Re: [PATCH 6/6] powerpc/fsl_soc: Isolate legacy fsl_spi support to mpc832x_rdb boards

[Peter Korsgaard <jacmet@sunsite.dk>]

>>>>> "Peter" == Peter Korsgaard <jacmet@sunsite.dk> writes:

Anyone? I've locally reverted the commit, but most likely I'm not the
only one using the spi_mpc83xx driver without direct gpio controlled
chip select handling.

 Anton> The advantages of this:
 Anton> - Don't encourage legacy support;
 Anton> - Less external symbols, less code to compile-in for !MPC832x_RDB
 Anton> platforms.

 Peter> It's nice with your cleanups, but I wonder how to handle more
 Peter> complicated chip select handling than simply toggling a single gpio.

 Peter> I have a board (or 2 actually, but they are similar in this regard)
 Peter> with a mpc8347 using SPI to a number of addon boards. For signal
 Peter> integrity reasons the SPI signals are routed to a MUX, so the chip
 Peter> select logic has to set the MUX in addition to controlling the CS line
 Peter> of the device.

 Peter> I've been using code like this since late 2007, but this patch
 Peter> ofcourse breaks it:

 Peter> static void thinx_spi_activate_cs(u8 cs, u8 polarity)
 Peter> {
 Peter> 	static u8 old_cs = 255;

 Peter> 	if (cs != old_cs) {
 Peter> 		/* mux setup (cs 2:1)*/
 Peter> 		gpio_set_value(gpio1 + GPIO_SPI_MUX_NOE, 1);
 Peter> 		gpio_set_value(gpio1 + GPIO_SPI_MUX_SEL0, cs&2);
 Peter> 		gpio_set_value(gpio1 + GPIO_SPI_MUX_SEL1, cs&4);
 Peter> 		gpio_set_value(gpio1 + GPIO_SPI_MUX_NOE, 0);
 Peter> 		old_cs = cs;
 Peter> 	}

 Peter> 	switch (cs) {
 Peter> 	case 0: gpio_set_value(gpio1 + GPIO_SPI_CS_BKL1, polarity); break;
 Peter> 	case 1: gpio_set_value(gpio1 + GPIO_SPI_CS_BKL2, polarity); break;
 Peter> 	case 2: gpio_set_value(gpio1 + GPIO_SPI_CS_OPT1, polarity); break;
 Peter> 	case 3: gpio_set_value(gpio1 + GPIO_SPI_CS_OPT2, polarity); break;
 Peter> 	}
 Peter> }

 Peter> static void thinx_spi_deactivate_cs(u8 cs, u8 polarity)
 Peter> {
 Peter> 	switch (cs) {
 Peter> 	case 0: gpio_set_value(gpio1 + GPIO_SPI_CS_BKL1, !polarity); break;
 Peter> 	case 1: gpio_set_value(gpio1 + GPIO_SPI_CS_BKL2, !polarity); break;
 Peter> 	case 2: gpio_set_value(gpio1 + GPIO_SPI_CS_OPT1, !polarity); break;
 Peter> 	case 3: gpio_set_value(gpio1 + GPIO_SPI_CS_OPT2, !polarity); break;
 Peter> 	}
 Peter> }

 Peter> static __init int thinx_spi_init(void)
 Peter> {
 Peter> 	struct device_node *np;
 Peter> 	struct of_gpio_chip *gc;
 Peter> 	static const int gpios[] = {
 Peter> 		GPIO_SPI_CS_BKL1,
 Peter> 		GPIO_SPI_CS_BKL2,
 Peter> 		GPIO_SPI_CS_OPT1,
 Peter> 		GPIO_SPI_CS_OPT2,
 Peter> 		GPIO_SPI_MUX_NOE,
 Peter> 		GPIO_SPI_MUX_SEL0,
 Peter> 		GPIO_SPI_MUX_SEL1
 Peter> 	};
 Peter> 	int i;

 Peter> 	np = of_find_node_by_name(NULL, "gpio-controller");
 Peter> 	if (!np || !np->data) {
 Peter> 		printk(KERN_ERR
 Peter> 		       "gpio1 node not found or controller not registerred\n");
 Peter> 		return -ENODEV;
 Peter> 	}
 Peter> 	gc = np->data;
 Peter> 	gpio1 = gc->gc.base;

 Peter> 	for (i=0; i<ARRAY_SIZE(gpios); i++) {
 Peter> 		gpio_request(gpio1 + gpios[i], "spi");
 Peter> 		gpio_direction_output(gpio1 + gpios[i], 1);
 Peter> 	}

 Peter> 	fsl_spi_init(thinx_spi_boardinfo, ARRAY_SIZE(thinx_spi_boardinfo),
 Peter> 		     thinx_spi_activate_cs, thinx_spi_deactivate_cs);

 Peter> 	return 0;
 Peter> }

 Peter> Now, I don't quite see how to handle this with the new OF bindings -
 Peter> Any ideas?

-- 
Bye, Peter Korsgaard