[ Shorten the Cc list to just those that maintain this ]

This series does not make any user space visible changes.
It only adds the necessary infrastructure of the deferred unwinder
and makes a few helpful cleanups to perf.

 Based off of tip/master: 252d33c92dbc23bcc1e662a889787c09a02eeccc

Peter,

Would you be willing to take this series? I'd like to get this part
in the kernel in the next merge window and then we can focus on getting perf
and ftrace to use it in the next merge window.

Perf exposes a lot of the interface to user space as the perf tool needs
to handle the merging of the stacks, I figured it would be better to just
get the kernel side mostly done and then work out the kinks of the code
between user and kernel.

Are you OK with this?

This series combines the non user interface of:

 [v7] perf: Deferred unwinding of user space stack traces
 https://lore.kernel.org/linux-trace-kernel/20250430195746.827125963@goodmis.org/

which had no changes, with:

 [v6] perf: Deferred unwinding of user space stack traces for per CPU events
 https://lore.kernel.org/linux-trace-kernel/20250501013202.997535180@goodmis.org/

With the following changes:

- Have unwind_deferred_request() return positive if already queued

- Check (current->flags & PF_KTHREAD | PF_EXITING) in
  unwind_deferred_request(), as the task_work will fail to be added in the
  exit code.

Hence, this is called v7.

Josh Poimboeuf (13):
      unwind_user: Add user space unwinding API
      unwind_user: Add frame pointer support
      unwind_user/x86: Enable frame pointer unwinding on x86
      perf/x86: Rename and move get_segment_base() and make it global
      unwind_user: Add compat mode frame pointer support
      unwind_user/x86: Enable compat mode frame pointer unwinding on x86
      unwind_user/deferred: Add unwind cache
      unwind_user/deferred: Add deferred unwinding interface
      unwind_user/deferred: Make unwind deferral requests NMI-safe
      perf: Remove get_perf_callchain() init_nr argument
      perf: Have get_perf_callchain() return NULL if crosstask and user are set
      perf: Simplify get_perf_callchain() user logic
      perf: Skip user unwind if the task is a kernel thread.

Steven Rostedt (4):
      unwind_user/deferred: Add unwind_deferred_trace()
      unwind deferred: Use bitmask to determine which callbacks to call
      unwind deferred: Use SRCU unwind_deferred_task_work()
      perf: Use current->flags & PF_KTHREAD instead of current->mm == NULL

----
 MAINTAINERS                              |   8 +
 arch/Kconfig                             |  11 +
 arch/x86/Kconfig                         |   2 +
 arch/x86/events/core.c                   |  44 +---
 arch/x86/include/asm/ptrace.h            |   2 +
 arch/x86/include/asm/unwind_user.h       |  61 ++++++
 arch/x86/include/asm/unwind_user_types.h |  17 ++
 arch/x86/kernel/ptrace.c                 |  38 ++++
 include/asm-generic/Kbuild               |   2 +
 include/asm-generic/unwind_user.h        |  24 +++
 include/asm-generic/unwind_user_types.h  |   9 +
 include/linux/entry-common.h             |   2 +
 include/linux/perf_event.h               |   2 +-
 include/linux/sched.h                    |   6 +
 include/linux/unwind_deferred.h          |  50 +++++
 include/linux/unwind_deferred_types.h    |  18 ++
 include/linux/unwind_user.h              |  15 ++
 include/linux/unwind_user_types.h        |  35 ++++
 kernel/Makefile                          |   1 +
 kernel/bpf/stackmap.c                    |   4 +-
 kernel/events/callchain.c                |  38 ++--
 kernel/events/core.c                     |   7 +-
 kernel/fork.c                            |   4 +
 kernel/unwind/Makefile                   |   1 +
 kernel/unwind/deferred.c                 | 349 +++++++++++++++++++++++++++++++
 kernel/unwind/user.c                     | 130 ++++++++++++
 26 files changed, 815 insertions(+), 65 deletions(-)
 create mode 100644 arch/x86/include/asm/unwind_user.h
 create mode 100644 arch/x86/include/asm/unwind_user_types.h
 create mode 100644 include/asm-generic/unwind_user.h
 create mode 100644 include/asm-generic/unwind_user_types.h
 create mode 100644 include/linux/unwind_deferred.h
 create mode 100644 include/linux/unwind_deferred_types.h
 create mode 100644 include/linux/unwind_user.h
 create mode 100644 include/linux/unwind_user_types.h
 create mode 100644 kernel/unwind/Makefile
 create mode 100644 kernel/unwind/deferred.c
 create mode 100644 kernel/unwind/user.c

* Steven Rostedt <rostedt@goodmis.org> wrote:

>       unwind_user: Add user space unwinding API
>       unwind_user: Add frame pointer support
>       unwind_user/x86: Enable frame pointer unwinding on x86
>       perf/x86: Rename and move get_segment_base() and make it global
>       unwind_user: Add compat mode frame pointer support
>       unwind_user/x86: Enable compat mode frame pointer unwinding on x86
>       unwind_user/deferred: Add unwind cache

What is the cost of 'caching' here? Will we double-buffer the tracing 
data before it reaches its single primary tooling user, with no use of 
any actual 'caching', which will be scenario in like 99.9% of the 
everyday usecases when this facility is used?

>       unwind_user/deferred: Add deferred unwinding interface
>       unwind_user/deferred: Make unwind deferral requests NMI-safe
>       perf: Remove get_perf_callchain() init_nr argument
>       perf: Have get_perf_callchain() return NULL if crosstask and user are set
>       perf: Simplify get_perf_callchain() user logic
>       perf: Skip user unwind if the task is a kernel thread.

Please don't leave periods in titles.

Thanks,

	Ingo

On Sun, 4 May 2025 11:41:56 +0200
Ingo Molnar <mingo@kernel.org> wrote:

> * Steven Rostedt <rostedt@goodmis.org> wrote:
> 
> >       unwind_user: Add user space unwinding API
> >       unwind_user: Add frame pointer support
> >       unwind_user/x86: Enable frame pointer unwinding on x86
> >       perf/x86: Rename and move get_segment_base() and make it global
> >       unwind_user: Add compat mode frame pointer support
> >       unwind_user/x86: Enable compat mode frame pointer unwinding on x86
> >       unwind_user/deferred: Add unwind cache  
> 
> What is the cost of 'caching' here? Will we double-buffer the tracing 
> data before it reaches its single primary tooling user, with no use of 
> any actual 'caching', which will be scenario in like 99.9% of the 
> everyday usecases when this facility is used?

I'm sorry, I may not understand the question here.

The cache doesn't add any extra buffer. The previous patch (Add unwind
deferred trace) allocates "entries" the first time a trace is done to
save the user stacktrace into the buffer. It will not free the entries
(until exit of the task) to save from having to allocate the entries
again.

If for some reason an interrupt happens while it is recording the trace
and the interrupt requests another trace, without the cache, it will do
the work of walking the user stack trace again.

The "cache" code, simply keeps information around to know that the
current trace is still valid, and that it doesn't need to do the work
of walking the user stack to produce the stack again.

> 
> >       unwind_user/deferred: Add deferred unwinding interface
> >       unwind_user/deferred: Make unwind deferral requests NMI-safe
> >       perf: Remove get_perf_callchain() init_nr argument
> >       perf: Have get_perf_callchain() return NULL if crosstask and user are set
> >       perf: Simplify get_perf_callchain() user logic
> >       perf: Skip user unwind if the task is a kernel thread.  
> 
> Please don't leave periods in titles.

OK, will fix.

Thanks for looking at this Ingo!

-- Steve