[PATCH v3 00/12] tracing: Add support for in-kernel dynamic event API

From: Tom Zanussi
Date: Fri Jan 24 2020 - 17:56:33 EST


Hi,

This is v3 of 'tracing: Add support for in-kernel dynamic event API',
incorporating changes based on suggestions from Masami and Steven.

- Rebased to trace/for-next

- Regularized the entire API to use synth_event_*, kprobe_event_*,
dynevent_*, and added some new macros and functions to make things
more consistent

- Introduced trace_array_find_get() and used it in
trace_array_get_file() as suggested

- Removed exports from dynevent_cmd functions that didn't need to be
exported

- Switched the param order of __kprobe_event_gen_cmd_start() to fix
a problem found building with clang. Apparently varargs and
implicit promotion of types like bool don't mix. Thanks to Nick
Desaulniers for pointing that out.

- Updated the documentation for all of the above

Text from the v2 posting:

This is v2 of the previous 'tracing: Add support for in-kernel
synthetic event API', and is largely a rewrite based on suggestions
from Masami to expand the scope to include other types of dynamic
events such as kprobe events, in addition to the original sythetic
event focus.

The functionality of the original API remains available, though it's
in a slightly different form due to the use of the new dynevent_cmd
API it's now based on. The dynevent_cmd API provides a common dynamic
event command generation capability that both the synthetic event API
and the kprobe event API are now based on, and there are now test
modules demonstrating both the synthetic event and kprobes APIs.

A couple of the patches are snippets from Masami's 'tracing:
bootconfig: Boot-time tracing and Extra boot config' series, and the
patch implementing the dynevent_cmd API includes some of the
spnprintf() generating code from that patchset.

Because I used Masami's gen_*_cmd() naming suggestion for generating
the commands, the previous patchset's generate_*() functions were
renamed to trace_*() to avoid confusion, and probably is better naming
anyway.

An overview of the user-visible changes in comparison to v1:

- create_synth_event() using an array of synth_desc_fields remains
unchanged and works the same way as previously

- gen_synth_cmd() takes a variable-length number of args which
represent 'type field;' pairs. Using this with no field args
basically replaces the previous 'create_empty_synth_event()'

- The 'add_synth_field()' and 'add_synth_fields()' function from v1
are essentially the same except that they now take a dynevent_cmd
instead of a synth_event pointer

- The finalize_synth_event() from v1 is replaced by
create_dynevent() in the new version.

- The new v2 API includes some additional functions to initialize
the dynevent_cmd - synth_dynevent_cmd() is used to do that. While
it's an extra step, it makes it easier to do error handling.

- There's a new trace_synth_event() function that traces a synthetic
event using a variable-arg list of values.

- The original generate_synth_event() using an array of values is
now called trace_synth_event_array().

- For piecewise event tracing, the original
generate_synth_event_start() and generate_synth_event_end() have
now been renamed to trace_synth_event_end().

- add_next_synth_val() and add_synth_val() remain the same.

- A similar API and test module demonstrating the API has been added
for kprobe events

- Both the synthetic events and kprobe events API is based on the
dynevent_cmd API, newly added

- The Documentation for all of the above has been updated

Text from the orginal v1 posting:

I've recently had several requests and suggestions from users to add
support for the creation and generation of synthetic events from
kernel code such as modules, and not just from the available command
line commands.

This patchset adds support for that. The first three patches add some
minor preliminary setup, followed by the two main patches that add the
ability to create and generate synthetic events from the kernel. The
next patch adds a test module that demonstrates actual use of the API
and verifies that it works as intended, followed by Documentation.

Special thanks to Artem Bityutskiy, who worked with me over several
iterations of the API, and who had many great suggestions on the
details of the interface, and pointed out several problems with the
code itself.

The following changes since commit 659ded30272d67a04b3692f0bfa12263be20d790:

trace/kprobe: Remove unused MAX_KPROBE_CMDLINE_SIZE (2020-01-22 07:07:38 -0500)

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/zanussi/linux-trace.git ftrace/synth-event-gen-v3

Tom Zanussi (12):
tracing: Add trace_array_find/_get() to find instance trace arrays
tracing: Add trace_get/put_event_file()
tracing: Add synth_event_delete()
tracing: Add dynamic event command creation interface
tracing: Add synthetic event command generation functions
tracing: Change trace_boot to use synth_event interface
tracing: Add synth_event_trace() and related functions
tracing: Add synth event generation test module
tracing: Add kprobe event command generation functions
tracing: Change trace_boot to use kprobe_event interface
tracing: Add kprobe event command generation test module
tracing: Documentation for in-kernel synthetic event API

Documentation/trace/events.rst | 515 ++++++++++++++++++++
include/linux/trace_events.h | 124 +++++
kernel/trace/Kconfig | 25 +
kernel/trace/Makefile | 2 +
kernel/trace/kprobe_event_gen_test.c | 225 +++++++++
kernel/trace/synth_event_gen_test.c | 523 ++++++++++++++++++++
kernel/trace/trace.c | 43 +-
kernel/trace/trace.h | 36 ++
kernel/trace/trace_boot.c | 66 ++-
kernel/trace/trace_events.c | 325 +++++++++++++
kernel/trace/trace_events_hist.c | 896 ++++++++++++++++++++++++++++++++++-
kernel/trace/trace_kprobe.c | 160 ++++++-
12 files changed, 2868 insertions(+), 72 deletions(-)
create mode 100644 kernel/trace/kprobe_event_gen_test.c
create mode 100644 kernel/trace/synth_event_gen_test.c

--
2.14.1