This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: systemtap 2.2.1 installcheck => kernel BUG at .. kprobes.c:707
- From: Masami Hiramatsu <masami dot hiramatsu dot pt at hitachi dot com>
- To: "Frank Ch. Eigler" <fche at redhat dot com>
- Cc: Timo Juhani Lindfors <timo dot lindfors at iki dot fi>, systemtap at sourceware dot org, srostedt at redhat dot com, "yrl dot pp-manager dot tt at hitachi dot com" <yrl dot pp-manager dot tt at hitachi dot com>
- Date: Thu, 23 May 2013 11:01:47 +0900
- Subject: Re: systemtap 2.2.1 installcheck => kernel BUG at .. kprobes.c:707
- References: <84obc93lq6 dot fsf at sauna dot l dot org> <y0m61yhn6ff dot fsf at fche dot csb> <519ADFB6 dot 9090409 at hitachi dot com> <20130522142149 dot GK28935 at redhat dot com>
(2013/05/22 23:21), Frank Ch. Eigler wrote:
> Hi, Masami -
>
>> I've found the root cause of this bug at last.
>
> Thanks for looking into it!
>
>> [...]
>> This shouldn't happen if the probe unregistration is done AFTER
>> kprobes_module_callback is called (because at that time the op
>> is gone), and kprobe-tracer does it.
>> So, ftrace never gets this bug, but systemtap does.
>
> I don't quite understand. In systemtap, as per runtime/transport/transport.c,
> we use a notifier priority of 1 for the _stp_module_notifier, so that we get
> notified after the kprobes_module_callback. I believe this behavior matches
> ftrace.
OK, even though, a probe have been removed right before the target module
is unloaded, it also hits the same problem. So, could you try to apply
the patch I posted on LKML and this ML yesterday?
https://lkml.org/lkml/2013/5/22/172
I hope it fixed the bug ...
Thank you,
--
Masami HIRAMATSU
IT Management Research Dept. Linux Technology Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: masami.hiramatsu.pt@hitachi.com