[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20160208085025.GA30075@pali>
Date: Mon, 8 Feb 2016 09:50:25 +0100
From: Pali Rohár <pali.rohar@...il.com>
To: Sebastian Reichel <sre@...nel.org>
Cc: Tony Lindgren <tony@...mide.com>,
Benoît Cousson <bcousson@...libre.com>,
Aaro Koskinen <aaro.koskinen@....fi>,
Pavel Machek <pavel@....cz>, linux-omap@...r.kernel.org,
linux-kernel@...r.kernel.org,
Ivaylo Dimitrov <ivo.g.dimitrov.75@...il.com>
Subject: Re: [PATCH 4/5] HSI: nokia-modem: drop support for disabled pm
On Sunday 31 January 2016 17:10:52 Sebastian Reichel wrote:
> Hi,
>
> On Sun, Jan 31, 2016 at 12:24:34PM +0100, Pali Rohár wrote:
> > On Sunday 31 January 2016 02:19:46 Sebastian Reichel wrote:
> > > Disabled power management means, that the driver can only be
> > > used together with further out-of-tree kernel patches. There
> > > is no reason to support this in the mainline kernel and not
> > > having support for it means, that userspace can automatically
> > > detect if we are running kernel based power management.
> >
> > I'm against this change. It will break support for (patched) Maemo and
> > proprietary Nokia sscd modem daemon.
>
> Do you have some links for this? So far my assumption was, that sscd
> requires /sys/devices/platform/gpio-switch, so you either need to
> patch sscd somehow, or patch the kernel to provide gpio-switch, or
> provide some kind of userspace wrapper.
I can either patch kernel to provide gpio-switch interface or use some
kind of user space hacks to provide this interface (e.g mount bind or
fuse fs mounted to that location).
> > Also you can export GPIOs manually via sysfs and touch that state. Which
> > means that you can do power management with mainline kernel even if you
> > set nokia-modem.pm=0.
>
> Yeah... You export GPIOs manually. Did you actually try this? You
Yes, this is what preinit script is doing.
> will end up with exactly the same result as nokia-modem.pm=1, just
> with a few less symlinks. Also you have to make sure, that you
> actually exported the right gpios, since the global gpio number is
> dependent on the gpio-controller registration order.
>
> > So please drop this patch.
>
> Please provide facts, that Maemo can use pm=0 without additional
> kernel patches and does not work with pm=1.
Currently we use simple kernel driver which register gpio-switch
interface. Why to remove something which is 1) working and is 2) useful
for users & developers?
--
Pali Rohár
pali.rohar@...il.com
Powered by blists - more mailing lists