From mboxrd@z Thu Jan  1 00:00:00 1970
From: Andy Gross <agross@codeaurora.org>
Subject: Re: [PATCH] firmware: qcom: scm: Convert to platform driver
Date: Fri, 4 Sep 2015 15:34:25 -0500
Message-ID: <20150904203425.GA31824@qualcomm.com>
References: <1437431152-6730-1-git-send-email-agross@codeaurora.org>
 <20150903213322.GI15099@codeaurora.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <20150903213322.GI15099@codeaurora.org>
Sender: linux-kernel-owner@vger.kernel.org
To: Stephen Boyd <sboyd@codeaurora.org>
Cc: linux-arm-msm@vger.kernel.org, linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org, Bjorn Andersson <bjorn.andersson@sonymobile.com>, devicetree@vger.kernel.org, Rob Clark <robdclark@gmail.com>
List-Id: devicetree@vger.kernel.org

On Thu, Sep 03, 2015 at 02:33:22PM -0700, Stephen Boyd wrote:
> On 07/20, Andy Gross wrote:
> > This patch creates a platform driver for the SCM so that we can adequately
> > manage resources.  This removes clients having to carry the necessary
> > clocks to use the SCM resources.
> > 
> > Signed-off-by: Andy Gross <agross@codeaurora.org>
> > ---
> 
> It would be nice if we could use this platform device for doing
> the DMAish memory allocations that we do in this driver too. I
> guess one complication there is that we would need to allocate
> memory with the DMA APIs before CPUs are brought up
> (early_initcall level).

Yeah that's one thing we could do but we'd have to defer the memory stuff until
it's used the first time (specific calls require it).

> 
> > diff --git a/drivers/firmware/qcom_scm.c b/drivers/firmware/qcom_scm.c
> > index 45c008d..5dd0514 100644
> > --- a/drivers/firmware/qcom_scm.c
> > +++ b/drivers/firmware/qcom_scm.c
> > @@ -15,14 +15,57 @@
> >   * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
> >   * 02110-1301, USA.
> >   */
> > -
> > +#include <linux/platform_device.h>
> > +#include <linux/module.h>
> > +#include <linux/platform_device.h>
> 
> This include is here twice.
> 
> >  #include <linux/cpumask.h>
> >  #include <linux/export.h>
> >  #include <linux/types.h>
> >  #include <linux/qcom_scm.h>
> > +#include <linux/of.h>
> > +#include <linux/clk.h>
> >  
> [...]
> > +
> > +/**
> > + * qcom_scm_is_available() - Checks if SCM is available
> > + */
> > +bool qcom_scm_is_available(void)
> > +{
> > +	return !!__scm;
> > +}
> > +EXPORT_SYMBOL(qcom_scm_is_available);
> > +
> > +static int qcom_scm_remove(struct platform_device *pdev)
> > +{
> > +	__scm = NULL;
> > +
> > +	return 0;
> > +}
> 
> Maybe we just shouldn't allow this? The firmware isn't going
> anywhere at runtime, and this driver is currently marked as
> bool in the Kconfig.

Fair enough.

> 
> > +
> > +static const struct of_device_id qcom_scm_dt_match[] = {
> > +	{ .compatible = "qcom,scm",},
> > +	{},
> > +};
> > +
> > +MODULE_DEVICE_TABLE(of, qcom_scm_dt_match);
> > +
> > +static struct platform_driver qcom_scm_driver = {
> > +	.driver = {
> > +		.name	= "scm",
> 
> Maybe 'qcom_scm' ?

yeah i should have used that.

> 
> > +		.of_match_table = qcom_scm_dt_match,
> > +	},
> > +	.probe = qcom_scm_probe,
> > +	.remove = qcom_scm_remove,
> > +};
> > +
> > +module_platform_driver(qcom_scm_driver);
> 
> Isn't there some sort of builtin_platform_driver() macro for
> builtin modules?

I'll take a look and convert.


-- 
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project