linux-kernel - Re: [PATCH] dt-bindings: arm: Fix cpu capacity mismatch in example

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

Message-ID: <20181029063450.3o3ay7pazik5tgne@vireshk-i7>
Date:   Mon, 29 Oct 2018 12:04:50 +0530
From:   Viresh Kumar <viresh.kumar@...aro.org>
To:     Daniel Lezcano <daniel.lezcano@...aro.org>
Cc:     catalin.marinas@....com, Rob Herring <robh+dt@...nel.org>,
        Mark Rutland <mark.rutland@....com>,
        linux-arm-kernel@...ts.infradead.org,
        Vincent Guittot <vincent.guittot@...aro.org>,
        devicetree@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] dt-bindings: arm: Fix cpu capacity mismatch in example

On 26-10-18, 10:30, Daniel Lezcano wrote:
> On 26/10/2018 06:11, Viresh Kumar wrote:
> > On 25-10-18, 14:04, Daniel Lezcano wrote:
> >> I think it is actually correct. The example is confusing on what the
> >> numbers are. IIUC, it is:
> >>
> >> (after normalizing)
> >>
> >> dhrystone result on big CPU is 1024 at 1100MHz
> >> dhrystone result on little CPU is 446 at 850MHz
> >>
> >> We have to scale the result of the little for 1100MHz in order to compare.
> >>
> >> 1100/850 = 1.294
> >>
> >> dhrystone result on little CPU is 446 * 1.294 = 577 at 1100MHz
> >>
> >> So we put the normalized values 1024 and 577. The arch_topology will
> >> scale 577 back to 446 as it will compute the max capacity based on the
> >> max freq which is 850MHz.
> >>
> >> The *final* capacities are 1024 for cluster0 and 446 for cluster1 (read
> >> the cpu max capacity, the ones showed in the sysfs).
> >>
> >> Did I miss something ?
> > 
> > No. What about making it more clear in the example to save the next idiot (like
> > me) from wasting time :)
> > 
> > diff --git a/Documentation/devicetree/bindings/arm/cpu-capacity.txt b/Documentation/devicetree/bindings/arm/cpu-capacity.txt
> > index 9b5685a1d15d..d061e6575bde 100644
> > --- a/Documentation/devicetree/bindings/arm/cpu-capacity.txt
> > +++ b/Documentation/devicetree/bindings/arm/cpu-capacity.txt
> > @@ -61,7 +61,10 @@ mhz values (normalized w.r.t. the highest value found while parsing the DT).
> >  Example 1 (ARM 64-bit, 6-cpu system, two clusters):
> >  capacities-dmips-mhz are scaled w.r.t. 1024 (cpu@0 and cpu@1)
> >  supposing cluster0@...-freq=1100 and custer1@...-freq=850,
> > -final capacities are 1024 for cluster0 and 446 for cluster1
> > +final capacities are 1024 for cluster0 and 446 for cluster1.
> > +Note that the values mentioned below in the example (1024 and 578)
> > +aren't normalized based on max frequency of each cluster and that is
> > +left for the operating system to do.
> 
> Yes, it will help but if you want to make things even more clear, I
> suggest to elaborate the text a bit and give the numbers above to
> explain (1100/850=1.294, 446 * 1.294 = 577, ...)

I didn't wanted to explain way too much, but how about this..

diff --git a/Documentation/devicetree/bindings/arm/cpu-capacity.txt b/Documentation/devicetree/bindings/arm/cpu-capacity.txt
index 9b5685a1d15d..84262cdb8d29 100644
--- a/Documentation/devicetree/bindings/arm/cpu-capacity.txt
+++ b/Documentation/devicetree/bindings/arm/cpu-capacity.txt
@@ -59,9 +59,11 @@ mhz values (normalized w.r.t. the highest value found while parsing the DT).
 ===========================================
 
 Example 1 (ARM 64-bit, 6-cpu system, two clusters):
-capacities-dmips-mhz are scaled w.r.t. 1024 (cpu@0 and cpu@1)
-supposing cluster0@...-freq=1100 and custer1@...-freq=850,
-final capacities are 1024 for cluster0 and 446 for cluster1
+The capacities-dmips-mhz or DMIPS/MHz values (scaled to 1024)
+are 1024 and 578 for cluster0 and cluster1. Further normalization
+is done by the operating system based on cluster0@...-freq=1100 and
+custer1@...-freq=850, final capacities are 1024 for cluster0 and
+446 for cluster1 (576*850/1100).
 
 cpus {
        #address-cells = <2>;


-- 
viresh