Hi,

On 8/20/19 11:38 PM, Sylwester Nawrocki wrote:
> On 8/20/19 21:37, Krzysztof Kozlowski wrote:
>>>> diff --git a/drivers/soc/samsung/exynos-chipid.c 
>>>> b/drivers/soc/samsung/exynos-chipid.c
> 
>>>> @@ -51,29 +48,24 @@ static const char * __init 
>>>> product_id_to_soc_id(unsigned int product_id)
>>>>   int __init exynos_chipid_early_init(void)
>>>>   {
>>>>        struct soc_device_attribute *soc_dev_attr;
>>>> -     void __iomem *exynos_chipid_base;
>>>>        struct soc_device *soc_dev;
>>>>        struct device_node *root;
>>>> -     struct device_node *np;
>>>> +     struct regmap *regmap;
>>>>        u32 product_id;
>>>>        u32 revision;
>>>> +     int ret;
>>>>
>>>> -     /* look up for chipid node */
>>>> -     np = of_find_compatible_node(NULL, NULL, 
>>>> "samsung,exynos4210-chipid");
>>>> -     if (!np)
>>>> -             return -ENODEV;
>>>> -
>>>> -     exynos_chipid_base = of_iomap(np, 0);
>>>> -     of_node_put(np);
>>>> -
>>>> -     if (!exynos_chipid_base) {
>>>> -             pr_err("Failed to map SoC chipid\n");
>>>> -             return -ENXIO;
>>>> +     regmap = 
>>>> syscon_regmap_lookup_by_compatible("samsung,exynos4210-chipid");
>>>> +     if (IS_ERR(regmap)) {
>>>> +             pr_err("Failed to get CHIPID regmap\n");
>>>> +             return PTR_ERR(regmap);
>>>>        }
>>> Following this change, I am now seeing the above error on our Tegra
>>> boards where this driver is enabled. This is triggering a kernel
>>> warnings test we have to fail. Hence, I don't think that you can remove
>>> the compatible node test here, unless you have a better way to determine
>>> if this is a samsung device.
>>
>> Right, this is really wrong... I missed that it is not a probe but
>> early init. And this init will be called on every board... Probably it
>> should be converted to a regular driver.

Early initialization is needed for SoC driver to be used from within
arch/arm/mach-exynos/ and _initcall() usage is the usual way for SoC
drivers to be initialized:

drivers/soc/amlogic/meson-gx-socinfo.c
drivers/soc/amlogic/meson-mx-socinfo.c
drivers/soc/atmel/soc.c
drivers/soc/bcm/brcmstb/common.c
drivers/soc/imx/soc-imx-scu.c
drivers/soc/imx/soc-imx8.c
drivers/soc/renesas/renesas-soc.c
drivers/soc/tegra/fuse/fuse-tegra.c
drivers/soc/ux500/ux500-soc-id.c
drivers/soc/versatile/soc-integrator.c
drivers/soc/versatile/soc-integrator.c

The only SoC drivers that are regular drivers are:

drivers/soc/fsl/guts.c
drivers/soc/versatile/soc-realview.c

> I'm also inclined to have it converted to a regular driver.  We already
> have "exynos-asv" driver matching on the chipid node (patch 3/9). 
> The ASV patches will not be merged soon anyway, all this needs some more
> thought. Krzysztof, can we abandon the chipid patches for now? Your

chipid driver is good and useful on its own. The preferred solution
IMHO would be to just revert "soc: samsung: Convert exynos-chipid
driver to use the regmap API" commit.

> pull request doesn't appear to be merged to arm-soc yet. Sorry about
> that.
Best regards,
--
Bartlomiej Zolnierkiewicz
Samsung R&D Institute Poland
Samsung Electronics

Reply via email to