- /* Now, we do the remaining detection. There is no identification-
- dedicated register so we have to rely on several tricks:
- unused bits, registers cycling over 8-address boundaries,
- addresses 0x04-0x07 returning the last read value.
- The cycling+unused addresses combination is not tested,
- since it would significantly slow the detection down and would
- hardly add any value.
-
- The National Semiconductor LM75A is different than earlier
- LM75s. It has an ID byte of 0xaX (where X is the chip
- revision, with 1 being the only revision in existence) in
- register 7, and unused registers return 0xff rather than the
- last read value. */
+ /*
+ * Now, we do the remaining detection. There is no identification-
+ * dedicated register so we have to rely on several tricks:
+ * unused bits, registers cycling over 8-address boundaries,
+ * addresses 0x04-0x07 returning the last read value.
+ * The cycling+unused addresses combination is not tested,
+ * since it would significantly slow the detection down and would
+ * hardly add any value.
+ *
+ * The National Semiconductor LM75A is different than earlier
+ * LM75s. It has an ID byte of 0xaX (where X is the chip
+ * revision, with 1 being the only revision in existence) in
+ * register 7, and unused registers return 0xff rather than the
+ * last read value.
+ *
+ * Note that this function only detects the original National
+ * Semiconductor LM75 and the LM75A. Clones from other vendors
+ * aren't detected, on purpose, because they are typically never
+ * found on PC hardware. They are found on embedded designs where
+ * they can be instantiated explicitly so detection is not needed.
+ * The absence of identification registers on all these clones
+ * would make their exhaustive detection very difficult and weak,
+ * and odds are that the driver would bind to unsupported devices.
+ */