Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1319175 > unrolled thread

Re: [PATCH v3 3/3] pci: dra7xx: use pdata callbacks to perform reset

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v3 3/3] pci: dra7xx: use pdata callbacks to perform reset Tony Lindgren <tony@atomide.com> - 2016-01-27 18:40 +0100
    Re: [PATCH v3 3/3] pci: dra7xx: use pdata callbacks to perform reset Suman Anna <s-anna@ti.com> - 2016-01-27 19:20 +0100
      Re: [PATCH v3 3/3] pci: dra7xx: use pdata callbacks to perform reset Tony Lindgren <tony@atomide.com> - 2016-01-27 20:00 +0100

#1319175 — Re: [PATCH v3 3/3] pci: dra7xx: use pdata callbacks to perform reset

* Kishon Vijay Abraham I <kishon@ti.com> [160114 06:12]:
> Use assert/deassert callbacks populated in the platform data to
> to perform reset of PCIe.
> 
> Use these callbacks until a reset controller driver is
> is available in the kernel to reset PCIe.
...

> --- a/drivers/pci/host/pci-dra7xx.c
> +++ b/drivers/pci/host/pci-dra7xx.c
> @@ -347,6 +404,10 @@ static int __init dra7xx_pcie_probe(struct platform_device *pdev)
>  	enum of_gpio_flags flags;
>  	unsigned long gpio_flags;
>  
> +	ret = dra7xx_pcie_reset(pdev);
> +	if (ret)
> +		return ret;
> +
>  	dra7xx = devm_kzalloc(dev, sizeof(*dra7xx), GFP_KERNEL);
>  	if (!dra7xx)
>  		return -ENOMEM;

With the hwmod data properly configured the reset already happens
for the device by the bus driver, the hwmod code in this case?

> @@ -457,6 +518,7 @@ static int __exit dra7xx_pcie_remove(struct platform_device *pdev)
>  	struct pcie_port *pp = &dra7xx->pp;
>  	struct device *dev = &pdev->dev;
>  	int count = dra7xx->phy_count;
> +	int ret;
>  
>  	if (pp->irq_domain)
>  		irq_domain_remove(pp->irq_domain);
> @@ -467,6 +529,10 @@ static int __exit dra7xx_pcie_remove(struct platform_device *pdev)
>  		phy_exit(dra7xx->phy[count]);
>  	}
>  
> +	ret = dra7xx_pcie_assert_reset(pdev);
> +	if (ret < 0)
> +		return ret;
> +
>  	return 0;
>  }

Why do you need another reset here? Can't you just implement PM runtime
in the driver and do the usual pm_runtime_put_sync followed by
pm_runtime_disable?

Basically I'm wondering how come we need these platform data callbacks
at all.

Regards,

Tony

[toc] | [next] | [standalone]

#1319226

Hi Tony,

On 01/27/2016 11:31 AM, Tony Lindgren wrote:
> * Kishon Vijay Abraham I <kishon@ti.com> [160114 06:12]:
>> Use assert/deassert callbacks populated in the platform data to
>> to perform reset of PCIe.
>>
>> Use these callbacks until a reset controller driver is
>> is available in the kernel to reset PCIe.
> ...
> 
>> --- a/drivers/pci/host/pci-dra7xx.c
>> +++ b/drivers/pci/host/pci-dra7xx.c
>> @@ -347,6 +404,10 @@ static int __init dra7xx_pcie_probe(struct platform_device *pdev)
>>  	enum of_gpio_flags flags;
>>  	unsigned long gpio_flags;
>>  
>> +	ret = dra7xx_pcie_reset(pdev);
>> +	if (ret)
>> +		return ret;
>> +
>>  	dra7xx = devm_kzalloc(dev, sizeof(*dra7xx), GFP_KERNEL);
>>  	if (!dra7xx)
>>  		return -ENOMEM;
> 
> With the hwmod data properly configured the reset already happens
> for the device by the bus driver, the hwmod code in this case?
> 
>> @@ -457,6 +518,7 @@ static int __exit dra7xx_pcie_remove(struct platform_device *pdev)
>>  	struct pcie_port *pp = &dra7xx->pp;
>>  	struct device *dev = &pdev->dev;
>>  	int count = dra7xx->phy_count;
>> +	int ret;
>>  
>>  	if (pp->irq_domain)
>>  		irq_domain_remove(pp->irq_domain);
>> @@ -467,6 +529,10 @@ static int __exit dra7xx_pcie_remove(struct platform_device *pdev)
>>  		phy_exit(dra7xx->phy[count]);
>>  	}
>>  
>> +	ret = dra7xx_pcie_assert_reset(pdev);
>> +	if (ret < 0)
>> +		return ret;
>> +
>>  	return 0;
>>  }
> 
> Why do you need another reset here? Can't you just implement PM runtime
> in the driver and do the usual pm_runtime_put_sync followed by
> pm_runtime_disable?

The omap_hwmod_enable/disable code does not deal with hardresets (PRCM
reset lines) and so the pm_runtime_get_sync/put_sync only end up dealing
with clocks, and we need to invoke the reset functions separately.
Modules with softresets in SYSCONFIG are ok, as they are dealt with
properly.

> Basically I'm wondering how come we need these platform data callbacks
> at all.

The hardresets are controlled through the
omap_device_assert(deassert)_hardreset functions, and since these are
limited to mach-omap2, we are invoking them through platform data callbacks.


regards
Suman

[toc] | [prev] | [next] | [standalone]

#1319348

* Suman Anna <s-anna@ti.com> [160127 10:17]:
> On 01/27/2016 11:31 AM, Tony Lindgren wrote:
> > Why do you need another reset here? Can't you just implement PM runtime
> > in the driver and do the usual pm_runtime_put_sync followed by
> > pm_runtime_disable?
> 
> The omap_hwmod_enable/disable code does not deal with hardresets (PRCM
> reset lines) and so the pm_runtime_get_sync/put_sync only end up dealing
> with clocks, and we need to invoke the reset functions separately.
> Modules with softresets in SYSCONFIG are ok, as they are dealt with
> properly.

Hmm _reset() in omap_hwmod.c has this to call _assert_hardreset:

	if (oh->class->reset) {
		r = oh->class->reset(oh);
	} else {
		if (oh->rst_lines_cnt > 0) {
			for (i = 0; i < oh->rst_lines_cnt; i++)
				_assert_hardreset(oh, oh->rst_lines[i].name);
			return 0;
		} else {
			r = _ocp_softreset(oh);
			if (r == -ENOENT)
				r = 0;
		}
	}

Care to explain what exactly the problem with the hwmod code not doing
the reset on init?

And why do you need to do another reset in dra7xx_pcie_remove()?

> > Basically I'm wondering how come we need these platform data callbacks
> > at all.
> 
> The hardresets are controlled through the
> omap_device_assert(deassert)_hardreset functions, and since these are
> limited to mach-omap2, we are invoking them through platform data callbacks.

Right.. But I'm wondering about the why you need to do this in the
driver at all part :)

Regards,

Tony

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web