Thread (47 messages) 47 messages, 6 authors, 2018-05-18

[PATCH v10 03/27] clk: davinci: psc: allow for dev == NULL

From: Sekhar Nori <hidden>
Date: 2018-05-15 13:43:51
Also in: linux-clk, linux-devicetree, lkml

On Wednesday 09 May 2018 10:55 PM, David Lechner wrote:
quoted hunk ↗ jump to hunk
@@ -261,10 +263,14 @@ davinci_lpsc_clk_register(struct device *dev, const char *name,
 	lpsc->pd = pd;
 	lpsc->flags = flags;
 
-	ret = devm_clk_hw_register(dev, &lpsc->hw);
+	ret = clk_hw_register(dev, &lpsc->hw);
 	if (ret < 0)
 		return ERR_PTR(ret);
 
+	/* for now, genpd is only registered when using device-tree */
+	if (!dev || !dev->of_node)
+		return lpsc;
+
 	/* genpd attach needs a way to look up this clock */
 	ret = clk_hw_register_clkdev(&lpsc->hw, name, best_dev_name(dev));
 
@@ -378,11 +384,11 @@ __davinci_psc_register_clocks(struct device *dev,
 	struct regmap *regmap;
 	int i, ret;
 
-	psc = devm_kzalloc(dev, sizeof(*psc), GFP_KERNEL);
+	psc = kzalloc(sizeof(*psc), GFP_KERNEL);
 	if (!psc)
 		return ERR_PTR(-ENOMEM);
 
-	clks = devm_kmalloc_array(dev, num_clks, sizeof(*clks), GFP_KERNEL);
+	clks = kmalloc_array(num_clks, sizeof(*clks), GFP_KERNEL);
 	if (!clks)
 		return ERR_PTR(-ENOMEM);
 
@@ -396,14 +402,14 @@ __davinci_psc_register_clocks(struct device *dev,
 	for (i = 0; i < num_clks; i++)
 		clks[i] = ERR_PTR(-ENOENT);
 
-	pm_domains = devm_kcalloc(dev, num_clks, sizeof(*pm_domains), GFP_KERNEL);
+	pm_domains = kcalloc(num_clks, sizeof(*pm_domains), GFP_KERNEL);
 	if (!pm_domains)
 		return ERR_PTR(-ENOMEM);
 
 	psc->pm_data.domains = pm_domains;
 	psc->pm_data.num_domains = num_clks;
 
-	regmap = devm_regmap_init_mmio(dev, base, &davinci_psc_regmap_config);
+	regmap = regmap_init_mmio(dev, base, &davinci_psc_regmap_config);
 	if (IS_ERR(regmap))
 		return ERR_CAST(regmap);
Here and in the PLL driver, you have dropped the devm_* variants (like
agreed upon), but not added any error path handling. For the clocks
needed for boot, its probably fine, but the same code path is used for
non-essential clocks too. Just from a code completeness perspective, it
will be nice to see the error path correctly handled.

Apart from that both 1/27 and 3/27 look good to me.

Thanks,
Sekhar
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help