
On 3/27/2018 8:56 PM, Jiri Olsa wrote:
On Tue, Mar 27, 2018 at 09:44:23AM +0800, Jin, Yao wrote:
On 3/26/2018 5:39 PM, Jiri Olsa wrote:
On Tue, Mar 27, 2018 at 12:07:03AM +0800, Jin Yao wrote:
This patch checks the values passed by CFLAGS (-DXXX) and then
print the status of libraries
On 3/27/2018 8:38 PM, Jiri Olsa wrote:
On Mon, Mar 26, 2018 at 09:51:03PM +0800, Jin, Yao wrote:
On 3/26/2018 5:39 PM, Jiri Olsa wrote:
On Tue, Mar 27, 2018 at 12:07:03AM +0800, Jin Yao wrote:
This patch checks the values passed by CFLAGS (-DXXX) and then
print the status of libraries
get_cpuid: [ on ] # HAVE_AUXTRACE_SUPPORT
bpf: [ on ] # HAVE_LIBBPF_SUPPORT
The line "syscall_table: [ on ] # HAVE_SYSCALL_TABLE_SUPPORT" is
new created.
Signed-off-by: Jin Yao
---
tools/perf/builtin-version.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/tools/perf/built
add a line for HAVE_SYSCALL_TABLE_SUPPORT.
Jin Yao (3):
perf script: Use HAVE_LIBXXX_SUPPORT to replace NO_LIBXXX
perf: Rename HAVE_SYSCALL_TABLE to HAVE_SYSCALL_TABLE_SUPPORT
perf version: Print status for syscall_table
tools/perf/Makefile.config | 2 +-
tools/perf
To make consistent with other HAVE_XXX_SUPPORT in Makefile.config,
this patch renames HAVE_SYSCALL_TABLE to HAVE_SYSCALL_TABLE_SUPPORT
and update the C code accordingly.
Signed-off-by: Jin Yao
---
tools/perf/Makefile.config | 2 +-
tools/perf/builtin-help.c | 2 +-
tools/perf
time on getting the root cause. If with this
patch, it should be very easily to know that.
Thanks
Jin Yao
On 1/12/2018 10:22 AM, Jin, Yao wrote:
On 1/11/2018 11:30 PM, Jiri Olsa wrote:
On Thu, Jan 11, 2018 at 07:03:06PM +0800, Jin Yao wrote:
We keep having bug reports that when users build per
On 3/21/2018 11:38 PM, Jiri Olsa wrote:
On Wed, Mar 21, 2018 at 10:11:10AM +0800, Jin, Yao wrote:
Hi Jiri,
I'm still thinking it's worth displaying the warning when perf missing some
libraries.
Somebody just told me that perf didn't work well. While after some
investigation
04:38:07PM +0100, Jiri Olsa escreveu:
On Wed, Mar 21, 2018 at 10:11:10AM +0800, Jin, Yao wrote:
Hi Jiri,
I'm still thinking it's worth displaying the warning when perf missing some
libraries.
Somebody just told me that perf didn't work well. While after some
investigations, I
On 3/22/2018 4:51 PM, Jiri Olsa wrote:
On Thu, Mar 22, 2018 at 09:04:10AM +0800, Jin, Yao wrote:
On 3/21/2018 11:38 PM, Jiri Olsa wrote:
On Wed, Mar 21, 2018 at 10:11:10AM +0800, Jin, Yao wrote:
Hi Jiri,
I'm still thinking it's worth displaying the warning when perf mi
Hi,
Could this patch be accepted?
I'm working on the supporting for stdio mode. I just wish to post the
patch series after this patch being merged.
Thanks
Jin Yao
On 2/27/2018 5:38 PM, Jin Yao wrote:
Unlike the perf report interactive annotate mode, the perf annotate
doesn't d
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 displaying the IPC/Cycle.
In previous patch, this is supported in TUI mode. While it's not convenient
for users since
[.] rand@plt 0.00 [ 0.0%]
Note that, stdio mode doesn't support this feature.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-report.txt | 1 +
tools/perf/builtin-report.c | 11
tools/perf/util/hist.h | 1 +
tools/p
,__abort_msg@@GLIBC_PRIVATE+0x8a0
...
Jin Yao (2):
perf annotate: Compute average IPC and IPC coverage per symbol
perf report: Display average IPC and IPC coverage per symbol
tools/perf/Documentation/perf-report.txt | 1 +
tools/perf/builtin-report.c | 11
tools/perf
__lll_unlock_wake_private
add$0x80,%rsp
21.56 2.908a: movslq 0xc(%rsp),%rax
2.90 add$0x18,%rsp
9.03 2.90 1 ← retq
It shows for this symbol the average IPC is 2.30 and the IPC coverage
is 54.8%.
Signed-off-by: Jin Yao
---
tools/perf
On 11/26/2018 5:55 PM, Jiri Olsa wrote:
On Mon, Nov 26, 2018 at 05:40:54PM +0800, Jin Yao wrote:
SNIP
diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c
index f96c005..94f62c8 100644
--- a/tools/perf/util/sort.c
+++ b/tools/perf/util/sort.c
@@ -13,6 +13,7 @@
#include "strl
On 11/26/2018 5:52 PM, Jiri Olsa wrote:
On Mon, Nov 26, 2018 at 05:40:54PM +0800, Jin Yao wrote:
Support displaying the average IPC and IPC coverage for symbol
in perf report TUI browser. We create a new sort-key 'ipc' for
that.
For example,
$ perf record -g -b ...
$ perf report
On 11/26/2018 5:53 PM, Jiri Olsa wrote:
On Mon, Nov 26, 2018 at 05:40:54PM +0800, Jin Yao wrote:
SNIP
diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c
index f96c005..94f62c8 100644
--- a/tools/perf/util/sort.c
+++ b/tools/perf/util/sort.c
@@ -13,6 +13,7 @@
#include "strl
__lll_unlock_wake_private
add$0x80,%rsp
21.56 2.908a: movslq 0xc(%rsp),%rax
2.90 add$0x18,%rsp
9.03 2.90 1 ← retq
It shows for this symbol the average IPC is 2.30 and the IPC coverage
is 54.8%.
Signed-off-by: Jin Yao
---
tools/perf
0.5%]
6.85% [.] rand@plt 0.00 [ 0.0%]
0.02% [k] run_timer_softirq 1.60 [ 57.2%]
v2:
---
Merge in Jiri's patch to support stdio mode
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-report.txt | 1 +
tools/perf/builtin-report.c | 1
We often use the symbol__annotate2() to annotate a specified symbol.
While annotating may take some time, so in order to avoid annotating
the same symbol repeatedly, the patch creates a new flag to indicate
the symbol has been annotated.
Signed-off-by: Jin Yao
---
tools/perf/util/annotate.c | 1
3. Minor update such as adding { } for multiline code in 'if'
condition.
Jin Yao (3):
perf annotate: Compute average IPC and IPC coverage per symbol
perf annotate: Create a annotate2 flag in struct symbol
perf report: Display average IPC and IPC coverage per symbol
tool
On 11/28/2018 2:33 AM, Andi Kleen wrote:
On Tue, Nov 27, 2018 at 08:09:48PM +0800, Jin Yao wrote:
Support displaying the average IPC and IPC coverage for symbol
in perf report TUI browser. We create a new sort-key 'ipc' for
that.
Another thing that would be nice would be to aut
We often use the symbol__annotate2() to annotate a specified symbol.
While annotating may take some time, so in order to avoid annotating
the same symbol repeatedly, the patch creates a new flag to indicate
the symbol has been annotated.
Signed-off-by: Jin Yao
---
tools/perf/util/annotate.c | 1
date such as adding { } for multiline code in 'if'
condition.
Jin Yao (3):
perf annotate: Compute average IPC and IPC coverage per symbol
perf annotate: Create a annotate2 flag in struct symbol
perf report: Display average IPC and IPC coverage per symbol
tools/perf/
bled when "symbol"
is specified.
v2:
---
Merge in Jiri's patch to support stdio mode
Signed-off-by: Jin Yao
---
tools/perf/builtin-report.c | 26 ---
tools/perf/util/hist.h | 1 +
tools/perf/util/sort.c | 61 +
too
__lll_unlock_wake_private
add$0x80,%rsp
21.56 2.908a: movslq 0xc(%rsp),%rax
2.90 add$0x18,%rsp
9.03 2.90 1 ← retq
It shows for this symbol the average IPC is 2.30 and the IPC coverage
is 54.8%.
Signed-off-by: Jin Yao
---
tools/perf
On 11/28/2018 5:10 PM, Ingo Molnar wrote:
* Jin Yao wrote:
Add supporting of displaying the average IPC and IPC coverage
percentage per function.
For example,
$ perf record -b ...
$ perf report -s symbol or
perf report -s symbol --stdio
Overhead Symbol IPC
On 11/28/2018 6:18 PM, Jiri Olsa wrote:
On Wed, Nov 28, 2018 at 11:17:57AM +0100, Jiri Olsa wrote:
On Wed, Nov 28, 2018 at 11:14:55PM +0800, Jin Yao wrote:
Add supporting of displaying the average IPC and IPC coverage
percentage per function.
For example,
$ perf record -b ...
$ perf
__lll_unlock_wake_private
add$0x80,%rsp
21.56 2.908a: movslq 0xc(%rsp),%rax
2.90 add$0x18,%rsp
9.03 2.90 1 ← retq
It shows for this symbol the average IPC is 2.30 and the IPC coverage
is 54.8%.
Signed-off-by: Jin Yao
---
tools/perf
Add explanations for new columns "IPC" and "IPC coverage" in perf
documentation.
Signed-off-by: Jin Yao
---
tools/perf/Documentation/perf-report.txt | 8
1 file changed, 8 insertions(+)
diff --git a/tools/perf/Documentation/perf-report.txt
b/tools/perf/Documenta
bled when "symbol"
is specified.
v2:
---
Merge in Jiri's patch to support stdio mode
Signed-off-by: Jin Yao
---
tools/perf/builtin-report.c | 26 ---
tools/perf/util/hist.h | 1 +
tools/perf/util/sort.c | 61 +
too
1 - 100 of 1316 matches
Mail list logo
