[Wolfgang] > > > >This means you have a RTC on the I2C bus, right? Same here. It works > >fine on all baords except those with a I2C based RTC. Which is why we > >detected the problem so late. >
[Eugene] > It looks like a bug in m41t00_set_rtc_time. > > Generic I2C layer is very high-level subsystem and can not be > used from the interrupt context. > If the i2c layer cannot be used in the (e.g. timer) interrupt context, wouldn't that preclude the use of the i2c layer to provide support (i.e. clock updates) to an i2c-based rtc? Any suggestions on how to work around this? Wolfgang: I know you had some examples of i2c-based rtc support in your tree. Assuming the above is the cause behind the oops, does your driver preclude rtc updates from the timer interrupt context? Some time ago I queried the list on the best way to support an i2c-based RTC; at that time, there was no consensus. Any more thoughts along these lines? Regards, Tim ** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
