> -----Original Message----- > From: Manivannan Sadhasivam <manivannan.sadhasivam@xxxxxxxxxx> > Sent: 2024年10月23日 1:18 > To: Hongxing Zhu <hongxing.zhu@xxxxxxx> > Cc: kw@xxxxxxxxx; bhelgaas@xxxxxxxxxx; lpieralisi@xxxxxxxxxx; Frank Li > <frank.li@xxxxxxx>; l.stach@xxxxxxxxxxxxxx; robh+dt@xxxxxxxxxx; > conor+dt@xxxxxxxxxx; shawnguo@xxxxxxxxxx; > krzysztof.kozlowski+dt@xxxxxxxxxx; festevam@xxxxxxxxx; > s.hauer@xxxxxxxxxxxxxx; linux-pci@xxxxxxxxxxxxxxx; > linux-arm-kernel@xxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; > devicetree@xxxxxxxxxxxxxxx; kernel@xxxxxxxxxxxxxx; imx@xxxxxxxxxxxxxxx > Subject: Re: [PATCH v4 7/9] PCI: imx6: Use dwc common suspend resume > method > > On Tue, Oct 15, 2024 at 04:33:31PM +0800, Richard Zhu wrote: > > From: Frank Li <Frank.Li@xxxxxxx> > > > > Call common dwc suspend/resume function. Use dwc common iATU > method to > > send out PME_TURN_OFF message. Old platform such as iMX6SX and > iMX6QP, > > iATU > > CTRL2 bit 22 (PCIE_ATU_INHIBIT_PAYLOAD) are reserved. So can't send > > out MSG without data by dummy MMIO write. Without > > PCIE_ATU_INHIBIT_PAYLOAD, MSGD will be sent out instead of MSG. So > > keep old method to send PME_TURN_OFF MSG. > > > > This PME_Turn_Off implementation is the only difference between the DWC > common ops and the custom one here? I don't think so. Please describe all > the differences. > Hi Manivannan: Thanks for your comments. On i.MX6SX and i.MX6QP platforms, the PME_TURN_OFF would be issued by the hooked callback pp.ops->pme_turn_off(). They are same to the previous operations. In the other words, these two platforms don't use the dwc generic PME_TURN_OFF implementation. Do you means to describe all the differences between dwc common suspend/resume and custom suspend/resume in commit message? In dwc suspend, the l1ss and link stats are checked before suspend entry. And after PME_TURN_OFF is issued, the L2 stat is polling. These two differences are not contained in the custom PM operations. > > Signed-off-by: Frank Li <Frank.Li@xxxxxxx> > > Signed-off-by: Richard Zhu <hongxing.zhu@xxxxxxx> > > --- > > drivers/pci/controller/dwc/pci-imx6.c | 97 > > ++++++++++----------------- > > 1 file changed, 36 insertions(+), 61 deletions(-) > > > > diff --git a/drivers/pci/controller/dwc/pci-imx6.c > > b/drivers/pci/controller/dwc/pci-imx6.c > > index 161daad34a94..baa853d84b4d 100644 > > --- a/drivers/pci/controller/dwc/pci-imx6.c > > +++ b/drivers/pci/controller/dwc/pci-imx6.c > > @@ -33,6 +33,7 @@ > > #include <linux/pm_domain.h> > > #include <linux/pm_runtime.h> > > > > +#include "../../pci.h" > > #include "pcie-designware.h" > > > > #define IMX8MQ_GPR_PCIE_REF_USE_PAD BIT(9) > > @@ -82,6 +83,7 @@ enum imx_pcie_variants { > > #define IMX_PCIE_FLAG_HAS_SERDES BIT(6) > > #define IMX_PCIE_FLAG_SUPPORT_64BIT BIT(7) > > #define IMX_PCIE_FLAG_CPU_ADDR_FIXUP BIT(8) > > +#define IMX_PCIE_FLAG_CUSTOM_PME_TURNOFF BIT(9) > > > > #define imx_check_flag(pci, val) (pci->drvdata->flags & val) > > > > @@ -106,19 +108,18 @@ struct imx_pcie_drvdata { > > int (*init_phy)(struct imx_pcie *pcie); > > int (*enable_ref_clk)(struct imx_pcie *pcie, bool enable); > > int (*core_reset)(struct imx_pcie *pcie, bool assert); > > + const struct dw_pcie_host_ops *ops; > > }; > > > > struct imx_pcie { > > struct dw_pcie *pci; > > struct gpio_desc *reset_gpiod; > > - bool link_is_up; > > struct clk_bulk_data clks[IMX_PCIE_MAX_CLKS]; > > struct regmap *iomuxc_gpr; > > u16 msi_ctrl; > > u32 controller_id; > > struct reset_control *pciephy_reset; > > struct reset_control *apps_reset; > > - struct reset_control *turnoff_reset; > > u32 tx_deemph_gen1; > > u32 tx_deemph_gen2_3p5db; > > u32 tx_deemph_gen2_6db; > > @@ -898,13 +899,11 @@ static int imx_pcie_start_link(struct dw_pcie > *pci) > > dev_info(dev, "Link: Only Gen1 is enabled\n"); > > } > > > > - imx_pcie->link_is_up = true; > > tmp = dw_pcie_readw_dbi(pci, offset + PCI_EXP_LNKSTA); > > dev_info(dev, "Link up, Gen%i\n", tmp & PCI_EXP_LNKSTA_CLS); > > return 0; > > > > err_reset_phy: > > - imx_pcie->link_is_up = false; > > dev_dbg(dev, "PHY DEBUG_R0=0x%08x DEBUG_R1=0x%08x\n", > > dw_pcie_readl_dbi(pci, PCIE_PORT_DEBUG0), > > dw_pcie_readl_dbi(pci, PCIE_PORT_DEBUG1)); @@ -1023,9 > +1022,33 @@ > > static u64 imx_pcie_cpu_addr_fixup(struct dw_pcie *pcie, u64 cpu_addr) > > return cpu_addr - entry->offset; > > } > > > > +/* > > + * Old dwc iATU ctrl2 bit 22 (PCIE_ATU_INHIBIT_PAYLOAD) are reserved. > > +So can't > > + * send out MSG without data by dummy MMIO write. Without > > + * PCIE_ATU_INHIBIT_PAYLOAD, MSGD will be sent out. So have to keep > > +old method > > + * to send PME_TURN_OFF MSG. > > Please reword the comments: > > "In Old DWC implementations, PCIE_ATU_INHIBIT_PAYLOAD bit in iATU Ctrl2 > register is reserved. So the generic DWC implementation of sending the > PME_Turn_Off message using a dummy MMIO write cannot be used." > > > + */ > > +static void imx_pcie_pm_turn_off(struct dw_pcie_rp *pp) > > s/pm/pme > > > +{ > > + struct dw_pcie *pci = to_dw_pcie_from_pp(pp); > > + struct imx_pcie *imx_pcie = to_imx_pcie(pci); > > + > > + regmap_set_bits(imx_pcie->iomuxc_gpr, IOMUXC_GPR12, > IMX6SX_GPR12_PCIE_PM_TURN_OFF); > > + regmap_clear_bits(imx_pcie->iomuxc_gpr, IOMUXC_GPR12, > > +IMX6SX_GPR12_PCIE_PM_TURN_OFF); > > + > > + usleep_range(PCIE_PME_TO_L2_TIMEOUT_US/10, > > +PCIE_PME_TO_L2_TIMEOUT_US); } > > + > > + > > static const struct dw_pcie_host_ops imx_pcie_host_ops = { > > .init = imx_pcie_host_init, > > .deinit = imx_pcie_host_exit, > > + .pme_turn_off = imx_pcie_pm_turn_off, }; > > + > > +static const struct dw_pcie_host_ops imx_pcie_host_dw_pme_ops = { > > + .init = imx_pcie_host_init, > > + .deinit = imx_pcie_host_exit, > > }; > > > > static const struct dw_pcie_ops dw_pcie_ops = { @@ -1146,43 +1169,6 > > @@ static int imx_add_pcie_ep(struct imx_pcie *imx_pcie, > > return 0; > > } > > > > -static void imx_pcie_pm_turnoff(struct imx_pcie *imx_pcie) -{ > > - struct device *dev = imx_pcie->pci->dev; > > - > > - /* Some variants have a turnoff reset in DT */ > > - if (imx_pcie->turnoff_reset) { > > - reset_control_assert(imx_pcie->turnoff_reset); > > - reset_control_deassert(imx_pcie->turnoff_reset); > > - goto pm_turnoff_sleep; > > - } > > What about this part of the code? Don't you need it now? > > > - > > - /* Others poke directly at IOMUXC registers */ > > - switch (imx_pcie->drvdata->variant) { > > - case IMX6SX: > > - case IMX6QP: > > - regmap_update_bits(imx_pcie->iomuxc_gpr, IOMUXC_GPR12, > > - IMX6SX_GPR12_PCIE_PM_TURN_OFF, > > - IMX6SX_GPR12_PCIE_PM_TURN_OFF); > > - regmap_update_bits(imx_pcie->iomuxc_gpr, IOMUXC_GPR12, > > - IMX6SX_GPR12_PCIE_PM_TURN_OFF, 0); > > - break; > > - default: > > - dev_err(dev, "PME_Turn_Off not implemented\n"); > > - return; > > - } > > - > > - /* > > - * Components with an upstream port must respond to > > - * PME_Turn_Off with PME_TO_Ack but we can't check. > > - * > > - * The standard recommends a 1-10ms timeout after which to > > - * proceed anyway as if acks were received. > > - */ > > -pm_turnoff_sleep: > > - usleep_range(1000, 10000); > > -} > > - > > static void imx_pcie_msi_save_restore(struct imx_pcie *imx_pcie, bool > > save) { > > u8 offset; > > @@ -1206,36 +1192,26 @@ static void imx_pcie_msi_save_restore(struct > > imx_pcie *imx_pcie, bool save) static int > > imx_pcie_suspend_noirq(struct device *dev) { > > struct imx_pcie *imx_pcie = dev_get_drvdata(dev); > > - struct dw_pcie_rp *pp = &imx_pcie->pci->pp; > > > > if (!(imx_pcie->drvdata->flags & IMX_PCIE_FLAG_SUPPORTS_SUSPEND)) > > return 0; > > > > imx_pcie_msi_save_restore(imx_pcie, true); > > - imx_pcie_pm_turnoff(imx_pcie); > > - imx_pcie_stop_link(imx_pcie->pci); > > - imx_pcie_host_exit(pp); > > - > > - return 0; > > + return dw_pcie_suspend_noirq(imx_pcie->pci); > > } > > > > static int imx_pcie_resume_noirq(struct device *dev) { > > int ret; > > struct imx_pcie *imx_pcie = dev_get_drvdata(dev); > > - struct dw_pcie_rp *pp = &imx_pcie->pci->pp; > > > > if (!(imx_pcie->drvdata->flags & IMX_PCIE_FLAG_SUPPORTS_SUSPEND)) > > return 0; > > > > - ret = imx_pcie_host_init(pp); > > + ret = dw_pcie_resume_noirq(imx_pcie->pci); > > if (ret) > > return ret; > > imx_pcie_msi_save_restore(imx_pcie, false); > > - dw_pcie_setup_rc(pp); > > - > > - if (imx_pcie->link_is_up) > > - imx_pcie_start_link(imx_pcie->pci); > > So this is also not needed? Why? Please explain in the commit message. These codes would be invoked from dw_pcie_resume_noirq() too. Frank had pointed the invoke procedure. Would add them into commit message later. Best Regards Richard > > - Mani > > -- > மணிவண்ணன் சதாசிவம்