Re: [PATCH 3/4] PCI: qcom: Read back PARF_LTSSM register

[Manivannan Sadhasivam]

On Wed, Dec 27, 2023 at 11:17:21PM +0100, Konrad Dybcio wrote:
> To ensure write completion, read the PARF_LTSSM register after setting
> the LTSSM enable bit before polling for "link up".
> 
> Signed-off-by: Konrad Dybcio <konrad.dybcio@xxxxxxxxxx>

I'd consider this as a bug since if the LTSSM write gets cached in Write Buffer,
then the polling time becomes wrong in dw_pcie_wait_for_link(), leading to
false link_up failure.

Although both of the write/read (LTSSM in qcom_pcie_2_3_2_ltssm_enable(), and
PCI_EXP_LNKSTA in qcom_pcie_link_up()) belong to PCIe domain, they belong to
different regions (PARF, DBI). So I'm not sure we can safely ignore the write
completion issue. So,

Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@xxxxxxxxxx>

- Mani

> ---
>  drivers/pci/controller/dwc/pcie-qcom.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c
> index a02dc197c495..3d77269e70da 100644
> --- a/drivers/pci/controller/dwc/pcie-qcom.c
> +++ b/drivers/pci/controller/dwc/pcie-qcom.c
> @@ -540,6 +540,7 @@ static void qcom_pcie_2_3_2_ltssm_enable(struct qcom_pcie *pcie)
>  	val = readl(pcie->parf + PARF_LTSSM);
>  	val |= LTSSM_EN;
>  	writel(val, pcie->parf + PARF_LTSSM);
> +	readl(pcie->parf + PARF_LTSSM);
>  }
>  
>  static int qcom_pcie_get_resources_2_3_2(struct qcom_pcie *pcie)
> 
> -- 
> 2.43.0
> 

-- 
மணிவண்ணன் சதாசிவம்