kernel event tracing support about uftrace

[Taeung]

I thought a trivial thing..
We can tell reviewers review/kernel-event-v1 branch compared to master branch,
instead of just review/kernel-event-v1. How about that ?

[namhyung]

I cannot understand.. could you please elaborate?

[Taeung]

When you tell reviewers branch name at the end of comment on this issue like below,

Pushed the code to review/kernel-event-v1 branch.

I thought it is better to tell reviewers a branch URL(i.e. master...review/kernel-event-v1) compared to master branch
instead of just review/kernel-event-v1 IMHO

[Taeung]

I found NG from the test 148 event_kernel but is it just due to difference of kernel version ?
Don't we need to consider it ?

$ uname -r
4.5.0-rc4+

$ sudo ./runtest.py -d event_kernel
Test case                 pg             finstrument-fu
------------------------: O0 O1 O2 O3 Os O0 O1 O2 O3 Os
t148_event_kernel: diff result of -pg -O0
--- expect	2017-06-20 23:56:55.544535059 +0900
+++ result	2017-06-20 23:56:55.544535059 +0900
@@ -7,2 +7,3 @@
        usleep() {
+         syscall_trace_enter_phase1();
          sys_nanosleep() {
@@ -10,2 +11,3 @@
          } /* sys_nanosleep */
+         syscall_slow_exit_work();
        } /* usleep */

[namhyung]

I thought it is better to tell reviewers a branch URL(i.e. master...review/kernel-event-v1) compared to master branch

Ah, ok. Thanks for the info! :)

I found NG from the test 148 event_kernel but is it just due to difference of kernel version ?

I guess it's because you enabled kernel syscall tracing?

[Taeung]

Ah.. I got it.. will recheck it.

[namhyung]

Pushed to review/kernel-event-v2.

Changelog:

• enabled kernel tracing without -k option
• setup event-fork option to follow children

[namhyung]

Pushed to review/kernel-event-v3.

Changelog:

• adjust print format in accordance with the user event
• clean-up error handing during record

[namhyung]

Pushed review/kernel-event-v4.

Changelog:

• support (partial) glob matching (e.g. "sched:sched_process_*")
• fixup pid filter update on fork/vfork
• do not imply kernel function tracing

[namhyung]

Pushed review/kernel-event-v5

Changelog:

• further fixup for pid filter update