[U-Boot] [PATCH v4] at91: Update MEESC board support

[Tom]

Daniel Gorsulowski wrote:
> Hello Tom,
> 
> Tom wrote:
>> Daniel Gorsulowski wrote:
>>> This patch implements several updates:
>>> -disable CONFIG_ENV_OVERWRITE
>>> -add new hardware style variants and set the arch numbers appropriate
>>> -pass the serial# and hardware revision to the kernel
>>> -removed unused macros from include/configs/meesc.h
>>> -fixed multiline comment style
>>>
>>> Signed-off-by: Daniel Gorsulowski <Daniel.Gorsulowski at esd.eu>
>>> ---
>>> v2: - don't write the ethernet address to the EMAC module anymore
>>>
>>> v3: - removed function meesc_set_arch_number and moved code to checkboard()
>>>     - reworked function get_board_serial()
>>>     - removed unused macros from include/configs/meesc.h
>>>
>>> v4: - fixed commit message (to many characters)
>>>     - fixed indentation in switch-case statement
>>>     - fixed multiline comment style
>>>
>>>  board/esd/meesc/meesc.c |   65 +++++++++++++++++++++++++++++++++++++++++-----
>>>  include/configs/meesc.h |   25 ++++++------------
>>>  2 files changed, 66 insertions(+), 24 deletions(-)
>>>
>>> diff --git a/board/esd/meesc/meesc.c b/board/esd/meesc/meesc.c
>>> index 636d0ed..7cdc04a 100644
>>> --- a/board/esd/meesc/meesc.c
>>> +++ b/board/esd/meesc/meesc.c
>>> @@ -126,8 +126,10 @@ static void meesc_ethercat_hw_init(void)
>>>  		AT91_SMC_NRDPULSE_(4) | AT91_SMC_NCS_RDPULSE_(9));
>>>  	at91_sys_write(AT91_SMC1_CYCLE(0),
>>>  		AT91_SMC_NWECYCLE_(10) | AT91_SMC_NRDCYCLE_(5));
>>> -	/* Configure behavior at external wait signal, byte-select mode, 16 bit
>>> -	data bus width, none data float wait states and TDF optimization */
>>> +	/*
>>> +	 * Configure behavior at external wait signal, byte-select mode, 16 bit
>>> +	 * data bus width, none data float wait states and TDF optimization
>>> +	 */
>>>  	at91_sys_write(AT91_SMC1_MODE(0),
>>>  		AT91_SMC_READMODE | AT91_SMC_EXNWMODE_READY |
>>>  		AT91_SMC_BAT_SELECT | AT91_SMC_DBW_16 | AT91_SMC_TDF_(0) |
>>> @@ -156,8 +158,32 @@ int board_eth_init(bd_t *bis)
>>>  int checkboard(void)
>>>  {
>>>  	char str[32];
>>> -
>>> -	puts("Board: esd CAN-EtherCAT Gateway");
>>> +	u_char hw_type;	/* hardware type */
>>> +
>>> +	/* read the "Type" register of the ET1100 controller */
>>> +	hw_type = readb(CONFIG_ET1100_BASE);
>>> +
>>> +	switch (hw_type) {
>>> +	case 0x11:
>>> +	case 0x3F:
>>> +		/* ET1100 present, arch number of MEESC-Board */
>>> +		gd->bd->bi_arch_number = MACH_TYPE_MEESC;
>>> +		puts("Board: CAN-EtherCAT Gateway");
>>> +		break;
>>> +	case 0xFF:
>>> +		/* no ET1100 present, arch number of EtherCAN/2-Board */
>>> +		gd->bd->bi_arch_number = MACH_TYPE_ETHERCAN2;
>>> +		puts("Board: EtherCAN/2 Gateway");
>>> +		/* switch on LED1D */
>>> +		at91_set_gpio_output(AT91_PIN_PB12, 1);
>>> +		break;
>>> +	default:
>>> +		/* assume, no ET1100 present, arch number of EtherCAN/2-Board */
>>> +		gd->bd->bi_arch_number = MACH_TYPE_ETHERCAN2;
>>> +		printf("FATAL! Read invalid hw_type: %02X\n", hw_type);
>>> +		puts("Board: EtherCAN/2 Gateway");
>> You may want to soften this "FATAL" to "ERROR"
>> I would expect a "FATAL" to be followed by a call to hang()
>>
> Ok, I'll change this to "ERROR"
>>> +		break;
>>> +	}
>>>  	if (getenv_r("serial#", str, sizeof(str)) > 0) {
>>>  		puts(", serial# ");
>>>  		puts(str);
>>> @@ -167,6 +193,32 @@ int checkboard(void)
>>>  	return 0;
>>>  }
>>>  
>>> +#ifdef CONFIG_SERIAL_TAG
>>> +void get_board_serial(struct tag_serialnr *serialnr)
>>> +{
>>> +	char *str;
>>> +
>>> +	char *serial = getenv("serial#");
>>> +	if (serial) {
>>> +		str = strchr(serial, '_');
>>> +		if (str && (strlen(str) >= 4)) {
>>> +			serialnr->high = (*(str + 1) << 8) | *(str + 2);
>>> +			serialnr->low = simple_strtoul(str + 3, NULL, 16);
>> How is serial# set?
>> It seems like this is available if the user explicitly does a `setenv 
>> serial`
>> So at best this is a placeholder for future function.
>> The logic that decodes the variable assumes a format that a user is
>> not going, in general, to use.
> The serial# is set by factory start-up operation, so the expected format is
> guaranteed.
> And because of disabled CONFIG_ENV_OVERWRITE, it can never be changed
> again by the user. (Moreover, the user has no access to the serial console)

Excellent!
This is good enough for me.
If I see I newer version before the weekend, I will push that otherwise this
one is fine.

Ack
Tom

>>
>> Tom
> 
> Regards, Daniel