On 03/12/19 1:47 PM, Geert Uytterhoeven wrote:
> Hi Harish,
>
> On Tue, Dec 3, 2019 at 6:42 AM Harish Jenny K N
> <harish_kandiga@xxxxxxxxxx> wrote:
>>> +static int gpio_aggregator_probe(struct platform_device *pdev)
>>> +{
>>> + struct device *dev = &pdev->dev;
>>> + struct gpio_desc **descs;
>>> + struct gpiochip_fwd *fwd;
>>> + int i, n;
>>> +
>>> + n = gpiod_count(dev, NULL);
>>> + if (n < 0)
>>> + return n;
>>> +
>>> + descs = devm_kmalloc_array(dev, n, sizeof(*descs), GFP_KERNEL);
>>> + if (!descs)
>>> + return -ENOMEM;
>>> +
>>> + for (i = 0; i < n; i++) {
>>> + descs[i] = devm_gpiod_get_index(dev, NULL, i, GPIOD_ASIS);
>> can you please add this check as well as we need to return EPROBE_DEFER.
>>
>> if (desc[i] == ERR_PTR(-ENOENT))
>> < return -EPROBE_DEFER;
> So gpiod_get_index() nevers return -EPROBE_DEFER, but returns -ENOENT
> instead?
> How can a driver distinguish between "GPIO not found" and "gpiochip driver
> not yet initialized"?
> Worse, so the *_optional() variants will return NULL in both cases, too, so
> the caller will always fall back to optional GPIO not present?
>
> Or am I missing something?
>
> Gr{oetje,eeting}s,
>
> Geert
We had earlier tested our changes on 4.14 kernel and the explicit return of -EPROBE_DEFER was needed in the inverter driver.
probably the commit 6662ae6af82df10259a70c7569b4c12ea7f3ba93 ( gpiolib: Keep returning EPROBE_DEFER when we should)
has fixed the issue and now it returns -EPROBE_DEFER. you can ignore this comment as of now. I will test and let you know if needed.
Thanks,
Harish
