On Mon, Jul 27, 2015 at 05:30:47PM +0300, Vladimir Zapolskiy wrote: > The series fixes i2c bus device refcounting for clients of i2c_get_adapter(), > of_find_i2c_adapter_by_node() and of_find_i2c_device_by_node() interfaces. Yay, thanks for working on that! > The v2 3/4 change adds and exports new of_get_i2c_adapter_by_node() > interface of i2c core, v2 4/4 is an unchanged version of v1 10/10, which > utilizes this new interface in order to fix the refcounting bug described > below in detail. And thanks for the good descriptions, too. > To fix existing users of these interfaces and to avoid any further > confusion and misusage in future, add one more interface > of_get_i2c_adapter_by_node(), it is similar to i2c_get_adapter() in > sense that an I2C bus device driver found and locked by user can be > correctly unlocked by i2c_put_adapter(). I tend to agree to the idea of the new function, should be less error prone to users. Need to think about it a day more, though. > Changes from v1 to v2: > * added two more patches 1/4 and 2/4 fixing i2c bus device refcounting, > thanks to Thierry for the idea of 2/4 change, I'll apply 1/4 to for-current, since this is a clear bugfix. 2/4 seems very worthwhile, too, but seems like less a bugfix to me; while it changes things to be more correct, it also is a preparation for the following patches. > * defer changes in 8 broken DRM and fbdev clients until i2c-core changes > are added to linux-next (or preferably 4.2.0-rc Linus' branch, if possible) > to improve signal/noise ratio on mailing lists. That was a clever thing to do. However, my gut feeling is that these changes to refcounting behaviour should go via the next merge window to get proper testing. If you guys want that in 4.2, then I would need a lot of Tested, Acked, and Reviewed-by tags very soon. Thanks again, Wolfram
Attachment:
signature.asc
Description: Digital signature
