6 Jan
2011
6 Jan
'11
6:01 p.m.
Dimitris Papastamos wrote:
The semantics behind this code is that if the driver provides a reg_cache_size and a reg_word_size it *should* provide a defaults register cache as well.
I *had* code which did this, but apparently it broke somewhere. Oh well.
If you want to manage your own register cache in the driver which is not advised, you will have to add similar functionality in your _priv struct. If you require more flexible functionality you need to consider implementing a sensible strategy and submitting it as a patch.
I will fix my driver to restore the functionality.
--
Timur Tabi
Linux kernel developer at Freescale