[PATCH] OMAPDSS: fix omapfb build error due missing feat functions declaration

[PATCH] OMAPDSS: fix omapfb build error due missing feat functions declaration

[Javier Martinez Canillas]

The omapfb is failing to build in -next due missing declarations for
dss_feat_get_supported_displays() and dss_feat_get_supported_outputs():

 CC [M]  drivers/video/fbdev/omap2//omapfb/dss/dss.o
drivers/video/fbdev/omap2//omapfb/dss/dss.c: In function 'dss_save_context':
drivers/video/fbdev/omap2//omapfb/dss/dss.c:144:2: error: implicit declaration of function 'dss_feat_get_supported_displays' [-Werror=implicit-function-declaration]

Add the declaration for these functions in the dss_features.h header
file to fix this compile error.

Also, remove the functions export since are not used outside the driver.

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>

---

Hello Tomi,

This seems to be caused by a wrong order in which some commits from
the omapdss tree [0] were merged, or rather a wrong base used for
the dss-drm-merge (4.4.0-rc4) and the for-next (4.4.0-rc5) branches.

The problem is that commit 2ea164a9af9c ("drm/omap: move omapdss &
displays under omapdrm") moves the dss_features.h header from the
drivers/video/fbdev/omap2/dss/ dir to drivers/gpu/drm/omapdrm}/dss.

And then commit 8b987a111b1a ("omapfb: copy omapdss & displays for
omapfb" copies the hader from the omapdrm dir to the omapfb one.

Now, since commit 66c78ddd7a2e ("OMAPDSS: make a two dss feat funcs
internal to omapdss") is the one that moves the function declaration
from the public include/video/omapdss.h header to the internal one,
this means that git will be smart enough to identify that the header
that was drivers/video/fbdev/omap2/dss/dss_features.h now is the one
in the omapdrm directory but that means that the omapfb files won't
get the changes since the move & copy happened before.

This happens AFAIU because dss-drm-merge having an older base so the
patches in for-next are rebased on top of dss-drm-merge when merging.

The issue can easily be avoided by first merging the patch that does
the feat funcs declaration move and then merging the ones that do the
per driver omapdss refactoring.

But I'm posting this patch anyways in case you can't redo your branches
and need a fix to apply on top of your current for-next.

[0]: git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux.git

Best regards,
Javier

 drivers/video/fbdev/omap2/omapfb/dss/dss_features.c | 2 --
 drivers/video/fbdev/omap2/omapfb/dss/dss_features.h | 3 +++
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/video/fbdev/omap2/omapfb/dss/dss_features.c b/drivers/video/fbdev/omap2/omapfb/dss/dss_features.c
index b0b6dfd657bf..a2e26b8d6ded 100644
--- a/drivers/video/fbdev/omap2/omapfb/dss/dss_features.c
+++ b/drivers/video/fbdev/omap2/omapfb/dss/dss_features.c
@@ -844,13 +844,11 @@ enum omap_display_type dss_feat_get_supported_displays(enum omap_channel channel
 {
 	return omap_current_dss_features->supported_displays[channel];
 }
-EXPORT_SYMBOL(dss_feat_get_supported_displays);
 
 enum omap_dss_output_id dss_feat_get_supported_outputs(enum omap_channel channel)
 {
 	return omap_current_dss_features->supported_outputs[channel];
 }
-EXPORT_SYMBOL(dss_feat_get_supported_outputs);
 
 enum omap_color_mode dss_feat_get_supported_color_modes(enum omap_plane plane)
 {
diff --git a/drivers/video/fbdev/omap2/omapfb/dss/dss_features.h b/drivers/video/fbdev/omap2/omapfb/dss/dss_features.h
index 100f7a2d0638..4474ebff8607 100644
--- a/drivers/video/fbdev/omap2/omapfb/dss/dss_features.h
+++ b/drivers/video/fbdev/omap2/omapfb/dss/dss_features.h
@@ -102,4 +102,7 @@ bool dss_feat_rotation_type_supported(enum omap_dss_rotation_type rot_type);
 bool dss_has_feature(enum dss_feat_id id);
 void dss_feat_get_reg_field(enum dss_feat_reg_field id, u8 *start, u8 *end);
 void dss_features_init(enum omapdss_version version);
+
+enum omap_display_type dss_feat_get_supported_displays(enum omap_channel channel);
+enum omap_dss_output_id dss_feat_get_supported_outputs(enum omap_channel channel);
 #endif
-- 
2.4.3

Re: [PATCH] OMAPDSS: fix omapfb build error due missing feat functions declaration

[Tomi Valkeinen]

[-- Attachment #1: Type: text/plain, Size: 1430 bytes --]

Hi Javier,

On 21/12/15 20:26, Javier Martinez Canillas wrote:
> The omapfb is failing to build in -next due missing declarations for
> dss_feat_get_supported_displays() and dss_feat_get_supported_outputs():
> 
>  CC [M]  drivers/video/fbdev/omap2//omapfb/dss/dss.o
> drivers/video/fbdev/omap2//omapfb/dss/dss.c: In function 'dss_save_context':
> drivers/video/fbdev/omap2//omapfb/dss/dss.c:144:2: error: implicit declaration of function 'dss_feat_get_supported_displays' [-Werror=implicit-function-declaration]
> 
> Add the declaration for these functions in the dss_features.h header
> file to fix this compile error.
> 
> Also, remove the functions export since are not used outside the driver.

Thanks!

Yep, I messed that up. I thought I had tested it, but apparently I only
tested the final for-next only for omapdrm, which does compile and work.

omapdss.h is the the only file still shared between omapdrm and omapfb
after the copy-omapdss-series (I'll work on omapdss.h later), and of
course there was a change to omapdss.h which broke the build.

So I rebased the copy-omapdss-series on top of the rest of the omapdss
patches, and updated the "omapfb: copy omapdss & displays for omapfb" to
make a fresh copy of omapdss for omapfb. I think it's better to update
the series, rather than applying fixes for already confusing series.

I've pushed new version to my for-next branch.

 Tomi


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [PATCH] OMAPDSS: fix omapfb build error due missing feat functions declaration

[Javier Martinez Canillas]

Hello Tomi,

On 12/22/2015 04:50 AM, Tomi Valkeinen wrote:
> Hi Javier,
> 
> On 21/12/15 20:26, Javier Martinez Canillas wrote:
>> The omapfb is failing to build in -next due missing declarations for
>> dss_feat_get_supported_displays() and dss_feat_get_supported_outputs():
>>
>>  CC [M]  drivers/video/fbdev/omap2//omapfb/dss/dss.o
>> drivers/video/fbdev/omap2//omapfb/dss/dss.c: In function 'dss_save_context':
>> drivers/video/fbdev/omap2//omapfb/dss/dss.c:144:2: error: implicit declaration of function 'dss_feat_get_supported_displays' [-Werror=implicit-function-declaration]
>>
>> Add the declaration for these functions in the dss_features.h header
>> file to fix this compile error.
>>
>> Also, remove the functions export since are not used outside the driver.
> 
> Thanks!
>

You are welcome.
 
> Yep, I messed that up. I thought I had tested it, but apparently I only
> tested the final for-next only for omapdrm, which does compile and work.
>

Yes, that's what I thought.
 
> omapdss.h is the the only file still shared between omapdrm and omapfb
> after the copy-omapdss-series (I'll work on omapdss.h later), and of
> course there was a change to omapdss.h which broke the build.
> 
> So I rebased the copy-omapdss-series on top of the rest of the omapdss
> patches, and updated the "omapfb: copy omapdss & displays for omapfb" to
> make a fresh copy of omapdss for omapfb. I think it's better to update
> the series, rather than applying fixes for already confusing series.
>

I completely agree, maintaining bisectability is important. As I mentioned
in my last email, I just posted in case you couldn't rework your branches
for whatever reason.
 
> I've pushed new version to my for-next branch.
>

Great, thanks!

>  Tomi
> 

Best regards,
-- 
Javier Martinez Canillas
Open Source Group
Samsung Research America