linux-kernel - Re: [PATCH v6 1/2] platform: Add driver for RAVE Supervisory Processor

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: <4eca3066-8194-6fae-fec5-498b94347a65@cogentembedded.com>
Date:   Thu, 31 Aug 2017 12:01:01 +0300
From:   Nikita Yushchenko <nikita.yoush@...entembedded.com>
To:     Pavel Machek <pavel@....cz>
Cc:     Andrey Smirnov <andrew.smirnov@...il.com>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        Chris Healy <cphealy@...il.com>,
        Lucas Stach <l.stach@...gutronix.de>,
        Lee Jones <lee.jones@...aro.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jeff White <Jeff.White@....aero>
Subject: Re: [PATCH v6 1/2] platform: Add driver for RAVE Supervisory
 Processor

>> I think that trying to make this generic is purely synthetic. This
>> information is board-specific per it's nature, it comes from what board
>> is designed for, different boards have quite different sets of possible
>> reset reasons. What is needed is - pass this board-specific information
>> to board-specific user space.
>>
>> What's proper API for that, if not a sysfs attribute?
> 
> Please go through the thread.
> 
> Sysfs attribute is okay, but:
> 
> 1) it should probably be a string
> 
> 2) it should certainly be superset of all the reasons
> 
> 3) it should be in generic place, say /sys/power/reset_reason
> 
> 4) it should be documented what each state means

What I'm concerned here is that a requirement appears for kernel driver
to keep and maintain knowledge of what all that codes mean. For me, it
looks like information locality breakage. Information in question is
application-specific, it definitely changes from board to board and
probably will change for particular board over time.

It's desirable to keep this information application-private, and have
kernel only passed it from hardware to application, unmodified and
uninterpreted.

Requiring kernel driver to interpret application-specific information
only increases complexity (i.e. any changes will have to be maintained
in two places - in driver and in application).

So question is - is there any proper API to communicate
application-private information from hardware through kernel to
userspace without any in-kernel interpretation?