linux-kernel - Re: [PATCH 5/6] media: sun4i: Add H3 deinterlace driver

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: <3227980.eWD6USAIP4@jernej-laptop>
Date:   Sat, 14 Sep 2019 08:42:22 +0200
From:   Jernej Škrabec <jernej.skrabec@...l.net>
To:     Maxime Ripard <mripard@...nel.org>
Cc:     wens@...e.org, robh+dt@...nel.org, mark.rutland@....com,
        mchehab@...nel.org, hverkuil@...all.nl, devicetree@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
        linux-media@...r.kernel.org, linux-sunxi@...glegroups.com
Subject: Re: [PATCH 5/6] media: sun4i: Add H3 deinterlace driver

Hi!

Dne četrtek, 12. september 2019 ob 22:26:47 CEST je Maxime Ripard napisal(a):
> Hi,
> 
> On Thu, Sep 12, 2019 at 07:51:31PM +0200, Jernej Skrabec wrote:
> > +	dev->regmap = devm_regmap_init_mmio(dev->dev, dev->base,
> > +					    
&deinterlace_regmap_config);
> > +	if (IS_ERR(dev->regmap)) {
> > +		dev_err(dev->dev, "Couldn't create deinterlace 
regmap\n");
> > +
> > +		return PTR_ERR(dev->regmap);
> > +	}
> > +
> > +	ret = clk_prepare_enable(dev->bus_clk);
> > +	if (ret) {
> > +		dev_err(dev->dev, "Failed to enable bus clock\n");
> > +
> > +		return ret;
> > +	}
> 
> Do you need to keep the bus clock enabled all the time? Usually, for
> the SoCs that have a reset line, you only need it to read / write to
> the registers, not to have the controller actually running.
> 
> If you don't, then regmap_init_mmio_clk will take care of that for
> you.
> 
> > +	clk_set_rate(dev->mod_clk, 300000000);
> > +
> > +	ret = clk_prepare_enable(dev->mod_clk);
> > +	if (ret) {
> > +		dev_err(dev->dev, "Failed to enable mod clock\n");
> > +
> > +		goto err_bus_clk;
> > +	}
> > +
> > +	ret = clk_prepare_enable(dev->ram_clk);
> > +	if (ret) {
> > +		dev_err(dev->dev, "Failed to enable ram clock\n");
> > +
> > +		goto err_mod_clk;
> > +	}
> > +
> > +	ret = reset_control_reset(dev->rstc);
> > +	if (ret) {
> > +		dev_err(dev->dev, "Failed to apply reset\n");
> > +
> > +		goto err_ram_clk;
> > +	}
> 
> This could be moved to a runtime_pm hook, with get_sync called in the
> open. That way you won't leave the device powered on if it's unused.

Currently I'm looking at sun4i_csi.c as an example of runtime ops, but it 
seems a bit wrong to have suspend and resume function marked with 
__maybe_unused because they are the only functions which enable needed clocks.
If CONFIG_PM is not enabled, then this driver simply won't work, because 
clocks will never get enabled. I guess I can implement runtime pm ops in the 
same way and add additional handling when CONFIG_PM is not enabled, right?

BTW, which callback is get_sync? I don't see it in dev_pm_ops. I suppose I 
need only runtime_suspend and runtime_resume.

Off topic: sun6i_csi.c includes linux/pm_runtime.h but it doesn't have any kind 
of power management as far as I can see.

Best regards,
Jernej

> 
> > +struct deinterlace_dev {
> > +	struct v4l2_device	v4l2_dev;
> > +	struct video_device	vfd;
> > +	struct device		*dev;
> > +	struct v4l2_m2m_dev	*m2m_dev;
> > +
> > +	/* Device file mutex */
> > +	struct mutex		dev_mutex;
> > +
> > +	void __iomem		*base;
> > +	struct regmap		*regmap;
> 
> Do you need to store the base address in that structure if you're
> using the regmap?
> 
> Maxime