[GIT PULL] Performance events changes for v6.12

[GIT PULL] Performance events changes for v6.12

[Ingo Molnar]

Linus,

Please pull the latest perf/core Git tree from:

   git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git perf-core-2024-09-18

   # HEAD: 5e645f31139183ac9a282238da18ca6bbc1c6f4a Merge branch 'perf/urgent' into perf/core, to pick up fixes

   [ Merge note: this pull request depends on you having pulled perf-urgent-2024-09-18 already. ]

Performance events changes for v6.12:

 - Implement per-PMU context rescheduling to significantly improve single-PMU
   performance, and related cleanups/fixes. (by Peter Zijlstra and Namhyung Kim)

 - Fix ancient bug resulting in a lot of events being dropped erroneously
   at higher sampling frequencies. (by Luo Gengkun)

 - uprobes enhancements:

     - Implement RCU-protected hot path optimizations for better performance:

         "For baseline vs SRCU, peak througput increased from 3.7 M/s (million uprobe
          triggerings per second) up to about 8 M/s. For uretprobes it's a bit more
          modest with bump from 2.4 M/s to 5 M/s.

          For SRCU vs RCU Tasks Trace, peak throughput for uprobes increases further from
          8 M/s to 10.3 M/s (+28%!), and for uretprobes from 5.3 M/s to 5.8 M/s (+11%),
          as we have more work to do on uretprobes side.

          Even single-thread (no contention) performance is slightly better: 3.276 M/s to
          3.396 M/s (+3.5%) for uprobes, and 2.055 M/s to 2.174 M/s (+5.8%)
          for uretprobes."

          (by Andrii Nakryiko et al)

     - Document mmap_lock, don't abuse get_user_pages_remote(). (by Oleg Nesterov)

     - Cleanups & fixes to prepare for future work:

        - Remove uprobe_register_refctr()
	- Simplify error handling for alloc_uprobe()
        - Make uprobe_register() return struct uprobe *
        - Fold __uprobe_unregister() into uprobe_unregister()
        - Shift put_uprobe() from delete_uprobe() to uprobe_unregister()
        - BPF: Fix use-after-free in bpf_uprobe_multi_link_attach()

          (by Oleg Nesterov)

 - New feature & ABI extension: allow events to use PERF_SAMPLE READ with
   inheritance, enabling sample based profiling of a group of counters over
   a hierarchy of processes or threads.  (by Ben Gainey)

 - Intel uncore & power events updates:

      - Add Arrow Lake and Lunar Lake support
      - Add PERF_EV_CAP_READ_SCOPE
      - Clean up and enhance cpumask and hotplug support

        (by Kan Liang)

      - Add LNL uncore iMC freerunning support
      - Use D0:F0 as a default device

        (by Zhenyu Wang)

 - Intel PT: fix AUX snapshot handling race. (by Adrian Hunter)

 - Misc fixes and cleanups. (by James Clark, Jiri Olsa, Oleg Nesterov and Peter Zijlstra)

Thanks,

	Ingo

------------------>

Adrian Hunter (1):
      perf/x86/intel/pt: Fix sampling synchronization

Andrii Nakryiko (7):
      perf,x86: avoid missing caller address in stack traces captured in uprobe
      uprobes: simplify error handling for alloc_uprobe()
      uprobes: revamp uprobe refcounting and lifetime management
      uprobes: protected uprobe lifetime with SRCU
      uprobes: get rid of enum uprobe_filter_ctx in uprobe filter callbacks
      uprobes: travers uprobe's consumer list locklessly under SRCU protection
      uprobes: perform lockless SRCU-protected uprobes_tree lookup

Ben Gainey (2):
      perf: Rename perf_event_context.nr_pending to nr_no_switch_fast.
      perf: Support PERF_SAMPLE_READ with inherit

Ingo Molnar (2):
      Merge branch 'perf/urgent' into perf/core, to pick up fixes
      Merge branch 'perf/urgent' into perf/core, to pick up fixes

James Clark (1):
      perf/x86/intel/bts: Fix comment about default perf_event_paranoid setting

Jiri Olsa (1):
      selftests/bpf: fix uprobe.path leak in bpf_testmod

Kan Liang (8):
      perf/x86/intel/uncore: Add Arrow Lake support
      perf/x86/intel/uncore: Factor out common MMIO init and ops functions
      perf/x86/intel/uncore: Add Lunar Lake support
      perf: Generic hotplug support for a PMU with a scope
      perf: Add PERF_EV_CAP_READ_SCOPE
      perf/x86/intel/cstate: Clean up cpumask and hotplug
      iommu/vt-d: Clean up cpumask and hotplug for perfmon
      dmaengine: idxd: Clean up cpumask and hotplug for perfmon

Luo Gengkun (1):
      perf/core: Fix small negative period being ignored

Namhyung Kim (1):
      perf: Really fix event_function_call() locking

Oleg Nesterov (8):
      uprobes: document the usage of mm->mmap_lock
      uprobes: is_trap_at_addr: don't use get_user_pages_remote()
      uprobes: kill uprobe_register_refctr()
      uprobes: make uprobe_register() return struct uprobe *
      uprobes: change uprobe_register() to use uprobe_unregister() instead of __uprobe_unregister()
      uprobes: fold __uprobe_unregister() into uprobe_unregister()
      uprobes: shift put_uprobe() from delete_uprobe() to uprobe_unregister()
      bpf: Fix use-after-free in bpf_uprobe_multi_link_attach()

Peter Zijlstra (8):
      perf/x86: Add hw_perf_event::aux_config
      perf: Optimize context reschedule for single PMU cases
      perf: Extract a few helpers
      perf: Fix event_function_call() locking
      perf: Add context time freeze
      perf: Optimize __pmu_ctx_sched_out()
      perf/uprobe: split uprobe_unregister()
      rbtree: provide rb_find_rcu() / rb_find_add_rcu()

Zhenyu Wang (2):
      perf/x86/intel/uncore: Add LNL uncore iMC freerunning support
      perf/x86/intel/uncore: Use D0:F0 as a default device

 arch/x86/events/core.c                                |  63 +++++++++++++++++++
 arch/x86/events/intel/bts.c                           |   3 -
 arch/x86/events/intel/cstate.c                        | 142 ++-----------------------------------------
 arch/x86/events/intel/pt.c                            |  29 +++++----
 arch/x86/events/intel/uncore.c                        |   9 +++
 arch/x86/events/intel/uncore.h                        |   2 +
 arch/x86/events/intel/uncore_snb.c                    | 185 ++++++++++++++++++++++++++++++++++++++++++++++++++------
 drivers/dma/idxd/idxd.h                               |   7 ---
 drivers/dma/idxd/init.c                               |   3 -
 drivers/dma/idxd/perfmon.c                            |  98 +-----------------------------
 drivers/iommu/intel/iommu.h                           |   2 -
 drivers/iommu/intel/perfmon.c                         | 111 +---------------------------------
 include/linux/cpuhotplug.h                            |   2 -
 include/linux/perf_event.h                            |  32 +++++++++-
 include/linux/rbtree.h                                |  67 +++++++++++++++++++++
 include/linux/uprobes.h                               |  48 ++++++++-------
 kernel/events/core.c                                  | 586 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-------------------------------------------------
 kernel/events/uprobes.c                               | 505 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--------------------------------------------------------------------
 kernel/trace/bpf_trace.c                              |  38 ++++++------
 kernel/trace/trace_uprobe.c                           |  44 +++++++-------
 tools/testing/selftests/bpf/bpf_testmod/bpf_testmod.c |  27 +++++----
 21 files changed, 1146 insertions(+), 857 deletions(-)

Re: [GIT PULL] Performance events changes for v6.12

[Linus Torvalds]

On Wed, Sep 18, 2024 at 2:45 PM Ingo Molnar <mingo@kernel.org> wrote:
>
> Please pull the latest perf/core Git tree from:
>
>    git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git perf-core-2024-09-18

This - and the perf events fixes pull - was in my spam-box.

The reason is because kernel.org has made its DMARC setup stricter.
and now requires that if you have a kernel.org "from" address, the
DKIM must also be from kernel.org.

But you seem to be using gmail.com as the smtp gateway, and so it has
the gmail DKIM, and so while that passes with the gmail hashes:

        dkim=pass header.i=@gmail.com header.s=20230601 header.b=Gi5JkPcO;

the DMARC does not pass, because your From and your DKIM do not match:

       dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=QUARANTINE)
header.from=kernel.org;

end the email gets correctly marked as spam.

You need to use the kernel.org smtp server in order to get the proper
kernel.org signature for a kernel.org source.

See

        https://korg.docs.kernel.org/mail.html

for proper setup.

Anyway, I did pick these two emails up from the spam filter, but I
don't guarantee that I always notice things getting caught as spam.

               Linus

Re: [GIT PULL] Performance events changes for v6.12

[Linus Torvalds]

On Wed, Sep 18, 2024 at 2:45 PM Ingo Molnar <mingo@kernel.org> wrote:
>
>  arch/x86/events/core.c                                |  63 +++++++++++++++++++
>  arch/x86/events/intel/bts.c                           |   3 -
[...]
>  kernel/events/core.c                                  | 586 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-------------------------------------------------
>  kernel/events/uprobes.c                               | 505 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--------------------------------------------------------------------

You seem to be cutting-and-pasting your diffstats from a *very* wide
terminal window, resulting in a very messy diffstat result.

Please either pipe the git request-pull output to some tool (I suggest
just piping to 'xsel' or similar that does the 'cut' part for you, but
anything works - you can just do "| cat" to make the stdout not be the
terminal).

Or just use "--stat=80" to set the output width to a fixed sane thing,
instead of that insane 250+ character window width that it seems you
are using and that causes those overlong diffstat lines and makes it
all hard to read.

               Linus

Re: [GIT PULL] Performance events changes for v6.12

[pr-tracker-bot]

The pull request you sent on Wed, 18 Sep 2024 14:45:39 +0200:

> git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git perf-core-2024-09-18

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/9f0c253ddddca608457a42e509267bed2dee0a50

Thank you!

-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/prtracker.html

Re: [GIT PULL] Performance events changes for v6.12

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 703 bytes --]

Hi all,

On Wed, 18 Sep 2024 14:45:39 +0200 Ingo Molnar <mingo@kernel.org> wrote:
>
> Please pull the latest perf/core Git tree from:
> 
>    git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git perf-core-2024-09-18
> 
>    # HEAD: 5e645f31139183ac9a282238da18ca6bbc1c6f4a Merge branch 'perf/urgent' into perf/core, to pick up fixes

>       perf: Generic hotplug support for a PMU with a scope

This commit introduces a (reported) build warning in some configurations:

https://lore.kernel.org/all/20240911153854.240bbc1f@canb.auug.org.au

That turns into a build failure for some configurations :-(

A suggested fix was posted a few days ago.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [GIT PULL] Performance events changes for v6.12

[Linus Torvalds]

On Thu, Sep 19, 2024 at 10:33 AM Ingo Molnar <mingo@kernel.org> wrote:
>
> Thanks a lot - I fixed up my SMTP sending path, so starting with this mail
> it shouldn't go to your spam folder anymore.

Well, that didn't work at all. It's still marked as spam, and the
headers of your email still say you go through gmail:

  Received: from gmail.com (1F2EF419.nat.pool.telekom.hu. [31.46.244.25])
        by smtp.gmail.com with ESMTPSA id
ffacd0b85a97d-378e73e80basm14420731f8f.26.2024.09.19.01.33.08
        (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256);
        Thu, 19 Sep 2024 01:33:09 -0700 (PDT)
  Sender: Ingo Molnar <mingo.kernel.org@gmail.com>
  Date: Thu, 19 Sep 2024 10:33:06 +0200
  From: Ingo Molnar <mingo@kernel.org>

with no sign of it having gone through mail.kernel.org (which is the
only way to get the right DKIM). So it still ends up with that

     dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=QUARANTINE)
header.from=kernel.org;

thing.

            Linus

Re: [GIT PULL] Performance events changes for v6.12

[Ingo Molnar]

* Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Wed, Sep 18, 2024 at 2:45 PM Ingo Molnar <mingo@kernel.org> wrote:
> >
> >  arch/x86/events/core.c                                |  63 +++++++++++++++++++
> >  arch/x86/events/intel/bts.c                           |   3 -
> [...]
> >  kernel/events/core.c                                  | 586 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-------------------------------------------------
> >  kernel/events/uprobes.c                               | 505 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--------------------------------------------------------------------
> 
> You seem to be cutting-and-pasting your diffstats from a *very* wide
> terminal window, resulting in a very messy diffstat result.

Yeah, sorry about that - it was a one-off manual diffstat due to the merge 
order between perf/urgent and perf/core, and indeed my terminal is ~240 
characters wide and I didn't notice the diffstat width-scaling weirdness.

> Please either pipe the git request-pull output to some tool (I suggest 
> just piping to 'xsel' or similar that does the 'cut' part for you, but 
> anything works - you can just do "| cat" to make the stdout not be the 
> terminal).
> 
> Or just use "--stat=80" to set the output width to a fixed sane thing, 
> instead of that insane 250+ character window width that it seems you are 
> using and that causes those overlong diffstat lines and makes it all hard 
> to read.

Will do next time it happens. Below is a belated diffstat with the correct 
width.

Thanks,

	Ingo

 arch/x86/events/core.c                             |  63 +++
 arch/x86/events/intel/bts.c                        |   3 -
 arch/x86/events/intel/cstate.c                     | 142 +----
 arch/x86/events/intel/pt.c                         |  29 +-
 arch/x86/events/intel/uncore.c                     |   9 +
 arch/x86/events/intel/uncore.h                     |   2 +
 arch/x86/events/intel/uncore_snb.c                 | 185 ++++++-
 drivers/dma/idxd/idxd.h                            |   7 -
 drivers/dma/idxd/init.c                            |   3 -
 drivers/dma/idxd/perfmon.c                         |  98 +---
 drivers/iommu/intel/iommu.h                        |   2 -
 drivers/iommu/intel/perfmon.c                      | 111 +---
 include/linux/cpuhotplug.h                         |   2 -
 include/linux/perf_event.h                         |  32 +-
 include/linux/rbtree.h                             |  67 +++
 include/linux/uprobes.h                            |  48 +-
 kernel/events/core.c                               | 586 +++++++++++++++------
 kernel/events/uprobes.c                            | 505 ++++++++++--------
 kernel/trace/bpf_trace.c                           |  38 +-
 kernel/trace/trace_uprobe.c                        |  44 +-
 .../selftests/bpf/bpf_testmod/bpf_testmod.c        |  27 +-
 21 files changed, 1146 insertions(+), 857 deletions(-)

Re: [GIT PULL] Performance events changes for v6.12

[Ingo Molnar]

* Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Thu, Sep 19, 2024 at 10:33 AM Ingo Molnar <mingo@kernel.org> wrote:
> >
> > Thanks a lot - I fixed up my SMTP sending path, so starting with this mail
> > it shouldn't go to your spam folder anymore.
> 
> Well, that didn't work at all. It's still marked as spam, and the
> headers of your email still say you go through gmail:

Because I'm a moron:

I fixed my setup, tested that it's working fine:

      Received: from nyc.source.kernel.org (nyc.source.kernel.org. [147.75.193.91])

... and then I sent you the email from a Mutt session I started yesterday. :-/

This mail should be fine. I hope.

Thanks,

	Ingo

Re: [GIT PULL] Performance events changes for v6.12

[Linus Torvalds]

On Thu, 19 Sept 2024 at 10:48, Ingo Molnar <mingo@kernel.org> wrote:
>
> I fixed my setup, tested that it's working fine:
>
>       Received: from nyc.source.kernel.org (nyc.source.kernel.org. [147.75.193.91])
>
> ... and then I sent you the email from a Mutt session I started yesterday. :-/
>
> This mail should be fine. I hope.

Heh. Yes. This looks A-Ok over here. Thanks,

              Linus

Re: [GIT PULL] Performance events changes for v6.12

[Ingo Molnar]

* Stephen Rothwell <sfr@canb.auug.org.au> wrote:

> Hi all,
> 
> On Wed, 18 Sep 2024 14:45:39 +0200 Ingo Molnar <mingo@kernel.org> wrote:
> >
> > Please pull the latest perf/core Git tree from:
> > 
> >    git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git perf-core-2024-09-18
> > 
> >    # HEAD: 5e645f31139183ac9a282238da18ca6bbc1c6f4a Merge branch 'perf/urgent' into perf/core, to pick up fixes
> 
> >       perf: Generic hotplug support for a PMU with a scope
> 
> This commit introduces a (reported) build warning in some configurations:
> 
> https://lore.kernel.org/all/20240911153854.240bbc1f@canb.auug.org.au
> 
> That turns into a build failure for some configurations :-(

Thanks: I've queued up the fix for this ARM warning in perf/urgent and will 
get it to Linus later today.

	Ingo

Re: [GIT PULL] Performance events changes for v6.12

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 257 bytes --]

Hi Ingo,

On Thu, 19 Sep 2024 11:53:20 +0200 Ingo Molnar <mingo@kernel.org> wrote:
>
> Thanks: I've queued up the fix for this ARM warning in perf/urgent and will 
> get it to Linus later today.

Excellent, thanks.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [GIT PULL] Performance events changes for v6.12

[Linus Torvalds]

On Thu, 19 Sept 2024 at 02:53, Ingo Molnar <mingo@kernel.org> wrote:
>
> Thanks: I've queued up the fix for this ARM warning in perf/urgent and will
> get it to Linus later today.

So this apparently never happened.

I'm back home now, and so I now do my arm64 builds too, and it is
indeed failing.

                Linus

Re: [GIT PULL] Performance events changes for v6.12

[Linus Torvalds]

On Sat, 21 Sept 2024 at 15:39, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> On Thu, 19 Sept 2024 at 02:53, Ingo Molnar <mingo@kernel.org> wrote:
> >
> > Thanks: I've queued up the fix for this ARM warning in perf/urgent and will
> > get it to Linus later today.
>
> So this apparently never happened.
>
> I'm back home now, and so I now do my arm64 builds too, and it is
> indeed failing.

In case people didn't notice - in order to now do my now usual full
builds I just took the fix directly (commit 673a5009cf2f: "perf: Fix
topology_sibling_cpumask check warning on ARM").

                Linus

Re: [GIT PULL] Performance events changes for v6.12

[Ingo Molnar]

* Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Sat, 21 Sept 2024 at 15:39, Linus Torvalds
> <torvalds@linux-foundation.org> wrote:
> >
> > On Thu, 19 Sept 2024 at 02:53, Ingo Molnar <mingo@kernel.org> wrote:
> > >
> > > Thanks: I've queued up the fix for this ARM warning in perf/urgent 
> > > and will get it to Linus later today.
> >
> > So this apparently never happened.

Sorry about that, was sick for a few days.

> > I'm back home now, and so I now do my arm64 builds too, and it is 
> > indeed failing.
> 
> In case people didn't notice - in order to now do my now usual full 
> builds I just took the fix directly (commit 673a5009cf2f: "perf: Fix 
> topology_sibling_cpumask check warning on ARM").

Looks good, thank you!

	Ingo