When none of the OMAP4-generation SoCs are enabled, we run into a link
error for am43xx/am43xx:
drivers/clk/ti/dpll.o: In function `of_ti_am3_dpll_x2_setup':
dpll.c:(.init.text+0xd8): undefined reference to `clkhwops_omap4_dpllmx'
This is easily fixed by adding another #ifdef.
While looking at the code, I also spotted another problem with the
assignment of hw_ops variable that is not used again later. I'm
changing this to setting clk_hw->ops instead, which I guess is what
was intended here.
Fixes: 0565fb168d63 ("clk: ti: dpll: move omap3 DPLL functionality to clock driver")
Signed-off-by: Arnd Bergmann <[email protected]>
---
drivers/clk/ti/dpll.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/clk/ti/dpll.c b/drivers/clk/ti/dpll.c
index 96d84888c6c5..fcf297fd2233 100644
--- a/drivers/clk/ti/dpll.c
+++ b/drivers/clk/ti/dpll.c
@@ -312,7 +312,6 @@ static void _register_dpll_x2(struct device_node *node,
struct clk_hw_omap *clk_hw;
const char *name = node->name;
const char *parent_name;
- int ret;
parent_name = of_clk_get_parent_name(node, 0);
if (!parent_name) {
@@ -332,16 +331,21 @@ static void _register_dpll_x2(struct device_node *node,
init.parent_names = &parent_name;
init.num_parents = 1;
+#if defined(CONFIG_ARCH_OMAP4) || defined(CONFIG_SOC_OMAP5) || \
+ defined(CONFIG_SOC_DRA7XX)
if (hw_ops == &clkhwops_omap4_dpllmx) {
+ int ret;
+
/* Check if register defined, if not, drop hw-ops */
ret = of_property_count_elems_of_size(node, "reg", 1);
if (ret <= 0) {
- hw_ops = NULL;
+ clk_hw->ops = NULL;
} else if (ti_clk_get_reg_addr(node, 0, &clk_hw->clksel_reg)) {
kfree(clk_hw);
return;
}
}
+#endif
/* register the clock */
clk = ti_clk_register(NULL, &clk_hw->hw, name);
--
2.9.0
On 20/04/17 00:43, Arnd Bergmann wrote:
> When none of the OMAP4-generation SoCs are enabled, we run into a link
> error for am43xx/am43xx:
>
> drivers/clk/ti/dpll.o: In function `of_ti_am3_dpll_x2_setup':
> dpll.c:(.init.text+0xd8): undefined reference to `clkhwops_omap4_dpllmx'
>
> This is easily fixed by adding another #ifdef.
>
> While looking at the code, I also spotted another problem with the
> assignment of hw_ops variable that is not used again later. I'm
> changing this to setting clk_hw->ops instead, which I guess is what
> was intended here.
Good catch... It seems the corner case is masked in the testing I did
locally so far.
>
> Fixes: 0565fb168d63 ("clk: ti: dpll: move omap3 DPLL functionality to clock driver")
However, I believe the fixes tag should point to this one in linux-next:
commit 473adbf4e02857a6b78dfb3d9fcf752638bbadb9
Author: Tero Kristo <[email protected]>
Date: Thu Feb 9 11:25:28 2017 +0200
clk: ti: dpll44xx: fix clksel register initialization
> Signed-off-by: Arnd Bergmann <[email protected]>
Other than that:
Acked-by: Tero Kristo <[email protected]>
> ---
> drivers/clk/ti/dpll.c | 8 ++++++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/clk/ti/dpll.c b/drivers/clk/ti/dpll.c
> index 96d84888c6c5..fcf297fd2233 100644
> --- a/drivers/clk/ti/dpll.c
> +++ b/drivers/clk/ti/dpll.c
> @@ -312,7 +312,6 @@ static void _register_dpll_x2(struct device_node *node,
> struct clk_hw_omap *clk_hw;
> const char *name = node->name;
> const char *parent_name;
> - int ret;
>
> parent_name = of_clk_get_parent_name(node, 0);
> if (!parent_name) {
> @@ -332,16 +331,21 @@ static void _register_dpll_x2(struct device_node *node,
> init.parent_names = &parent_name;
> init.num_parents = 1;
>
> +#if defined(CONFIG_ARCH_OMAP4) || defined(CONFIG_SOC_OMAP5) || \
> + defined(CONFIG_SOC_DRA7XX)
> if (hw_ops == &clkhwops_omap4_dpllmx) {
> + int ret;
> +
> /* Check if register defined, if not, drop hw-ops */
> ret = of_property_count_elems_of_size(node, "reg", 1);
> if (ret <= 0) {
> - hw_ops = NULL;
> + clk_hw->ops = NULL;
> } else if (ti_clk_get_reg_addr(node, 0, &clk_hw->clksel_reg)) {
> kfree(clk_hw);
> return;
> }
> }
> +#endif
>
> /* register the clock */
> clk = ti_clk_register(NULL, &clk_hw->hw, name);
>
On Thu, Apr 20, 2017 at 4:57 PM, Tero Kristo <[email protected]> wrote:
> On 20/04/17 00:43, Arnd Bergmann wrote:
>>
>> When none of the OMAP4-generation SoCs are enabled, we run into a link
>> error for am43xx/am43xx:
>>
>> drivers/clk/ti/dpll.o: In function `of_ti_am3_dpll_x2_setup':
>> dpll.c:(.init.text+0xd8): undefined reference to `clkhwops_omap4_dpllmx'
>>
>> This is easily fixed by adding another #ifdef.
>>
>> While looking at the code, I also spotted another problem with the
>> assignment of hw_ops variable that is not used again later. I'm
>> changing this to setting clk_hw->ops instead, which I guess is what
>> was intended here.
>
>
> Good catch... It seems the corner case is masked in the testing I did
> locally so far.
>
>>
>> Fixes: 0565fb168d63 ("clk: ti: dpll: move omap3 DPLL functionality to
>> clock driver")
>
>
> However, I believe the fixes tag should point to this one in linux-next:
>
> commit 473adbf4e02857a6b78dfb3d9fcf752638bbadb9
> Author: Tero Kristo <[email protected]>
> Date: Thu Feb 9 11:25:28 2017 +0200
>
> clk: ti: dpll44xx: fix clksel register initialization
Ah, right.
> Other than that:
>
> Acked-by: Tero Kristo <[email protected]>
Do you want me to resend, or can you forward the patch with the
fixed fixes line?
Arnd
On 20/04/17 18:06, Arnd Bergmann wrote:
> On Thu, Apr 20, 2017 at 4:57 PM, Tero Kristo <[email protected]> wrote:
>> On 20/04/17 00:43, Arnd Bergmann wrote:
>>>
>>> When none of the OMAP4-generation SoCs are enabled, we run into a link
>>> error for am43xx/am43xx:
>>>
>>> drivers/clk/ti/dpll.o: In function `of_ti_am3_dpll_x2_setup':
>>> dpll.c:(.init.text+0xd8): undefined reference to `clkhwops_omap4_dpllmx'
>>>
>>> This is easily fixed by adding another #ifdef.
>>>
>>> While looking at the code, I also spotted another problem with the
>>> assignment of hw_ops variable that is not used again later. I'm
>>> changing this to setting clk_hw->ops instead, which I guess is what
>>> was intended here.
>>
>>
>> Good catch... It seems the corner case is masked in the testing I did
>> locally so far.
>>
>>>
>>> Fixes: 0565fb168d63 ("clk: ti: dpll: move omap3 DPLL functionality to
>>> clock driver")
>>
>>
>> However, I believe the fixes tag should point to this one in linux-next:
>>
>> commit 473adbf4e02857a6b78dfb3d9fcf752638bbadb9
>> Author: Tero Kristo <[email protected]>
>> Date: Thu Feb 9 11:25:28 2017 +0200
>>
>> clk: ti: dpll44xx: fix clksel register initialization
>
> Ah, right.
>
>> Other than that:
>>
>> Acked-by: Tero Kristo <[email protected]>
>
> Do you want me to resend, or can you forward the patch with the
> fixed fixes line?
I guess thats up to Stephen, he typically picks single fixes like these.
I can also create a pull request with the fixes you provided if Stephen
prefers that, in that case I can fix the "Fixes:" tag myself.
Stephen any preference?
-Tero
PS: I'm still checking the other two patches you provided, as they are
more complex than this one.
On 04/20, Tero Kristo wrote:
> On 20/04/17 18:06, Arnd Bergmann wrote:
> >On Thu, Apr 20, 2017 at 4:57 PM, Tero Kristo <[email protected]> wrote:
> >>On 20/04/17 00:43, Arnd Bergmann wrote:
> >>>
> >>>When none of the OMAP4-generation SoCs are enabled, we run into a link
> >>>error for am43xx/am43xx:
> >>>
> >>>drivers/clk/ti/dpll.o: In function `of_ti_am3_dpll_x2_setup':
> >>>dpll.c:(.init.text+0xd8): undefined reference to `clkhwops_omap4_dpllmx'
> >>>
> >>>This is easily fixed by adding another #ifdef.
> >>>
> >>>While looking at the code, I also spotted another problem with the
> >>>assignment of hw_ops variable that is not used again later. I'm
> >>>changing this to setting clk_hw->ops instead, which I guess is what
> >>>was intended here.
> >>
> >>
> >>Good catch... It seems the corner case is masked in the testing I did
> >>locally so far.
> >>
> >>>
> >>>Fixes: 0565fb168d63 ("clk: ti: dpll: move omap3 DPLL functionality to
> >>>clock driver")
> >>
> >>
> >>However, I believe the fixes tag should point to this one in linux-next:
> >>
> >>commit 473adbf4e02857a6b78dfb3d9fcf752638bbadb9
> >>Author: Tero Kristo <[email protected]>
> >>Date: Thu Feb 9 11:25:28 2017 +0200
> >>
> >> clk: ti: dpll44xx: fix clksel register initialization
> >
> >Ah, right.
> >
> >>Other than that:
> >>
> >>Acked-by: Tero Kristo <[email protected]>
> >
> >Do you want me to resend, or can you forward the patch with the
> >fixed fixes line?
>
> I guess thats up to Stephen, he typically picks single fixes like
> these. I can also create a pull request with the fixes you provided
> if Stephen prefers that, in that case I can fix the "Fixes:" tag
> myself.
>
> Stephen any preference?
I'll just apply it right now.
--
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project
On 04/19, Arnd Bergmann wrote:
> When none of the OMAP4-generation SoCs are enabled, we run into a link
> error for am43xx/am43xx:
>
> drivers/clk/ti/dpll.o: In function `of_ti_am3_dpll_x2_setup':
> dpll.c:(.init.text+0xd8): undefined reference to `clkhwops_omap4_dpllmx'
>
> This is easily fixed by adding another #ifdef.
>
> While looking at the code, I also spotted another problem with the
> assignment of hw_ops variable that is not used again later. I'm
> changing this to setting clk_hw->ops instead, which I guess is what
> was intended here.
>
> Fixes: 0565fb168d63 ("clk: ti: dpll: move omap3 DPLL functionality to clock driver")
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
Applied to clk-next
--
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project