Re: [PATCH v7 1/4] spinlock: A new lockref structure for locklessupdate of refcount

[Linus Torvalds]

On Fri, Aug 30, 2013 at 10:11 AM, Sedat Dilek <sedat.dilek@xxxxxxxxx> wrote:
>
> There is no -f option for record but for report, so I swapped them:

That's odd. "perf record -f" very much works for me, and I use it all
the time to make sure that I don't mix perf data from older runs..

[ Time passes. I check current 'perf'. Dammit. This has changed. I
seldom rebuild "perf", so I never noticed. It was removed in commit
4a4d371a4dfb, apparently because the old "append" mode no longer even
exists ]

But never mind, that doesn't matter for your numbers:

> $ sudo ~/src/linux-kernel/linux/tools/perf/perf report -f
>
> Samples: 159K of event 'cycles:pp', Event count (approx.): 76535356682
>  84,10%  t_lockref_from-  [kernel.kallsyms]     [k] check_poison_obj
>   5,22%  t_lockref_from-  [kernel.kallsyms]     [k] memset
>   1,15%  t_lockref_from-  [kernel.kallsyms]     [k] irq_return
>   0,45%  t_lockref_from-  [kernel.kallsyms]     [k] kmem_cache_alloc
>   0,44%  t_lockref_from-  [kernel.kallsyms]     [k] kmem_cache_free

You're wasting all the time in slab debugging, probably due to the
object debug option.

None of the rest is even remotely interesting, and the spinlock hits
you do have are more likely to come from the same object debug code
than from pathname lookup.

                Linus
--
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/