linux-kernel - Re: [PATCH] cpufreq: Don't send callback pointer to cpufreq_add_update_util

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]

Message-ID: <6633140.4621GLIDvq@aspire.rjw.lan>
Date:   Fri, 18 Aug 2017 14:20:44 +0200
From:   "Rafael J. Wysocki" <rjw@...ysocki.net>
To:     Viresh Kumar <viresh.kumar@...aro.org>
Cc:     Srinivas Pandruvada <srinivas.pandruvada@...ux.intel.com>,
        Len Brown <lenb@...nel.org>, Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        linux-pm@...r.kernel.org,
        Vincent Guittot <vincent.guittot@...aro.org>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] cpufreq: Don't send callback pointer to cpufreq_add_update_util_hook()

On Friday, August 18, 2017 6:19:44 AM CEST Viresh Kumar wrote:
> On 17-08-17, 17:31, Rafael J. Wysocki wrote:
> > On Thursday, August 17, 2017 2:04:48 PM CEST Viresh Kumar wrote:
> > > The callers already have the structure (struct update_util_data) where
> > > the function pointer is saved by cpufreq_add_update_util_hook(). And its
> > > better if the callers fill it themselves, as they can do it from the
> > > governor->init() callback then, which is called only once per policy
> > > lifetime rather than doing it from governor->start which can get called
> > > multiple times.
> > 
> > So what problem exactly is this addressing?
> 
> Its not fixing any problem really, but is rather just a cleanup patch.
> I had a look at include/linux/sched/cpufreq.h and got confused for a
> moment:
> 
> struct update_util_data {
>        void (*func)(struct update_util_data *data, u64 time, unsigned int flags);
> };
> 
> void cpufreq_add_update_util_hook(int cpu, struct update_util_data *data,
>                        void (*func)(struct update_util_data *data, u64 time,
>                                     unsigned int flags));
> 
> 
> It wasn't quite straight-forward to understand why we needed to pass
> both "data" and "func", while "data" should already have "func" set
> within it. And then I realized that cpufreq_add_update_util_hook() is
> actually setting that field.
> 
> Filling the pointer from the callers is probably better because:
> - It makes it more readable.
> - We have to pass one less argument and the function prototype becomes
>   quite short.
> - The callers don't have to set the data->func pointer from the
>   governor->start() callback now and can do it only once from
>   governor->init(). ->start(), stop() callbacks can get called a lot,
>   for example with CPU hotplug.
> 
> But yeah, its all trivial stuff. No big problem solved.

Well, there is a reason to do it this way, at least for me.

If you want to look for all of the governor callbacks that can be
used with _update_util(), it is now sufficient to grep for
cpufreq_add_update_util_hook(), basically, but with the change you'd
need to find the initialization of the update_util structure in there
and see what function it points to.

Thanks,
Rafael