[RESEND PATCH] PCI: qcom: add runtime pm support to pcie_port

[RESEND PATCH] PCI: qcom: add runtime pm support to pcie_port

[Bjorn Andersson]

From: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>

This patch is required when the pcie controller sits on a bus with
its own power domain and clocks which are controlled via a bus driver
like simple pm bus. As these bus driver have runtime pm enabled, it makes
sense to update the usage counter so that the runtime pm does not suspend
the clks or power domain associated with the bus driver.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
---

This patch was posted 18 months ago but was somehow lost. linux-next recently
enabled PCIe on MSM8996 and DB820c, but the system stalls as the agnoc gdsc
isn't enabled. This patch still fixes this problem for me.

 drivers/pci/dwc/pcie-qcom.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/pci/dwc/pcie-qcom.c b/drivers/pci/dwc/pcie-qcom.c
index 5897af7d3355..3f35098b71b1 100644
--- a/drivers/pci/dwc/pcie-qcom.c
+++ b/drivers/pci/dwc/pcie-qcom.c
@@ -1088,6 +1088,8 @@ static int qcom_pcie_host_init(struct pcie_port *pp)
 	struct qcom_pcie *pcie = to_qcom_pcie(pci);
 	int ret;
 
+	pm_runtime_get_sync(pci->dev);
+
 	qcom_ep_reset_assert(pcie);
 
 	ret = pcie->ops->init(pcie);
@@ -1124,6 +1126,7 @@ static int qcom_pcie_host_init(struct pcie_port *pp)
 	phy_power_off(pcie->phy);
 err_deinit:
 	pcie->ops->deinit(pcie);
+	pm_runtime_put_sync(pci->dev);
 
 	return ret;
 }
@@ -1204,6 +1207,7 @@ static int qcom_pcie_probe(struct platform_device *pdev)
 	struct qcom_pcie *pcie;
 	int ret;
 
+
 	pcie = devm_kzalloc(dev, sizeof(*pcie), GFP_KERNEL);
 	if (!pcie)
 		return -ENOMEM;
@@ -1212,6 +1216,8 @@ static int qcom_pcie_probe(struct platform_device *pdev)
 	if (!pci)
 		return -ENOMEM;
 
+	pm_runtime_enable(dev);
+
 	pci->dev = dev;
 	pci->ops = &dw_pcie_ops;
 	pp = &pci->pp;
-- 
2.17.0

Re: [RESEND PATCH] PCI: qcom: add runtime pm support to pcie_port

[Vinod]

On 18-05-18, 00:07, Bjorn Andersson wrote:

> diff --git a/drivers/pci/dwc/pcie-qcom.c b/drivers/pci/dwc/pcie-qcom.c
> index 5897af7d3355..3f35098b71b1 100644
> --- a/drivers/pci/dwc/pcie-qcom.c
> +++ b/drivers/pci/dwc/pcie-qcom.c
> @@ -1088,6 +1088,8 @@ static int qcom_pcie_host_init(struct pcie_port *pp)
>  	struct qcom_pcie *pcie = to_qcom_pcie(pci);
>  	int ret;
>  
> +	pm_runtime_get_sync(pci->dev);
> +
>  	qcom_ep_reset_assert(pcie);
>  
>  	ret = pcie->ops->init(pcie);
> @@ -1124,6 +1126,7 @@ static int qcom_pcie_host_init(struct pcie_port *pp)
>  	phy_power_off(pcie->phy);
>  err_deinit:
>  	pcie->ops->deinit(pcie);
> +	pm_runtime_put_sync(pci->dev);

Wondering if there is a reason for sync put here?

-- 
~Vinod

Re: [RESEND PATCH] PCI: qcom: add runtime pm support to pcie_port

[Srinivas Kandagatla]

On 23/05/18 07:15, Vinod wrote:
> On 18-05-18, 00:07, Bjorn Andersson wrote:
> 
>> diff --git a/drivers/pci/dwc/pcie-qcom.c b/drivers/pci/dwc/pcie-qcom.c
>> index 5897af7d3355..3f35098b71b1 100644
>> --- a/drivers/pci/dwc/pcie-qcom.c
>> +++ b/drivers/pci/dwc/pcie-qcom.c
>> @@ -1088,6 +1088,8 @@ static int qcom_pcie_host_init(struct pcie_port *pp)
>>   	struct qcom_pcie *pcie = to_qcom_pcie(pci);
>>   	int ret;
>>   
>> +	pm_runtime_get_sync(pci->dev);
>> +
>>   	qcom_ep_reset_assert(pcie);
>>   
>>   	ret = pcie->ops->init(pcie);
>> @@ -1124,6 +1126,7 @@ static int qcom_pcie_host_init(struct pcie_port *pp)
>>   	phy_power_off(pcie->phy);
>>   err_deinit:
>>   	pcie->ops->deinit(pcie);
>> +	pm_runtime_put_sync(pci->dev);
> 
> Wondering if there is a reason for sync put here?
pm_runtime_get_sync() call paired with a pm_runtime_put() call will be
appropriate here I guess as per 
https://www.kernel.org/doc/Documentation/power/runtime_pm.txt

I will respin this patch.

Thanks,
srini


> 

Re: [RESEND PATCH] PCI: qcom: add runtime pm support to pcie_port

[Vinod]

On 23-05-18, 11:10, Srinivas Kandagatla wrote:
> 
> 
> On 23/05/18 07:15, Vinod wrote:
> > On 18-05-18, 00:07, Bjorn Andersson wrote:
> > 
> > > diff --git a/drivers/pci/dwc/pcie-qcom.c b/drivers/pci/dwc/pcie-qcom.c
> > > index 5897af7d3355..3f35098b71b1 100644
> > > --- a/drivers/pci/dwc/pcie-qcom.c
> > > +++ b/drivers/pci/dwc/pcie-qcom.c
> > > @@ -1088,6 +1088,8 @@ static int qcom_pcie_host_init(struct pcie_port *pp)
> > >   	struct qcom_pcie *pcie = to_qcom_pcie(pci);
> > >   	int ret;
> > > +	pm_runtime_get_sync(pci->dev);
> > > +
> > >   	qcom_ep_reset_assert(pcie);
> > >   	ret = pcie->ops->init(pcie);
> > > @@ -1124,6 +1126,7 @@ static int qcom_pcie_host_init(struct pcie_port *pp)
> > >   	phy_power_off(pcie->phy);
> > >   err_deinit:
> > >   	pcie->ops->deinit(pcie);
> > > +	pm_runtime_put_sync(pci->dev);
> > 
> > Wondering if there is a reason for sync put here?
> pm_runtime_get_sync() call paired with a pm_runtime_put() call will be
> appropriate here I guess as per
> https://www.kernel.org/doc/Documentation/power/runtime_pm.txt

Yeah for the get call sync() is mostly appropriate as you would want hardware to
be up after this. For put most of cases people don't need it to be bought down
synchronously unless you have something specific required which was my question
here. Looks like that is not the case :)

> I will respin this patch.

Sure..

-- 
~Vinod