Message ID | 20231106-refclk_always_on-v1-0-17a7fd8b532b@quicinc.com |
---|---|
Headers | show |
Series | phy: qcom-qmp-pcie: Add support to keep refclk always on | expand |
On Mon, 6 Nov 2023 at 13:53, Krishna chaitanya chundru <quic_krichai@quicinc.com> wrote: > > This series adds support to provide refclk to endpoint even in low > power states. > > Due to some platform specific issues with CLKREQ signal, it is not being > propagated to the host and as host doesn't know the clkreq signal host is > not sending refclk. Due to this endpoint is seeing linkdown and going > to bad state. Is this a board issue or an SoC issue? In the latter case this should go to the PHY configuration structure instead of being specified in the DT. > To avoid those ref clk should be provided always to the endpoint. The > issue is coming only when ep intiates the L1.1 or L1.2 exit and clkreq > is not being propagated properly to the host. > > Signed-off-by: Krishna chaitanya chundru <quic_krichai@quicinc.com> > --- > Krishna chaitanya chundru (2): > dt-bindings: phy: qcom,qmp: Add PCIe qcom,refclk-always-on property > phy: qcom-qmp-pcie: Add support for keeping refclk always on > > .../bindings/phy/qcom,sc8280xp-qmp-pcie-phy.yaml | 5 +++++ > drivers/phy/qualcomm/phy-qcom-qmp-pcie.c | 21 +++++++++++++++++---- > 2 files changed, 22 insertions(+), 4 deletions(-) > --- > base-commit: 71e68e182e382e951d6248bccc3c960dcec5a718 > change-id: 20231106-refclk_always_on-9beae8297cb8 > > Best regards, > -- > Krishna chaitanya chundru <quic_krichai@quicinc.com> >
On 11/6/2023 5:28 PM, Dmitry Baryshkov wrote: > On Mon, 6 Nov 2023 at 13:53, Krishna chaitanya chundru > <quic_krichai@quicinc.com> wrote: >> This series adds support to provide refclk to endpoint even in low >> power states. >> >> Due to some platform specific issues with CLKREQ signal, it is not being >> propagated to the host and as host doesn't know the clkreq signal host is >> not sending refclk. Due to this endpoint is seeing linkdown and going >> to bad state. > Is this a board issue or an SoC issue? In the latter case this should > go to the PHY configuration structure instead of being specified in > the DT. Hi Dmitry, This is not SOC level issue it is a board issue. - Krishna Chaitanya. >> To avoid those ref clk should be provided always to the endpoint. The >> issue is coming only when ep intiates the L1.1 or L1.2 exit and clkreq >> is not being propagated properly to the host. >> >> Signed-off-by: Krishna chaitanya chundru <quic_krichai@quicinc.com> >> --- >> Krishna chaitanya chundru (2): >> dt-bindings: phy: qcom,qmp: Add PCIe qcom,refclk-always-on property >> phy: qcom-qmp-pcie: Add support for keeping refclk always on >> >> .../bindings/phy/qcom,sc8280xp-qmp-pcie-phy.yaml | 5 +++++ >> drivers/phy/qualcomm/phy-qcom-qmp-pcie.c | 21 +++++++++++++++++---- >> 2 files changed, 22 insertions(+), 4 deletions(-) >> --- >> base-commit: 71e68e182e382e951d6248bccc3c960dcec5a718 >> change-id: 20231106-refclk_always_on-9beae8297cb8 >> >> Best regards, >> -- >> Krishna chaitanya chundru <quic_krichai@quicinc.com> >> >
On Mon, 6 Nov 2023 at 13:53, Krishna chaitanya chundru <quic_krichai@quicinc.com> wrote: > > In PCIe low power states like L1.1 or L1.2 the phy will stop > supplying refclk to endpoint. If endpoint asserts clkreq to bring > back link L0, then RC needs to provide refclk to endpoint. > > If there is some issues in platform with clkreq signal propagation > to host and due to that host will not send refclk which results PCIe link > down. For those platforms phy needs to provide refclk even in low power > states. > > Add a flag which indicates refclk is always supplied to endpoint. > > Signed-off-by: Krishna chaitanya chundru <quic_krichai@quicinc.com> > --- > drivers/phy/qualcomm/phy-qcom-qmp-pcie.c | 21 +++++++++++++++++---- > 1 file changed, 17 insertions(+), 4 deletions(-) > > diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c b/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c > index a63ca7424974..d7e377a7d96e 100644 > --- a/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c > +++ b/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c > @@ -43,6 +43,8 @@ > /* QPHY_PCS_STATUS bit */ > #define PHYSTATUS BIT(6) > #define PHYSTATUS_4_20 BIT(7) > +/* PCS_PCIE_ENDPOINT_REFCLK_CNTRL */ > +#define EPCLK_ALWAYS_ON_EN BIT(6) > > #define PHY_INIT_COMPLETE_TIMEOUT 10000 > > @@ -77,6 +79,7 @@ enum qphy_reg_layout { > QPHY_START_CTRL, > QPHY_PCS_STATUS, > QPHY_PCS_POWER_DOWN_CONTROL, > + QPHY_PCS_ENDPOINT_REFCLK_CNTRL, > /* Keep last to ensure regs_layout arrays are properly initialized */ > QPHY_LAYOUT_SIZE > }; > @@ -103,10 +106,11 @@ static const unsigned int sdm845_qhp_pciephy_regs_layout[QPHY_LAYOUT_SIZE] = { > }; > > static const unsigned int pciephy_v4_regs_layout[QPHY_LAYOUT_SIZE] = { > - [QPHY_SW_RESET] = QPHY_V4_PCS_SW_RESET, > - [QPHY_START_CTRL] = QPHY_V4_PCS_START_CONTROL, > - [QPHY_PCS_STATUS] = QPHY_V4_PCS_PCS_STATUS1, > - [QPHY_PCS_POWER_DOWN_CONTROL] = QPHY_V4_PCS_POWER_DOWN_CONTROL, > + [QPHY_SW_RESET] = QPHY_V4_PCS_SW_RESET, > + [QPHY_START_CTRL] = QPHY_V4_PCS_START_CONTROL, > + [QPHY_PCS_STATUS] = QPHY_V4_PCS_PCS_STATUS1, > + [QPHY_PCS_POWER_DOWN_CONTROL] = QPHY_V4_PCS_POWER_DOWN_CONTROL, No unnecessary whitespace changes, please. > + [QPHY_PCS_ENDPOINT_REFCLK_CNTRL] = QPHY_V4_PCS_PCIE_ENDPOINT_REFCLK_CNTRL, Any other platform having this register? > }; > > static const unsigned int pciephy_v5_regs_layout[QPHY_LAYOUT_SIZE] = { > @@ -2244,6 +2248,8 @@ struct qmp_pcie { > struct phy *phy; > int mode; > > + bool refclk_always_on; > + > struct clk_fixed_rate pipe_clk_fixed; > }; > > @@ -3159,6 +3165,10 @@ static void qmp_pcie_init_registers(struct qmp_pcie *qmp, const struct qmp_phy_c > qmp_pcie_configure(pcs, tbls->pcs, tbls->pcs_num); > qmp_pcie_configure(pcs_misc, tbls->pcs_misc, tbls->pcs_misc_num); > > + if (qmp->refclk_always_on && cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL]) > + qphy_setbits(pcs_misc, cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL], > + EPCLK_ALWAYS_ON_EN); > + > if (cfg->lanes >= 4 && qmp->tcsr_4ln_config) { > qmp_pcie_configure(serdes, cfg->serdes_4ln_tbl, cfg->serdes_4ln_num); > qmp_pcie_init_port_b(qmp, tbls); > @@ -3681,6 +3691,9 @@ static int qmp_pcie_probe(struct platform_device *pdev) > if (ret) > goto err_node_put; > > + qmp->refclk_always_on = of_property_read_bool(dev->of_node, > + "qcom,refclk-always-on"); Error out if !cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL]). Otherwise your DT value can be silently ignored. > + > ret = phy_pipe_clk_register(qmp, np); > if (ret) > goto err_node_put; > > -- > 2.42.0 >
On 11/6/2023 5:43 PM, Dmitry Baryshkov wrote: > On Mon, 6 Nov 2023 at 13:53, Krishna chaitanya chundru > <quic_krichai@quicinc.com> wrote: >> In PCIe low power states like L1.1 or L1.2 the phy will stop >> supplying refclk to endpoint. If endpoint asserts clkreq to bring >> back link L0, then RC needs to provide refclk to endpoint. >> >> If there is some issues in platform with clkreq signal propagation >> to host and due to that host will not send refclk which results PCIe link >> down. For those platforms phy needs to provide refclk even in low power >> states. >> >> Add a flag which indicates refclk is always supplied to endpoint. >> >> Signed-off-by: Krishna chaitanya chundru <quic_krichai@quicinc.com> >> --- >> drivers/phy/qualcomm/phy-qcom-qmp-pcie.c | 21 +++++++++++++++++---- >> 1 file changed, 17 insertions(+), 4 deletions(-) >> >> diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c b/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c >> index a63ca7424974..d7e377a7d96e 100644 >> --- a/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c >> +++ b/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c >> @@ -43,6 +43,8 @@ >> /* QPHY_PCS_STATUS bit */ >> #define PHYSTATUS BIT(6) >> #define PHYSTATUS_4_20 BIT(7) >> +/* PCS_PCIE_ENDPOINT_REFCLK_CNTRL */ >> +#define EPCLK_ALWAYS_ON_EN BIT(6) > >> #define PHY_INIT_COMPLETE_TIMEOUT 10000 >> >> @@ -77,6 +79,7 @@ enum qphy_reg_layout { >> QPHY_START_CTRL, >> QPHY_PCS_STATUS, >> QPHY_PCS_POWER_DOWN_CONTROL, >> + QPHY_PCS_ENDPOINT_REFCLK_CNTRL, >> /* Keep last to ensure regs_layout arrays are properly initialized */ >> QPHY_LAYOUT_SIZE >> }; >> @@ -103,10 +106,11 @@ static const unsigned int sdm845_qhp_pciephy_regs_layout[QPHY_LAYOUT_SIZE] = { >> }; >> >> static const unsigned int pciephy_v4_regs_layout[QPHY_LAYOUT_SIZE] = { >> - [QPHY_SW_RESET] = QPHY_V4_PCS_SW_RESET, >> - [QPHY_START_CTRL] = QPHY_V4_PCS_START_CONTROL, >> - [QPHY_PCS_STATUS] = QPHY_V4_PCS_PCS_STATUS1, >> - [QPHY_PCS_POWER_DOWN_CONTROL] = QPHY_V4_PCS_POWER_DOWN_CONTROL, >> + [QPHY_SW_RESET] = QPHY_V4_PCS_SW_RESET, >> + [QPHY_START_CTRL] = QPHY_V4_PCS_START_CONTROL, >> + [QPHY_PCS_STATUS] = QPHY_V4_PCS_PCS_STATUS1, >> + [QPHY_PCS_POWER_DOWN_CONTROL] = QPHY_V4_PCS_POWER_DOWN_CONTROL, > No unnecessary whitespace changes, please. I will remove above white space and keep below change as it is as it is throwing error as white space required there > >> + [QPHY_PCS_ENDPOINT_REFCLK_CNTRL] = QPHY_V4_PCS_PCIE_ENDPOINT_REFCLK_CNTRL, > Any other platform having this register? we have this register for other platforms also I will add that register in which ever versions it exits in next patch > >> }; >> >> static const unsigned int pciephy_v5_regs_layout[QPHY_LAYOUT_SIZE] = { >> @@ -2244,6 +2248,8 @@ struct qmp_pcie { >> struct phy *phy; >> int mode; >> >> + bool refclk_always_on; >> + >> struct clk_fixed_rate pipe_clk_fixed; >> }; >> >> @@ -3159,6 +3165,10 @@ static void qmp_pcie_init_registers(struct qmp_pcie *qmp, const struct qmp_phy_c >> qmp_pcie_configure(pcs, tbls->pcs, tbls->pcs_num); >> qmp_pcie_configure(pcs_misc, tbls->pcs_misc, tbls->pcs_misc_num); >> >> + if (qmp->refclk_always_on && cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL]) >> + qphy_setbits(pcs_misc, cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL], >> + EPCLK_ALWAYS_ON_EN); >> + >> if (cfg->lanes >= 4 && qmp->tcsr_4ln_config) { >> qmp_pcie_configure(serdes, cfg->serdes_4ln_tbl, cfg->serdes_4ln_num); >> qmp_pcie_init_port_b(qmp, tbls); >> @@ -3681,6 +3691,9 @@ static int qmp_pcie_probe(struct platform_device *pdev) >> if (ret) >> goto err_node_put; >> >> + qmp->refclk_always_on = of_property_read_bool(dev->of_node, >> + "qcom,refclk-always-on"); > Error out if !cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL]). > Otherwise your DT value can be silently ignored. sure I will add this in next patch series. - Krishna Chaitanya. >> + >> ret = phy_pipe_clk_register(qmp, np); >> if (ret) >> goto err_node_put; >> >> -- >> 2.42.0 >> >
On Mon, Nov 06, 2023 at 05:22:35PM +0530, Krishna chaitanya chundru wrote: > In PCIe low power states like L1.1 or L1.2 the phy will stop > supplying refclk to endpoint. If endpoint asserts clkreq to bring > back link L0, then RC needs to provide refclk to endpoint. > > If there is some issues in platform with clkreq signal propagation > to host and due to that host will not send refclk which results PCIe link > down. For those platforms phy needs to provide refclk even in low power Double <space> ------------^^ > states. > > Add a flag which indicates refclk is always supplied to endpoint. > The patch itself look good, the problem description looks good, but if you have an indication that the refclk "is always supplied to the endpoint", then you don't have a problem with refclk and I don't think you need this patch. Something to the tune of "keep refclk always supplied to endpoint" seems to more appropriately describe what this flag is doing. > Signed-off-by: Krishna chaitanya chundru <quic_krichai@quicinc.com> > --- > drivers/phy/qualcomm/phy-qcom-qmp-pcie.c | 21 +++++++++++++++++---- > 1 file changed, 17 insertions(+), 4 deletions(-) > > diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c b/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c > index a63ca7424974..d7e377a7d96e 100644 > --- a/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c > +++ b/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c > @@ -43,6 +43,8 @@ > /* QPHY_PCS_STATUS bit */ > #define PHYSTATUS BIT(6) > #define PHYSTATUS_4_20 BIT(7) > +/* PCS_PCIE_ENDPOINT_REFCLK_CNTRL */ > +#define EPCLK_ALWAYS_ON_EN BIT(6) > > #define PHY_INIT_COMPLETE_TIMEOUT 10000 > > @@ -77,6 +79,7 @@ enum qphy_reg_layout { > QPHY_START_CTRL, > QPHY_PCS_STATUS, > QPHY_PCS_POWER_DOWN_CONTROL, > + QPHY_PCS_ENDPOINT_REFCLK_CNTRL, > /* Keep last to ensure regs_layout arrays are properly initialized */ > QPHY_LAYOUT_SIZE > }; > @@ -103,10 +106,11 @@ static const unsigned int sdm845_qhp_pciephy_regs_layout[QPHY_LAYOUT_SIZE] = { > }; > > static const unsigned int pciephy_v4_regs_layout[QPHY_LAYOUT_SIZE] = { > - [QPHY_SW_RESET] = QPHY_V4_PCS_SW_RESET, > - [QPHY_START_CTRL] = QPHY_V4_PCS_START_CONTROL, > - [QPHY_PCS_STATUS] = QPHY_V4_PCS_PCS_STATUS1, > - [QPHY_PCS_POWER_DOWN_CONTROL] = QPHY_V4_PCS_POWER_DOWN_CONTROL, > + [QPHY_SW_RESET] = QPHY_V4_PCS_SW_RESET, > + [QPHY_START_CTRL] = QPHY_V4_PCS_START_CONTROL, > + [QPHY_PCS_STATUS] = QPHY_V4_PCS_PCS_STATUS1, > + [QPHY_PCS_POWER_DOWN_CONTROL] = QPHY_V4_PCS_POWER_DOWN_CONTROL, > + [QPHY_PCS_ENDPOINT_REFCLK_CNTRL] = QPHY_V4_PCS_PCIE_ENDPOINT_REFCLK_CNTRL, > }; > > static const unsigned int pciephy_v5_regs_layout[QPHY_LAYOUT_SIZE] = { > @@ -2244,6 +2248,8 @@ struct qmp_pcie { > struct phy *phy; > int mode; > > + bool refclk_always_on; > + > struct clk_fixed_rate pipe_clk_fixed; > }; > > @@ -3159,6 +3165,10 @@ static void qmp_pcie_init_registers(struct qmp_pcie *qmp, const struct qmp_phy_c > qmp_pcie_configure(pcs, tbls->pcs, tbls->pcs_num); > qmp_pcie_configure(pcs_misc, tbls->pcs_misc, tbls->pcs_misc_num); > > + if (qmp->refclk_always_on && cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL]) > + qphy_setbits(pcs_misc, cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL], > + EPCLK_ALWAYS_ON_EN); > + > if (cfg->lanes >= 4 && qmp->tcsr_4ln_config) { > qmp_pcie_configure(serdes, cfg->serdes_4ln_tbl, cfg->serdes_4ln_num); > qmp_pcie_init_port_b(qmp, tbls); > @@ -3681,6 +3691,9 @@ static int qmp_pcie_probe(struct platform_device *pdev) > if (ret) > goto err_node_put; > > + qmp->refclk_always_on = of_property_read_bool(dev->of_node, > + "qcom,refclk-always-on"); Leave this line unwrapped, for readability. Regards, Bjorn > + > ret = phy_pipe_clk_register(qmp, np); > if (ret) > goto err_node_put; > > -- > 2.42.0 >
On 11/6/2023 9:47 PM, Bjorn Andersson wrote: > On Mon, Nov 06, 2023 at 05:22:35PM +0530, Krishna chaitanya chundru wrote: >> In PCIe low power states like L1.1 or L1.2 the phy will stop >> supplying refclk to endpoint. If endpoint asserts clkreq to bring >> back link L0, then RC needs to provide refclk to endpoint. >> >> If there is some issues in platform with clkreq signal propagation >> to host and due to that host will not send refclk which results PCIe link >> down. For those platforms phy needs to provide refclk even in low power > Double <space> ------------^^ ACK >> states. >> >> Add a flag which indicates refclk is always supplied to endpoint. >> > The patch itself look good, the problem description looks good, but if > you have an indication that the refclk "is always supplied to the > endpoint", then you don't have a problem with refclk and I don't think > you need this patch. > > Something to the tune of "keep refclk always supplied to endpoint" seems > to more appropriately describe what this flag is doing. Sure I will change it in my next patch. >> Signed-off-by: Krishna chaitanya chundru <quic_krichai@quicinc.com> >> --- >> drivers/phy/qualcomm/phy-qcom-qmp-pcie.c | 21 +++++++++++++++++---- >> 1 file changed, 17 insertions(+), 4 deletions(-) >> >> diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c b/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c >> index a63ca7424974..d7e377a7d96e 100644 >> --- a/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c >> +++ b/drivers/phy/qualcomm/phy-qcom-qmp-pcie.c >> @@ -43,6 +43,8 @@ >> /* QPHY_PCS_STATUS bit */ >> #define PHYSTATUS BIT(6) >> #define PHYSTATUS_4_20 BIT(7) >> +/* PCS_PCIE_ENDPOINT_REFCLK_CNTRL */ >> +#define EPCLK_ALWAYS_ON_EN BIT(6) >> >> #define PHY_INIT_COMPLETE_TIMEOUT 10000 >> >> @@ -77,6 +79,7 @@ enum qphy_reg_layout { >> QPHY_START_CTRL, >> QPHY_PCS_STATUS, >> QPHY_PCS_POWER_DOWN_CONTROL, >> + QPHY_PCS_ENDPOINT_REFCLK_CNTRL, >> /* Keep last to ensure regs_layout arrays are properly initialized */ >> QPHY_LAYOUT_SIZE >> }; >> @@ -103,10 +106,11 @@ static const unsigned int sdm845_qhp_pciephy_regs_layout[QPHY_LAYOUT_SIZE] = { >> }; >> >> static const unsigned int pciephy_v4_regs_layout[QPHY_LAYOUT_SIZE] = { >> - [QPHY_SW_RESET] = QPHY_V4_PCS_SW_RESET, >> - [QPHY_START_CTRL] = QPHY_V4_PCS_START_CONTROL, >> - [QPHY_PCS_STATUS] = QPHY_V4_PCS_PCS_STATUS1, >> - [QPHY_PCS_POWER_DOWN_CONTROL] = QPHY_V4_PCS_POWER_DOWN_CONTROL, >> + [QPHY_SW_RESET] = QPHY_V4_PCS_SW_RESET, >> + [QPHY_START_CTRL] = QPHY_V4_PCS_START_CONTROL, >> + [QPHY_PCS_STATUS] = QPHY_V4_PCS_PCS_STATUS1, >> + [QPHY_PCS_POWER_DOWN_CONTROL] = QPHY_V4_PCS_POWER_DOWN_CONTROL, >> + [QPHY_PCS_ENDPOINT_REFCLK_CNTRL] = QPHY_V4_PCS_PCIE_ENDPOINT_REFCLK_CNTRL, >> }; >> >> static const unsigned int pciephy_v5_regs_layout[QPHY_LAYOUT_SIZE] = { >> @@ -2244,6 +2248,8 @@ struct qmp_pcie { >> struct phy *phy; >> int mode; >> >> + bool refclk_always_on; >> + >> struct clk_fixed_rate pipe_clk_fixed; >> }; >> >> @@ -3159,6 +3165,10 @@ static void qmp_pcie_init_registers(struct qmp_pcie *qmp, const struct qmp_phy_c >> qmp_pcie_configure(pcs, tbls->pcs, tbls->pcs_num); >> qmp_pcie_configure(pcs_misc, tbls->pcs_misc, tbls->pcs_misc_num); >> >> + if (qmp->refclk_always_on && cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL]) >> + qphy_setbits(pcs_misc, cfg->regs[QPHY_PCS_ENDPOINT_REFCLK_CNTRL], >> + EPCLK_ALWAYS_ON_EN); >> + >> if (cfg->lanes >= 4 && qmp->tcsr_4ln_config) { >> qmp_pcie_configure(serdes, cfg->serdes_4ln_tbl, cfg->serdes_4ln_num); >> qmp_pcie_init_port_b(qmp, tbls); >> @@ -3681,6 +3691,9 @@ static int qmp_pcie_probe(struct platform_device *pdev) >> if (ret) >> goto err_node_put; >> >> + qmp->refclk_always_on = of_property_read_bool(dev->of_node, >> + "qcom,refclk-always-on"); > Leave this line unwrapped, for readability. > > Regards, > Bjorn sure I will update this in next patch. - Krishna Chaitanya. >> + >> ret = phy_pipe_clk_register(qmp, np); >> if (ret) >> goto err_node_put; >> >> -- >> 2.42.0 >>
This series adds support to provide refclk to endpoint even in low power states. Due to some platform specific issues with CLKREQ signal, it is not being propagated to the host and as host doesn't know the clkreq signal host is not sending refclk. Due to this endpoint is seeing linkdown and going to bad state. To avoid those ref clk should be provided always to the endpoint. The issue is coming only when ep intiates the L1.1 or L1.2 exit and clkreq is not being propagated properly to the host. Signed-off-by: Krishna chaitanya chundru <quic_krichai@quicinc.com> --- Krishna chaitanya chundru (2): dt-bindings: phy: qcom,qmp: Add PCIe qcom,refclk-always-on property phy: qcom-qmp-pcie: Add support for keeping refclk always on .../bindings/phy/qcom,sc8280xp-qmp-pcie-phy.yaml | 5 +++++ drivers/phy/qualcomm/phy-qcom-qmp-pcie.c | 21 +++++++++++++++++---- 2 files changed, 22 insertions(+), 4 deletions(-) --- base-commit: 71e68e182e382e951d6248bccc3c960dcec5a718 change-id: 20231106-refclk_always_on-9beae8297cb8 Best regards,