[U-Boot-Users] Please pull u-boot-83xx.git

[Timur Tabi <timur@freescale.com>]

Wolfgang Denk wrote:
> In message <200611261446.24607.sr@denx.de> you wrote:
>>>     mpc83xx: fix TQM build by defining a CFG_FLASH_SIZE for it
>>>
>>> -------------------------- include/configs/TQM834x.h --------------------------
>>> index b1f033d..f0e4900 100644
>>> @@ -95,6 +95,7 @@
>>>  #define CFG_FLASH_CFI_DRIVER			/* use the CFI driver */
>>>  #undef CFG_FLASH_CHECKSUM
>>>  #define CFG_FLASH_BASE		0x80000000	/* start of FLASH   */
>>> +#define CFG_FLASH_SIZE		8		/* FLASH size in MB */
>> Do you only support sizes bigger or equal to 1 MByte? What if a board
>> only has 512kBytes?
> 
> I object against such changes which have not been discussed with
> the maintainer of the board, and most probably have not been tested
> either.
> 
> TQ has a pretty long list of configuration options for their modules,
> and the change above will probably break quite a few, or most.

The above change is necessary to make the board compatible with the flash-size 
code in mpc83xx/start.S:

	/* Store 0x80000012 + log2(CFG_FLASH_SIZE) into LBLAWAR1 */
	lis r4, (0x80000012)@h
	ori r4, r4, (0x80000012)@l
	li r5, CFG_FLASH_SIZE
1:	srawi. r5, r5, 1	/* r5 = r5 >> 1 */
	addi r4, r4, 1
	bne 1b

Previously, this code just assumed that flash was 8MB.  However, the 8349ITX has 
16MB of flash.  Rather than have #ifdefs in start.S, this code was changed to 
calculate the correct value for LBLAWAR1 based on CFG_FLASH_SIZE.

I believe that this code is the most elegant solution to the problem, however it 
does require that each board header file define a value for CFG_FLASH_SIZE.  The 
alternative would be to add these lines above the new code:

#ifndef CFG_FLASH_SIZE
#define CFG_FLASH_SIZE 8
#endif

If you prefer that we use this approach, then we can make that change.  However, 
I think our approach makes more sense.

-- 
Timur Tabi
Linux Kernel Developer @ Freescale