Hi Guenter,
On Tue, 09 Dec 2014 06:46:46 -0800, Guenter Roeck wrote:
> On 12/09/2014 02:45 AM, Jean Delvare wrote:
> > On Mon, 8 Dec 2014 19:42:40 -0800, Guenter Roeck wrote:
> >> + if ($chip == 8) { # TMP451
> >> + my $filter = i2c_smbus_read_byte_data($file, 0x24);
> >> +
> >> + return if $cid != 0x00;
> >> return if $rate > 0x09;
> >> - return if $mid != 0x55; # Texas Instruments
> >> - return 4 if ($cid == 0x00); # TMP451
> >> + return if $filter & 0xfc;
> >
> > If you want to improve reliability further, you could check for 4
> > unused bits in limit registers 0x12 and 0x15.
>
> Not sure if that would improve much. I would rather test for some
> non-zeros. I am going to get one of those chips and check if
> the consecutive alert register (0x22) has bit 0 set or not
> (the datasheet contradicts itself, and the bit is set in the
> register dump I have available).
Of course, testing for a non-zero bit is more valuable than testing for
a zero bit. But non-zero bits are rare.
> We could also check if (some of) the limit registers are non-zero.
> Would that make sense ?
Yes it would. For all chips without a reasonable chip ID register
value, every trick is welcome. You can also check that the low limit is
strictly lower than the high temperature limit, for example.
--
Jean Delvare
SUSE L3 Support
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
