[PATCH 0/3] clk: keystone: Fix some error handling paths

[PATCH 0/3] clk: keystone: Fix some error handling paths

[Christophe JAILLET]

The first of these 3 patches is a v2 of a patch posted earlier.
The main change in this v2 is the propagation of the error returned by
'clk_register_pll()'.

The 2nd one is completely un-related and fixes some typo in error messages
spoted while looking at the file.

Finaly, the 3rd one refactors the '_of_pll_clk_init' function in order to
factorize some cleanups in case of error and to add some missing 'iounmap()'
calls if 'clk_register_pll()' fails.

These patches are un-compiled. I get some errors for some reasons in my build
environment. I've not checked further the root cause of these build errors
but I guess that my plateform is not what is expected to build this code.

Christophe JAILLET (3):
  clk: keystone: Fix an error checking
  clk: keystone: Fix some error messages
  clk: keystone: Fix missing iounmap calls in an error handling path

 drivers/clk/keystone/pll.c | 20 ++++++++++++--------
 1 file changed, 12 insertions(+), 8 deletions(-)

-- 
2.9.3

[PATCH 1/3] clk: keystone: Fix an error checking

[Christophe JAILLET]

clk_register_pll() can return ERR_PTR(-ENOMEM) so checking the return value
against NULL only is not correct.
In order to fix it, update clk_register_pll() to always return an error
pointer in case of error and check the return value with IS_ERR.

While at it, also fix a tab vs space in the surrounding code.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
Un-compiled and un-tested.
---
 drivers/clk/keystone/pll.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/clk/keystone/pll.c b/drivers/clk/keystone/pll.c
index a26ba2184454..70d6fb2d23bc 100644
--- a/drivers/clk/keystone/pll.c
+++ b/drivers/clk/keystone/pll.c
@@ -140,7 +140,7 @@ static struct clk *clk_register_pll(struct device *dev,
 	init.parent_names = (parent_name ? &parent_name : NULL);
 	init.num_parents = (parent_name ? 1 : 0);
 
-	pll->pll_data	= pll_data;
+	pll->pll_data = pll_data;
 	pll->hw.init = &init;
 
 	clk = clk_register(NULL, &pll->hw);
@@ -150,7 +150,7 @@ static struct clk *clk_register_pll(struct device *dev,
 	return clk;
 out:
 	kfree(pll);
-	return NULL;
+	return clk;
 }
 
 /**
@@ -213,7 +213,7 @@ static void __init _of_pll_clk_init(struct device_node *node, bool pllctrl)
 	}
 
 	clk = clk_register_pll(NULL, node->name, parent_name, pll_data);
-	if (clk) {
+	if (!IS_ERR(clk)) {
 		of_clk_add_provider(node, of_clk_src_simple_get, clk);
 		return;
 	}
-- 
2.9.3

Re: [PATCH 1/3] clk: keystone: Fix an error checking

[Stephen Boyd]

On 10/24, Christophe JAILLET wrote:
> clk_register_pll() can return ERR_PTR(-ENOMEM) so checking the return value
> against NULL only is not correct.

The code just doesn't propagate the error up to the caller.
Instead the caller treats NULL as an error and non-NULL as valid.
If the callee detects an error it hides it and returns NULL.

> In order to fix it, update clk_register_pll() to always return an error
> pointer in case of error and check the return value with IS_ERR.
> 
> While at it, also fix a tab vs space in the surrounding code.
> 
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
> Un-compiled and un-tested.

Please at least compile test patches.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

Re: [PATCH 1/3] clk: keystone: Fix an error checking

[Christophe JAILLET]

Le 02/11/2016 à 01:22, Stephen Boyd a écrit :
> On 10/24, Christophe JAILLET wrote:
>> clk_register_pll() can return ERR_PTR(-ENOMEM) so checking the return value
>> against NULL only is not correct.
> The code just doesn't propagate the error up to the caller.
> Instead the caller treats NULL as an error and non-NULL as valid.
> If the callee detects an error it hides it and returns NULL.

Could you please clarify?
I thought that your point was that 'clk_register_pll()' was returning 
NULL when 'clk_register()' was returning an error.
The proposed patch fixes it and now 'clk_register_pll()' returns:
    - either  ERR_PTR(-ENOMEM) if zkalloc fails
    - or clk if 'clk_register()' fails. In this case it is an error pointer
    - or a valid, non NULL, pointer in case of success

The caller, '_of_pll_clk_init()' has also been updated to test the 
return value using IS_ERR.


So, which caller/callee do you refer to?
Would you like '_of_pll_clk_init()' to also propagate the error?

Or is it the phrasing of the log entry which is not clear enough and 
should be updated?

>
>> In order to fix it, update clk_register_pll() to always return an error
>> pointer in case of error and check the return value with IS_ERR.
>>
>> While at it, also fix a tab vs space in the surrounding code.
>>
>> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
>> ---
>> Un-compiled and un-tested.
> Please at least compile test patches.
Agreed, and I usually do.
But in this particular case, I don't have the build environment to do it.

I preferred to report what looks like a (small) bug to me and clearly 
state that I didn't even compile-tested it rather than just ignoring it.
Hoping that this approach, in such a case, is acceptable.

CJ

[PATCH 2/3] clk: keystone: Fix some error messages

[Christophe JAILLET]

Report 'bit-shift' in the error message as it is the property we are
looking for.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
 drivers/clk/keystone/pll.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/clk/keystone/pll.c b/drivers/clk/keystone/pll.c
index 70d6fb2d23bc..35c0e2b011d1 100644
--- a/drivers/clk/keystone/pll.c
+++ b/drivers/clk/keystone/pll.c
@@ -271,7 +271,7 @@ static void __init of_pll_div_clk_init(struct device_node *node)
 	}
 
 	if (of_property_read_u32(node, "bit-shift", &shift)) {
-		pr_err("%s: missing 'shift' property\n", __func__);
+		pr_err("%s: missing 'bit-shift' property\n", __func__);
 		return;
 	}
 
@@ -315,7 +315,7 @@ static void __init of_pll_mux_clk_init(struct device_node *node)
 	}
 
 	if (of_property_read_u32(node, "bit-shift", &shift)) {
-		pr_err("%s: missing 'shift' property\n", __func__);
+		pr_err("%s: missing 'bit-shift' property\n", __func__);
 		return;
 	}
 
-- 
2.9.3

[PATCH 3/3] clk: keystone: Fix missing iounmap calls in an error handling path

[Christophe JAILLET]

Factorize 'iounmap()' calls in the error handling path.
The main goal is to add these calls if 'clk_register_pll()' fails.

Add an error message if an 'of_iomap' call fails to be consistent.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
Un-compiled & un-tested
---
 drivers/clk/keystone/pll.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/drivers/clk/keystone/pll.c b/drivers/clk/keystone/pll.c
index 35c0e2b011d1..73a2558b29c0 100644
--- a/drivers/clk/keystone/pll.c
+++ b/drivers/clk/keystone/pll.c
@@ -191,7 +191,6 @@ static void __init _of_pll_clk_init(struct device_node *node, bool pllctrl)
 	pll_data->pll_ctl0 = of_iomap(node, i);
 	if (!pll_data->pll_ctl0) {
 		pr_err("%s: ioremap failed\n", __func__);
-		iounmap(pll_data->pllod);
 		goto out;
 	}
 
@@ -206,8 +205,7 @@ static void __init _of_pll_clk_init(struct device_node *node, bool pllctrl)
 		i = of_property_match_string(node, "reg-names", "multiplier");
 		pll_data->pllm = of_iomap(node, i);
 		if (!pll_data->pllm) {
-			iounmap(pll_data->pll_ctl0);
-			iounmap(pll_data->pllod);
+			pr_err("%s: ioremap failed\n", __func__);
 			goto out;
 		}
 	}
@@ -220,6 +218,12 @@ static void __init _of_pll_clk_init(struct device_node *node, bool pllctrl)
 
 out:
 	pr_err("%s: error initializing pll %s\n", __func__, node->name);
+	if (pll_data->pllm)
+		iounmap(pll_data->pllm);
+	if (pll_data->pll_ctl0)
+		iounmap(pll_data->pll_ctl0);
+	if (pll_data->pllod)
+		iounmap(pll_data->pllod);
 	kfree(pll_data);
 }
 
-- 
2.9.3

Re: [PATCH 3/3] clk: keystone: Fix missing iounmap calls in an error handling path

[walter harms]

Am 24.10.2016 22:43, schrieb Christophe JAILLET:
> Factorize 'iounmap()' calls in the error handling path.
> The main goal is to add these calls if 'clk_register_pll()' fails.
> 
> Add an error message if an 'of_iomap' call fails to be consistent.
> 
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
> Un-compiled & un-tested
> ---
>  drivers/clk/keystone/pll.c | 10 +++++++---
>  1 file changed, 7 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/clk/keystone/pll.c b/drivers/clk/keystone/pll.c
> index 35c0e2b011d1..73a2558b29c0 100644
> --- a/drivers/clk/keystone/pll.c
> +++ b/drivers/clk/keystone/pll.c
> @@ -191,7 +191,6 @@ static void __init _of_pll_clk_init(struct device_node *node, bool pllctrl)
>  	pll_data->pll_ctl0 = of_iomap(node, i);
>  	if (!pll_data->pll_ctl0) {
>  		pr_err("%s: ioremap failed\n", __func__);
> -		iounmap(pll_data->pllod);
>  		goto out;
>  	}
>  
> @@ -206,8 +205,7 @@ static void __init _of_pll_clk_init(struct device_node *node, bool pllctrl)
>  		i = of_property_match_string(node, "reg-names", "multiplier");
>  		pll_data->pllm = of_iomap(node, i);
>  		if (!pll_data->pllm) {
> -			iounmap(pll_data->pll_ctl0);
> -			iounmap(pll_data->pllod);
> +			pr_err("%s: ioremap failed\n", __func__);
>  			goto out;
>  		}
>  	}
> @@ -220,6 +218,12 @@ static void __init _of_pll_clk_init(struct device_node *node, bool pllctrl)
>  
>  out:
>  	pr_err("%s: error initializing pll %s\n", __func__, node->name);
> +	if (pll_data->pllm)
> +		iounmap(pll_data->pllm);
> +	if (pll_data->pll_ctl0)
> +		iounmap(pll_data->pll_ctl0);
> +	if (pll_data->pllod)
> +		iounmap(pll_data->pllod);
>  	kfree(pll_data);
>  }
>  

IMHO calles the iounmap() need no check for NULL.

re,
 wh

Re: [PATCH 3/3] clk: keystone: Fix missing iounmap calls in an error handling path

[Stephen Boyd]

On 10/25, walter harms wrote:
> Am 24.10.2016 22:43, schrieb Christophe JAILLET:
> > @@ -220,6 +218,12 @@ static void __init _of_pll_clk_init(struct device_node *node, bool pllctrl)
> >  
> >  out:
> >  	pr_err("%s: error initializing pll %s\n", __func__, node->name);
> > +	if (pll_data->pllm)
> > +		iounmap(pll_data->pllm);
> > +	if (pll_data->pll_ctl0)
> > +		iounmap(pll_data->pll_ctl0);
> > +	if (pll_data->pllod)
> > +		iounmap(pll_data->pllod);
> >  	kfree(pll_data);
> >  }
> >  
> 
> IMHO calles the iounmap() need no check for NULL.
> 

ARM doesn't seem to check for NULL there though. So that would be
a bug.

It would be nice to remove the checks though. Perhaps someone
could do that by unifying ionumap into asm-generic with the NULL
check and then have architecture specific functions for the rest
of it?

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project