Re: [Kernel] [PATCH] gpiolib: rewrite gpiochip_add_to_list

[Bamvor Zhang Jian <bamvor.zhangjian@xxxxxxxxxx>]

Hi, Julien

On 01/09/2016 06:56 AM, Julien Grossholtz wrote:
> Hi,
>
> The issue is in the break condition of the list_for_each_entry_safe loop.
> If there is a chip at the beginning it breaks even if there is another
> one after. You should check the next chip:
> if (chip->base > prev->base && chip->base + chip->ngpio <= next->base)
Thanks you suggestion. It indeed fix the issue you mentioned.
I will send the new version after I make sure there is no other corner case.

Regards

Bamvor
>
> You can also call list_add directly into this for_each.
>
> Julien
>
> ----- Original Message -----
> From: "Julien Grossholtz" <julien.grossholtz@xxxxxxxxxxxxxxxxxxxx>
> To: "Bamvor Jian Zhang" <bamvor.zhangjian@xxxxxxxxxx>
> Cc: linux-gpio@xxxxxxxxxxxxxxx, "linus walleij" <linus.walleij@xxxxxxxxxx>, "broonie" <broonie@xxxxxxxxxx>, "kernel" <kernel@xxxxxxxxxxxxxxxxxxxx>, arnd@xxxxxxxx
> Sent: Friday, January 8, 2016 4:47:03 PM
> Subject: Re: [Kernel] [PATCH] gpiolib: rewrite gpiochip_add_to_list
>
> Hello,
>
> I tested this patch. It is not working in the following situation:
> A first driver request manually base number from 0 to 31, 32 to 63 etc... for some chips (gpio-mxc.c in my case).
> After this, an other driver set base to -1. It works for the first chip, but fails for all others.
>
> It looks like there is still an issue in the gpio chips list order. I will put some debug and see if I can give you feedback.
>
> Julien
>
> ----- Original Message -----
> From: "Bamvor Jian Zhang" <bamvor.zhangjian@xxxxxxxxxx>
> To: linux-gpio@xxxxxxxxxxxxxxx
> Cc: "linus walleij" <linus.walleij@xxxxxxxxxx>, broonie@xxxxxxxxxx, "julien grossholtz" <julien.grossholtz@xxxxxxxxxxxxxxxxxxxx>, arnd@xxxxxxxx, "Bamvor Jian Zhang" <bamvor.zhangjian@xxxxxxxxxx>
> Sent: Friday, January 8, 2016 1:37:15 AM
> Subject: [PATCH] gpiolib: rewrite gpiochip_add_to_list
>
> The original code of gpiochip_add_to_list is not very clear which
> lead to bugs or compiling warning, reference the following patches:
> Bugs:
> 1.  Commit ef7c7553039b ("gpiolib: improve overlap check of range of
>     gpio").
> 2.  "[PATCH] gpiolib: fix chip order in gpio list" [1]
>
> Warning:
> 1.  Commit e28ecca6eac4 ("gpio: fix warning about iterator").
> of gpio").
>
> There is a off-list discussion about how to improve it consequently.
> This commit try to follow this by rewriting the whole functions.
>
> Tested pass with my gpio mockup driver and test scripts[2].
>
> [1] http://www.spinics.net/lists/kernel/msg2156917.html
> [2] http://www.spinics.net/lists/linux-gpio/msg09598.html
>
> Suggested-by: Arnd Bergmann <arnd@xxxxxxxx>
> Signed-off-by: Bamvor Jian Zhang <bamvor.zhangjian@xxxxxxxxxx>
> ---
>  drivers/gpio/gpiolib.c | 65 ++++++++++++++++++++++----------------------------
>  1 file changed, 28 insertions(+), 37 deletions(-)
>
> diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> index 3db34e7..b9cc01f 100644
> --- a/drivers/gpio/gpiolib.c
> +++ b/drivers/gpio/gpiolib.c
> @@ -189,55 +189,46 @@ EXPORT_SYMBOL_GPL(gpiod_get_direction);
>   */
>  static int gpiochip_add_to_list(struct gpio_chip *chip)
>  {
> - struct gpio_chip *iterator;
> - struct gpio_chip *previous = NULL;
> + struct gpio_chip *prev, *next;
>
>   if (list_empty(&gpio_chips)) {
> - list_add_tail(&chip->list, &gpio_chips);
> + /* initial entry in list */
> + list_add(&chip->list, &gpio_chips);
>   return 0;
>   }
>
> - list_for_each_entry(iterator, &gpio_chips, list) {
> - if (iterator->base >= chip->base + chip->ngpio) {
> - /*
> - * Iterator is the first GPIO chip so there is no
> - * previous one
> - */
> - if (!previous) {
> - goto found;
> - } else {
> - /*
> - * We found a valid range(means
> - * [base, base + ngpio - 1]) between previous
> - * and iterator chip.
> - */
> - if (previous->base + previous->ngpio
> - <= chip->base)
> - goto found;
> - }
> - }
> - previous = iterator;
> + next = list_entry(gpio_chips.next, struct gpio_chip, list);
> + if (chip->base + chip->ngpio <= next->base) {
> + /* add before first entry */
> + list_add(&chip->list, &gpio_chips);
> + return 0;
>   }
>
> - /*
> - * We are beyond the last chip in the list and iterator now
> - * points to the head.
> - * Let iterator point to the last chip in the list.
> - */
> + prev = list_entry(gpio_chips.prev, struct gpio_chip, list);
> + if (prev->base + prev->ngpio <= chip->base) {
> + /* add behind last entry */
> + list_add_tail(&chip->list, &gpio_chips);
> + return 0;
> + }
>
> - iterator = list_last_entry(&gpio_chips, struct gpio_chip, list);
> - if (iterator->base + iterator->ngpio <= chip->base)
> - goto found;
> + list_for_each_entry_safe(prev, next, &gpio_chips, list)
> + if (chip->base > prev->base)
> + break;
>
> - dev_err(chip->parent,
> -       "GPIO integer space overlap, cannot add chip\n");
> - return -EBUSY;
> + if (&prev->list != &gpio_chips &&
> +    &next->list != &gpio_chips &&
> +    prev->base + prev->ngpio <= chip->base &&
> +    chip->base + chip->ngpio <= next->base) {
> + /* add between prev and next */
> + list_add(&chip->list, &prev->list);
> + return 0;
> + }
>
> -found:
> - list_add_tail(&chip->list, &iterator->list);
> - return 0;
> + dev_err(chip->parent, "GPIO integer space overlap, cannot add chip\n");
> + return -EBUSY;
>  }
>
> +
>  /**
>   * Convert a GPIO name to its descriptor
>   */
>

-- 
-----------------------------
   blog: http://aarch64.me
-----------------------------
--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html