Re: [PATCH] rtc: rtc-hid-sensor-time: fix possible bug on driver_remove

From: Alexander Holler
Date: Sat Jun 08 2013 - 04:56:46 EST


sorry to bother again. I've now had the time to play with the new hctosys I proposed patches for and rtc-hid-sensor-time.

If the new hctosys mechanism will end up in the kernel, the following patches I've recently send and which already got merged to -mm won't work anymore and should disappear:

[PATCH 3/4] rtc: rtc-hid-sensor-time: add option hctosys to set time at boot
[PATCH 4/4] rtc: rtc-hid-sensor-time: add support for milliseconds
[PATCH] rtc: rtc-hid-sensor-time: fix possible bug on driver_remove

The other two patches

[PATCH 1/4] rtc: rtc-hid-sensor-time: allow full years (16bit) in HID reports
[PATCH 2/4] rtc: rtc-hid-sensor-time: allow 16 and 32 bit values for all attributes.

which got merged too are still fine.

I don't know if I should send reverts for those 3 patches, or ...

I've already have a patch which would be needed if the new hctosys would be used, but I will delay sending them until we figured out if and how that new hctosys will become real. But you might already revert or delete the above 3 patches in your tree.

That silly looking driver rtc-hid-sensor-time is really an endless story. :/

Besides writing the driver and an example firmware implementation I had to make patches for iio, hid-sensor-hub and hid, and now for timekeeping and rtc. All for nothing but the fun having to convince maintainers of all sorts and bothering them. ;)

And the todo-list still contains some stuff:

- rtc-hid-sensor-time still does not get loaded automatically if a matching hid-sensor-hub appears to the system. That needs changes to hid and/or hid-sensor-hub. Problem is that I don't have a real hid sensor hub and I don't know why the people which do produce such products and have written hid-sensor-hub didn't have implemented a way to do such. Or I've just missed something.

- Have to redo millisecond support.

- ntp functionality (e.g. module parameter adjtime) to make it possible to use a hardware clock with support for milliseconds to turn the system clock to a stratum 1 clock.

- and last but not least I hope the standard for those HID sensors will become a standardized way to set the time. I still wouldn't like it if I had to "invent" something myself (easy, but it would be linux-specific and "invented" by some fool like me who just wrote it's own firmware without having any knowledge about some real hid-sensor products ;) ), but ...


Alexander Holler
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at