This is an RFC patch set that also affects kprobes and perf.
At the Linux Collaboration Summit, I talked with Mathieu and others about
lowering the footprint of trace events. I spent all of last week
trying to get the size as small as I could.
Currently, each TRACE_EVENT() macro adds 1 - 5K per tracepoint. I got various
results by adding a TRACE_EVENT() with the compiler, depending on
config options that did not seem related. The new tracepoint I added
would add between 1 and 5K, but I did not investigate enough to
see what the true size was.
What was consistent, was the DEFINE_EVENT(). Currently, it adds
a little over 700 bytes per DEFINE_EVENT().
This patch series does not seem to affect TRACE_EVENT() much (had
the same various sizes), but consistently brings DEFINE_EVENT()s
down from 700 bytes to 250 bytes per DEFINE_EVENT(). Since syscalls
use one "class" and are equivalent to DEFINE_EVENT() this can
be a significant savings.
With events and syscalls (82 events and 616 syscalls), before this
patch series, the size of vmlinux was: 16161794, and afterward: 16058182.
That is 103,612 bytes in savings! (over 100K)
Without tracing syscalls (82 events), it brought the size of vmlinux
down from 1591046 to 15999394.
22,071 bytes in savings.
This is just an RFC (for now), to get peoples opinions on the changes.
It does a bit of rewriting of the CPP macros, just to warning you ;-)
-- Steve
The code can be found at:
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
tip/tracing/rfc-1
Steven Rostedt (10):
tracing: Create class struct for events
tracing: Let tracepoints have data passed to tracepoint callbacks
tracing: Convert TRACE_EVENT() to use the DECLARE_TRACE_DATA()
tracing: Remove per event trace registering
tracing: Move fields from event to class structure
tracing: Move raw_init from events to class
tracing: Allow events to share their print functions
tracing: Move print functions into event class
tracing: Remove duplicate id information in event structure
tracing: Combine event filter_active and enable into single flags field
----
include/linux/ftrace_event.h | 71 +++++++++---
include/linux/syscalls.h | 55 +++-------
include/linux/tracepoint.h | 119 ++++++++++++++++---
include/trace/ftrace.h | 215 ++++++++++------------------------
include/trace/syscall.h | 9 +-
kernel/trace/blktrace.c | 13 ++-
kernel/trace/kmemtrace.c | 28 +++--
kernel/trace/trace.c | 9 +-
kernel/trace/trace.h | 5 +-
kernel/trace/trace_event_perf.c | 17 ++-
kernel/trace/trace_events.c | 126 +++++++++++++-------
kernel/trace/trace_events_filter.c | 28 +++--
kernel/trace/trace_export.c | 16 ++-
kernel/trace/trace_functions_graph.c | 2 +-
kernel/trace/trace_kprobe.c | 104 ++++++++++-------
kernel/trace/trace_output.c | 137 +++++++++++++++-------
kernel/trace/trace_output.h | 2 +-
kernel/trace/trace_syscalls.c | 105 +++++++++++++++--
kernel/tracepoint.c | 91 ++++++++-------
19 files changed, 700 insertions(+), 452 deletions(-)
Steven Rostedt wrote:
> This is an RFC patch set that also affects kprobes and perf.
>
> At the Linux Collaboration Summit, I talked with Mathieu and others about
> lowering the footprint of trace events. I spent all of last week
> trying to get the size as small as I could.
>
> Currently, each TRACE_EVENT() macro adds 1 - 5K per tracepoint. I got various
> results by adding a TRACE_EVENT() with the compiler, depending on
> config options that did not seem related. The new tracepoint I added
> would add between 1 and 5K, but I did not investigate enough to
> see what the true size was.
>
> What was consistent, was the DEFINE_EVENT(). Currently, it adds
> a little over 700 bytes per DEFINE_EVENT().
>
> This patch series does not seem to affect TRACE_EVENT() much (had
> the same various sizes), but consistently brings DEFINE_EVENT()s
> down from 700 bytes to 250 bytes per DEFINE_EVENT(). Since syscalls
> use one "class" and are equivalent to DEFINE_EVENT() this can
> be a significant savings.
>
> With events and syscalls (82 events and 616 syscalls), before this
> patch series, the size of vmlinux was: 16161794, and afterward: 16058182.
>
> That is 103,612 bytes in savings! (over 100K)
>
>
> Without tracing syscalls (82 events), it brought the size of vmlinux
> down from 1591046 to 15999394.
>
> 22,071 bytes in savings.
>
> This is just an RFC (for now), to get peoples opinions on the changes.
> It does a bit of rewriting of the CPP macros, just to warning you ;-)
Hm, at least for kprobe tracer, this change is ok,
even though it isn't much worth as for tracepoints.
I think, if we have ftrace_event_ops, it will help reducing
the size of dynamic events a bit (not so much).
Thank you,
--
Masami Hiramatsu
e-mail: [email protected]
* Steven Rostedt ([email protected]) wrote:
> This is an RFC patch set that also affects kprobes and perf.
Hi Steven,
>
> At the Linux Collaboration Summit, I talked with Mathieu and others about
> lowering the footprint of trace events. I spent all of last week
> trying to get the size as small as I could.
>
> Currently, each TRACE_EVENT() macro adds 1 - 5K per tracepoint. I got various
> results by adding a TRACE_EVENT() with the compiler, depending on
> config options that did not seem related. The new tracepoint I added
> would add between 1 and 5K, but I did not investigate enough to
> see what the true size was.
Adding only one might not give an accurate picture, as some sections can
be aligned on 4k boundaries. So if the added TRACE_EVENT() brings you to
the next 4k page, your added size is bumped of an extra 4k.
>
> What was consistent, was the DEFINE_EVENT(). Currently, it adds
> a little over 700 bytes per DEFINE_EVENT().
>
> This patch series does not seem to affect TRACE_EVENT() much (had
> the same various sizes), but consistently brings DEFINE_EVENT()s
> down from 700 bytes to 250 bytes per DEFINE_EVENT(). Since syscalls
> use one "class" and are equivalent to DEFINE_EVENT() this can
> be a significant savings.
>
> With events and syscalls (82 events and 616 syscalls), before this
> patch series, the size of vmlinux was: 16161794, and afterward: 16058182.
>
> That is 103,612 bytes in savings! (over 100K)
>
>
> Without tracing syscalls (82 events), it brought the size of vmlinux
> down from 1591046 to 15999394.
Probably a cut n paste error on the line above. Should read:
down from 16021465 to 15999394.
if you want the "22071 bytes in savings" to hold.
Will look over your patchset.
Thanks,
Mathieu
>
> 22,071 bytes in savings.
>
> This is just an RFC (for now), to get peoples opinions on the changes.
> It does a bit of rewriting of the CPP macros, just to warning you ;-)
>
> -- Steve
>
> The code can be found at:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
> tip/tracing/rfc-1
>
>
> Steven Rostedt (10):
> tracing: Create class struct for events
> tracing: Let tracepoints have data passed to tracepoint callbacks
> tracing: Convert TRACE_EVENT() to use the DECLARE_TRACE_DATA()
> tracing: Remove per event trace registering
> tracing: Move fields from event to class structure
> tracing: Move raw_init from events to class
> tracing: Allow events to share their print functions
> tracing: Move print functions into event class
> tracing: Remove duplicate id information in event structure
> tracing: Combine event filter_active and enable into single flags field
>
> ----
> include/linux/ftrace_event.h | 71 +++++++++---
> include/linux/syscalls.h | 55 +++-------
> include/linux/tracepoint.h | 119 ++++++++++++++++---
> include/trace/ftrace.h | 215 ++++++++++------------------------
> include/trace/syscall.h | 9 +-
> kernel/trace/blktrace.c | 13 ++-
> kernel/trace/kmemtrace.c | 28 +++--
> kernel/trace/trace.c | 9 +-
> kernel/trace/trace.h | 5 +-
> kernel/trace/trace_event_perf.c | 17 ++-
> kernel/trace/trace_events.c | 126 +++++++++++++-------
> kernel/trace/trace_events_filter.c | 28 +++--
> kernel/trace/trace_export.c | 16 ++-
> kernel/trace/trace_functions_graph.c | 2 +-
> kernel/trace/trace_kprobe.c | 104 ++++++++++-------
> kernel/trace/trace_output.c | 137 +++++++++++++++-------
> kernel/trace/trace_output.h | 2 +-
> kernel/trace/trace_syscalls.c | 105 +++++++++++++++--
> kernel/tracepoint.c | 91 ++++++++-------
> 19 files changed, 700 insertions(+), 452 deletions(-)
>
--
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com