Message ID | 20240827063631.3932971-9-quic_qianyu@quicinc.com |
---|---|
State | New |
Headers | show |
Series | Add support for PCIe3 on x1e80100 | expand |
On 27.08.2024 8:36 AM, Qiang Yu wrote: > On platform x1e80100 QCP, PCIe3 is a standard x8 form factor. Hence, add > support to use 3.3v, 3.3v aux and 12v regulators. > > Signed-off-by: Qiang Yu <quic_qianyu@quicinc.com> > --- > drivers/pci/controller/dwc/pcie-qcom.c | 52 +++++++++++++++++++++++++- > 1 file changed, 50 insertions(+), 2 deletions(-) > > diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c > index 6f953e32d990..59fb415dfeeb 100644 > --- a/drivers/pci/controller/dwc/pcie-qcom.c > +++ b/drivers/pci/controller/dwc/pcie-qcom.c > @@ -248,6 +248,8 @@ struct qcom_pcie_cfg { > bool no_l0s; > }; > > +#define QCOM_PCIE_SLOT_MAX_SUPPLIES 3 > + > struct qcom_pcie { > struct dw_pcie *pci; > void __iomem *parf; /* DT parf */ > @@ -260,6 +262,7 @@ struct qcom_pcie { > struct icc_path *icc_cpu; > const struct qcom_pcie_cfg *cfg; > struct dentry *debugfs; > + struct regulator_bulk_data slot_supplies[QCOM_PCIE_SLOT_MAX_SUPPLIES]; > bool suspended; > bool use_pm_opp; > }; > @@ -1174,6 +1177,41 @@ static int qcom_pcie_link_up(struct dw_pcie *pci) > return !!(val & PCI_EXP_LNKSTA_DLLLA); > } > > +static int qcom_pcie_enable_slot_supplies(struct qcom_pcie *pcie) > +{ > + struct dw_pcie *pci = pcie->pci; > + int ret; > + > + ret = regulator_bulk_enable(ARRAY_SIZE(pcie->slot_supplies), > + pcie->slot_supplies); > + if (ret < 0) > + dev_err(pci->dev, "Failed to enable slot regulators\n"); return dev_err_probe would be a good call.. probably more so below, but won't hurt to use here too > + > + return ret; > +} > + > +static void qcom_pcie_disable_slot_supplies(struct qcom_pcie *pcie) > +{ > + regulator_bulk_disable(ARRAY_SIZE(pcie->slot_supplies), > + pcie->slot_supplies); > +} This I feel like is overly abstracted Konrad
On Tue, Aug 27, 2024 at 02:44:09PM +0300, Dmitry Baryshkov wrote: > On Tue, 27 Aug 2024 at 09:36, Qiang Yu <quic_qianyu@quicinc.com> wrote: > > > > On platform x1e80100 QCP, PCIe3 is a standard x8 form factor. Hence, add > > support to use 3.3v, 3.3v aux and 12v regulators. > > First of all, I don't see corresponding bindings change. > > Second, these supplies power up the slot, not the host controller > itself. As such these supplies do not belong to the host controller > entry. Please consider using the pwrseq framework instead. > Indeed. For legacy reasons, slot power supplies were populated in the host bridge node itself until recently Rob started objecting it [1]. And it makes real sense to put these supplies in the root port node and handle them in the relevant driver. I'm still evaluating whether the handling should be done in the portdrv or pwrctl driver, but haven't reached the conclusion. Pwrctl seems to be the ideal choice, but I see a few issues related to handling the OF node for the root port. Hope I'll come to a conclusion in the next few days and will update this thread. - Mani [1] https://lore.kernel.org/lkml/20240604235806.GA1903493-robh@kernel.org/ > > > > Signed-off-by: Qiang Yu <quic_qianyu@quicinc.com> > > --- > > drivers/pci/controller/dwc/pcie-qcom.c | 52 +++++++++++++++++++++++++- > > 1 file changed, 50 insertions(+), 2 deletions(-) > > > > diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c > > index 6f953e32d990..59fb415dfeeb 100644 > > --- a/drivers/pci/controller/dwc/pcie-qcom.c > > +++ b/drivers/pci/controller/dwc/pcie-qcom.c > > @@ -248,6 +248,8 @@ struct qcom_pcie_cfg { > > bool no_l0s; > > }; > > > > +#define QCOM_PCIE_SLOT_MAX_SUPPLIES 3 > > + > > struct qcom_pcie { > > struct dw_pcie *pci; > > void __iomem *parf; /* DT parf */ > > @@ -260,6 +262,7 @@ struct qcom_pcie { > > struct icc_path *icc_cpu; > > const struct qcom_pcie_cfg *cfg; > > struct dentry *debugfs; > > + struct regulator_bulk_data slot_supplies[QCOM_PCIE_SLOT_MAX_SUPPLIES]; > > bool suspended; > > bool use_pm_opp; > > }; > > @@ -1174,6 +1177,41 @@ static int qcom_pcie_link_up(struct dw_pcie *pci) > > return !!(val & PCI_EXP_LNKSTA_DLLLA); > > } > > > > +static int qcom_pcie_enable_slot_supplies(struct qcom_pcie *pcie) > > +{ > > + struct dw_pcie *pci = pcie->pci; > > + int ret; > > + > > + ret = regulator_bulk_enable(ARRAY_SIZE(pcie->slot_supplies), > > + pcie->slot_supplies); > > + if (ret < 0) > > + dev_err(pci->dev, "Failed to enable slot regulators\n"); > > + > > + return ret; > > +} > > + > > +static void qcom_pcie_disable_slot_supplies(struct qcom_pcie *pcie) > > +{ > > + regulator_bulk_disable(ARRAY_SIZE(pcie->slot_supplies), > > + pcie->slot_supplies); > > +} > > + > > +static int qcom_pcie_get_slot_supplies(struct qcom_pcie *pcie) > > +{ > > + struct dw_pcie *pci = pcie->pci; > > + int ret; > > + > > + pcie->slot_supplies[0].supply = "vpcie12v"; > > + pcie->slot_supplies[1].supply = "vpcie3v3"; > > + pcie->slot_supplies[2].supply = "vpcie3v3aux"; > > + ret = devm_regulator_bulk_get(pci->dev, ARRAY_SIZE(pcie->slot_supplies), > > + pcie->slot_supplies); > > + if (ret < 0) > > + dev_err(pci->dev, "Failed to get slot regulators\n"); > > + > > + return ret; > > +} > > + > > static int qcom_pcie_host_init(struct dw_pcie_rp *pp) > > { > > struct dw_pcie *pci = to_dw_pcie_from_pp(pp); > > @@ -1182,10 +1220,14 @@ static int qcom_pcie_host_init(struct dw_pcie_rp *pp) > > > > qcom_ep_reset_assert(pcie); > > > > - ret = pcie->cfg->ops->init(pcie); > > + ret = qcom_pcie_enable_slot_supplies(pcie); > > if (ret) > > return ret; > > > > + ret = pcie->cfg->ops->init(pcie); > > + if (ret) > > + goto err_disable_slot; > > + > > ret = phy_set_mode_ext(pcie->phy, PHY_MODE_PCIE, PHY_MODE_PCIE_RC); > > if (ret) > > goto err_deinit; > > @@ -1216,7 +1258,8 @@ static int qcom_pcie_host_init(struct dw_pcie_rp *pp) > > phy_power_off(pcie->phy); > > err_deinit: > > pcie->cfg->ops->deinit(pcie); > > - > > +err_disable_slot: > > + qcom_pcie_disable_slot_supplies(pcie); > > return ret; > > } > > > > @@ -1228,6 +1271,7 @@ static void qcom_pcie_host_deinit(struct dw_pcie_rp *pp) > > qcom_ep_reset_assert(pcie); > > phy_power_off(pcie->phy); > > pcie->cfg->ops->deinit(pcie); > > + qcom_pcie_disable_slot_supplies(pcie); > > } > > > > static void qcom_pcie_host_post_init(struct dw_pcie_rp *pp) > > @@ -1602,6 +1646,10 @@ static int qcom_pcie_probe(struct platform_device *pdev) > > goto err_pm_runtime_put; > > } > > > > + ret = qcom_pcie_get_slot_supplies(pcie); > > + if (ret) > > + goto err_pm_runtime_put; > > + > > ret = pcie->cfg->ops->get_resources(pcie); > > if (ret) > > goto err_pm_runtime_put; > > -- > > 2.34.1 > > > > > -- > With best wishes > Dmitry
On Wed, Sep 11, 2024 at 04:17:41PM +0800, Qiang Yu wrote: > > On 8/28/2024 12:58 AM, Manivannan Sadhasivam wrote: > > On Tue, Aug 27, 2024 at 02:44:09PM +0300, Dmitry Baryshkov wrote: > > > On Tue, 27 Aug 2024 at 09:36, Qiang Yu <quic_qianyu@quicinc.com> wrote: > > > > On platform x1e80100 QCP, PCIe3 is a standard x8 form factor. Hence, add > > > > support to use 3.3v, 3.3v aux and 12v regulators. > > > First of all, I don't see corresponding bindings change. > > > > > > Second, these supplies power up the slot, not the host controller > > > itself. As such these supplies do not belong to the host controller > > > entry. Please consider using the pwrseq framework instead. > > > > > Indeed. For legacy reasons, slot power supplies were populated in the host > > bridge node itself until recently Rob started objecting it [1]. And it makes > > real sense to put these supplies in the root port node and handle them in the > > relevant driver. > > > > I'm still evaluating whether the handling should be done in the portdrv or > > pwrctl driver, but haven't reached the conclusion. Pwrctl seems to be the ideal > > choice, but I see a few issues related to handling the OF node for the root > > port. > > > > Hope I'll come to a conclusion in the next few days and will update this thread. > > > > - Mani > > > > [1] https://lore.kernel.org/lkml/20240604235806.GA1903493-robh@kernel.org/ > Hi Mani, do you have any updates? > I'm working with Bartosz to add a new pwrctl driver for rootports. And we are debugging an issue currently. Unfortunately, the progress is very slow as I'm on vacation still. Will post the patches once it got resolved. - Mani > Thanks, > Qiang > > > > > > Signed-off-by: Qiang Yu <quic_qianyu@quicinc.com> > > > > --- > > > > drivers/pci/controller/dwc/pcie-qcom.c | 52 +++++++++++++++++++++++++- > > > > 1 file changed, 50 insertions(+), 2 deletions(-) > > > > > > > > diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c > > > > index 6f953e32d990..59fb415dfeeb 100644 > > > > --- a/drivers/pci/controller/dwc/pcie-qcom.c > > > > +++ b/drivers/pci/controller/dwc/pcie-qcom.c > > > > @@ -248,6 +248,8 @@ struct qcom_pcie_cfg { > > > > bool no_l0s; > > > > }; > > > > > > > > +#define QCOM_PCIE_SLOT_MAX_SUPPLIES 3 > > > > + > > > > struct qcom_pcie { > > > > struct dw_pcie *pci; > > > > void __iomem *parf; /* DT parf */ > > > > @@ -260,6 +262,7 @@ struct qcom_pcie { > > > > struct icc_path *icc_cpu; > > > > const struct qcom_pcie_cfg *cfg; > > > > struct dentry *debugfs; > > > > + struct regulator_bulk_data slot_supplies[QCOM_PCIE_SLOT_MAX_SUPPLIES]; > > > > bool suspended; > > > > bool use_pm_opp; > > > > }; > > > > @@ -1174,6 +1177,41 @@ static int qcom_pcie_link_up(struct dw_pcie *pci) > > > > return !!(val & PCI_EXP_LNKSTA_DLLLA); > > > > } > > > > > > > > +static int qcom_pcie_enable_slot_supplies(struct qcom_pcie *pcie) > > > > +{ > > > > + struct dw_pcie *pci = pcie->pci; > > > > + int ret; > > > > + > > > > + ret = regulator_bulk_enable(ARRAY_SIZE(pcie->slot_supplies), > > > > + pcie->slot_supplies); > > > > + if (ret < 0) > > > > + dev_err(pci->dev, "Failed to enable slot regulators\n"); > > > > + > > > > + return ret; > > > > +} > > > > + > > > > +static void qcom_pcie_disable_slot_supplies(struct qcom_pcie *pcie) > > > > +{ > > > > + regulator_bulk_disable(ARRAY_SIZE(pcie->slot_supplies), > > > > + pcie->slot_supplies); > > > > +} > > > > + > > > > +static int qcom_pcie_get_slot_supplies(struct qcom_pcie *pcie) > > > > +{ > > > > + struct dw_pcie *pci = pcie->pci; > > > > + int ret; > > > > + > > > > + pcie->slot_supplies[0].supply = "vpcie12v"; > > > > + pcie->slot_supplies[1].supply = "vpcie3v3"; > > > > + pcie->slot_supplies[2].supply = "vpcie3v3aux"; > > > > + ret = devm_regulator_bulk_get(pci->dev, ARRAY_SIZE(pcie->slot_supplies), > > > > + pcie->slot_supplies); > > > > + if (ret < 0) > > > > + dev_err(pci->dev, "Failed to get slot regulators\n"); > > > > + > > > > + return ret; > > > > +} > > > > + > > > > static int qcom_pcie_host_init(struct dw_pcie_rp *pp) > > > > { > > > > struct dw_pcie *pci = to_dw_pcie_from_pp(pp); > > > > @@ -1182,10 +1220,14 @@ static int qcom_pcie_host_init(struct dw_pcie_rp *pp) > > > > > > > > qcom_ep_reset_assert(pcie); > > > > > > > > - ret = pcie->cfg->ops->init(pcie); > > > > + ret = qcom_pcie_enable_slot_supplies(pcie); > > > > if (ret) > > > > return ret; > > > > > > > > + ret = pcie->cfg->ops->init(pcie); > > > > + if (ret) > > > > + goto err_disable_slot; > > > > + > > > > ret = phy_set_mode_ext(pcie->phy, PHY_MODE_PCIE, PHY_MODE_PCIE_RC); > > > > if (ret) > > > > goto err_deinit; > > > > @@ -1216,7 +1258,8 @@ static int qcom_pcie_host_init(struct dw_pcie_rp *pp) > > > > phy_power_off(pcie->phy); > > > > err_deinit: > > > > pcie->cfg->ops->deinit(pcie); > > > > - > > > > +err_disable_slot: > > > > + qcom_pcie_disable_slot_supplies(pcie); > > > > return ret; > > > > } > > > > > > > > @@ -1228,6 +1271,7 @@ static void qcom_pcie_host_deinit(struct dw_pcie_rp *pp) > > > > qcom_ep_reset_assert(pcie); > > > > phy_power_off(pcie->phy); > > > > pcie->cfg->ops->deinit(pcie); > > > > + qcom_pcie_disable_slot_supplies(pcie); > > > > } > > > > > > > > static void qcom_pcie_host_post_init(struct dw_pcie_rp *pp) > > > > @@ -1602,6 +1646,10 @@ static int qcom_pcie_probe(struct platform_device *pdev) > > > > goto err_pm_runtime_put; > > > > } > > > > > > > > + ret = qcom_pcie_get_slot_supplies(pcie); > > > > + if (ret) > > > > + goto err_pm_runtime_put; > > > > + > > > > ret = pcie->cfg->ops->get_resources(pcie); > > > > if (ret) > > > > goto err_pm_runtime_put; > > > > -- > > > > 2.34.1 > > > > > > > > > > -- > > > With best wishes > > > Dmitry
On 9/11/2024 11:32 PM, Manivannan Sadhasivam wrote: > On Wed, Sep 11, 2024 at 04:17:41PM +0800, Qiang Yu wrote: >> On 8/28/2024 12:58 AM, Manivannan Sadhasivam wrote: >>> On Tue, Aug 27, 2024 at 02:44:09PM +0300, Dmitry Baryshkov wrote: >>>> On Tue, 27 Aug 2024 at 09:36, Qiang Yu <quic_qianyu@quicinc.com> wrote: >>>>> On platform x1e80100 QCP, PCIe3 is a standard x8 form factor. Hence, add >>>>> support to use 3.3v, 3.3v aux and 12v regulators. >>>> First of all, I don't see corresponding bindings change. >>>> >>>> Second, these supplies power up the slot, not the host controller >>>> itself. As such these supplies do not belong to the host controller >>>> entry. Please consider using the pwrseq framework instead. >>>> >>> Indeed. For legacy reasons, slot power supplies were populated in the host >>> bridge node itself until recently Rob started objecting it [1]. And it makes >>> real sense to put these supplies in the root port node and handle them in the >>> relevant driver. >>> >>> I'm still evaluating whether the handling should be done in the portdrv or >>> pwrctl driver, but haven't reached the conclusion. Pwrctl seems to be the ideal >>> choice, but I see a few issues related to handling the OF node for the root >>> port. >>> >>> Hope I'll come to a conclusion in the next few days and will update this thread. >>> >>> - Mani >>> >>> [1] https://lore.kernel.org/lkml/20240604235806.GA1903493-robh@kernel.org/ >> Hi Mani, do you have any updates? >> > I'm working with Bartosz to add a new pwrctl driver for rootports. And we are > debugging an issue currently. Unfortunately, the progress is very slow as I'm on > vacation still. > > Will post the patches once it got resolved. > > - Mani OK, thanks for your update. Thanks, Qiang >> Thanks, >> Qiang >>>>> Signed-off-by: Qiang Yu <quic_qianyu@quicinc.com> >>>>> --- >>>>> drivers/pci/controller/dwc/pcie-qcom.c | 52 +++++++++++++++++++++++++- >>>>> 1 file changed, 50 insertions(+), 2 deletions(-) >>>>> >>>>> diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c >>>>> index 6f953e32d990..59fb415dfeeb 100644 >>>>> --- a/drivers/pci/controller/dwc/pcie-qcom.c >>>>> +++ b/drivers/pci/controller/dwc/pcie-qcom.c >>>>> @@ -248,6 +248,8 @@ struct qcom_pcie_cfg { >>>>> bool no_l0s; >>>>> }; >>>>> >>>>> +#define QCOM_PCIE_SLOT_MAX_SUPPLIES 3 >>>>> + >>>>> struct qcom_pcie { >>>>> struct dw_pcie *pci; >>>>> void __iomem *parf; /* DT parf */ >>>>> @@ -260,6 +262,7 @@ struct qcom_pcie { >>>>> struct icc_path *icc_cpu; >>>>> const struct qcom_pcie_cfg *cfg; >>>>> struct dentry *debugfs; >>>>> + struct regulator_bulk_data slot_supplies[QCOM_PCIE_SLOT_MAX_SUPPLIES]; >>>>> bool suspended; >>>>> bool use_pm_opp; >>>>> }; >>>>> @@ -1174,6 +1177,41 @@ static int qcom_pcie_link_up(struct dw_pcie *pci) >>>>> return !!(val & PCI_EXP_LNKSTA_DLLLA); >>>>> } >>>>> >>>>> +static int qcom_pcie_enable_slot_supplies(struct qcom_pcie *pcie) >>>>> +{ >>>>> + struct dw_pcie *pci = pcie->pci; >>>>> + int ret; >>>>> + >>>>> + ret = regulator_bulk_enable(ARRAY_SIZE(pcie->slot_supplies), >>>>> + pcie->slot_supplies); >>>>> + if (ret < 0) >>>>> + dev_err(pci->dev, "Failed to enable slot regulators\n"); >>>>> + >>>>> + return ret; >>>>> +} >>>>> + >>>>> +static void qcom_pcie_disable_slot_supplies(struct qcom_pcie *pcie) >>>>> +{ >>>>> + regulator_bulk_disable(ARRAY_SIZE(pcie->slot_supplies), >>>>> + pcie->slot_supplies); >>>>> +} >>>>> + >>>>> +static int qcom_pcie_get_slot_supplies(struct qcom_pcie *pcie) >>>>> +{ >>>>> + struct dw_pcie *pci = pcie->pci; >>>>> + int ret; >>>>> + >>>>> + pcie->slot_supplies[0].supply = "vpcie12v"; >>>>> + pcie->slot_supplies[1].supply = "vpcie3v3"; >>>>> + pcie->slot_supplies[2].supply = "vpcie3v3aux"; >>>>> + ret = devm_regulator_bulk_get(pci->dev, ARRAY_SIZE(pcie->slot_supplies), >>>>> + pcie->slot_supplies); >>>>> + if (ret < 0) >>>>> + dev_err(pci->dev, "Failed to get slot regulators\n"); >>>>> + >>>>> + return ret; >>>>> +} >>>>> + >>>>> static int qcom_pcie_host_init(struct dw_pcie_rp *pp) >>>>> { >>>>> struct dw_pcie *pci = to_dw_pcie_from_pp(pp); >>>>> @@ -1182,10 +1220,14 @@ static int qcom_pcie_host_init(struct dw_pcie_rp *pp) >>>>> >>>>> qcom_ep_reset_assert(pcie); >>>>> >>>>> - ret = pcie->cfg->ops->init(pcie); >>>>> + ret = qcom_pcie_enable_slot_supplies(pcie); >>>>> if (ret) >>>>> return ret; >>>>> >>>>> + ret = pcie->cfg->ops->init(pcie); >>>>> + if (ret) >>>>> + goto err_disable_slot; >>>>> + >>>>> ret = phy_set_mode_ext(pcie->phy, PHY_MODE_PCIE, PHY_MODE_PCIE_RC); >>>>> if (ret) >>>>> goto err_deinit; >>>>> @@ -1216,7 +1258,8 @@ static int qcom_pcie_host_init(struct dw_pcie_rp *pp) >>>>> phy_power_off(pcie->phy); >>>>> err_deinit: >>>>> pcie->cfg->ops->deinit(pcie); >>>>> - >>>>> +err_disable_slot: >>>>> + qcom_pcie_disable_slot_supplies(pcie); >>>>> return ret; >>>>> } >>>>> >>>>> @@ -1228,6 +1271,7 @@ static void qcom_pcie_host_deinit(struct dw_pcie_rp *pp) >>>>> qcom_ep_reset_assert(pcie); >>>>> phy_power_off(pcie->phy); >>>>> pcie->cfg->ops->deinit(pcie); >>>>> + qcom_pcie_disable_slot_supplies(pcie); >>>>> } >>>>> >>>>> static void qcom_pcie_host_post_init(struct dw_pcie_rp *pp) >>>>> @@ -1602,6 +1646,10 @@ static int qcom_pcie_probe(struct platform_device *pdev) >>>>> goto err_pm_runtime_put; >>>>> } >>>>> >>>>> + ret = qcom_pcie_get_slot_supplies(pcie); >>>>> + if (ret) >>>>> + goto err_pm_runtime_put; >>>>> + >>>>> ret = pcie->cfg->ops->get_resources(pcie); >>>>> if (ret) >>>>> goto err_pm_runtime_put; >>>>> -- >>>>> 2.34.1 >>>>> >>>> -- >>>> With best wishes >>>> Dmitry
On 12.09.2024 3:39 PM, Qiang Yu wrote: > > On 9/11/2024 11:32 PM, Manivannan Sadhasivam wrote: >> On Wed, Sep 11, 2024 at 04:17:41PM +0800, Qiang Yu wrote: >>> On 8/28/2024 12:58 AM, Manivannan Sadhasivam wrote: >>>> On Tue, Aug 27, 2024 at 02:44:09PM +0300, Dmitry Baryshkov wrote: >>>>> On Tue, 27 Aug 2024 at 09:36, Qiang Yu <quic_qianyu@quicinc.com> wrote: >>>>>> On platform x1e80100 QCP, PCIe3 is a standard x8 form factor. Hence, add >>>>>> support to use 3.3v, 3.3v aux and 12v regulators. >>>>> First of all, I don't see corresponding bindings change. >>>>> >>>>> Second, these supplies power up the slot, not the host controller >>>>> itself. As such these supplies do not belong to the host controller >>>>> entry. Please consider using the pwrseq framework instead. >>>>> >>>> Indeed. For legacy reasons, slot power supplies were populated in the host >>>> bridge node itself until recently Rob started objecting it [1]. And it makes >>>> real sense to put these supplies in the root port node and handle them in the >>>> relevant driver. >>>> >>>> I'm still evaluating whether the handling should be done in the portdrv or >>>> pwrctl driver, but haven't reached the conclusion. Pwrctl seems to be the ideal >>>> choice, but I see a few issues related to handling the OF node for the root >>>> port. >>>> >>>> Hope I'll come to a conclusion in the next few days and will update this thread. >>>> >>>> - Mani >>>> >>>> [1] https://lore.kernel.org/lkml/20240604235806.GA1903493-robh@kernel.org/ >>> Hi Mani, do you have any updates? >>> >> I'm working with Bartosz to add a new pwrctl driver for rootports. And we are >> debugging an issue currently. Unfortunately, the progress is very slow as I'm on >> vacation still. >> >> Will post the patches once it got resolved. >> >> - Mani > OK, thanks for your update. Qiang, you can still resubmit the rest of the patches without having to wait on that to be resolved Konrad
On Thu, Sep 12, 2024 at 04:15:56PM +0200, Konrad Dybcio wrote: > On 12.09.2024 3:39 PM, Qiang Yu wrote: > > > > On 9/11/2024 11:32 PM, Manivannan Sadhasivam wrote: > >> On Wed, Sep 11, 2024 at 04:17:41PM +0800, Qiang Yu wrote: > >>> On 8/28/2024 12:58 AM, Manivannan Sadhasivam wrote: > >>>> On Tue, Aug 27, 2024 at 02:44:09PM +0300, Dmitry Baryshkov wrote: > >>>>> On Tue, 27 Aug 2024 at 09:36, Qiang Yu <quic_qianyu@quicinc.com> wrote: > >>>>>> On platform x1e80100 QCP, PCIe3 is a standard x8 form factor. Hence, add > >>>>>> support to use 3.3v, 3.3v aux and 12v regulators. > >>>>> First of all, I don't see corresponding bindings change. > >>>>> > >>>>> Second, these supplies power up the slot, not the host controller > >>>>> itself. As such these supplies do not belong to the host controller > >>>>> entry. Please consider using the pwrseq framework instead. > >>>>> > >>>> Indeed. For legacy reasons, slot power supplies were populated in the host > >>>> bridge node itself until recently Rob started objecting it [1]. And it makes > >>>> real sense to put these supplies in the root port node and handle them in the > >>>> relevant driver. > >>>> > >>>> I'm still evaluating whether the handling should be done in the portdrv or > >>>> pwrctl driver, but haven't reached the conclusion. Pwrctl seems to be the ideal > >>>> choice, but I see a few issues related to handling the OF node for the root > >>>> port. > >>>> > >>>> Hope I'll come to a conclusion in the next few days and will update this thread. > >>>> > >>>> - Mani > >>>> > >>>> [1] https://lore.kernel.org/lkml/20240604235806.GA1903493-robh@kernel.org/ > >>> Hi Mani, do you have any updates? > >>> > >> I'm working with Bartosz to add a new pwrctl driver for rootports. And we are > >> debugging an issue currently. Unfortunately, the progress is very slow as I'm on > >> vacation still. > >> > >> Will post the patches once it got resolved. > >> > >> - Mani > > OK, thanks for your update. > > Qiang, you can still resubmit the rest of the patches without having > to wait on that to be resolved > In that case, the slot supplies should be described in the PCIe bridge. - Mani > Konrad
On Thu, 12 Sept 2024 at 17:45, Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> wrote: > > On Thu, Sep 12, 2024 at 04:15:56PM +0200, Konrad Dybcio wrote: > > On 12.09.2024 3:39 PM, Qiang Yu wrote: > > > > > > On 9/11/2024 11:32 PM, Manivannan Sadhasivam wrote: > > >> On Wed, Sep 11, 2024 at 04:17:41PM +0800, Qiang Yu wrote: > > >>> On 8/28/2024 12:58 AM, Manivannan Sadhasivam wrote: > > >>>> On Tue, Aug 27, 2024 at 02:44:09PM +0300, Dmitry Baryshkov wrote: > > >>>>> On Tue, 27 Aug 2024 at 09:36, Qiang Yu <quic_qianyu@quicinc.com> wrote: > > >>>>>> On platform x1e80100 QCP, PCIe3 is a standard x8 form factor. Hence, add > > >>>>>> support to use 3.3v, 3.3v aux and 12v regulators. > > >>>>> First of all, I don't see corresponding bindings change. > > >>>>> > > >>>>> Second, these supplies power up the slot, not the host controller > > >>>>> itself. As such these supplies do not belong to the host controller > > >>>>> entry. Please consider using the pwrseq framework instead. > > >>>>> > > >>>> Indeed. For legacy reasons, slot power supplies were populated in the host > > >>>> bridge node itself until recently Rob started objecting it [1]. And it makes > > >>>> real sense to put these supplies in the root port node and handle them in the > > >>>> relevant driver. > > >>>> > > >>>> I'm still evaluating whether the handling should be done in the portdrv or > > >>>> pwrctl driver, but haven't reached the conclusion. Pwrctl seems to be the ideal > > >>>> choice, but I see a few issues related to handling the OF node for the root > > >>>> port. > > >>>> > > >>>> Hope I'll come to a conclusion in the next few days and will update this thread. > > >>>> > > >>>> - Mani > > >>>> > > >>>> [1] https://lore.kernel.org/lkml/20240604235806.GA1903493-robh@kernel.org/ > > >>> Hi Mani, do you have any updates? > > >>> > > >> I'm working with Bartosz to add a new pwrctl driver for rootports. And we are > > >> debugging an issue currently. Unfortunately, the progress is very slow as I'm on > > >> vacation still. > > >> > > >> Will post the patches once it got resolved. > > >> > > >> - Mani > > > OK, thanks for your update. > > > > Qiang, you can still resubmit the rest of the patches without having > > to wait on that to be resolved > > > > In that case, the slot supplies should be described in the PCIe bridge. Patches 1-6 don't seem to depend on slot supplies, so they can be submitted separately. > > - Mani > > > Konrad > > -- > மணிவண்ணன் சதாசிவம்
On 9/12/2024 10:49 PM, Dmitry Baryshkov wrote: > On Thu, 12 Sept 2024 at 17:45, Manivannan Sadhasivam > <manivannan.sadhasivam@linaro.org> wrote: >> On Thu, Sep 12, 2024 at 04:15:56PM +0200, Konrad Dybcio wrote: >>> On 12.09.2024 3:39 PM, Qiang Yu wrote: >>>> On 9/11/2024 11:32 PM, Manivannan Sadhasivam wrote: >>>>> On Wed, Sep 11, 2024 at 04:17:41PM +0800, Qiang Yu wrote: >>>>>> On 8/28/2024 12:58 AM, Manivannan Sadhasivam wrote: >>>>>>> On Tue, Aug 27, 2024 at 02:44:09PM +0300, Dmitry Baryshkov wrote: >>>>>>>> On Tue, 27 Aug 2024 at 09:36, Qiang Yu <quic_qianyu@quicinc.com> wrote: >>>>>>>>> On platform x1e80100 QCP, PCIe3 is a standard x8 form factor. Hence, add >>>>>>>>> support to use 3.3v, 3.3v aux and 12v regulators. >>>>>>>> First of all, I don't see corresponding bindings change. >>>>>>>> >>>>>>>> Second, these supplies power up the slot, not the host controller >>>>>>>> itself. As such these supplies do not belong to the host controller >>>>>>>> entry. Please consider using the pwrseq framework instead. >>>>>>>> >>>>>>> Indeed. For legacy reasons, slot power supplies were populated in the host >>>>>>> bridge node itself until recently Rob started objecting it [1]. And it makes >>>>>>> real sense to put these supplies in the root port node and handle them in the >>>>>>> relevant driver. >>>>>>> >>>>>>> I'm still evaluating whether the handling should be done in the portdrv or >>>>>>> pwrctl driver, but haven't reached the conclusion. Pwrctl seems to be the ideal >>>>>>> choice, but I see a few issues related to handling the OF node for the root >>>>>>> port. >>>>>>> >>>>>>> Hope I'll come to a conclusion in the next few days and will update this thread. >>>>>>> >>>>>>> - Mani >>>>>>> >>>>>>> [1] https://lore.kernel.org/lkml/20240604235806.GA1903493-robh@kernel.org/ >>>>>> Hi Mani, do you have any updates? >>>>>> >>>>> I'm working with Bartosz to add a new pwrctl driver for rootports. And we are >>>>> debugging an issue currently. Unfortunately, the progress is very slow as I'm on >>>>> vacation still. >>>>> >>>>> Will post the patches once it got resolved. >>>>> >>>>> - Mani >>>> OK, thanks for your update. >>> Qiang, you can still resubmit the rest of the patches without having >>> to wait on that to be resolved >>> >> In that case, the slot supplies should be described in the PCIe bridge. > Patches 1-6 don't seem to depend on slot supplies, so they can be > submitted separately. OK, let me send v2 patch. Hi Mani, if you need any supports, please let me know. Thanks, Qiang > >> - Mani >> >>> Konrad >> -- >> மணிவண்ணன் சதாசிவம் > >
diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c index 6f953e32d990..59fb415dfeeb 100644 --- a/drivers/pci/controller/dwc/pcie-qcom.c +++ b/drivers/pci/controller/dwc/pcie-qcom.c @@ -248,6 +248,8 @@ struct qcom_pcie_cfg { bool no_l0s; }; +#define QCOM_PCIE_SLOT_MAX_SUPPLIES 3 + struct qcom_pcie { struct dw_pcie *pci; void __iomem *parf; /* DT parf */ @@ -260,6 +262,7 @@ struct qcom_pcie { struct icc_path *icc_cpu; const struct qcom_pcie_cfg *cfg; struct dentry *debugfs; + struct regulator_bulk_data slot_supplies[QCOM_PCIE_SLOT_MAX_SUPPLIES]; bool suspended; bool use_pm_opp; }; @@ -1174,6 +1177,41 @@ static int qcom_pcie_link_up(struct dw_pcie *pci) return !!(val & PCI_EXP_LNKSTA_DLLLA); } +static int qcom_pcie_enable_slot_supplies(struct qcom_pcie *pcie) +{ + struct dw_pcie *pci = pcie->pci; + int ret; + + ret = regulator_bulk_enable(ARRAY_SIZE(pcie->slot_supplies), + pcie->slot_supplies); + if (ret < 0) + dev_err(pci->dev, "Failed to enable slot regulators\n"); + + return ret; +} + +static void qcom_pcie_disable_slot_supplies(struct qcom_pcie *pcie) +{ + regulator_bulk_disable(ARRAY_SIZE(pcie->slot_supplies), + pcie->slot_supplies); +} + +static int qcom_pcie_get_slot_supplies(struct qcom_pcie *pcie) +{ + struct dw_pcie *pci = pcie->pci; + int ret; + + pcie->slot_supplies[0].supply = "vpcie12v"; + pcie->slot_supplies[1].supply = "vpcie3v3"; + pcie->slot_supplies[2].supply = "vpcie3v3aux"; + ret = devm_regulator_bulk_get(pci->dev, ARRAY_SIZE(pcie->slot_supplies), + pcie->slot_supplies); + if (ret < 0) + dev_err(pci->dev, "Failed to get slot regulators\n"); + + return ret; +} + static int qcom_pcie_host_init(struct dw_pcie_rp *pp) { struct dw_pcie *pci = to_dw_pcie_from_pp(pp); @@ -1182,10 +1220,14 @@ static int qcom_pcie_host_init(struct dw_pcie_rp *pp) qcom_ep_reset_assert(pcie); - ret = pcie->cfg->ops->init(pcie); + ret = qcom_pcie_enable_slot_supplies(pcie); if (ret) return ret; + ret = pcie->cfg->ops->init(pcie); + if (ret) + goto err_disable_slot; + ret = phy_set_mode_ext(pcie->phy, PHY_MODE_PCIE, PHY_MODE_PCIE_RC); if (ret) goto err_deinit; @@ -1216,7 +1258,8 @@ static int qcom_pcie_host_init(struct dw_pcie_rp *pp) phy_power_off(pcie->phy); err_deinit: pcie->cfg->ops->deinit(pcie); - +err_disable_slot: + qcom_pcie_disable_slot_supplies(pcie); return ret; } @@ -1228,6 +1271,7 @@ static void qcom_pcie_host_deinit(struct dw_pcie_rp *pp) qcom_ep_reset_assert(pcie); phy_power_off(pcie->phy); pcie->cfg->ops->deinit(pcie); + qcom_pcie_disable_slot_supplies(pcie); } static void qcom_pcie_host_post_init(struct dw_pcie_rp *pp) @@ -1602,6 +1646,10 @@ static int qcom_pcie_probe(struct platform_device *pdev) goto err_pm_runtime_put; } + ret = qcom_pcie_get_slot_supplies(pcie); + if (ret) + goto err_pm_runtime_put; + ret = pcie->cfg->ops->get_resources(pcie); if (ret) goto err_pm_runtime_put;
On platform x1e80100 QCP, PCIe3 is a standard x8 form factor. Hence, add support to use 3.3v, 3.3v aux and 12v regulators. Signed-off-by: Qiang Yu <quic_qianyu@quicinc.com> --- drivers/pci/controller/dwc/pcie-qcom.c | 52 +++++++++++++++++++++++++- 1 file changed, 50 insertions(+), 2 deletions(-)