[U-Boot] [PATCH V2 03/15] imx: mx6ul: Update imx registers head file

[Peng Fan]

Hi Marek,

On Sun, Jul 12, 2015 at 01:50:55PM +0200, Marek Vasut wrote:
>On Sunday, July 12, 2015 at 12:32:28 PM, Peng Fan wrote:
>> Hi Marek,
>> 
>> On Sun, Jul 12, 2015 at 12:11:42AM +0200, Marek Vasut wrote:
>> >On Saturday, July 11, 2015 at 10:07:11 AM, Peng Fan wrote:
>> >> 1. Update imx register base address for i.MX6UL.
>> >> 2. Remove duplicated MXS_APBH/GPMI/BCH_BASE.
>> >> 3. Remove #ifdef for register addresses that equal to
>> >> 
>> >>    "AIPS2_OFF_BASE_ADDR + 0x34000" for different chips.
>> >> 
>> >> 4. According fuse map, complete fuse_bank4_regs.
>> >> 5. Move AIPS3_ARB_BASE_ADDR and AIPS3_ARB_END_ADDR out of #ifdef
>> >> CONFIG_MX6SX, because we can use runtime check
>> >> 
>> >> Signed-off-by: Peng Fan <Peng.Fan at freescale.com>
>> >> ---
>> >
>> >[...]
>> >
>> >> @@ -309,11 +311,16 @@
>> >> 
>> >>  #define PWM8_BASE_ADDR              (AIPS3_ARB_BASE_ADDR + 0xB0000)
>> >>  #endif
>> >> 
>> >> +/* only for i.MX6SX/UL */
>> >> +#define WDOG3_BASE_ADDR (is_cpu_type(MXC_CPU_MX6UL) ?		\
>> >> +			 (AIPS2_OFF_BASE_ADDR + 0x64000) :	\
>> >> +			 (AIPS3_ARB_BASE_ADDR + 0x88000))
>> >> +
>> >
>> >This might not work if used in assembler.
>> 
>> I did not see any scenario that it will be used in assembler now.
>
>Does that mean we will be fine with adding code which we know might easily 
>break?
>

My bad.

>> If we truely need it in assembler, then we can not avoid so many "ifdef"s.
>
>errr ... ifdef __ASSEMBLY__ would suffice here I guess, but to be really smart
>about it, I'd say you might want to define something like MX6UL_WDOG3_BASE,
>MX6S_WDOG3_BASE and then define WDOG3_BASE_ADDR as (is_cpu_type(...) ? 
>MX6UL_WDOG3_BASE : MX6S.... ) . Then you could access the WDOG3 from assembler
>and from C code with the benefit that from C code, you'd be able to access it
>in a bit smoother way. I don't know if there's a way to do it easily in asm
>too, sorry.

Good suggestion. Later, will try to use this way in v3, after waiting
for more comments in the whole patch set. 

>
>Best regards,
>Marek Vasut

Thanks,
Peng
--