From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Ian Rogers <irogers@google.com>,
Arnaldo Carvalho de Melo <arnaldo.melo@gmail.com>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
Linux Next Mailing List <linux-next@vger.kernel.org>,
namhyung@kernel.org
Subject: Re: linux-next: build failure after merge of the perf tree
Date: Thu, 23 Mar 2023 10:27:36 -0300 [thread overview]
Message-ID: <ZBxTyLqkIaoVhIXU@kernel.org> (raw)
In-Reply-To: <20230323095437.1ecccec1@canb.auug.org.au>
Em Thu, Mar 23, 2023 at 09:54:37AM +1100, Stephen Rothwell escreveu:
> Hi Ian,
>
> On Wed, 22 Mar 2023 11:37:31 -0700 Ian Rogers <irogers@google.com> wrote:
> >
> > On Tue, Mar 21, 2023 at 2:40 PM Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> > >
> > > Hi all,
> > >
> > > On Fri, 17 Mar 2023 09:50:25 +1100 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> > > >
> > > > After merging the perf tree, today's linux-next build (native perf)
> > > > failed like this:
> > > >
> > > > Auto-detecting system features:
> > > > ... clang-bpf-co-re: [ [32mon [m ]
> > > > ... llvm: [ [31mOFF [m ]
> > > > ... libcap: [ [32mon [m ]
> > > > ... libbfd: [ [32mon [m ]
> > > >
> > > > make[1]: *** Deleting file '/home/sfr/next/perf/util/bpf_skel/vmlinux.h'
> > > > libbpf: failed to find '.BTF' ELF section in /boot/vmlinux-6.0.0-5-powerpc64le
> > > > Error: failed to load BTF from /boot/vmlinux-6.0.0-5-powerpc64le: No data available
> > > > make[1]: *** [Makefile.perf:1075: /home/sfr/next/perf/util/bpf_skel/vmlinux.h] Error 195
> > > > make[1]: *** Waiting for unfinished jobs....
> > > > make: *** [Makefile.perf:236: sub-make] Error 2
> > > > Command exited with non-zero status 2
> > > >
> > > > To be clear this is a native build of perf on a PPC64le host using this
> > > > command line:
> > > >
> > > > make -C tools/perf -f Makefile.perf -s -O -j60 O=../perf EXTRA_CFLAGS=-Wno-psabi
> > > >
> > > > (I could probably remove the EXTRA_CLFAGS now that I am building with
> > > > gcc 12.2)
> > > >
> > > > I don't know which commit caused this.
> > > >
> > > > I have used the perf tree from next-20230316 for today.
> > >
> > > I am still getting this build failure.
> > The build failure is intentional as not having BPF skeleton support in
> > the perf tool will remove features. I've just sent:
> > https://lore.kernel.org/lkml/20230322183108.1380882-1-irogers@google.com/
> > Which will recommend adding NO_BPF_SKEL=1 to your build options when
> > this failure occurs. I didn't think a features test was appropriate
> > for this as the feature test would basically replicate the vmlinux.h
> > generation and I didn't want to move that support through the build
> > system.
> I was trying to understand why this step fails, but from the error
> messages, it seems to require something to be present in the distro
> supplied kernel image? Is there something missing from the ppc
> build process? Or toolchain? Why is it looking at the installed
> kernel and not the built kernel? Does the perf build now depend on the
> kernel being built first?
>
> I will add NO_BPF_SKEL=1 to my build from tomorrow, but surely that
> means that we miss some perf build testing :-(
Before this BUILD_BPF_SKEL wasn't on by default, so you're back testing
as much as before.
Having said that, we need to improve the warning and I processed a patch
from Ian to that extent:
https://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git/commit/?h=tmp.perf-tools-next&id=12a83df53444165d39d3e09fcd9627e7bec7828e
$(SKEL_OUT)/vmlinux.h: $(VMLINUX_BTF) $(BPFTOOL)
ifeq ($(VMLINUX_H),)
- $(QUIET_GEN)$(BPFTOOL) btf dump file $< format c > $@
+ $(QUIET_GEN)$(BPFTOOL) btf dump file $< format c > $@ || \
+ (echo "Failure to generate vmlinux.h needed for the recommended BPF skeleton support." && \
+ echo "To disable this use the build option NO_BPF_SKEL=1." && \
+ echo "Alternatively point at a pre-generated vmlinux.h with VMLINUX_H=<path>." && \
+ false)
else
$(Q)cp "$(VMLINUX_H)" $@
endif
Which improves a bit the situation.
We could check if bpftool is available and if not, suggest installing
it.
If it is available, we could check if /sys/kernel/bpf/ is available, if
not suggest using a kernel with CONFIG_DEBUG_INFO_BTF=y, as most distros
have by now.
As to the features this enables:
⬢[acme@toolbox perf-tools-next]$ ls -la tools/perf/util/bpf_skel/*.c
-rw-r--r--. 1 acme acme 5581 Oct 17 09:07 tools/perf/util/bpf_skel/bperf_cgroup.bpf.c
-rw-r--r--. 1 acme acme 1764 May 5 2022 tools/perf/util/bpf_skel/bperf_follower.bpf.c
-rw-r--r--. 1 acme acme 1438 May 5 2022 tools/perf/util/bpf_skel/bperf_leader.bpf.c
-rw-r--r--. 1 acme acme 2290 May 5 2022 tools/perf/util/bpf_skel/bpf_prog_profiler.bpf.c
-rw-r--r--. 1 acme acme 2164 May 5 2022 tools/perf/util/bpf_skel/func_latency.bpf.c
-rw-r--r--. 1 acme acme 9017 Aug 26 2022 tools/perf/util/bpf_skel/kwork_trace.bpf.c
-rw-r--r--. 1 acme acme 9251 Mar 14 08:33 tools/perf/util/bpf_skel/lock_contention.bpf.c
-rw-r--r--. 1 acme acme 6109 Feb 27 17:47 tools/perf/util/bpf_skel/off_cpu.bpf.c
-rw-r--r--. 1 acme acme 4310 Mar 15 11:08 tools/perf/util/bpf_skel/sample_filter.bpf.c
⬢[acme@toolbox perf-tools-next]$
For instance, take a look at these csets:
⬢[acme@toolbox perf-tools-next]$ git log --oneline tools/perf/util/bpf_skel/lock_contention.bpf.c
d24c0144b1dde00f perf lock contention: Show per-cpu rq_lock with address
1811e82767dcc6eb perf lock contention: Track and show siglock with address
3ace2435bb93445e perf lock contention: Track and show mmap_lock with address
17535a33a9c1e4fb perf lock contention: Fix compiler builtin detection
1bece1351c653c3d perf lock contention: Support old rw_semaphore type
3477f079fe70b3c9 perf lock contention: Add -o/--lock-owner option
ebab291641bed48f perf lock contention: Support filters for different aggregation
5e3febe7b7b99f94 perf lock contention: Support lock addr/name filtering for BPF
529772c4df286159 perf lock contention: Support lock type filtering for BPF
688d2e8de231c54e perf lock contention: Add -l/--lock-addr option
eca949b2b4addd94 perf lock contention: Implement -t/--threads option for BPF
fd507d3e359c7e06 perf lock contention: Add lock_data.h for common data
c66a36af7ba3a628 perf lock contention: Do not use BPF task local storage
433b31fa00797a2a perf lock contention: Fix a build error on 32-bit
c1da8dd5c11dabd5 perf lock contention: Skip stack trace from BPF
6d499a6b3d90277d perf lock: Print the number of lost entries for BPF
6fda2405f414b24a perf lock: Implement cpu and task filters for BPF
407b36f69efbdccf perf lock: Use BPF for lock contention analysis
⬢[acme@toolbox perf-tools-next]$
or:
⬢[acme@toolbox perf-tools-next]$ git log --oneline tools/perf/util/bpf_skel/kwork_trace.bpf.c
acfb65fe1d11a97f perf kwork: Add workqueue trace BPF support
5a81927a407c050a perf kwork: Add softirq trace BPF support
420298aefe94840f perf kwork: Add IRQ trace BPF support
daf07d220710a3c8 perf kwork: Implement BPF trace
⬢[acme@toolbox perf-tools-next]$
So we need to fine tune this detection of needed components to build
these features, in time for the v6.4 merge window, that is why we
decided to make this opt-out to hammer out problems.
Other arches probably will hit some of these problems, lets try to
encourage others to try what is in linux-next.
- Arnaldo
next prev parent reply other threads:[~2023-03-23 13:27 UTC|newest]
Thread overview: 53+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-03-16 22:50 linux-next: build failure after merge of the perf tree Stephen Rothwell
2023-03-16 23:11 ` Arnaldo Carvalho de Melo
2023-03-16 23:55 ` Stephen Rothwell
2023-03-17 3:23 ` Ian Rogers
2023-03-21 21:39 ` Stephen Rothwell
2023-03-22 18:37 ` Ian Rogers
2023-03-22 22:54 ` Stephen Rothwell
2023-03-23 13:27 ` Arnaldo Carvalho de Melo [this message]
2023-03-27 20:31 ` Ian Rogers
2023-03-28 1:33 ` Stephen Rothwell
2023-03-28 19:47 ` Arnaldo Carvalho de Melo
2023-03-28 23:06 ` Stephen Rothwell
2023-03-29 12:28 ` Arnaldo Carvalho de Melo
2023-04-03 5:15 ` Stephen Rothwell
-- strict thread matches above, loose matches on Subject: below --
2024-04-01 22:41 Stephen Rothwell
2024-04-02 20:17 ` Arnaldo Carvalho de Melo
2024-04-12 15:22 ` Arnaldo Carvalho de Melo
2024-04-13 2:43 ` Stephen Rothwell
2024-03-21 21:41 Stephen Rothwell
2024-03-21 23:18 ` Arnaldo Carvalho de Melo
2024-03-21 23:19 ` Arnaldo Carvalho de Melo
2024-03-21 23:32 ` Stephen Rothwell
2024-03-22 2:13 ` Stephen Rothwell
2024-03-22 14:32 ` Arnaldo Carvalho de Melo
2024-03-21 23:36 ` Stephen Rothwell
2024-03-21 23:40 ` arnaldo.melo
2024-03-21 23:48 ` Arnaldo Carvalho de Melo
2024-02-21 23:06 Stephen Rothwell
2024-02-23 19:50 ` Arnaldo Carvalho de Melo
2024-02-06 23:40 Stephen Rothwell
2024-02-07 0:02 ` Ian Rogers
2023-10-24 23:26 Stephen Rothwell
2023-10-25 0:34 ` Namhyung Kim
2023-05-29 23:05 Stephen Rothwell
2023-05-30 2:16 ` Ian Rogers
2022-08-30 22:34 Stephen Rothwell
2022-09-01 5:11 ` Anshuman Khandual
2022-09-06 19:05 ` Stephen Rothwell
2022-09-07 2:31 ` Anshuman Khandual
2022-09-07 3:00 ` Stephen Rothwell
2022-09-07 4:22 ` Anshuman Khandual
2022-09-08 15:21 ` Arnaldo Carvalho de Melo
2022-07-20 23:05 Stephen Rothwell
2022-07-21 2:35 ` Ian Rogers
2022-07-21 3:55 ` Stephen Rothwell
2022-07-21 5:33 ` Ian Rogers
2022-01-05 22:19 Stephen Rothwell
2022-01-07 8:58 ` kajoljain
2022-01-11 21:45 ` Stephen Rothwell
2022-01-11 22:01 ` Arnaldo Carvalho de Melo
2022-01-12 5:30 ` kajoljain
2022-01-12 6:16 ` Stephen Rothwell
2022-01-14 12:03 ` Michael Ellerman
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=ZBxTyLqkIaoVhIXU@kernel.org \
--to=acme@kernel.org \
--cc=arnaldo.melo@gmail.com \
--cc=irogers@google.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-next@vger.kernel.org \
--cc=namhyung@kernel.org \
--cc=sfr@canb.auug.org.au \
/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;
as well as URLs for NNTP newsgroup(s).