Hi,

This series aims to fix the usage of dw_pcie_link_up() API to check for Link up
during system suspend. The motivation for this series comes from recent
discussions [1] [2], where developers wanted to skip PME_Turn_Off broadcast in
dw_pcie_suspend_noirq() API when devices are not attached to the bus. They ended
up using dw_pcie_link_up() to check for the device presence due to the bad
example in the pcie-qcom driver which does the same. The usage of
dw_pcie_link_up() API here would be racy as the link can go down at any time
after the check.

So to properly check for the device presence, this series introduces an API, 
pci_root_ports_have_device(), that accepts the Root bus pointer and checks for
the presence of a device under any of the Root Ports. This API is used to
replace the dw_pcie_link_up() check in suspend path of pcie-qcom driver and also
used to skip the PME_Turn_Off brodcast message in dwc_pcie_suspend_noirq() API
and to skip waiting for the link up in dwc_pcie_resume_noirq() API.

Testing
=======

This series is tested on Qualcomm Lenovo Thinkpad T14s and observed no
functional change during the system suspend path.

- Mani

[1] https://lore.kernel.org/linux-pci/CAKfTPtCtHquxtK=Zx2WSNm15MmqeUXO8XXi8FkS4EpuP80PP7g@mail.gmail.com/
[2] https://lore.kernel.org/linux-pci/27516921.17f2.1997bb2a498.Coremail.zhangsenchuan@eswincomputing.com/

Changes in v2:

* Skipped waiting for link up in dwc_pcie_resume_noirq() if there was no device
  before suspend.
* Fixed the kdoc for pci_root_ports_have_device()

Manivannan Sadhasivam (3):
  PCI: host-common: Add an API to check for any device under the Root
    Ports
  PCI: qcom: Check for the presence of a device instead of Link up
    during suspend
  PCI: dwc: Check for the device presence during suspend and resume

 .../pci/controller/dwc/pcie-designware-host.c | 13 ++++++++++++
 drivers/pci/controller/dwc/pcie-qcom.c        |  6 ++++--
 drivers/pci/controller/pci-host-common.c      | 21 +++++++++++++++++++
 drivers/pci/controller/pci-host-common.h      |  2 ++
 4 files changed, 40 insertions(+), 2 deletions(-)

-- 
2.48.1

On Fri, 7 Nov 2025 at 05:43, Manivannan Sadhasivam
<manivannan.sadhasivam@oss.qualcomm.com> wrote:
>
> Hi,
>
> This series aims to fix the usage of dw_pcie_link_up() API to check for Link up
> during system suspend. The motivation for this series comes from recent
> discussions [1] [2], where developers wanted to skip PME_Turn_Off broadcast in
> dw_pcie_suspend_noirq() API when devices are not attached to the bus. They ended
> up using dw_pcie_link_up() to check for the device presence due to the bad
> example in the pcie-qcom driver which does the same. The usage of
> dw_pcie_link_up() API here would be racy as the link can go down at any time
> after the check.
>
> So to properly check for the device presence, this series introduces an API,
> pci_root_ports_have_device(), that accepts the Root bus pointer and checks for
> the presence of a device under any of the Root Ports. This API is used to
> replace the dw_pcie_link_up() check in suspend path of pcie-qcom driver and also
> used to skip the PME_Turn_Off brodcast message in dwc_pcie_suspend_noirq() API
> and to skip waiting for the link up in dwc_pcie_resume_noirq() API.
>
> Testing
> =======
>
> This series is tested on Qualcomm Lenovo Thinkpad T14s and observed no
> functional change during the system suspend path.
>
> - Mani
>
> [1] https://lore.kernel.org/linux-pci/CAKfTPtCtHquxtK=Zx2WSNm15MmqeUXO8XXi8FkS4EpuP80PP7g@mail.gmail.com/
> [2] https://lore.kernel.org/linux-pci/27516921.17f2.1997bb2a498.Coremail.zhangsenchuan@eswincomputing.com/
>
> Changes in v2:
>
> * Skipped waiting for link up in dwc_pcie_resume_noirq() if there was no device
>   before suspend.
> * Fixed the kdoc for pci_root_ports_have_device()
>
> Manivannan Sadhasivam (3):
>   PCI: host-common: Add an API to check for any device under the Root
>     Ports
>   PCI: qcom: Check for the presence of a device instead of Link up
>     during suspend
>   PCI: dwc: Check for the device presence during suspend and resume

You already queued it but FWIW
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>

>
>  .../pci/controller/dwc/pcie-designware-host.c | 13 ++++++++++++
>  drivers/pci/controller/dwc/pcie-qcom.c        |  6 ++++--
>  drivers/pci/controller/pci-host-common.c      | 21 +++++++++++++++++++
>  drivers/pci/controller/pci-host-common.h      |  2 ++
>  4 files changed, 40 insertions(+), 2 deletions(-)
>
> --
> 2.48.1
>

On Fri, 07 Nov 2025 10:13:16 +0530, Manivannan Sadhasivam wrote:
> This series aims to fix the usage of dw_pcie_link_up() API to check for Link up
> during system suspend. The motivation for this series comes from recent
> discussions [1] [2], where developers wanted to skip PME_Turn_Off broadcast in
> dw_pcie_suspend_noirq() API when devices are not attached to the bus. They ended
> up using dw_pcie_link_up() to check for the device presence due to the bad
> example in the pcie-qcom driver which does the same. The usage of
> dw_pcie_link_up() API here would be racy as the link can go down at any time
> after the check.
> 
> [...]

Applied, thanks!

[1/3] PCI: host-common: Add an API to check for any device under the Root Ports
      commit: 64a4b36d229120a71ede91aed16519164a181c13
[2/3] PCI: qcom: Check for the presence of a device instead of Link up during suspend
      commit: b6811cd64ac54fb2426b6ce35ba7c63588402687
[3/3] PCI: dwc: Check for the device presence during suspend and resume
      commit: e8fe6b3413a1b92b4bc0f0182ea4b49ee369541b

Best regards,
-- 
Manivannan Sadhasivam <mani@kernel.org>