Re: [PATCH 1/4] [BUGFIX] ftrace/kprobes: Fix not to delete probesif in use

[Masami Hiramatsu]

(2011/10/08 14:00), Steven Rostedt wrote:
> On Tue, 2011-10-04 at 19:44 +0900, Masami Hiramatsu wrote:
>> Fix kprobe-tracer not to delete a probe if the probe is in use.
>> In that case, delete operation will return -EBUSY.
>>
>> This bug can cause a kernel panic if enabled probes are deleted
>> during perf record.
>>
>> (Add some probes on functions)
>> # perf record -e probe:\* -aR sh
>> sh-4.2# perf probe --del probe:\*
>> sh-4.2# exit
>> (kernel panic)
> 
> 
> Hi Masami,
> 
> I was able to reproduce the panic. I'm currently running patch 1 and 4
> through my standard tests before pushing this out as urgent.
> 
> Does this exist in previous kernels? If so, I'll also add the stable tag
> to it.

I guess so, since the perf's tracepoint handler locks target module
while recording, it is required for perf not to remove undergo events.

> Could you resend the trivial patches when you make your necessary
> corrections. They are not important as these are and I'll add them to a
> separate queue.

Right, that's not a matter. The first patch should be merged soon.

Thank you,

> 
> Thanks!
> 
> -- Steve
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/


-- 
Masami HIRAMATSU
Software Platform Research Dept. Linux Technology Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: masami.hiramatsu.pt@xxxxxxxxxxx
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/