On 02/02/16 13:04, Thomas Gleixner wrote:

On Tue, 2 Feb 2016, Vladimir Murzin wrote:

quoted

+static int __init mps2_clockevent_init(struct device_node *np)
+{
+	void __iomem *base;
+	struct clk *clk;
+	struct clockevent_mps2 *ce;
+	u32 rate;
+	int irq, ret;
+	const char *name = "mps2-clkevt";
+
+	ret = of_property_read_u32(np, "clock-frequency", &rate);
+	if (ret) {
+		clk = of_clk_get(np, 0);
+		if (IS_ERR(clk)) {
+			ret = PTR_ERR(clk);
+			pr_err("failed to get clock for clockevent: %d\n", ret);
+			goto err_clk_get;
+		}
+
+		ret = clk_prepare_enable(clk);
+		if (ret) {
+			pr_err("failed to enable clock for clockevent: %d\n", ret);
+			clk_put(clk);
+			goto err_clk_enable;
+		}
+
+		rate = clk_get_rate(clk);
+	}

So in case that "clock-frequency" is available in DT, what enables and
configures the clock?

They are fixed-clock, so I ended up this *cough* mess *cough* shortcut.
I'd be glad if you share your thoughts what is preferred way to cope
with it (is it worth to do at all?)

quoted

+err_ia_alloc:
+	kfree(ce);
+err_ce_alloc:
+err_get_irq:
+	iounmap(base);
+err_iomap:
+	clk_disable_unprepare(clk);

clk is not initialized when "clock-frequency" is available in DT. The compiler
should have told you ....

Shame on me, I've completely messed it up :( Thanks for pointing at it,
I'll fix this an other places you pointed at...

quoted

+err_clk_enable:
+	clk_put(clk);

Put without get ....

or double clk_put() if clk_prepare_enable() failed... shame, shame, shame

quoted

+static int __init mps2_clocksource_init(struct device_node *np)
+{
+	void __iomem *base;
+	struct clk *clk;
+	u32 rate;
+	int ret;
+	const char *name = "mps2-clksrc";
+
+	ret = of_property_read_u32(np, "clock-frequency", &rate);
+	if (ret) {
+		clk = of_clk_get(np, 0);
+		if (IS_ERR(clk)) {
+			ret = PTR_ERR(clk);
+			pr_err("failed to get clock for clocksource: %d\n", ret);
+			goto err_clk_get;
+		}
+
+		ret = clk_prepare_enable(clk);
+		if (ret) {
+			pr_err("failed to enable clock for clocksource: %d\n", ret);
+			clk_put(clk);
+			goto err_clk_enable;
+		}
+
+		rate = clk_get_rate(clk);
+	}

Same problem as above.

quoted

+err_clocksource_init:
+	iounmap(base);
+err_iomap:
+	clk_disable_unprepare(clk);

Ditto.

quoted

+err_clk_enable:
+	clk_put(clk);

Ditto.

quoted

+err_clk_get:
+	return ret;
+}
+
+static void __init mps2_timer_init(struct device_node *np)
+{
+	static int has_clocksource, has_clockevent;
+	int ret;
+
+	if (!has_clocksource) {
+		ret = mps2_clocksource_init(np);
+		if (!ret) {
+			has_clocksource = 1;
+			return;
+		}
+	}
+
+	if (!has_clockevent) {
+		ret = mps2_clockevent_init(np);
+		if (!ret) {
+			has_clockevent = 1;
+			return;
+		}
+	}

You take randomly the first timer as clocksource and the second one as
clockevent. If clocksource init fails, you try to install it as clockevent.

Sorry, but I really cannot follow the logic here.

It has been done as a response on Daniel's comment [1]. I'm quite happy
to get known what init sequence is expected here.

[1]
http://lists.infradead.org/pipermail/linux-arm-kernel/2015-November/388223.html

Thanks for your time, Thomas!

Vladimir

Thanks,

	tglx