| Message ID | 1476190715-16884-4-git-send-email-kishon@ti.com (mailing list archive) |
|---|---|
| State | New, archived |
| Delegated to: | Bjorn Helgaas |
| Headers | show |
On Tuesday 11 October 2016 06:28 PM, Kishon Vijay Abraham I wrote: > PCIe in AM57x/DRA7x devices is by default > configured to work in GEN2 mode. However there > may be situations when working in GEN1 mode is > desired. One example is limitation i925 (PCIe GEN2 > mode not supported at junction temperatures < 0C). > > Add support to force Root Complex to work in GEN1 > mode if so desired, but don't force GEN1 mode on > any board just yet. > > Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> > Signed-off-by: Sekhar Nori <nsekhar@ti.com> > --- > Documentation/devicetree/bindings/pci/ti-pci.txt | 1 + > drivers/pci/host/pci-dra7xx.c | 27 ++++++++++++++++++++++ > 2 files changed, 28 insertions(+) > > diff --git a/Documentation/devicetree/bindings/pci/ti-pci.txt b/Documentation/devicetree/bindings/pci/ti-pci.txt > index 60e2516..a3d6ca3 100644 > --- a/Documentation/devicetree/bindings/pci/ti-pci.txt > +++ b/Documentation/devicetree/bindings/pci/ti-pci.txt > @@ -25,6 +25,7 @@ PCIe Designware Controller > > Optional Property: > - gpios : Should be added if a gpio line is required to drive PERST# line > + - ti,pcie-is-gen1 : Force the PCIe controller to work in GEN1 (2.5 GT/s). > > Example: > axi { > diff --git a/drivers/pci/host/pci-dra7xx.c b/drivers/pci/host/pci-dra7xx.c > index 4ccba6d..2a669cb 100644 > --- a/drivers/pci/host/pci-dra7xx.c > +++ b/drivers/pci/host/pci-dra7xx.c > @@ -63,11 +63,14 @@ > #define LINK_UP BIT(16) > #define DRA7XX_CPU_TO_BUS_ADDR 0x0FFFFFFF > > +#define EXP_CAP_ID_OFFSET 0x70 > + > struct dra7xx_pcie { > struct pcie_port pp; > void __iomem *base; /* DT ti_conf */ > int phy_count; /* DT phy-names count */ > struct phy **phy; > + bool is_gen1; I think it will be nice to reference the errata number as comments in the code along with references to TI public documentation that explains the said errata. Or may be this can be added to the binding documentation. Somewhere 'git grep' can find. BTW, per Documentation/devicetree/bindings/submitting-patches.txt, the binding documentation should be a separate patch. Thanks, Sekhar -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2016/10/11 20:58, Kishon Vijay Abraham I wrote: > PCIe in AM57x/DRA7x devices is by default > configured to work in GEN2 mode. However there > may be situations when working in GEN1 mode is > desired. One example is limitation i925 (PCIe GEN2 > mode not supported at junction temperatures < 0C). > Just a drive-by comment. Seems there are already much more requirments for host drivers to know the limitation of link speed, so I pushed patches recently to consolidate it for host drivers[0] suggested by Rob. So maybe you could use that instead. :) [0] https://patchwork.kernel.org/patch/9371931/ https://patchwork.kernel.org/patch/9371929/ > Add support to force Root Complex to work in GEN1 > mode if so desired, but don't force GEN1 mode on > any board just yet. > > Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> > Signed-off-by: Sekhar Nori <nsekhar@ti.com> > --- > Documentation/devicetree/bindings/pci/ti-pci.txt | 1 + > drivers/pci/host/pci-dra7xx.c | 27 ++++++++++++++++++++++ > 2 files changed, 28 insertions(+) > > diff --git a/Documentation/devicetree/bindings/pci/ti-pci.txt b/Documentation/devicetree/bindings/pci/ti-pci.txt > index 60e2516..a3d6ca3 100644 > --- a/Documentation/devicetree/bindings/pci/ti-pci.txt > +++ b/Documentation/devicetree/bindings/pci/ti-pci.txt > @@ -25,6 +25,7 @@ PCIe Designware Controller > > Optional Property: > - gpios : Should be added if a gpio line is required to drive PERST# line > + - ti,pcie-is-gen1 : Force the PCIe controller to work in GEN1 (2.5 GT/s). > > Example: > axi { > diff --git a/drivers/pci/host/pci-dra7xx.c b/drivers/pci/host/pci-dra7xx.c > index 4ccba6d..2a669cb 100644 > --- a/drivers/pci/host/pci-dra7xx.c > +++ b/drivers/pci/host/pci-dra7xx.c > @@ -63,11 +63,14 @@ > #define LINK_UP BIT(16) > #define DRA7XX_CPU_TO_BUS_ADDR 0x0FFFFFFF > > +#define EXP_CAP_ID_OFFSET 0x70 > + > struct dra7xx_pcie { > struct pcie_port pp; > void __iomem *base; /* DT ti_conf */ > int phy_count; /* DT phy-names count */ > struct phy **phy; > + bool is_gen1; > }; > > #define to_dra7xx_pcie(x) container_of((x), struct dra7xx_pcie, pp) > @@ -96,12 +99,33 @@ static int dra7xx_pcie_establish_link(struct dra7xx_pcie *dra7xx_pcie) > struct pcie_port *pp = &dra7xx_pcie->pp; > struct device *dev = pp->dev; > u32 reg; > + u32 exp_cap_off = EXP_CAP_ID_OFFSET; > > if (dw_pcie_link_up(pp)) { > dev_err(dev, "link is already up\n"); > return 0; > } > > + if (dra7xx_pcie->is_gen1) { > + dw_pcie_cfg_read(pp->dbi_base + exp_cap_off + PCI_EXP_LNKCAP, > + 4, ®); > + if ((reg & PCI_EXP_LNKCAP_SLS) != PCI_EXP_LNKCAP_SLS_2_5GB) { > + reg &= ~((u32)PCI_EXP_LNKCAP_SLS); > + reg |= PCI_EXP_LNKCAP_SLS_2_5GB; > + dw_pcie_cfg_write(pp->dbi_base + exp_cap_off + > + PCI_EXP_LNKCAP, 4, reg); > + } > + > + dw_pcie_cfg_read(pp->dbi_base + exp_cap_off + PCI_EXP_LNKCTL2, > + 2, ®); > + if ((reg & PCI_EXP_LNKCAP_SLS) != PCI_EXP_LNKCAP_SLS_2_5GB) { > + reg &= ~((u32)PCI_EXP_LNKCAP_SLS); > + reg |= PCI_EXP_LNKCAP_SLS_2_5GB; > + dw_pcie_cfg_write(pp->dbi_base + exp_cap_off + > + PCI_EXP_LNKCTL2, 2, reg); > + } > + } > + > reg = dra7xx_pcie_readl(dra7xx_pcie, PCIECTRL_DRA7XX_CONF_DEVICE_CMD); > reg |= LTSSM_EN; > dra7xx_pcie_writel(dra7xx_pcie, PCIECTRL_DRA7XX_CONF_DEVICE_CMD, reg); > @@ -402,6 +426,9 @@ static int __init dra7xx_pcie_probe(struct platform_device *pdev) > reg &= ~LTSSM_EN; > dra7xx_pcie_writel(dra7xx_pcie, PCIECTRL_DRA7XX_CONF_DEVICE_CMD, reg); > > + if (of_property_read_bool(np, "ti,pcie-is-gen1")) > + dra7xx_pcie->is_gen1 = true; > + > ret = dra7xx_add_pcie_port(dra7xx_pcie, pdev); > if (ret < 0) > goto err_gpio; >
Hi Kishon, On Tue, Oct 11, 2016 at 06:28:34PM +0530, Kishon Vijay Abraham I wrote: > PCIe in AM57x/DRA7x devices is by default > configured to work in GEN2 mode. However there > may be situations when working in GEN1 mode is > desired. One example is limitation i925 (PCIe GEN2 > mode not supported at junction temperatures < 0C). > > Add support to force Root Complex to work in GEN1 > mode if so desired, but don't force GEN1 mode on > any board just yet. > > Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> > Signed-off-by: Sekhar Nori <nsekhar@ti.com> > --- > Documentation/devicetree/bindings/pci/ti-pci.txt | 1 + > drivers/pci/host/pci-dra7xx.c | 27 ++++++++++++++++++++++ > 2 files changed, 28 insertions(+) > > diff --git a/Documentation/devicetree/bindings/pci/ti-pci.txt b/Documentation/devicetree/bindings/pci/ti-pci.txt > index 60e2516..a3d6ca3 100644 > --- a/Documentation/devicetree/bindings/pci/ti-pci.txt > +++ b/Documentation/devicetree/bindings/pci/ti-pci.txt > @@ -25,6 +25,7 @@ PCIe Designware Controller > > Optional Property: > - gpios : Should be added if a gpio line is required to drive PERST# line > + - ti,pcie-is-gen1 : Force the PCIe controller to work in GEN1 (2.5 GT/s). Can we use "max-link-speed" so it's similar to imx6? > Example: > axi { > diff --git a/drivers/pci/host/pci-dra7xx.c b/drivers/pci/host/pci-dra7xx.c > index 4ccba6d..2a669cb 100644 > --- a/drivers/pci/host/pci-dra7xx.c > +++ b/drivers/pci/host/pci-dra7xx.c > @@ -63,11 +63,14 @@ > #define LINK_UP BIT(16) > #define DRA7XX_CPU_TO_BUS_ADDR 0x0FFFFFFF > > +#define EXP_CAP_ID_OFFSET 0x70 > + > struct dra7xx_pcie { > struct pcie_port pp; > void __iomem *base; /* DT ti_conf */ > int phy_count; /* DT phy-names count */ > struct phy **phy; > + bool is_gen1; > }; > > #define to_dra7xx_pcie(x) container_of((x), struct dra7xx_pcie, pp) > @@ -96,12 +99,33 @@ static int dra7xx_pcie_establish_link(struct dra7xx_pcie *dra7xx_pcie) > struct pcie_port *pp = &dra7xx_pcie->pp; > struct device *dev = pp->dev; > u32 reg; > + u32 exp_cap_off = EXP_CAP_ID_OFFSET; > > if (dw_pcie_link_up(pp)) { > dev_err(dev, "link is already up\n"); > return 0; > } > > + if (dra7xx_pcie->is_gen1) { > + dw_pcie_cfg_read(pp->dbi_base + exp_cap_off + PCI_EXP_LNKCAP, > + 4, ®); > + if ((reg & PCI_EXP_LNKCAP_SLS) != PCI_EXP_LNKCAP_SLS_2_5GB) { > + reg &= ~((u32)PCI_EXP_LNKCAP_SLS); > + reg |= PCI_EXP_LNKCAP_SLS_2_5GB; > + dw_pcie_cfg_write(pp->dbi_base + exp_cap_off + > + PCI_EXP_LNKCAP, 4, reg); > + } > + > + dw_pcie_cfg_read(pp->dbi_base + exp_cap_off + PCI_EXP_LNKCTL2, > + 2, ®); > + if ((reg & PCI_EXP_LNKCAP_SLS) != PCI_EXP_LNKCAP_SLS_2_5GB) { > + reg &= ~((u32)PCI_EXP_LNKCAP_SLS); > + reg |= PCI_EXP_LNKCAP_SLS_2_5GB; > + dw_pcie_cfg_write(pp->dbi_base + exp_cap_off + > + PCI_EXP_LNKCTL2, 2, reg); > + } > + } > + > reg = dra7xx_pcie_readl(dra7xx_pcie, PCIECTRL_DRA7XX_CONF_DEVICE_CMD); > reg |= LTSSM_EN; > dra7xx_pcie_writel(dra7xx_pcie, PCIECTRL_DRA7XX_CONF_DEVICE_CMD, reg); > @@ -402,6 +426,9 @@ static int __init dra7xx_pcie_probe(struct platform_device *pdev) > reg &= ~LTSSM_EN; > dra7xx_pcie_writel(dra7xx_pcie, PCIECTRL_DRA7XX_CONF_DEVICE_CMD, reg); > > + if (of_property_read_bool(np, "ti,pcie-is-gen1")) > + dra7xx_pcie->is_gen1 = true; > + > ret = dra7xx_add_pcie_port(dra7xx_pcie, pdev); > if (ret < 0) > goto err_gpio; > -- > 1.7.9.5 > -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi, On Saturday 12 November 2016 02:45 AM, Bjorn Helgaas wrote: > Hi Kishon, > > On Tue, Oct 11, 2016 at 06:28:34PM +0530, Kishon Vijay Abraham I wrote: >> PCIe in AM57x/DRA7x devices is by default >> configured to work in GEN2 mode. However there >> may be situations when working in GEN1 mode is >> desired. One example is limitation i925 (PCIe GEN2 >> mode not supported at junction temperatures < 0C). >> >> Add support to force Root Complex to work in GEN1 >> mode if so desired, but don't force GEN1 mode on >> any board just yet. >> >> Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> >> Signed-off-by: Sekhar Nori <nsekhar@ti.com> >> --- >> Documentation/devicetree/bindings/pci/ti-pci.txt | 1 + >> drivers/pci/host/pci-dra7xx.c | 27 ++++++++++++++++++++++ >> 2 files changed, 28 insertions(+) >> >> diff --git a/Documentation/devicetree/bindings/pci/ti-pci.txt b/Documentation/devicetree/bindings/pci/ti-pci.txt >> index 60e2516..a3d6ca3 100644 >> --- a/Documentation/devicetree/bindings/pci/ti-pci.txt >> +++ b/Documentation/devicetree/bindings/pci/ti-pci.txt >> @@ -25,6 +25,7 @@ PCIe Designware Controller >> >> Optional Property: >> - gpios : Should be added if a gpio line is required to drive PERST# line >> + - ti,pcie-is-gen1 : Force the PCIe controller to work in GEN1 (2.5 GT/s). > > Can we use "max-link-speed" so it's similar to imx6? yeah, maybe we should make it a generic PCI property? Thanks Kishon -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
[+cc Shawn] On Sat, Nov 12, 2016 at 12:40:10PM +0530, Kishon Vijay Abraham I wrote: > Hi, > > On Saturday 12 November 2016 02:45 AM, Bjorn Helgaas wrote: > > Hi Kishon, > > > > On Tue, Oct 11, 2016 at 06:28:34PM +0530, Kishon Vijay Abraham I wrote: > >> PCIe in AM57x/DRA7x devices is by default > >> configured to work in GEN2 mode. However there > >> may be situations when working in GEN1 mode is > >> desired. One example is limitation i925 (PCIe GEN2 > >> mode not supported at junction temperatures < 0C). > >> > >> Add support to force Root Complex to work in GEN1 > >> mode if so desired, but don't force GEN1 mode on > >> any board just yet. > >> > >> Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> > >> Signed-off-by: Sekhar Nori <nsekhar@ti.com> > >> --- > >> Documentation/devicetree/bindings/pci/ti-pci.txt | 1 + > >> drivers/pci/host/pci-dra7xx.c | 27 ++++++++++++++++++++++ > >> 2 files changed, 28 insertions(+) > >> > >> diff --git a/Documentation/devicetree/bindings/pci/ti-pci.txt b/Documentation/devicetree/bindings/pci/ti-pci.txt > >> index 60e2516..a3d6ca3 100644 > >> --- a/Documentation/devicetree/bindings/pci/ti-pci.txt > >> +++ b/Documentation/devicetree/bindings/pci/ti-pci.txt > >> @@ -25,6 +25,7 @@ PCIe Designware Controller > >> > >> Optional Property: > >> - gpios : Should be added if a gpio line is required to drive PERST# line > >> + - ti,pcie-is-gen1 : Force the PCIe controller to work in GEN1 (2.5 GT/s). > > > > Can we use "max-link-speed" so it's similar to imx6? > > yeah, maybe we should make it a generic PCI property? I forgot that Shawn has already done this! I had already merged those patches on pci/host-rockchip, but I moved them to pci/host since they're not Rockchip-specific. Can you take a look at that and see if you can do what you need based on that pci/host branch? Bjorn -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Bjorn, On Tuesday 15 November 2016 02:54 AM, Bjorn Helgaas wrote: > [+cc Shawn] > > On Sat, Nov 12, 2016 at 12:40:10PM +0530, Kishon Vijay Abraham I wrote: >> Hi, >> >> On Saturday 12 November 2016 02:45 AM, Bjorn Helgaas wrote: >>> Hi Kishon, >>> >>> On Tue, Oct 11, 2016 at 06:28:34PM +0530, Kishon Vijay Abraham I wrote: >>>> PCIe in AM57x/DRA7x devices is by default >>>> configured to work in GEN2 mode. However there >>>> may be situations when working in GEN1 mode is >>>> desired. One example is limitation i925 (PCIe GEN2 >>>> mode not supported at junction temperatures < 0C). >>>> >>>> Add support to force Root Complex to work in GEN1 >>>> mode if so desired, but don't force GEN1 mode on >>>> any board just yet. >>>> >>>> Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> >>>> Signed-off-by: Sekhar Nori <nsekhar@ti.com> >>>> --- >>>> Documentation/devicetree/bindings/pci/ti-pci.txt | 1 + >>>> drivers/pci/host/pci-dra7xx.c | 27 ++++++++++++++++++++++ >>>> 2 files changed, 28 insertions(+) >>>> >>>> diff --git a/Documentation/devicetree/bindings/pci/ti-pci.txt b/Documentation/devicetree/bindings/pci/ti-pci.txt >>>> index 60e2516..a3d6ca3 100644 >>>> --- a/Documentation/devicetree/bindings/pci/ti-pci.txt >>>> +++ b/Documentation/devicetree/bindings/pci/ti-pci.txt >>>> @@ -25,6 +25,7 @@ PCIe Designware Controller >>>> >>>> Optional Property: >>>> - gpios : Should be added if a gpio line is required to drive PERST# line >>>> + - ti,pcie-is-gen1 : Force the PCIe controller to work in GEN1 (2.5 GT/s). >>> >>> Can we use "max-link-speed" so it's similar to imx6? >> >> yeah, maybe we should make it a generic PCI property? > > I forgot that Shawn has already done this! I had already merged those > patches on pci/host-rockchip, but I moved them to pci/host since > they're not Rockchip-specific. Can you take a look at that and see if > you can do what you need based on that pci/host branch? Sure, I'll look at that on Monday when I'm back from vacation. Cheers Kishon -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/Documentation/devicetree/bindings/pci/ti-pci.txt b/Documentation/devicetree/bindings/pci/ti-pci.txt index 60e2516..a3d6ca3 100644 --- a/Documentation/devicetree/bindings/pci/ti-pci.txt +++ b/Documentation/devicetree/bindings/pci/ti-pci.txt @@ -25,6 +25,7 @@ PCIe Designware Controller Optional Property: - gpios : Should be added if a gpio line is required to drive PERST# line + - ti,pcie-is-gen1 : Force the PCIe controller to work in GEN1 (2.5 GT/s). Example: axi { diff --git a/drivers/pci/host/pci-dra7xx.c b/drivers/pci/host/pci-dra7xx.c index 4ccba6d..2a669cb 100644 --- a/drivers/pci/host/pci-dra7xx.c +++ b/drivers/pci/host/pci-dra7xx.c @@ -63,11 +63,14 @@ #define LINK_UP BIT(16) #define DRA7XX_CPU_TO_BUS_ADDR 0x0FFFFFFF +#define EXP_CAP_ID_OFFSET 0x70 + struct dra7xx_pcie { struct pcie_port pp; void __iomem *base; /* DT ti_conf */ int phy_count; /* DT phy-names count */ struct phy **phy; + bool is_gen1; }; #define to_dra7xx_pcie(x) container_of((x), struct dra7xx_pcie, pp) @@ -96,12 +99,33 @@ static int dra7xx_pcie_establish_link(struct dra7xx_pcie *dra7xx_pcie) struct pcie_port *pp = &dra7xx_pcie->pp; struct device *dev = pp->dev; u32 reg; + u32 exp_cap_off = EXP_CAP_ID_OFFSET; if (dw_pcie_link_up(pp)) { dev_err(dev, "link is already up\n"); return 0; } + if (dra7xx_pcie->is_gen1) { + dw_pcie_cfg_read(pp->dbi_base + exp_cap_off + PCI_EXP_LNKCAP, + 4, ®); + if ((reg & PCI_EXP_LNKCAP_SLS) != PCI_EXP_LNKCAP_SLS_2_5GB) { + reg &= ~((u32)PCI_EXP_LNKCAP_SLS); + reg |= PCI_EXP_LNKCAP_SLS_2_5GB; + dw_pcie_cfg_write(pp->dbi_base + exp_cap_off + + PCI_EXP_LNKCAP, 4, reg); + } + + dw_pcie_cfg_read(pp->dbi_base + exp_cap_off + PCI_EXP_LNKCTL2, + 2, ®); + if ((reg & PCI_EXP_LNKCAP_SLS) != PCI_EXP_LNKCAP_SLS_2_5GB) { + reg &= ~((u32)PCI_EXP_LNKCAP_SLS); + reg |= PCI_EXP_LNKCAP_SLS_2_5GB; + dw_pcie_cfg_write(pp->dbi_base + exp_cap_off + + PCI_EXP_LNKCTL2, 2, reg); + } + } + reg = dra7xx_pcie_readl(dra7xx_pcie, PCIECTRL_DRA7XX_CONF_DEVICE_CMD); reg |= LTSSM_EN; dra7xx_pcie_writel(dra7xx_pcie, PCIECTRL_DRA7XX_CONF_DEVICE_CMD, reg); @@ -402,6 +426,9 @@ static int __init dra7xx_pcie_probe(struct platform_device *pdev) reg &= ~LTSSM_EN; dra7xx_pcie_writel(dra7xx_pcie, PCIECTRL_DRA7XX_CONF_DEVICE_CMD, reg); + if (of_property_read_bool(np, "ti,pcie-is-gen1")) + dra7xx_pcie->is_gen1 = true; + ret = dra7xx_add_pcie_port(dra7xx_pcie, pdev); if (ret < 0) goto err_gpio;