Re: [PATCH v3 3/4] Watchdog: sp5100_tco: Add initialization using EFCH MMIO

[Terry Bowman]

On 1/24/22 11:36 AM, Jean Delvare wrote:
...
>> +	res = request_mem_region(EFCH_PM_ACPI_MMIO_PM_ADDR,
>> +				 EFCH_PM_ACPI_MMIO_PM_SIZE,
>> +				 "sp5100_tco");
>> +
>> +	if (!res) {
>> +		dev_err(dev,
>> +			"SMB base address memory region 0x%x already in use.\n",
> 
> SMB -> SMBus
> 

Yes, I will update.

>> +			EFCH_PM_ACPI_MMIO_PM_ADDR);
>> +		return -EBUSY;
>> +	}
>> +
>> +	addr = ioremap(EFCH_PM_ACPI_MMIO_PM_ADDR,
>> +		       EFCH_PM_ACPI_MMIO_PM_SIZE);
>> +	if (!addr) {
>> +		release_resource(res);
>> +		dev_err(dev, "SMB base address mapping failed.\n");
> 
> SMB -> SMBus
> 

Yes, I will update.

>> +		return -ENOMEM;
>> +	}
>> +
> 
> A short comment saying what the next command is doing would be
> appreciated.
> 

I will add a brief comment explaining.

>> +	if (!(efch_read_pm_reg8(addr, EFCH_PM_DECODEEN) &
>> +	      EFCH_PM_DECODEEN_WDT_TMREN)) {
> 
> I find such splits hard to read. If checkpatch complains when you don't
> split it (but I think it no longer does, right?) then just introduce a
> local variable to store the register value. Same for the 2 occurrences
> below.
> 

I will update for readability.

>> +		efch_update_pm_reg8(addr, EFCH_PM_DECODEEN,
>> +				    0xff,
>> +				    EFCH_PM_DECODEEN_WDT_TMREN);
> 
> Easily fits in one fewer line.
> 

I will change to 2 lines.

>> +	}
>> +
>> +	/* Determine MMIO base address */
>> +	if (efch_read_pm_reg8(addr, EFCH_PM_DECODEEN) &
>> +	    EFCH_PM_DECODEEN_WDT_TMREN)
>> +		mmio_addr = EFCH_PM_WDT_ADDR;
>> +
>> +	/* Determine alternate MMIO base address */
>> +	if (efch_read_pm_reg8(addr, EFCH_PM_ISACONTROL) &
>> +	    EFCH_PM_ISACONTROL_MMIOEN)
>> +		alt_mmio_addr = EFCH_PM_ACPI_MMIO_ADDR +
>> +			EFCH_PM_ACPI_MMIO_WDT_OFFSET;
>> +
>> +	ret = sp5100_tco_prepare_base(tco, mmio_addr, alt_mmio_addr, dev_name);
>> +	if (!ret) {
>> +		tco_timer_enable_mmio(addr);
>> +		ret = sp5100_tco_timer_init(tco);
>> +	}
>> +
>> +	iounmap(addr);
>> +	release_resource(res);
>> +
>> +	return ret;
>> +}
>> +
>>  static int sp5100_tco_setupdevice(struct device *dev,
>>  				  struct watchdog_device *wdd)
>>  {
>> @@ -327,6 +410,9 @@ static int sp5100_tco_setupdevice(struct device *dev,
>>  	u32 alt_mmio_addr = 0;
>>  	int ret;
>>  
>> +	if (tco->tco_reg_layout == efch_mmio)
>> +		return sp5100_tco_setupdevice_mmio(dev, wdd);
>> +
>>  	/* Request the IO ports used by this driver */
>>  	if (!request_muxed_region(SP5100_IO_PM_INDEX_REG,
>>  				  SP5100_PM_IOPORTS_SIZE, "sp5100_tco")) {
>> diff --git a/drivers/watchdog/sp5100_tco.h b/drivers/watchdog/sp5100_tco.h
>> index adf015aa4126..2df8f8b2c55b 100644
>> --- a/drivers/watchdog/sp5100_tco.h
>> +++ b/drivers/watchdog/sp5100_tco.h
>> @@ -83,3 +83,8 @@
>>  
>>  #define EFCH_PM_ACPI_MMIO_ADDR		0xfed80000
>>  #define EFCH_PM_ACPI_MMIO_WDT_OFFSET	0x00000b00
>> +#define EFCH_PM_ACPI_MMIO_PM_OFFSET	0x00000300
>> +
>> +#define EFCH_PM_ACPI_MMIO_PM_ADDR	(EFCH_PM_ACPI_MMIO_ADDR +	\
>> +					 EFCH_PM_ACPI_MMIO_PM_OFFSET)
>> +#define EFCH_PM_ACPI_MMIO_PM_SIZE       8
> 
> Other than these minor details, patch looks good to me, thanks.
> 
> Tested-by: Jean Delvare <jdelvare@xxxxxxx>
> 

I'm finishing up revisions for this and the i2c patch. I expect to send both tomorrow.

Regards,
Teryr