Re: [PATCH v3 7/7] clocksource/drivers/arm_arch_timer: validate arch_timer_rate

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Thomas,

On Friday 14 Feb 2020 at 01:35:58 (+0100), Thomas Gleixner wrote:
> Ionela Voinescu <ionela.voinescu@xxxxxxx> writes:
> 
> > From: Valentin Schneider <valentin.schneider@xxxxxxx>
> >
> > Using an arch timer with a frequency of less than 1MHz can result in an
> > incorrect functionality of the system which assumes a reasonable rate.
> >
> > One example is the use of activity monitors for frequency invariance
> > which uses the rate of the arch timer as the known rate of the constant
> > cycle counter in computing its ratio compared to the maximum frequency
> > of a CPU. For arch timer frequencies less than 1MHz this ratio could
> > end up being 0 which is an invalid value for its use.
> >
> > Therefore, warn if the arch timer rate is below 1MHz which contravenes
> > the recommended architecture interval of 1 to 50MHz.
> >
> > Signed-off-by: Ionela Voinescu <ionela.voinescu@xxxxxxx>
> 
> So this patch is from Valentin. Where is his Signed-off-by?
> 

Yes, sorry about this. This was based on a diff that Valentin provided
in v2. I'll change the author as agreed at:
https://lore.kernel.org/lkml/20200212103249.GA19041@xxxxxxx/

> >  
> > +static int validate_timer_rate(void)
> > +{
> > +	if (!arch_timer_rate)
> > +		return -EINVAL;
> > +
> > +	/* Arch timer frequency < 1MHz can cause trouble */
> > +	WARN_ON(arch_timer_rate < 1000000);
> 
> This does not make sense to me. If the rate is out of bounds then why
> warn an just continue instead of making it fail?
> 

Because it's not a hard restriction, it's just atypical for the rate to
be below 1Mhz. The spec only mentions a typical range of 1 to 50MHz and
the warning is only here to flag a potentially problematic rate, below
what is assumed typical in the spec.

In [1], where I'm actually relying on arch_timer_rate being higher than
than 1/SCHED_CAPACITY_SCALE² of the maximum frequency, I am making it
fail, as, for that scenario, it is a hard restriction.


+	 * We use a factor of 2 * SCHED_CAPACITY_SHIFT -> SCHED_CAPACITY_SCALE²
+	 * in order to ensure a good resolution for arch_max_freq_scale for
+	 * very low arch timer frequencies (up to the KHz range which should be
+	 * unlikely).
+	 */
+	ratio = (u64)arch_timer_get_rate() << (2 * SCHED_CAPACITY_SHIFT);
+	ratio = div64_u64(ratio, max_freq_hz);
+	if (!ratio) {
+		pr_err("System timer frequency too low.\n");
+		return -EINVAL;
+	}
+

[1] https://lore.kernel.org/lkml/89339501-5ee4-e871-3076-c8b02c6fbf6e@xxxxxxx/

Thanks,
Ionela.

> Thanks,
> 
>         tglx



[Index of Archives]     [Kernel Newbies]     [Security]     [Netfilter]     [Bugtraq]     [Linux FS]     [Yosemite Forum]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Samba]     [Video 4 Linux]     [Device Mapper]     [Linux Resources]

  Powered by Linux