Re: [PATCH] x86/oprofile: disable preemption in nmi_shutdown

[Johannes Weiner <hannes@saeurebad.de>]

Hi Vegard,

Vegard Nossum <vegard.nossum@gmail.com> writes:

> Hi,
>
> Does this look correct? I didn't really play with preemption before, but
> as far as I can tell, this is the right thing to do.
>
> I don't really get why model->shutdown(msrs) is done only for one of the
> CPUs, but my patch assumes that this is correct. (If that had been done
> from inside nmi_shutdown() for each CPU, we wouldn't have had to get the
> cpu var, and not needed to disable preemption.)
>
> Please comment :-)
>
>
> Vegard
>
>
> From: Vegard Nossum <vegard.nossum@gmail.com>
> Date: Sat, 21 Jun 2008 23:44:19 +0200
> Subject: [PATCH] x86/oprofile: disable preemption in nmi_shutdown
>
>     BUG: using smp_processor_id() in preemptible [00000000] code: oprofiled/27301
>     caller is nmi_shutdown+0x11/0x60
>     Pid: 27301, comm: oprofiled Not tainted 2.6.26-rc7 #25
>      [<c028a90d>] debug_smp_processor_id+0xbd/0xc0
>      [<c045fba1>] nmi_shutdown+0x11/0x60
>      [<c045dd4a>] oprofile_shutdown+0x2a/0x60
>
> Note that we don't need this for the other functions, since they are all
> called with on_each_cpu() (which disables preemption for us anyway).
>
> Signed-off-by: Vegard Nossum <vegard.nossum@gmail.com>
> ---
>  arch/x86/oprofile/nmi_int.c |    6 +++++-
>  1 files changed, 5 insertions(+), 1 deletions(-)
>
> diff --git a/arch/x86/oprofile/nmi_int.c b/arch/x86/oprofile/nmi_int.c
> index cc48d3f..4a177b4 100644
> --- a/arch/x86/oprofile/nmi_int.c
> +++ b/arch/x86/oprofile/nmi_int.c
> @@ -269,12 +269,16 @@ static void nmi_cpu_shutdown(void *dummy)
>  
>  static void nmi_shutdown(void)
>  {
> -	struct op_msrs *msrs = &__get_cpu_var(cpu_msrs);
> +	struct op_msrs *msrs;
> +
> +	preempt_disable();
> +	msrs = &__get_cpu_var(cpu_msrs);
>  	nmi_enabled = 0;
>  	on_each_cpu(nmi_cpu_shutdown, NULL, 0, 1);
>  	unregister_die_notifier(&profile_exceptions_nb);
>  	model->shutdown(msrs);
>  	free_msrs();
> +	preempt_enable();

Have a look at get_cpu_var() and put_cpu_var(), that is exactly the
pattern.

	Hannes