On Tue, Jan 31, 2017 at 4:29 PM, Paul Cercueil [off-list ref] wrote:

Le 2017-01-31 15:20, Linus Walleij a écrit :

quoted

quoted

+ of_property_read_u32(dev->of_node, "base", &jzgc->gc.base);

Remove this. Dynamic allocation should be fine, if you're using the
new userspace ABI like tools/gpio/* or libgpiod and only that and
in-kernel
consumers, dynamic numbers are just fine.

The problem is that the QI_LB60 board code still have a lot of references
to global GPIO numbers. Just grep for JZ_GPIO_PORT in
arch/mips/jz4740/board-qi_lb60.c to see what I mean...

OK I understand we might need a compromise here to get the code moving.

But we need to keep it out of the device tree.

I think it's better to put a base table relative to the memory base in
the driver in that case:

unsigned int gpio_global_base;

switch (memory_base_address) {
case 0x41000000:
    gpio_global_base = 0x00;
    break;
case 0x42000000:
    gpio_global_base = 0x20;

(...)

etc. (Those are not your base addresses but you get the idea).

Include a few comments like:

/*
 * DO NOT EXPAND THIS: FOR BACKWARD GPIO NUMBERSPACE
 * COMPATIBIBILITY ONLY: WORK TO TRANSITION CONSUMERS TO
 * USE THE GPIO DESCRIPTOR API IN <linux/gpio/consumer.h> INSTEAD.
 */

Then I'll be happy :)

Yours,
Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html