On Mon, Feb 29, 2016 at 11:36:49AM +0800, Shawn Guo wrote:

On Mon, Feb 22, 2016 at 09:01:53AM -0300, Fabio Estevam wrote:

quoted

From: Fabio Estevam <redacted>

Currently we have an incorrect behaviour when multiple devices
are present under the weim node. For example:

&weim {
	...
	status = "okay";
	
	sram at 0,0 {
		...
        	status = "okay";
	};

	mram at 0,0 {
		...
        	status = "disabled";
    	};
};

In this case only the 'sram' device should be probed and not 'mram'.

However what happens currently is that the status variable is ignored,
causing the 'sram' device to be disabled and 'mram' to be enabled.  

Change the weim_parse_dt() function to use
for_each_available_child_of_node()so that the devices marked with
'status = disabled' are not probed.

Cc: <redacted>
Suggested-by: Wolfgang Netbal <redacted>
Signed-off-by: Fabio Estevam <redacted>

Acked-by: Shawn Guo <shawnguo@kernel.org>

Arnd, Olof,

I do not have any other 'driver' patches queued, so please help directly
apply this one.  Considering this fixes a real problem, it would be good
if we can merge this through -rc.  But we understand that it's -rc6 now,
and this doesn't fix a regression or so-critical issue, so it should be
fine to queue the patch for the next release as well.

Sorry for the high latency here, I had missed this patch and noticed now when
I did a sweep. I've merged it into next/drivers for 4.6.


-Olof