From: Yang Jihong <yangjihong1@huawei.com>
To: Arnaldo Carvalho de Melo <acme@kernel.org>,
Andrii Nakryiko <andrii.nakryiko@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
Song Liu <song@kernel.org>, Andrii Nakryiko <andrii@kernel.org>,
Ingo Molnar <mingo@kernel.org>,
Thomas Gleixner <tglx@linutronix.de>,
Jiri Olsa <jolsa@kernel.org>,
Clark Williams <williams@redhat.com>,
Kate Carcia <kcarcia@redhat.com>, <linux-kernel@vger.kernel.org>,
<linux-perf-users@vger.kernel.org>,
Adrian Hunter <adrian.hunter@intel.com>,
Changbin Du <changbin.du@huawei.com>, Hao Luo <haoluo@google.com>,
Ian Rogers <irogers@google.com>,
James Clark <james.clark@arm.com>,
Kan Liang <kan.liang@linux.intel.com>,
Roman Lozko <lozko.roma@gmail.com>,
Stephane Eranian <eranian@google.com>,
Thomas Richter <tmricht@linux.ibm.com>,
Arnaldo Carvalho de Melo <acme@redhat.com>,
bpf <bpf@vger.kernel.org>
Subject: Re: BPF skels in perf .Re: [GIT PULL] perf tools changes for v6.4
Date: Sat, 6 May 2023 09:13:39 +0800 [thread overview]
Message-ID: <bf082a2a-f360-95fd-9d18-0702c421b296@huawei.com> (raw)
In-Reply-To: <ZFUCDVpqXkpGjXBK@kernel.org>
Hello,
On 2023/5/5 21:18, Arnaldo Carvalho de Melo wrote:
> Em Thu, May 04, 2023 at 07:01:51PM -0300, Arnaldo Carvalho de Melo escreveu:
>> Em Thu, May 04, 2023 at 06:48:50PM -0300, Arnaldo Carvalho de Melo escreveu:
>>> Em Thu, May 04, 2023 at 04:07:29PM -0300, Arnaldo Carvalho de Melo escreveu:
>>>> Em Thu, May 04, 2023 at 11:50:07AM -0700, Andrii Nakryiko escreveu:
>>>>> On Thu, May 4, 2023 at 10:52 AM Arnaldo Carvalho de Melo <acme@kernel.org> wrote:
>>>>>> Andrii, can you add some more information about the usage of vmlinux.h
>>>>>> instead of using kernel headers?
>>>>
>>>>> I'll just say that vmlinux.h is not a hard requirement to build BPF
>>>>> programs, it's more a convenience allowing easy access to definitions
>>>>> of both UAPI and kernel-internal structures for tracing needs and
>>>>> marking them relocatable using BPF CO-RE machinery. Lots of real-world
>>>>> applications just check-in pregenerated vmlinux.h to avoid build-time
>>>>> dependency on up-to-date host kernel and such.
>>>>
>>>>> If vmlinux.h generation and usage is causing issues, though, given
>>>>> that perf's BPF programs don't seem to be using many different kernel
>>>>> types, it might be a better option to just use UAPI headers for public
>>>>> kernel type definitions, and just define CO-RE-relocatable minimal
>>>>> definitions locally in perf's BPF code for the other types necessary.
>>>>> E.g., if perf needs only pid and tgid from task_struct, this would
>>>>> suffice:
>>>>
>>>>> struct task_struct {
>>>>> int pid;
>>>>> int tgid;
>>>>> } __attribute__((preserve_access_index));
>>>>
>>>> Yeah, that seems like a way better approach, no vmlinux involved, libbpf
>>>> CO-RE notices that task_struct changed from this two integers version
>>>> (of course) and does the relocation to where it is in the running kernel
>>>> by using /sys/kernel/btf/vmlinux.
>>>
>>> Doing it for one of the skels, build tested, runtime untested, but not
>>> using any vmlinux, BTF to help, not that bad, more verbose, but at least
>>> we state what are the fields we actually use, have those attribute
>>> documenting that those offsets will be recorded for future use, etc.
>
> Yang, can you please check that this works?
>
Yes, I've tested this patch and it works :)
Tested-by: Yang Jihong <yangjihong1@huawei.com>
# perf kwork report -b
Starting trace, Hit <Ctrl+C> to stop and report
^C
Kwork Name | Cpu | Total Runtime | Count |
Max runtime | Max runtime start | Max runtime end |
--------------------------------------------------------------------------------------------------------------------------------
(s)SCHED:7 | 0005 | 1.440 ms | 3 |
1.377 ms | 165822.963188 s | 165822.964565 s |
(s)SCHED:7 | 0001 | 1.388 ms | 2 |
1.377 ms | 165822.963188 s | 165822.964565 s |
(w)e1000_watchdog | 0002 | 0.532 ms | 1 |
0.532 ms | 165823.806777 s | 165823.807309 s |
(w)flush_to_ldisc | 0005 | 0.524 ms | 1 |
0.524 ms | 165824.255219 s | 165824.255743 s |
(s)NET_RX:3 | 0002 | 0.512 ms | 4 |
0.247 ms | 165824.254103 s | 165824.254350 s |
(w)wq_barrier_func | 0003 | 0.394 ms | 1 |
0.394 ms | 165822.964179 s | 165822.964573 s |
eth0:10 | 0002 | 0.250 ms | 4 |
0.115 ms | 165822.964814 s | 165822.964929 s |
(w)wb_workfn | 0005 | 0.208 ms | 1 |
0.208 ms | 165823.998558 s | 165823.998766 s |
(s)TIMER:1 | 0002 | 0.205 ms | 4 |
0.077 ms | 165823.806548 s | 165823.806626 s |
(s)SCHED:7 | 0002 | 0.185 ms | 5 |
0.049 ms | 165824.255119 s | 165824.255168 s |
(s)TIMER:1 | 0003 | 0.167 ms | 4 |
0.079 ms | 165823.998326 s | 165823.998405 s |
(s)SCHED:7 | 0000 | 0.165 ms | 4 |
0.073 ms | 165823.807235 s | 165823.807308 s |
(w)flush_to_ldisc | 0003 | 0.156 ms | 1 |
0.156 ms | 165824.255723 s | 165824.255879 s |
(s)TIMER:1 | 0007 | 0.152 ms | 2 |
0.100 ms | 165823.358416 s | 165823.358517 s |
(w)vmstat_update | 0001 | 0.094 ms | 1 |
0.094 ms | 165824.256460 s | 165824.256554 s |
(s)RCU:9 | 0001 | 0.088 ms | 4 |
0.040 ms | 165822.964728 s | 165822.964768 s |
(w)vmstat_shepherd | 0000 | 0.078 ms | 1 |
0.078 ms | 165824.256321 s | 165824.256399 s |
(s)SCHED:7 | 0007 | 0.066 ms | 2 |
0.038 ms | 165823.358528 s | 165823.358566 s |
(s)SCHED:7 | 0003 | 0.057 ms | 2 |
0.034 ms | 165823.998412 s | 165823.998446 s |
virtio0-requests:25 | 0000 | 0.056 ms | 1 |
0.056 ms | 165824.255959 s | 165824.256016 s |
(s)TIMER:1 | 0000 | 0.055 ms | 1 |
0.055 ms | 165824.256164 s | 165824.256219 s |
(s)RCU:9 | 0005 | 0.045 ms | 3 |
0.023 ms | 165822.964728 s | 165822.964751 s |
(s)RCU:9 | 0002 | 0.028 ms | 2 |
0.017 ms | 165823.174539 s | 165823.174556 s |
(s)RCU:9 | 0007 | 0.016 ms | 1 |
0.016 ms | 165823.358571 s | 165823.358587 s |
(s)RCU:9 | 0000 | 0.013 ms | 1 |
0.013 ms | 165824.256264 s | 165824.256278 s |
(s)RCU:9 | 0003 | 0.011 ms | 1 |
0.011 ms | 165822.973142 s | 165822.973153 s |
--------------------------------------------------------------------------------------------------------------------------------
# perf kwork lat -b
Starting trace, Hit <Ctrl+C> to stop and report
^C
Kwork Name | Cpu | Avg delay | Count |
Max delay | Max delay start | Max delay end |
--------------------------------------------------------------------------------------------------------------------------------
(w)neigh_periodic_work | 0001 | 0.568 ms | 1 |
0.568 ms | 165839.038372 s | 165839.038941 s |
(s)RCU:9 | 0000 | 0.292 ms | 1 |
0.292 ms | 165840.176224 s | 165840.176515 s |
(s)TIMER:1 | 0000 | 0.260 ms | 1 |
0.260 ms | 165840.176206 s | 165840.176466 s |
(w)disk_events_workfn | 0001 | 0.256 ms | 1 |
0.256 ms | 165839.038327 s | 165839.038583 s |
(s)RCU:9 | 0001 | 0.204 ms | 1 |
0.204 ms | 165839.038213 s | 165839.038417 s |
(s)SCHED:7 | 0001 | 0.153 ms | 1 |
0.153 ms | 165839.038231 s | 165839.038384 s |
(s)NET_RX:3 | 0002 | 0.132 ms | 3 |
0.184 ms | 165840.175984 s | 165840.176168 s |
(w)ata_sff_pio_task | 0001 | 0.124 ms | 1 |
0.124 ms | 165839.038852 s | 165839.038976 s |
(s)SCHED:7 | 0000 | 0.122 ms | 2 |
0.193 ms | 165840.176244 s | 165840.176437 s |
(s)RCU:9 | 0007 | 0.106 ms | 1 |
0.106 ms | 165838.982220 s | 165838.982326 s |
(s)RCU:9 | 0003 | 0.092 ms | 1 |
0.092 ms | 165839.040098 s | 165839.040189 s |
(s)TIMER:1 | 0001 | 0.085 ms | 1 |
0.085 ms | 165839.038179 s | 165839.038264 s |
(s)SCHED:7 | 0007 | 0.078 ms | 3 |
0.086 ms | 165839.990180 s | 165839.990265 s |
(s)TIMER:1 | 0007 | 0.077 ms | 3 |
0.081 ms | 165839.990136 s | 165839.990216 s |
(s)TIMER:1 | 0003 | 0.076 ms | 1 |
0.076 ms | 165839.040066 s | 165839.040142 s |
(s)RCU:9 | 0002 | 0.075 ms | 2 |
0.088 ms | 165839.118209 s | 165839.118297 s |
(w)flush_to_ldisc | 0006 | 0.070 ms | 1 |
0.070 ms | 165840.175332 s | 165840.175402 s |
(s)TIMER:1 | 0006 | 0.067 ms | 2 |
0.085 ms | 165838.908056 s | 165838.908142 s |
(s)TIMER:1 | 0002 | 0.067 ms | 2 |
0.074 ms | 165840.175076 s | 165840.175149 s |
(s)SCHED:7 | 0006 | 0.065 ms | 1 |
0.065 ms | 165838.908107 s | 165838.908171 s |
(s)SCHED:7 | 0003 | 0.051 ms | 1 |
0.051 ms | 165839.040115 s | 165839.040166 s |
(s)SCHED:7 | 0002 | 0.046 ms | 2 |
0.048 ms | 165840.175123 s | 165840.175171 s |
(s)RCU:9 | 0006 | 0.035 ms | 2 |
0.036 ms | 165838.913051 s | 165838.913087 s |
(s)BLOCK:4 | 0001 | 0.023 ms | 1 |
0.023 ms | 165839.039281 s | 165839.039303 s |
--------------------------------------------------------------------------------------------------------------------------------
Thanks,
Yang
next prev parent reply other threads:[~2023-05-06 1:13 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-05-03 21:18 Arnaldo Carvalho de Melo
2023-05-04 3:00 ` Linus Torvalds
2023-05-04 3:12 ` Linus Torvalds
2023-05-04 5:51 ` Ian Rogers
2023-05-04 18:36 ` Jiri Olsa
2023-05-04 11:09 ` BPF skels in perf .Re: " Arnaldo Carvalho de Melo
2023-05-04 17:25 ` Linus Torvalds
2023-05-04 17:52 ` Arnaldo Carvalho de Melo
2023-05-04 18:50 ` Andrii Nakryiko
2023-05-04 19:07 ` Arnaldo Carvalho de Melo
2023-05-04 21:48 ` Arnaldo Carvalho de Melo
2023-05-04 22:01 ` Arnaldo Carvalho de Melo
2023-05-05 13:18 ` Arnaldo Carvalho de Melo
2023-05-06 1:13 ` Yang Jihong [this message]
2023-05-05 13:20 ` Arnaldo Carvalho de Melo
2023-05-04 22:03 ` Ian Rogers
2023-05-04 23:03 ` Jiri Olsa
2023-05-04 23:15 ` Namhyung Kim
2023-05-05 9:36 ` Jiri Olsa
2023-05-04 23:19 ` Ian Rogers
2023-05-05 9:39 ` Jiri Olsa
2023-05-05 11:42 ` Jiri Olsa
2023-05-05 13:33 ` Arnaldo Carvalho de Melo
2023-05-05 15:14 ` Alexei Starovoitov
2023-05-05 16:56 ` [PATCH RFC/RFT] perf bpf skels: Stop using vmlinux.h generated from BTF, use subset of used structs + CO-RE. was " Arnaldo Carvalho de Melo
2023-05-05 17:04 ` Ian Rogers
2023-05-05 20:43 ` Jiri Olsa
2023-05-05 20:46 ` Ian Rogers
2023-05-05 20:48 ` Namhyung Kim
2023-05-10 18:56 ` Arnaldo Carvalho de Melo
2023-05-05 20:49 ` Arnaldo Carvalho de Melo
2023-05-05 21:15 ` Jiri Olsa
2023-05-05 21:21 ` Andrii Nakryiko
2023-05-05 21:52 ` Arnaldo Carvalho de Melo
2023-05-05 21:55 ` Andrii Nakryiko
2023-05-05 21:33 ` Arnaldo Carvalho de Melo
2023-05-08 21:53 ` Ian Rogers
2023-05-04 22:46 ` Namhyung Kim
2023-05-07 19:15 ` pr-tracker-bot
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=bf082a2a-f360-95fd-9d18-0702c421b296@huawei.com \
--to=yangjihong1@huawei.com \
--cc=acme@kernel.org \
--cc=acme@redhat.com \
--cc=adrian.hunter@intel.com \
--cc=andrii.nakryiko@gmail.com \
--cc=andrii@kernel.org \
--cc=bpf@vger.kernel.org \
--cc=changbin.du@huawei.com \
--cc=eranian@google.com \
--cc=haoluo@google.com \
--cc=irogers@google.com \
--cc=james.clark@arm.com \
--cc=jolsa@kernel.org \
--cc=kan.liang@linux.intel.com \
--cc=kcarcia@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=lozko.roma@gmail.com \
--cc=mingo@kernel.org \
--cc=song@kernel.org \
--cc=tglx@linutronix.de \
--cc=tmricht@linux.ibm.com \
--cc=torvalds@linux-foundation.org \
--cc=williams@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®