Re: [PATCH v5 3.1.0-rc4-tip 8/26] x86: analyze instruction anddetermine fixups.

[Masami Hiramatsu]

(2011/09/21 5:53), Stefan Hajnoczi wrote:
> On Tue, Sep 20, 2011 at 02:12:25PM -0400, Christoph Hellwig wrote:
>> On Tue, Sep 20, 2011 at 06:13:10PM +0100, Stefan Hajnoczi wrote:
>>> You've probably thought of this but it would be nice to skip XOL for
>>> nops.  This would be a common case with static probes (e.g. sdt.h) where
>>> the probe template includes a nop where we can easily plant int $0x3.
>>
>> Do we now have sdt.h support for uprobes?  That's one of the killer
>> features that always seemed to get postponed.
> 
> Not yet but it's a question of doing roughly what SystemTap does to
> parse the appropriate ELF sections and then putting those probes into
> uprobes.
> 
> Masami looked at this and found that SystemTap sdt.h currently requires
> an extra userspace memory store in order to activate probes.  Each probe
> has a "semaphore" 16-bit counter which applications may test before
> hitting the probe itself.  This is used to avoid overhead in
> applications that do expensive argument processing (e.g. creating
> strings) for probes.

Indeed, originally, those semaphores designed for such use cases.
However, some applications *always* use it (e.g. qemu-kvm).

> 
> But this should be solvable so it would be possible to use perf-probe(1)
> on a std.h-enabled binary.  Some distros already ship such binaries!

I'm not sure that we should stick on the current implementation
of the sdt.h. I think we'd better modify the sdt.h to replace
such semaphores with checking whether the tracepoint is changed from nop.

Or, we can introduce an add-hoc ptrace code to perftools for modifying
those semaphores. However, this means that user always has to use
perf to trace applications, and it's hard to trace multiple applications
at a time (can we attach all of them?)...

Thank you,

-- 
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/