Hello ksi,
ksi@koi8.net wrote:
On Fri, 13 Feb 2009, Heiko Schocher wrote:
ksi@koi8.net wrote:
Signed-off-by: Sergey Kubushyn ksi@koi8.net
diff -purN u-boot-i2c.orig/drivers/i2c/soft_i2c.c u-boot-i2c/drivers/i2c/soft_i2c.c --- u-boot-i2c.orig/drivers/i2c/soft_i2c.c 2009-02-12 10:43:41.000000000 -0800 +++ u-boot-i2c/drivers/i2c/soft_i2c.c 2009-02-12 10:46:00.000000000 -0800 @@ -1,4 +1,8 @@ /*
- Copyright (c) 2009 Sergey Kubushyn ksi@koi8.net
- Changes for multibus/multiadapter I2C support.
- (C) Copyright 2001, 2002
- Wolfgang Denk, DENX Software Engineering, wd@denx.de.
[...]
The following patch is based on your patches without 7/12 and adds multibus support for the soft_i2c driver without doing such a big change as you did. Maybe it is not yet perfect, because it is just a fast try, but I think we should go this way. What do you/others think?
The reason behind this patch is making SEVERAL different SOFT_I2C ADAPTERS available. Not BUSSES but separate PHYSICAL I2C ADAPTERS made of different pin pairs from different chips.
This you can also do with "my" suggestion ...
OK, please explain how are you going to make different functions for different adapters? Let's say you want to use 2 on-SoC GPIO pins for
You can do now the following for example in your include/configs/MPC8548CDS.h example:
you only have to define
#define I2C_SDA(bit) (printf("hwadap: %d sda1: %d", cur_adap_nr->hwadapnr, bit))
if this is a real driver you can make a function in your board code say (just a fast thought):
void i2c_soft_sda (int bit) { switch(cur_adap_nr->hwadapnr) { case 0: /* adapter specfic code 0 */ break; case 1: /* adapter specfic code 1 */ break; [...] } }
and define in config file
#define I2C_SDA(bit) i2c_soft_sda (bit)
adapter #0, 2 GPIOs from a PCI-PCI bridge for adapter #1, and 2 pins from some chip sitting behind that bridge for adapter #2 if all those pin sets are accessed totally different. I won't even start about using pins from different chips for SDA and SCL (let's say you only have one GPIO available on your SoC and another one on PCI Bridge.)
What your patch creates is just aliases to the SAME physical adapter.
No, it is not! I only use the same functions, but in the board specific code it is possible to made a switch and access the Pins where ever they are.
bye Heiko