On Wed, Oct 6, 2010 at 12:31 PM, Guenter Roeck <guenter.roeck@xxxxxxxxxxxx> wrote: > On Wed, 2010-10-06 at 14:23 -0400, Paul Thomas wrote: >> On Wed, Oct 6, 2010 at 11:24 AM, Jonathan Cameron <jic23@xxxxxxxxx> wrote: >> > On 10/06/10 19:13, Paul Thomas wrote: >> >> I'm planning a rev of the ads7871 driver. I want to add in the ability >> >> to set the gain on the chip as well as do differential reads. To do >> >> this I'm adding a bunch of SENSOR_DEVICE_ATTR lines. Awhile ago I had >> >> got a suggestion from Jonathan to use a local macro for these >> >> statements. Would that be a macro that looped through calling the >> >> other macro 8 times with values 0-7? Could someone point me to an >> >> example where this is used? >> > I suspect I meant something like >> > >> > #define ADS7871_VOLTAGE_ATTR(n) SENSOR_DEVICE_ATTR(in##n##_input, S_IRUGO, show_voltage, NULL, n); >> > But looking back at the usage, it really is a matter of taste, and I'm >> > not sure my suggestion was a good one in this case! >> > >> > Jonathan >> > >> >> OK, it's not much to just put in all the lines. So I'm adding >> attributes inX_gain and inX_diff for each of the 8 channels. The diff >> option sets whether it's a differential read or a single ended read to >> ground. >> > Since you plan to amend the ABI, it might make sense to discuss the > proposed attributes and their usage first. > > I would assume that hardware monitoring sensors have a static gain, ie a > gain which does not have to be set during runtime, but is fixed for a > specific board. I don't know about the diff attribute, but it also seems > to be a board option, not a runtime option. > > For such parameters, it might make more sense to define a common hwmon > platform parameter structure and expect it to be set during board > initialization. Works quite nicely for other drivers. > > Thanks, > Guenter No, both of those are runtime options internal to the ads7871, and because we initiate a read by setting that register anyway there isn't any additional setup cost (at least there isn't any spi cost, there is a little additional code in show_voltage). Additionally gain & diff are initialized to a gain of 1 and single ended reads which will match how the driver works now. I'm open to other suggestions also. thanks, Paul _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
