Re: [PATCH v2 1/2] of: Ensure unique names without sacrificing determinism

[Rob Herring]

On Thu, May 22, 2014 at 6:36 PM, Grant Likely <grant.likely@xxxxxxxxxx> wrote:
> The way the driver core is implemented, every device using the same bus
> type is required to have a unique name because a symlink to each device
> is created in the appropriate /sys/bus/*/devices directory, and two
> identical names causes a collision.
>
> The current code handles the requirement by using an globally
> incremented counter that is appended to the device name. It works, but
> it means any change to device registration will change the assigned
> numbers. Instead, if we build up the name by using information from the
> parent nodes, then it can be guaranteed to be unique without adding a
> random number to the end of it.
>
> Signed-off-by: Grant Likely <grant.likely@xxxxxxxxxx>
> Cc: Ezequiel Garcia <ezequiel@xxxxxxxxxxxxxxxxxxxx>
> Cc: Rob Herring <robh@xxxxxxxxxx>

Acked-by: Rob Herring <robh@xxxxxxxxxx>


> ---
>  drivers/of/platform.c | 40 +++++++++++++++++++---------------------
>  1 file changed, 19 insertions(+), 21 deletions(-)
>
> diff --git a/drivers/of/platform.c b/drivers/of/platform.c
> index d0009b3614af..95c133a0554b 100644
> --- a/drivers/of/platform.c
> +++ b/drivers/of/platform.c
> @@ -68,17 +68,15 @@ EXPORT_SYMBOL(of_find_device_by_node);
>   * of_device_make_bus_id - Use the device node data to assign a unique name
>   * @dev: pointer to device structure that is linked to a device tree node
>   *
> - * This routine will first try using either the dcr-reg or the reg property
> - * value to derive a unique name.  As a last resort it will use the node
> - * name followed by a unique number.
> + * This routine will first try using the translated bus address to
> + * derive a unique name. If it cannot, then it will prepend names from
> + * parent nodes until a unique name can be derived.
>   */
>  void of_device_make_bus_id(struct device *dev)
>  {
> -       static atomic_t bus_no_reg_magic;
>         struct device_node *node = dev->of_node;
>         const __be32 *reg;
>         u64 addr;
> -       int magic;
>
>  #ifdef CONFIG_PPC_DCR
>         /*
> @@ -100,25 +98,25 @@ void of_device_make_bus_id(struct device *dev)
>         }
>  #endif /* CONFIG_PPC_DCR */
>
> -       /*
> -        * For MMIO, get the physical address
> -        */
> -       reg = of_get_property(node, "reg", NULL);
> -       if (reg) {
> -               addr = of_translate_address(node, reg);
> -               if (addr != OF_BAD_ADDR) {
> -                       dev_set_name(dev, "%llx.%s",
> -                                    (unsigned long long)addr, node->name);
> +       /* Construct the name, using parent nodes if necessary to ensure uniqueness */
> +       while (node->parent) {
> +               /*
> +                * If the address can be translated, then that is as much
> +                * uniqueness as we need. Make it the first component and return
> +                */
> +               reg = of_get_property(node, "reg", NULL);
> +               if (reg && (addr = of_translate_address(node, reg)) != OF_BAD_ADDR) {
> +                       dev_set_name(dev, dev_name(dev) ? "%llx.%s:%s" : "%llx.%s",
> +                                    (unsigned long long)addr, node->name,
> +                                    dev_name(dev));
>                         return;
>                 }
> -       }
>
> -       /*
> -        * No BusID, use the node name and add a globally incremented
> -        * counter (and pray...)
> -        */
> -       magic = atomic_add_return(1, &bus_no_reg_magic);
> -       dev_set_name(dev, "%s.%d", node->name, magic - 1);
> +               /* format arguments only used if dev_name() resolves to NULL */
> +               dev_set_name(dev, dev_name(dev) ? "%s:%s" : "%s",
> +                            strrchr(node->full_name, '/') + 1, dev_name(dev));
> +               node = node->parent;
> +       }
>  }
>
>  /**
> --
> 1.9.1
>
--
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/