On Wed, Nov 21, 2012 at 12:48:43PM +0100, Steffen Trumtrar wrote:

Hi!

On Wed, Nov 21, 2012 at 10:12:43AM +0000, Manjunathappa, Prakash wrote:

quoted

Hi Steffen,

On Tue, Nov 20, 2012 at 21:24:52, Steffen Trumtrar wrote:

quoted

+/**
+ * of_get_display_timings - parse all display_timing entries from a device_node
+ * @np: device_node with the subnodes
+ **/
+struct display_timings *of_get_display_timings(const struct device_node *np)
+{
+	struct device_node *timings_np;
+	struct device_node *entry;
+	struct device_node *native_mode;
+	struct display_timings *disp;
+
+	if (!np) {
+		pr_err("%s: no devicenode given\n", __func__);
+		return NULL;
+	}
+
+	timings_np = of_find_node_by_name(np, "display-timings");

I get below build warnings on this line
drivers/video/of_display_timing.c: In function 'of_get_display_timings':
drivers/video/of_display_timing.c:109:2: warning: passing argument 1 of 'of_find_node_by_name' discards qualifiers from pointer target type
include/linux/of.h:167:28: note: expected 'struct device_node *' but argument is of type 'const struct device_node *'

quoted

+ * of_display_timings_exists - check if a display-timings node is provided
+ * @np: device_node with the timing
+ **/
+int of_display_timings_exists(const struct device_node *np)
+{
+	struct device_node *timings_np;
+
+	if (!np)
+		return -EINVAL;
+
+	timings_np = of_parse_phandle(np, "display-timings", 0);

Also here:
drivers/video/of_display_timing.c: In function 'of_display_timings_exists':
drivers/video/of_display_timing.c:209:2: warning: passing argument 1 of 'of_parse_phandle' discards qualifiers from pointer target type
include/linux/of.h:258:28: note: expected 'struct device_node *' but argument is of type 'const struct device_node *'

The warnings are because the of-functions do not use const pointers where they
should. I had two options: don't use const pointers even if they should be and
have no warnings or use const pointers and have a correct API. (Third option:
send patches for of-functions). I chose the second option.

Maybe a better approach would be a combination of 1 and 3: don't use
const pointers for struct device_node for now and bring the issue up
with the OF maintainers, possibly with patches attached that fix the
problematic functions.

Thierry