Bug 219302

Summary:	Question about trace-cmd reset
Product:	Tracing/Profiling	Reporter:	Metin Kaya (metin.kaya)
Component:	Ftrace	Assignee:	Steven Rostedt (rostedt)
Status:	RESOLVED CODE_FIX
Severity:	normal	CC:	metin.kaya
Priority:	P3
Hardware:	All
OS:	Linux
Kernel Version:		Subsystem:
Regression:	No	Bisected commit-id:

Description Metin Kaya 2024-09-23 12:06:24 UTC

Hello,

Running "trace-cmd reset" command destroys existing kprobe events:

```
# echo 'p do_sys_open' > /sys/kernel/tracing/kprobe_events
# echo 1 > /sys/kernel/tracing/events/kprobes/p_do_sys_open_0/enable
# cat /sys/kernel/tracing/kprobe_events
p:kprobes/p_do_sys_open_0 do_sys_open
# trace-cmd reset
# cat /sys/kernel/tracing/kprobe_events
#
```

I wonder if introducing a new flag to preserve kprobe events during the reset operation would make sense at all.

Thanks,

Comment 1 Steven Rostedt 2024-09-26 09:51:19 UTC

Yes it does make sense.

I'm open to taking patches that add options to trace-cmd reset that limit what it does. We should update libtracecmd's reset option too.

Comment 2 Metin Kaya 2024-09-26 10:18:17 UTC

Gotcha. Will send patch(es).
Thank you!

Comment 3 Metin Kaya 2024-10-09 15:40:56 UTC

> We should update libtracecmd's reset option too.

Hi Steven,

Could you please share a pointer about reset option in libtracecmd?

I implemented necessary changes, but only needed to modify clear_all_dynamic_events() and trace_reset() functions in trace-record.c.

Thanks,

Comment 4 Metin Kaya 2024-10-10 08:56:17 UTC

Just for the records: posted these patches on the mailing list [1].

[1] https://lore.kernel.org/linux-trace-devel/20241010085350.3479149-1-metin.kaya@arm.com/T/#t

Comment 5 Metin Kaya 2024-10-15 08:39:08 UTC

v2 [1] of the patch set is merged. Resolving the ticket.

[1] https://lore.kernel.org/linux-trace-devel/20241014123136.3890807-1-metin.kaya@arm.com/#t