linux-kernel - Re: [PATCH 4/5] oprofile: Remove exit function for timer mode

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:	Wed, 19 Oct 2011 20:07:41 +0200
From:	Robert Richter <robert.richter@....com>
To:	Ingo Molnar <mingo@...e.hu>
CC:	LKML <linux-kernel@...r.kernel.org>,
	oprofile-list <oprofile-list@...ts.sourceforge.net>,
	Peter Zijlstra <a.p.zijlstra@...llo.nl>
Subject: Re: [PATCH 4/5] oprofile: Remove exit function for timer mode

See below for Some more comments after testing and reviewing the
code. Will send a new version with these comments incorporated.

-Robert

On 17.10.11 10:27:04, Robert Richter wrote:
> +static struct notifier_block __cpuinitdata nmi_timer_cpu_nb = {

The __cpuinitdata attribute causes a section mismatch, will fix it.

> +	.notifier_call = nmi_timer_cpu_notifier
> +};

> +static int nmi_timer_setup(void)
> +{
> +	int cpu, err;
> +
> +	/* clock cycles per tick: */
> +	nmi_timer_attr.sample_period = (u64)cpu_khz * TICK_NSEC / NSEC_PER_MSEC;

This is broken on 32 bit.

Will replace it with:

        /* clock cycles per tick: */
        period = (u64)cpu_khz * 1000;
        do_div(period, HZ);
        nmi_timer_attr.sample_period = period;

> +	pr_info("sample_period: %lld, cpu_khz: %d, nsec/tick: %ld\n",
> +		nmi_timer_attr.sample_period, cpu_khz, TICK_NSEC);

I will remove this pr_info().

> +
> +	get_online_cpus();
> +	err = register_cpu_notifier(&nmi_timer_cpu_nb);
> +	if (err)
> +		goto out;

> diff --git a/drivers/oprofile/oprof.c b/drivers/oprofile/oprof.c
> index f7cd069..af9cbea 100644
> --- a/drivers/oprofile/oprof.c
> +++ b/drivers/oprofile/oprof.c
> @@ -248,15 +248,21 @@ static int __init oprofile_init(void)
>  	/* always init architecture to setup backtrace support */
>  	err = oprofile_arch_init(&oprofile_ops);
>  
> -	timer_mode = err || timer;	/* fall back to timer mode on errors */
> +	timer_mode = err || timer;
>  	if (timer_mode) {
>  		if (!err)
>  			oprofile_arch_exit();
> -		err = oprofile_timer_init(&oprofile_ops);
> -		if (err)
> -			return err;
> +		/* no nmi timer mode if oprofile.timer is set: */
> +		if (!timer)
> +			err = op_nmi_timer_init(&oprofile_ops);

As nmi timer is initialized here, we can remove it in arch/x86. This
eases setup functions for x86.

> +		/* fall back to timer mode on errors: */
> +		if (err || timer)
> +			err = oprofile_timer_init(&oprofile_ops);
>  	}
>  
> +	if (err)
> +		return err;
> +
>  	err = oprofilefs_register();
>  	if (!err)
>  		return 0;

-- 
Advanced Micro Devices, Inc.
Operating System Research Center

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