[U-Boot] [PATCH 20/28] armv8/ls2085ardb: Add support of LS2085ARDB platform

[York Sun]

On 03/20/2015 04:01 PM, Scott Wood wrote:
> On Thu, 2015-03-19 at 09:45 -0700, York Sun wrote:
>> The LS2080ARDB is a evaluation platform that supports LS2080A
>> family SoCs. This patch add sbasic support for the platform.
> 
> s/2080/2085/
> 
>> diff --git a/board/freescale/ls2085aqds/Makefile b/board/freescale/ls2085ardb/Makefile
>> similarity index 81%
>> copy from board/freescale/ls2085aqds/Makefile
>> copy to board/freescale/ls2085ardb/Makefile
>> index f174f33..0bfe21c 100644
>> --- a/board/freescale/ls2085aqds/Makefile
>> +++ b/board/freescale/ls2085ardb/Makefile
>> @@ -4,5 +4,5 @@
>>  # SPDX-License-Identifier:	GPL-2.0+
>>  #
>>  
>> -obj-y += ls2085aqds.o
>> +obj-y += ls2085ardb.o
>>  obj-y += ddr.o
>> diff --git a/board/freescale/ls2085aqds/README b/board/freescale/ls2085ardb/README
>> similarity index 73%
>> copy from board/freescale/ls2085aqds/README
>> copy to board/freescale/ls2085ardb/README
>> index a4d7b53..19f9d2a 100644
>> --- a/board/freescale/ls2085aqds/README
>> +++ b/board/freescale/ls2085ardb/README
>> @@ -1,10 +1,8 @@
>>  Overview
>>  --------
>> -The LS2080A Development System (QDS) is a high-performance computing,
>> +The LS2080A Reference Design (RDB) is a high-performance computing,
>>  evaluation, and development platform that supports the QorIQ LS2080A
>> -LayerScape Architecture processor. The LS2080AQDS provides validation and
>> -SW development platform for the Freescale LS2080A processor series, with
>> -a complete debugging environment.
>> +LayerScape Architecture processor.
> 
> I don't think Layerscape is supposed to be capitalized that way.
> 
>>  LS2085A SoC Overview
>>  ------------------
>> @@ -50,14 +48,11 @@ The LS2085A SoC includes the following function and features:
>>   - Service processor (SP) provides pre-boot initialization and secure-boot
>>    capabilities
>>  
>> - LS2080AQDS board Overview
>> + LS2080aRDB board Overview
> 
> Inconsistent capitalization
> 
>>  #ifndef __ASSEMBLY__
>>  unsigned long get_board_sys_clk(void);
>> -unsigned long get_board_ddr_clk(void);
>>  #endif
>>  
>>  #define CONFIG_SYS_CLK_FREQ		get_board_sys_clk()
>> -#define CONFIG_DDR_CLK_FREQ		get_board_ddr_clk()
>> +#define CONFIG_DDR_CLK_FREQ		133333333
>>  #define COUNTER_FREQUENCY_REAL		(CONFIG_SYS_CLK_FREQ/4)
> 
> Why?

The timer clock source is not a fixed clock. It derives from system clock,
dividing by 4. For this board, the system clock is fixed on board. But I don't
think the timer clock should be hard-coded.

> 
>> @@ -30,8 +32,8 @@ unsigned long get_board_ddr_clk(void);
>>  #define CONFIG_MEM_INIT_VALUE		0xdeadbeef
>>  #define SPD_EEPROM_ADDRESS1	0x51
>>  #define SPD_EEPROM_ADDRESS2	0x52
>> -#define SPD_EEPROM_ADDRESS3	0x53
>> -#define SPD_EEPROM_ADDRESS4	0x54
>> +#define SPD_EEPROM_ADDRESS3	0x54
>> +#define SPD_EEPROM_ADDRESS4	0x53	/* Bard error */
>>  #define SPD_EEPROM_ADDRESS5	0x55
>>  #define SPD_EEPROM_ADDRESS6	0x56	/* dummy address */
>>  #define SPD_EEPROM_ADDRESS	SPD_EEPROM_ADDRESS1
> 
> I suspect this should be s/Bard/Board/

Right.

> 
>> @@ -114,24 +106,24 @@ unsigned long get_board_ddr_clk(void);
>>  				| CSOR_NAND_ECC_DEC_EN  /* ECC on decode */ \
>>  				| CSOR_NAND_ECC_MODE_4  /* 4-bit ECC */ \
>>  				| CSOR_NAND_RAL_3	/* RAL = 2Byes */ \
>> -				| CSOR_NAND_PGS_2K	/* Page Size = 2K */ \
>> -				| CSOR_NAND_SPRZ_64/* Spare size = 64 */ \
>> -				| CSOR_NAND_PB(64))	/*Pages Per Block = 64*/
>> +				| CSOR_NAND_PGS_4K	/* Page Size = 2K */ \
>> +				| CSOR_NAND_SPRZ_224	/* Spare size = 64 */ \
>> +				| CSOR_NAND_PB(128))	/*Pages Per Block = 64*/
> 
> Comments don't match code.

Result of squashing. Will fix in next version.

> 
>> @@ -148,6 +140,7 @@ unsigned long get_board_ddr_clk(void);
>>  #define QIXIS_LBMAP_DFLTBANK		0x00
>>  #define QIXIS_LBMAP_ALTBANK		0x04
>>  #define QIXIS_RST_CTL_RESET		0x31
>> +#define QIXIS_RST_CTL_RESET_EN		0x30
>>  #define QIXIS_RCFG_CTL_RECONFIG_IDLE	0x20
>>  #define QIXIS_RCFG_CTL_RECONFIG_START	0x21
>>  #define QIXIS_RCFG_CTL_WATCHDOG_ENBLE	0x08
> 
> Is this really a difference between the two boards?
> 

RDB is designed to be compatible in many ways to QDS, but it is still different.
The QIXIS is compatible with most common registers, but not all of them.

I will send out a new version soon.

York