Re: [PATCH for 4.12] Revert "pinctrl: rockchip: avoid hardirq-unsafe functions in irq_chip"

From: Brian Norris <briannorris@chromium.org>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: "Heiko Stuebner" <heiko@sntech.de>,
	"Linus Walleij" <linus.walleij@linaro.org>,
	linux-rockchip@lists.infradead.org,
	"Julia Cartwright" <julia@ni.com>,
	LKML <linux-kernel@vger.kernel.org>,
	linux-gpio@vger.kernel.org, "John Keeping" <john@metanate.com>,
	linux-pm@vger.kernel.org, "Doug Anderson" <dianders@chromium.org>,
	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
	"Peter Zijlstra" <peterz@infradead.org>,
	"Tony Lindgren" <tony@atomide.com>,
	"David.Wu" <david.wu@rock-chips.com>,
	'黄涛' <huangtao@rock-chips.com>
Subject: Re: [PATCH for 4.12] Revert "pinctrl: rockchip: avoid hardirq-unsafe functions in irq_chip"
Date: Tue, 27 Jun 2017 10:23:09 -0700	[thread overview]
Message-ID: <20170627172306.GA91202@google.com> (raw)
In-Reply-To: <alpine.DEB.2.20.1706271504521.1798@nanos>

Hi Linus,

I'm not sure I follow all of Thomas's suggestions on what should be done
in the future yet, but I agree that can be done in parallel:

On Tue, Jun 27, 2017 at 03:06:26PM +0200, Thomas Gleixner wrote:
> On Tue, 27 Jun 2017, Thomas Gleixner wrote:
> > That aside, looking at the commit which caused this discussion:
> > 
> > 88bb94216f59e pinctrl: rockchip: avoid hardirq-unsafe functions in irq_chip
> > 
> > I assume (the changelog lacks details) that the patch want's to avoid a
> > might sleep splat from the irq callbacks caused by the regmap spinlock,
> > which gets converted into a sleeping lock on RT. It does this by abusing
> > the irq_bus_lock() mechanism, which is wrong to begin with.
> > 
> > The only irq chip function which uses the regmap magic is the
> > irq_set_type() callback. Now, I have a hard time to understand (though I'm
> > no regmap/pinctrl expert) why that regmap stuff needs to be called in the
> > first place. The level and the polarity are programmed via:
> > 
> >         writel_relaxed(level, gc->reg_base + GPIO_INTTYPE_LEVEL);
> >         writel_relaxed(polarity, gc->reg_base + GPIO_INT_POLARITY);
> > 
> > Why needs the regmap machinery to be invoked there? The GPIO is already
> > muxed and configured as interrupt, otherwise none of the irq functions
> > could be invoked. Hmm?
> 
> That said, the commit should be reverted and the issue needs to analyzed
> proper. We still need the RCU -> SCRU conversion, but that's a different
> problem.

Can we consider this an "ack" for the $subject then? Heiko also gave his
approval. How can this get merged? It's running a bit late for 4.12,
though it really shouldn't be risky (at least for non-RT stuff that was
working warning-free already in 4.11), but a 4.13-rc1 with -stable tag
could work as well.

I suppose I didn't put a proper 'Fixes' tag (though it should be obvious
from the commit message), so here goes:

Fixes: 88bb94216f59 ("pinctrl: rockchip: avoid hardirq-unsafe functions in irq_chip")

Regards,
Brian