On Fri, Jun 29, 2012 at 07:16:19PM +0530, Shilimkar, Santosh wrote:

Greg,

On Tue, Jun 26, 2012 at 1:19 PM, Shilimkar, Santosh
[off-list ref] wrote:

quoted

On Tue, Jun 26, 2012 at 10:56 AM, Rajendra Nayak [off-list ref] wrote:

quoted

On Tuesday 26 June 2012 10:53 AM, Shilimkar, Santosh wrote:

quoted

On Tue, Jun 26, 2012 at 10:49 AM, Rajendra Nayak[off-list ref] ?wrote:

quoted

On Tuesday 26 June 2012 10:39 AM, Shilimkar, Santosh wrote:

quoted

quoted

quoted

+#if defined(CONFIG_OF)

quoted

?> ? ?+static const struct of_device_id emif_of_match[] = {
?> ? ?+ ? ? ? ? ? ? { .compatible = "ti,emif-4d" },
?> ? ?+ ? ? ? ? ? ? { .compatible = "ti,emif-4d5" },
?> ? ?+ ? ? ? ? ? ? {},
?> ? ?+};
?> ? ?+MODULE_DEVICE_TABLE(of, emif_of_match);
?> ? ?+#endif
?> ? ?+
?> ? ? ?static struct platform_driver emif_driver = {
?> ? ? ? ? ? .remove ? ? ? ? = __exit_p(emif_remove),
?> ? ? ? ? ? .shutdown ? ? ? = emif_shutdown,
?> ? ? ? ? ? .driver = {
?> ? ? ? ? ? ? ? ? ? .name = "emif",
?> ? ?+#if defined(CONFIG_OF)
?> ? ?+ ? ? ? ? ? ? .of_match_table = of_match_ptr(emif_of_match),
?> ? ?+#endif

quoted

?Are these two still #if defined needed?

?Also, what about the IS_DEFINED macro?

Yes. To avoid the build break in !DT build.

No, infact of_match_ptr is there just so you can avoid a
#ifdef around and not break !CONFIG_OF.

Need to recheck but the build was throwing error becasue of
"emif_of_match"
mostly.

You still need the first one. I meant the second one around
'of_match_ptr' should be removed.

With !CONFIG_OF, of_match_ptr(_ptr) is just defined as NULL.

Yep. That make sense too.
Will drop the second check.

Posted the entire series again with updated $subject patch.
With your ack, I will ask Tony to pull this series for 3.6.

Sent them where?  I didn't see a copy here :(