Re: [PATCH v3 6/8] drivers: soc: add support for exynos SROM driver

From: LABBE Corentin
Date: Tue Oct 13 2015 - 11:00:14 EST


+static struct exynos_srom_reg_dump *exynos_srom_alloc_reg_dump(
+ const unsigned long *rdump,
+ unsigned long nr_rdump)
+{
+ struct exynos_srom_reg_dump *rd;
+ unsigned int i;
+
+ rd = kcalloc(nr_rdump, sizeof(*rd), GFP_KERNEL);
+ if (!rd)
+ return NULL;
+
+ for (i = 0; i < nr_rdump; ++i)
+ rd[i].offset = rdump[i];
+
+ return rd;
+}

You do not free rd anywhere in your code.

+static int exynos_srom_probe(struct platform_device *pdev)
+{
+ struct device_node *np;
+ struct device *dev = &pdev->dev;
+
+ np = dev->of_node;

Are you sure that dev->of_node will be always set ?
I see lots of driver who if (dev->of_node) {}

+ exynos_srom_base = of_iomap(np, 0);
+
+ if (!exynos_srom_base) {
+ pr_err("iomap of exynos srom controller failed\n");
+ return -ENOMEM;
+ }

You can use dev_err(dev, "") insted of pr_err

+
+ exynos_srom_regs = exynos_srom_alloc_reg_dump(exynos_srom_offsets,
+ sizeof(exynos_srom_offsets));
+
+ if (!exynos_srom_regs) {
+ iounmap(exynos_srom_regs);
+ return -ENOMEM;
+ }
+
+ return 0;
+}

Instead of using a global static exynos_srom_base/exynos_srom_regs, why you do not use platform_set_drvdata() ?

Regards
LABBE Corentin

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/