linux-kernel - Re: [PATCH 4/8] regulator: da9055: Add DT support

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: <20140206120336.GL32298@sirena.org.uk>
Date:	Thu, 6 Feb 2014 12:03:36 +0000
From:	Mark Brown <broonie@...nel.org>
To:	"Opensource [Adam Thomson]" <Adam.Thomson.Opensource@...semi.com>
Cc:	Lee Jones <lee.jones@...aro.org>,
	"alsa-devel@...a-project.org" <alsa-devel@...a-project.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
	Rob Herring <robh+dt@...nel.org>,
	Linus Walleij <linus.walleij@...aro.org>,
	Dmitry Torokhov <dmitry.torokhov@...il.com>,
	Alessandro Zummo <a.zummo@...ertech.it>,
	Guenter Roeck <linux@...ck-us.net>
Subject: Re: [PATCH 4/8] regulator: da9055: Add DT support

On Thu, Feb 06, 2014 at 11:31:13AM +0000, Opensource [Adam Thomson] wrote:
> On Wed, Feb 05, 2014 at 18:37:21PM +0000, Mark Brown wrote:

> > > +	for_each_child_of_node(nproot, np) {
> > > +		if (!of_node_cmp(np->name,
> > > +				 regulator->info->reg_desc.name)) {
> > > +			config->init_data = of_get_regulator_init_data(
> > > +				&pdev->dev, np);
> > > +			config->of_node = np;
> > > +			break;
> > > +		}
> > > +	}

> > I think you're looking for of_regulator_match() here.

> Used another driver as an example for this, but if there's a better method
> then I'm happy to use it. Will have a look.

That's probably an older driver - the code was factored out at some
point but lots of drivers don't get much love.

> > >  	/* Only LDO 5 and 6 has got the over current interrupt */
> > >  	if (pdev->id == DA9055_ID_LDO5 || pdev->id ==  DA9055_ID_LDO6) {
> > > -		irq = platform_get_irq_byname(pdev, "REGULATOR");
> > > -		irq = regmap_irq_get_virq(da9055->irq_data, irq);
> > > +		irq = regmap_irq_get_virq(da9055->irq_data,
> > > +					  DA9055_IRQ_REGULATOR);

> > This seems like a bit of a step backwards - what happened in the MFD
> > (and why didn't it update the users to avoid breaking bisection)?

> I tested this on target, when doing tests for devicetree. What was happening was
> that platform_get_irq_byname() was returning the VIRQ number already (368 in one
> test case where onkey was being probed) rather than the local IRQ number for the
> device (the resource information seemed to have been updated with the VIRQ
> number instead of the local IRQ number). So when that was passed to
> regmap_irq_get_virq() it would then return an incorrect IRQ number (0 in the
> same scenario, when I enabled DEBUG in irqdomain.c, I would see the message
> "error: hwirq 0x170 is too large for da9055_irq"). That incorrect irq was then
> being passed to devm_request_threaded_irq() which subsequently failed. This is
> why I made the change. Is it preferrable to use platform_get_irq_byname()
> instead of regmap_irq_get_virq() as using them both doesn't seem to work, unless
> I'm missing something fundamental here.

What's happening here is that the MFD framework has done the lookup for
you when passing the interrupt resource through - you should just use
platform_get_irq_byname() and save a little code in the driver.  If it's
behaving differently on DT and non-DT systems it seems better to figure
out why and then make it consistent.

Download attachment "signature.asc" of type "application/pgp-signature" (837 bytes)