linux-kernel - Re: [PATCH v1 2/2] thermal: sysfs: Eliminate unnecessary zone locking

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:   Fri, 1 Dec 2023 12:05:01 +0100
From:   "Rafael J. Wysocki" <rafael@...nel.org>
To:     Daniel Lezcano <daniel.lezcano@...aro.org>
Cc:     "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Lukasz Luba <lukasz.luba@....com>,
        Linux PM <linux-pm@...r.kernel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Srinivas Pandruvada <srinivas.pandruvada@...ux.intel.com>,
        Zhang Rui <rui.zhang@...el.com>
Subject: Re: [PATCH v1 2/2] thermal: sysfs: Eliminate unnecessary zone locking

On Fri, Dec 1, 2023 at 10:16 AM Daniel Lezcano
<daniel.lezcano@...aro.org> wrote:
>
> On 30/11/2023 19:37, Rafael J. Wysocki wrote:
> > From: Rafael J. Wysocki <rafael.j.wysocki@...el.com>
> >
> > The _show() callback functions of the trip point sysfs attributes,
> > temperature, hysteresis and type, need not use thermal zone locking,
> > because the layout of the data structures they access does not change
> > after the thermal zone registration.
> >
> > Namely, they all need to access a specific entry in the thermal
> > zone's trips[] table that is always present for non-tripless thermal
> > zones and its size cannot change after the thermal zone has been
> > registered.  Thus it is always safe to access the trips[] table of a
> > registered thermal zone from each of the sysfs attributes in question.
> >
> > Moreover, the type of a trip point does not change after registering its
> > thermal zone, and while its temperature and hysteresis can change, for
> > example due to a firmware-induced thermal zone update, holding the zone
> > lock around reading them is pointless, because it does not prevent stale
> > values from being returned to user space.  For example, a trip point
> > temperature can always change ater trip_point_temp_show() has read it
> > and before the function's return statement is executed, regardless of
> > whether or not zone locking is used.
> >
> > For this reason, drop the zone locking from trip_point_type_show(),
> > trip_point_temp_show(), and trip_point_hyst_show().
>
> Isn't the lock used to protect against device_del() from
> thermal_zone_device_unregister() ?

Ah, I missed the zone locking around device_del() in there.

OK, please scratch this series for now.