[PATCH V2 1/2] clocksource: imx-gpt: add support for ARM64

[PATCH V2 1/2] clocksource: imx-gpt: add support for ARM64

[Anson Huang]

This patch allows building and compile-testing the i.MX
GPT driver also for ARM64. The delay_timer is only
supported on ARMv7.

Signed-off-by: Anson Huang <Anson.Huang@nxp.com>
---
no change since V1.
 drivers/clocksource/Kconfig         | 2 +-
 drivers/clocksource/timer-imx-gpt.c | 4 ++++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/clocksource/Kconfig b/drivers/clocksource/Kconfig
index 55c77e4..5e892ca 100644
--- a/drivers/clocksource/Kconfig
+++ b/drivers/clocksource/Kconfig
@@ -580,7 +580,7 @@ config H8300_TPU
 
 config CLKSRC_IMX_GPT
 	bool "Clocksource using i.MX GPT" if COMPILE_TEST
-	depends on ARM && CLKDEV_LOOKUP
+	depends on (ARM || ARM64) && CLKDEV_LOOKUP
 	select CLKSRC_MMIO
 
 config CLKSRC_IMX_TPM
diff --git a/drivers/clocksource/timer-imx-gpt.c b/drivers/clocksource/timer-imx-gpt.c
index 165fbbb..a3d6ccb 100644
--- a/drivers/clocksource/timer-imx-gpt.c
+++ b/drivers/clocksource/timer-imx-gpt.c
@@ -141,21 +141,25 @@ static u64 notrace mxc_read_sched_clock(void)
 	return sched_clock_reg ? readl_relaxed(sched_clock_reg) : 0;
 }
 
+#if defined(CONFIG_ARM)
 static struct delay_timer imx_delay_timer;
 
 static unsigned long imx_read_current_timer(void)
 {
 	return readl_relaxed(sched_clock_reg);
 }
+#endif
 
 static int __init mxc_clocksource_init(struct imx_timer *imxtm)
 {
 	unsigned int c = clk_get_rate(imxtm->clk_per);
 	void __iomem *reg = imxtm->base + imxtm->gpt->reg_tcn;
 
+#if defined(CONFIG_ARM)
 	imx_delay_timer.read_current_timer = &imx_read_current_timer;
 	imx_delay_timer.freq = c;
 	register_current_timer_delay(&imx_delay_timer);
+#endif
 
 	sched_clock_reg = reg;
 
-- 
2.7.4

[PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree to avoid resource leak

[Anson Huang]

kfree should be called to free resource in error path
before return.

Signed-off-by: Anson Huang <Anson.Huang@nxp.com>
---
 drivers/clocksource/timer-imx-gpt.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/clocksource/timer-imx-gpt.c b/drivers/clocksource/timer-imx-gpt.c
index a3d6ccb..0f78b30 100644
--- a/drivers/clocksource/timer-imx-gpt.c
+++ b/drivers/clocksource/timer-imx-gpt.c
@@ -477,12 +477,16 @@ static int __init mxc_timer_init_dt(struct device_node *np,  enum imx_gpt_type t
 		return -ENOMEM;
 
 	imxtm->base = of_iomap(np, 0);
-	if (!imxtm->base)
+	if (!imxtm->base) {
+		kfree(imxtm);
 		return -ENXIO;
+	}
 
 	imxtm->irq = irq_of_parse_and_map(np, 0);
-	if (imxtm->irq <= 0)
+	if (imxtm->irq <= 0) {
+		kfree(imxtm);
 		return -EINVAL;
+	}
 
 	imxtm->clk_ipg = of_clk_get_by_name(np, "ipg");
 
-- 
2.7.4

Re: [PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree to avoid resource leak

[Daniel Lezcano]

Hi Anson,

On 05/11/2018 02:10, Anson Huang wrote:
> kfree should be called to free resource in error path
> before return.
> 
> Signed-off-by: Anson Huang <Anson.Huang@nxp.com>
> ---
>  drivers/clocksource/timer-imx-gpt.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/clocksource/timer-imx-gpt.c b/drivers/clocksource/timer-imx-gpt.c
> index a3d6ccb..0f78b30 100644
> --- a/drivers/clocksource/timer-imx-gpt.c
> +++ b/drivers/clocksource/timer-imx-gpt.c
> @@ -477,12 +477,16 @@ static int __init mxc_timer_init_dt(struct device_node *np,  enum imx_gpt_type t
>  		return -ENOMEM;
>  
>  	imxtm->base = of_iomap(np, 0);
> -	if (!imxtm->base)
> +	if (!imxtm->base) {
> +		kfree(imxtm);
>  		return -ENXIO;
> +	}
>  
>  	imxtm->irq = irq_of_parse_and_map(np, 0);
> -	if (imxtm->irq <= 0)
> +	if (imxtm->irq <= 0) {
> +		kfree(imxtm);
>  		return -EINVAL;
> +	}
>  
>  	imxtm->clk_ipg = of_clk_get_by_name(np, "ipg");

Please convert to timer-of API.

Thanks

  -- Daniel


-- 
 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

RE: [PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree to avoid resource leak

[Anson Huang]

Hi, Daniel

Best Regards!
Anson Huang

> -----Original Message-----
> From: Daniel Lezcano [mailto:daniel.lezcano@linaro.org]
> Sent: 2018年11月5日 21:35
> To: Anson Huang <anson.huang@nxp.com>; tglx@linutronix.de;
> linux-kernel@vger.kernel.org
> Cc: dl-linux-imx <linux-imx@nxp.com>
> Subject: Re: [PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree to avoid
> resource leak
> 
> Hi Anson,
> 
> On 05/11/2018 02:10, Anson Huang wrote:
> > kfree should be called to free resource in error path before return.
> >
> > Signed-off-by: Anson Huang <Anson.Huang@nxp.com>
> > ---
> >  drivers/clocksource/timer-imx-gpt.c | 8 ++++++--
> >  1 file changed, 6 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/clocksource/timer-imx-gpt.c
> > b/drivers/clocksource/timer-imx-gpt.c
> > index a3d6ccb..0f78b30 100644
> > --- a/drivers/clocksource/timer-imx-gpt.c
> > +++ b/drivers/clocksource/timer-imx-gpt.c
> > @@ -477,12 +477,16 @@ static int __init mxc_timer_init_dt(struct
> device_node *np,  enum imx_gpt_type t
> >  		return -ENOMEM;
> >
> >  	imxtm->base = of_iomap(np, 0);
> > -	if (!imxtm->base)
> > +	if (!imxtm->base) {
> > +		kfree(imxtm);
> >  		return -ENXIO;
> > +	}
> >
> >  	imxtm->irq = irq_of_parse_and_map(np, 0);
> > -	if (imxtm->irq <= 0)
> > +	if (imxtm->irq <= 0) {
> > +		kfree(imxtm);
> >  		return -EINVAL;
> > +	}
> >
> >  	imxtm->clk_ipg = of_clk_get_by_name(np, "ipg");
> 
> Please convert to timer-of API.

This patch mainly to fix the potential resource leak issue, for converting
to timer-of API, should I add another patch for it? Since it is a different
subject.

Thanks.

Anson

> 
> Thanks
> 
>   -- Daniel
> 
> 
> --
> 
> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww
> .linaro.org%2F&amp;data=02%7C01%7Canson.huang%40nxp.com%7Cb389e6
> 71eafb4400148208d643237383%7C686ea1d3bc2b4c6fa92cd99c5c301635%
> 7C0%7C0%7C636770216874548293&amp;sdata=toynTer%2Bq7DLRizwW7JKY
> qCrA8WQHeDmJRwLjXUu6XY%3D&amp;reserved=0> Linaro.org │ Open source
> software for ARM SoCs
> 
> Follow Linaro:
> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww
> .facebook.com%2Fpages%2FLinaro&amp;data=02%7C01%7Canson.huang%40
> nxp.com%7Cb389e671eafb4400148208d643237383%7C686ea1d3bc2b4c6fa
> 92cd99c5c301635%7C0%7C0%7C636770216874548293&amp;sdata=JJOClhL
> pER5JMuXHs65fJD%2FPC18%2FJBCyiRw3d41ei48%3D&amp;reserved=0>
> Facebook |
> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Ftwitt
> er.com%2F%23!%2Flinaroorg&amp;data=02%7C01%7Canson.huang%40nxp.c
> om%7Cb389e671eafb4400148208d643237383%7C686ea1d3bc2b4c6fa92cd
> 99c5c301635%7C0%7C0%7C636770216874548293&amp;sdata=W1QXRwgXt
> e3wbbjETuDinUniIDeA8THW%2Bsz6SSPWl30%3D&amp;reserved=0> Twitter |
> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww
> .linaro.org%2Flinaro-blog%2F&amp;data=02%7C01%7Canson.huang%40nxp.c
> om%7Cb389e671eafb4400148208d643237383%7C686ea1d3bc2b4c6fa92cd
> 99c5c301635%7C0%7C0%7C636770216874548293&amp;sdata=9JdVMOpdiI
> HdxVrFbiDgblAunod9RYkzwYRaHcsgzUo%3D&amp;reserved=0> Blog

Re: [PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree to avoid resource leak

[Daniel Lezcano]

Hi Anson,

On 06/11/2018 02:38, Anson Huang wrote:
> Hi, Daniel
> 
> Best Regards!
> Anson Huang
> 
>> -----Original Message-----
>> From: Daniel Lezcano [mailto:daniel.lezcano@linaro.org]
>> Sent: 2018年11月5日 21:35
>> To: Anson Huang <anson.huang@nxp.com>; tglx@linutronix.de;
>> linux-kernel@vger.kernel.org
>> Cc: dl-linux-imx <linux-imx@nxp.com>
>> Subject: Re: [PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree to avoid
>> resource leak
>>
>> Hi Anson,
>>
>> On 05/11/2018 02:10, Anson Huang wrote:
>>> kfree should be called to free resource in error path before return.
>>>
>>> Signed-off-by: Anson Huang <Anson.Huang@nxp.com>
>>> ---
>>>  drivers/clocksource/timer-imx-gpt.c | 8 ++++++--
>>>  1 file changed, 6 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/clocksource/timer-imx-gpt.c
>>> b/drivers/clocksource/timer-imx-gpt.c
>>> index a3d6ccb..0f78b30 100644
>>> --- a/drivers/clocksource/timer-imx-gpt.c
>>> +++ b/drivers/clocksource/timer-imx-gpt.c
>>> @@ -477,12 +477,16 @@ static int __init mxc_timer_init_dt(struct
>> device_node *np,  enum imx_gpt_type t
>>>  		return -ENOMEM;
>>>
>>>  	imxtm->base = of_iomap(np, 0);
>>> -	if (!imxtm->base)
>>> +	if (!imxtm->base) {
>>> +		kfree(imxtm);
>>>  		return -ENXIO;
>>> +	}
>>>
>>>  	imxtm->irq = irq_of_parse_and_map(np, 0);
>>> -	if (imxtm->irq <= 0)
>>> +	if (imxtm->irq <= 0) {
>>> +		kfree(imxtm);
>>>  		return -EINVAL;
>>> +	}
>>>
>>>  	imxtm->clk_ipg = of_clk_get_by_name(np, "ipg");
>>
>> Please convert to timer-of API.
> 
> This patch mainly to fix the potential resource leak issue, for converting
> to timer-of API, should I add another patch for it? Since it is a different
> subject.

Actually I was unclear. The patch is duplicating the kfree in the error
path, usually a rollback routine should be provided.

So instead of creating this rollback path, you can directly use the
timer-of which contains the error check and rollback.



-- 
 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

RE: [PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree to avoid resource leak

[Anson Huang]

Hi, Daniel

Best Regards!
Anson Huang

> -----Original Message-----
> From: Daniel Lezcano [mailto:daniel.lezcano@linaro.org]
> Sent: 2018年11月6日 17:06
> To: Anson Huang <anson.huang@nxp.com>; tglx@linutronix.de;
> linux-kernel@vger.kernel.org
> Cc: dl-linux-imx <linux-imx@nxp.com>
> Subject: Re: [PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree to avoid
> resource leak
> 
> 
> Hi Anson,
> 
> On 06/11/2018 02:38, Anson Huang wrote:
> > Hi, Daniel
> >
> > Best Regards!
> > Anson Huang
> >
> >> -----Original Message-----
> >> From: Daniel Lezcano [mailto:daniel.lezcano@linaro.org]
> >> Sent: 2018年11月5日 21:35
> >> To: Anson Huang <anson.huang@nxp.com>; tglx@linutronix.de;
> >> linux-kernel@vger.kernel.org
> >> Cc: dl-linux-imx <linux-imx@nxp.com>
> >> Subject: Re: [PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree
> >> to avoid resource leak
> >>
> >> Hi Anson,
> >>
> >> On 05/11/2018 02:10, Anson Huang wrote:
> >>> kfree should be called to free resource in error path before return.
> >>>
> >>> Signed-off-by: Anson Huang <Anson.Huang@nxp.com>
> >>> ---
> >>>  drivers/clocksource/timer-imx-gpt.c | 8 ++++++--
> >>>  1 file changed, 6 insertions(+), 2 deletions(-)
> >>>
> >>> diff --git a/drivers/clocksource/timer-imx-gpt.c
> >>> b/drivers/clocksource/timer-imx-gpt.c
> >>> index a3d6ccb..0f78b30 100644
> >>> --- a/drivers/clocksource/timer-imx-gpt.c
> >>> +++ b/drivers/clocksource/timer-imx-gpt.c
> >>> @@ -477,12 +477,16 @@ static int __init mxc_timer_init_dt(struct
> >> device_node *np,  enum imx_gpt_type t
> >>>  		return -ENOMEM;
> >>>
> >>>  	imxtm->base = of_iomap(np, 0);
> >>> -	if (!imxtm->base)
> >>> +	if (!imxtm->base) {
> >>> +		kfree(imxtm);
> >>>  		return -ENXIO;
> >>> +	}
> >>>
> >>>  	imxtm->irq = irq_of_parse_and_map(np, 0);
> >>> -	if (imxtm->irq <= 0)
> >>> +	if (imxtm->irq <= 0) {
> >>> +		kfree(imxtm);
> >>>  		return -EINVAL;
> >>> +	}
> >>>
> >>>  	imxtm->clk_ipg = of_clk_get_by_name(np, "ipg");
> >>
> >> Please convert to timer-of API.
> >
> > This patch mainly to fix the potential resource leak issue, for
> > converting to timer-of API, should I add another patch for it? Since
> > it is a different subject.
> 
> Actually I was unclear. The patch is duplicating the kfree in the error path,
> usually a rollback routine should be provided.
> 
> So instead of creating this rollback path, you can directly use the timer-of
> which contains the error check and rollback.

I understand the timer-of can save some code in timer driver, but current GPT
driver is kind of complicated in order to support so many i.MX platforms with
different GPT design, change it to timer-of will need some effort and I think we
can do it later. If the duplicating kfree in error path confuses you, I can change
it to rollback path you said. Below are 2 options in my mind:

1. add kfree to rollback path;
2. skip this patch, just apply 1/2 patch, and let me find some time to change GPT
driver using timer-of. 

Which one do you prefer? Thanks.

Anson.

> 
> 
> 
> --
> 
> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww
> .linaro.org%2F&amp;data=02%7C01%7Canson.huang%40nxp.com%7Cc675bb
> b2063b43c3be1508d643c715f3%7C686ea1d3bc2b4c6fa92cd99c5c301635%7
> C0%7C0%7C636770919692475408&amp;sdata=pHaN%2BBguQo2lmVJzLYXBJ
> 12l6DwX%2BxSXo9Bil9SwYxs%3D&amp;reserved=0> Linaro.org │ Open source
> software for ARM SoCs
> 
> Follow Linaro:
> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww
> .facebook.com%2Fpages%2FLinaro&amp;data=02%7C01%7Canson.huang%40
> nxp.com%7Cc675bbb2063b43c3be1508d643c715f3%7C686ea1d3bc2b4c6fa9
> 2cd99c5c301635%7C0%7C0%7C636770919692475408&amp;sdata=EgyadNU
> A1XLncN5p90lkIIiiMgQ0CIO405biLpKszvE%3D&amp;reserved=0> Facebook |
> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Ftwitt
> er.com%2F%23!%2Flinaroorg&amp;data=02%7C01%7Canson.huang%40nxp.c
> om%7Cc675bbb2063b43c3be1508d643c715f3%7C686ea1d3bc2b4c6fa92cd9
> 9c5c301635%7C0%7C0%7C636770919692475408&amp;sdata=NIgQhAPZBh9
> jKcrvNrEY005pZpLZP8pWIVlKXslQ%2B5U%3D&amp;reserved=0> Twitter |
> <https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww
> .linaro.org%2Flinaro-blog%2F&amp;data=02%7C01%7Canson.huang%40nxp.c
> om%7Cc675bbb2063b43c3be1508d643c715f3%7C686ea1d3bc2b4c6fa92cd9
> 9c5c301635%7C0%7C0%7C636770919692475408&amp;sdata=MCBh9dJbjN
> GtK6Vpsj3EfrF7OGBRAawA3jkCmZpgm0I%3D&amp;reserved=0> Blog

Re: [PATCH V2 2/2] clocksource: imx-gpt: add necessary kfree to avoid resource leak

[Daniel Lezcano]

Hi Anson,

On 06/11/2018 10:12, Anson Huang wrote:

[ ... ]

>>>> Please convert to timer-of API.
>>>
>>> This patch mainly to fix the potential resource leak issue, for
>>> converting to timer-of API, should I add another patch for it? Since
>>> it is a different subject.
>>
>> Actually I was unclear. The patch is duplicating the kfree in the error path,
>> usually a rollback routine should be provided.
>>
>> So instead of creating this rollback path, you can directly use the timer-of
>> which contains the error check and rollback.
> 
> I understand the timer-of can save some code in timer driver, but current GPT
> driver is kind of complicated in order to support so many i.MX platforms with
> different GPT design, change it to timer-of will need some effort and I think we
> can do it later. If the duplicating kfree in error path confuses you, I can change
> it to rollback path you said. Below are 2 options in my mind:
> 
> 1. add kfree to rollback path;
> 2. skip this patch, just apply 1/2 patch, and let me find some time to change GPT
> driver using timer-of. 
> 
> Which one do you prefer? Thanks.

I'm fine with option 2.

Thanks

  -- Daniel

-- 
 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

Re: [PATCH V2 1/2] clocksource: imx-gpt: add support for ARM64

[Daniel Lezcano]

On 05/11/2018 02:10, Anson Huang wrote:
> This patch allows building and compile-testing the i.MX
> GPT driver also for ARM64. The delay_timer is only
> supported on ARMv7.
> 
> Signed-off-by: Anson Huang <Anson.Huang@nxp.com>
> ---
> no change since V1.

Applied, thanks.

  -- Daniel


-- 
 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog