On 05/26/2011 09:33 PM, Jean Delvare wrote:
> Hi Per,
Hi Jean,
I'll fix the sensors-detect and MAX6642 driver during (or right after)
the weekend in the way you suggest. Thanks for the tips.
>
> On Thu, 26 May 2011 11:27:27 +0200, Per Dalén wrote:
>> Changed according to the suggestions from Guenter Roeck:
>>
>> * Check the manufacturer ID directly. Fastens the test if it's not a
>> Maxim chip.
>> * Read the other non-existing registers after the manufacturer ID.
>
> THis is better but I'm not completely happy yet.
>
>> This patch adds detection of MAX6642 to sensors-detect.
>>
>> Signed-off-by: Per Dalen <per.dalen@xxxxxxxxxxxx>
>> ---
>>
>> --- prog/detect/sensors-detect (revision 5975)
>> +++ prog/detect/sensors-detect (working copy)
>> @@ -848,6 +848,11 @@
>> i2c_addrs => [0x2c, 0x2e, 0x2f],
>> i2c_detect => sub { max6639_detect(@_); },
>> }, {
>> + name => "Maxim MAX6642",
>> + driver => "max6642",
>> + i2c_addrs => [0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f],
>
> Better written: [0x48..0x4f]
>
>> + i2c_detect => sub { max6642_detect(@_); },
>> + }, {
>> name => "Maxim MAX6655/MAX6656",
>> driver => "max6655",
>> i2c_addrs => [0x18..0x1a, 0x29..0x2b, 0x4c..0x4e],
>> @@ -5779,6 +5784,36 @@
>> return 6;
>> }
>>
>> +# Chip to detect: MAX6642
>> +# Registers used:
>> +# 0x02: Status register
>> +# 0x03: Configuration register
>> +# 0xfe: Manufacturer ID
>> +# 0x04,0x06,0xff: No registers
>> +# We use the 0x04,0x06 and 0xff addresses (unused) to improve the
>
> Space after comma please.
>
>> reliability.
>
> Please disable line wrapping in your e-mail client before sending patches.
>
>> +# These are not real registers and will always return the last returned
>> value.
>> +# This isn't documented.
>> +sub max6642_detect
>> +{
>> + my ($file, $addr) = @_;
>> + my ($man_id, $conf, $status);
>> +
>> + $man_id = i2c_smbus_read_byte_data($file, 0xfe, NO_CACHE);
>> + return unless $man_id == 0x4d; # Maxim
>> + return if i2c_smbus_read_byte_data($file, 0x04, NO_CACHE) != $man_id;
>> + return if i2c_smbus_read_byte_data($file, 0x06, NO_CACHE) != $man_id;
>> + return if i2c_smbus_read_byte_data($file, 0xff, NO_CACHE) != $man_id;
>> + $status = i2c_smbus_read_byte_data($file, 0x02, NO_CACHE);
>
> At this point you should check the unused status bits immediately (so
> that you can exit if they don't meet the expectations.) Then you should
> read non-registers 0x04, 0x06 and 0xff again. This is the only reliable
> way to make sure that there are no registers at these addresses. It is
> perfectly possible (though unlikely) that a given chip does have values
> 0x4d for all of registers 0x04, 0x06, 0xfe and 0xff.
>
> See max6657_detect() for an example.
>
>> + $conf = i2c_smbus_read_byte_data($file, 0x03, NO_CACHE);
>
> NO_CACHE is not needed here, as this register 0x03 does exist. Only use
> NO_CACHE when you really need it, it's expensive!
>
>> +
>> + # Bit 5, 3, 1 and 0 should be zero
>> + return unless ($status & 0x2b) == 0x00;
>> + # The 4 lower bits should be zero
>> + return unless ($conf & 0x0f) == 0x00;
>> +
>> + return 6;
>> +}
>> +
>> sub max6655_detect
>> {
>> my ($file, $addr) = @_;
>
>
BR
Per
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
