Hi Tomi,
After merging the omap_dss2 tree, today's linux-next build (powerpc
allyesconfig) failed like this:
drivers/video/fbdev/simplefb.c: In function 'simplefb_clocks_init':
drivers/video/fbdev/simplefb.c:207:2: error: implicit declaration of function 'of_clk_get_parent_count' [-Werror=implicit-function-declaration]
par->clk_count = of_clk_get_parent_count(np);
^
Caused by commit fc219bfd5b2e ("simplefb: add clock handling code").
This build does not have CONFIG_COMMON_CLK set.
I have added the following fix for today (there is probably a better
one):
From: Stephen Rothwell <[email protected]>
Date: Mon, 17 Nov 2014 19:05:42 +1100
Subject: [PATCH] simplefb: make it depend on CONFIG_COMMON_CLK
Signed-off-by: Stephen Rothwell <[email protected]>
---
drivers/video/fbdev/Kconfig | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/video/fbdev/Kconfig b/drivers/video/fbdev/Kconfig
index 025b439d4fe1..1f0298e166c3 100644
--- a/drivers/video/fbdev/Kconfig
+++ b/drivers/video/fbdev/Kconfig
@@ -2457,6 +2457,7 @@ config FB_HYPERV
config FB_SIMPLE
bool "Simple framebuffer support"
depends on (FB = y)
+ depends on COMMON_CLK
select FB_CFB_FILLRECT
select FB_CFB_COPYAREA
select FB_CFB_IMAGEBLIT
--
2.1.3
--
Cheers,
Stephen Rothwell [email protected]
Hi,
Thanks Stephen!
Hans, have you had a look at this? kbuild test bot also reported this.
Tomi
On 17/11/14 10:09, Stephen Rothwell wrote:
> Hi Tomi,
>
> After merging the omap_dss2 tree, today's linux-next build (powerpc
> allyesconfig) failed like this:
>
> drivers/video/fbdev/simplefb.c: In function 'simplefb_clocks_init':
> drivers/video/fbdev/simplefb.c:207:2: error: implicit declaration of function 'of_clk_get_parent_count' [-Werror=implicit-function-declaration]
> par->clk_count = of_clk_get_parent_count(np);
> ^
>
> Caused by commit fc219bfd5b2e ("simplefb: add clock handling code").
> This build does not have CONFIG_COMMON_CLK set.
>
> I have added the following fix for today (there is probably a better
> one):
>
> From: Stephen Rothwell <[email protected]>
> Date: Mon, 17 Nov 2014 19:05:42 +1100
> Subject: [PATCH] simplefb: make it depend on CONFIG_COMMON_CLK
>
> Signed-off-by: Stephen Rothwell <[email protected]>
> ---
> drivers/video/fbdev/Kconfig | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/video/fbdev/Kconfig b/drivers/video/fbdev/Kconfig
> index 025b439d4fe1..1f0298e166c3 100644
> --- a/drivers/video/fbdev/Kconfig
> +++ b/drivers/video/fbdev/Kconfig
> @@ -2457,6 +2457,7 @@ config FB_HYPERV
> config FB_SIMPLE
> bool "Simple framebuffer support"
> depends on (FB = y)
> + depends on COMMON_CLK
> select FB_CFB_FILLRECT
> select FB_CFB_COPYAREA
> select FB_CFB_IMAGEBLIT
>
Hi
On Tue, Nov 18, 2014 at 3:53 PM, Tomi Valkeinen <[email protected]> wrote:
> Hi,
>
> Thanks Stephen!
>
> Hans, have you had a look at this? kbuild test bot also reported this.
The fix from Stephen should not go to main-line. It breaks simplefb on
x86 without OF. Imo, linux/clk.h should provide a fallback if OF is
not defined (like it already does for of_clk_get() and friends).
Hans, any comments?
Thanks
David
On 18/11/14 17:07, David Herrmann wrote:
> Hi
>
> On Tue, Nov 18, 2014 at 3:53 PM, Tomi Valkeinen <[email protected]> wrote:
>> Hi,
>>
>> Thanks Stephen!
>>
>> Hans, have you had a look at this? kbuild test bot also reported this.
>
> The fix from Stephen should not go to main-line. It breaks simplefb on
> x86 without OF. Imo, linux/clk.h should provide a fallback if OF is
> not defined (like it already does for of_clk_get() and friends).
>
> Hans, any comments?
Hans, ping.
Tomi
Hi Tomi,
On 11/25/2014 12:00 PM, Tomi Valkeinen wrote:
> On 18/11/14 17:07, David Herrmann wrote:
>> Hi
>>
>> On Tue, Nov 18, 2014 at 3:53 PM, Tomi Valkeinen <[email protected]> wrote:
>>> Hi,
>>>
>>> Thanks Stephen!
>>>
>>> Hans, have you had a look at this? kbuild test bot also reported this.
>>
>> The fix from Stephen should not go to main-line. It breaks simplefb on
>> x86 without OF. Imo, linux/clk.h should provide a fallback if OF is
>> not defined (like it already does for of_clk_get() and friends).
>>
>> Hans, any comments?
>
> Hans, ping.
Sorry I somehow missed this thread so far, I did see the kbuild mail, but
it looked like a false positive to me, for the compile error case I already fixed
in the last revision of the patch, adding #ifdef CONFIG_OF around the offending code.
But on closer inspection I see that the entirety of linux/clk-provider.h is wrapped
in one huge #ifdef CONFIG_COMMON_CLK and then later on also has:
#ifdef CONFIG_OF
...
int of_clk_get_parent_count(struct device_node *np);
...
#endif
So we need both CONFIG_OF and CONFIG_COMMON_CLK, the attached patch
should fix this.
Regards,
Hans
On 25/11/14 13:20, Hans de Goede wrote:
> Hi Tomi,
>
> On 11/25/2014 12:00 PM, Tomi Valkeinen wrote:
>> On 18/11/14 17:07, David Herrmann wrote:
>>> Hi
>>>
>>> On Tue, Nov 18, 2014 at 3:53 PM, Tomi Valkeinen
>>> <[email protected]> wrote:
>>>> Hi,
>>>>
>>>> Thanks Stephen!
>>>>
>>>> Hans, have you had a look at this? kbuild test bot also reported this.
>>>
>>> The fix from Stephen should not go to main-line. It breaks simplefb on
>>> x86 without OF. Imo, linux/clk.h should provide a fallback if OF is
>>> not defined (like it already does for of_clk_get() and friends).
>>>
>>> Hans, any comments?
>>
>> Hans, ping.
>
> Sorry I somehow missed this thread so far, I did see the kbuild mail, but
> it looked like a false positive to me, for the compile error case I
> already fixed
> in the last revision of the patch, adding #ifdef CONFIG_OF around the
> offending code.
>
> But on closer inspection I see that the entirety of linux/clk-provider.h
> is wrapped
> in one huge #ifdef CONFIG_COMMON_CLK and then later on also has:
>
> #ifdef CONFIG_OF
> ...
> int of_clk_get_parent_count(struct device_node *np);
> ...
> #endif
>
> So we need both CONFIG_OF and CONFIG_COMMON_CLK, the attached patch
> should fix this.
Thanks, applied.
Tomi
Hi Tomi,
On Tue, 25 Nov 2014 14:11:59 +0200 Tomi Valkeinen <[email protected]> wrote:
>
> On 25/11/14 13:20, Hans de Goede wrote:
> >
> > On 11/25/2014 12:00 PM, Tomi Valkeinen wrote:
> >> On 18/11/14 17:07, David Herrmann wrote:
> >>>
> >>> On Tue, Nov 18, 2014 at 3:53 PM, Tomi Valkeinen
> >>> <[email protected]> wrote:
> >>>>
> >>>> Thanks Stephen!
> >>>>
> >>>> Hans, have you had a look at this? kbuild test bot also reported this.
> >>>
> >>> The fix from Stephen should not go to main-line. It breaks simplefb on
> >>> x86 without OF. Imo, linux/clk.h should provide a fallback if OF is
> >>> not defined (like it already does for of_clk_get() and friends).
> >>>
> >>> Hans, any comments?
> >>
> >> Hans, ping.
> >
> > Sorry I somehow missed this thread so far, I did see the kbuild mail, but
> > it looked like a false positive to me, for the compile error case I
> > already fixed
> > in the last revision of the patch, adding #ifdef CONFIG_OF around the
> > offending code.
> >
> > But on closer inspection I see that the entirety of linux/clk-provider.h
> > is wrapped
> > in one huge #ifdef CONFIG_COMMON_CLK and then later on also has:
> >
> > #ifdef CONFIG_OF
> > ...
> > int of_clk_get_parent_count(struct device_node *np);
> > ...
> > #endif
> >
> > So we need both CONFIG_OF and CONFIG_COMMON_CLK, the attached patch
> > should fix this.
>
> Thanks, applied.
But not pushed out, yet? Just checking that you haven't pushed on
purpose.
--
Cheers,
Stephen Rothwell [email protected]
On 26/11/14 00:07, Stephen Rothwell wrote:
>>> So we need both CONFIG_OF and CONFIG_COMMON_CLK, the attached patch
>>> should fix this.
>>
>> Thanks, applied.
>
> But not pushed out, yet? Just checking that you haven't pushed on
> purpose.
Not yet. I'll be pushing today.
Tomi