On 11/25/2013 05:41 PM, Heikki Krogerus wrote:
On Sat, Nov 23, 2013 at 05:59:30PM +0900, Alexandre Courbot wrote:
Wouldn't it be possible (and simpler) to move patch 2 of your series
to first position, and then to merge patch 1 and 3 together in second
position? It seems to me that you are basically undoing much of the
work of your first patch here (notably by removing
rfkill_gpio_convert_to_desc() which ends up having a very short life)
and that this could be avoided if you defined the platform lookup
tables first.
Doing so would avoid prevent you from using gpio_to_desc() which you
should never ever use anyway. :P
Adding the lookup table in first patch and then changing the driver in
the second creates a point to the history where this driver stops
working on this platform, which is something I'm not willing to do.
Does it? If you just add a lookup table and keep using the integer-based
GPIO interface, then your lookup table will not be used by anyone and
will basically be a no-op. Then you can switch to the GPIO descriptor
interface and take advantage of the lookup table. Unless I missed
something there should not be any point that breaks in the git history.
(to be clear: the first patch should *only* contain the lookup table,
and the second be a merge of the current patches 1 and 3 of this series.)
But, we can make one patch out of all three if everybody is OK with
that.
IIUC platform changes should be distinct from drivers whenever possible,
so this is probably not the best choice here.
Alex.
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
