Re: [PATCH 4/4] arm64: implement KPROBES_ON_FTRACE

2019-08-19 Thread Jisheng Zhang
On Mon, 19 Aug 2019 22:22:12 +0530 "Naveen N. Rao" wrote:


> 
> 
> Jisheng Zhang wrote:
> > This patch implements KPROBES_ON_FTRACE for arm64.
> >
> > ~ # mount -t debugfs debugfs /sys/kernel/debug/
> > ~ # cd /sys/kernel/debug/
> > /sys/kernel/debug # echo 'p _do_fork' > tracing/kprobe_events
> >
> > before the patch:
> >
> > /sys/kernel/debug # cat kprobes/list
> > ff801009ff7c  k  _do_fork+0x4[DISABLED]  
> 
> This looks wrong -- we should not be allowing kprobe to be registered on

Yes. I made a mistake when dumping this log. The kernel isn't as clean
as "before the patch".


> ftrace address without KPROBES_ON_FTRACE. Is _do_fork+0x4 the location
> of ftrace entry on arm64?

Indeed, w/o KPROBES_ON_FTRACE, it should be _do_fork+0x0

Thanks




Re: [PATCH 4/4] arm64: implement KPROBES_ON_FTRACE

2019-08-19 Thread Naveen N. Rao

Jisheng Zhang wrote:

This patch implements KPROBES_ON_FTRACE for arm64.

~ # mount -t debugfs debugfs /sys/kernel/debug/
~ # cd /sys/kernel/debug/
/sys/kernel/debug # echo 'p _do_fork' > tracing/kprobe_events

before the patch:

/sys/kernel/debug # cat kprobes/list
ff801009ff7c  k  _do_fork+0x4[DISABLED]


This looks wrong -- we should not be allowing kprobe to be registered on 
ftrace address without KPROBES_ON_FTRACE. Is _do_fork+0x4 the location 
of ftrace entry on arm64?


- Naveen