Re: [U-Boot] [PATCH] cmd_pci: Check for VendorID earlier

Hi Fabio,

On Thu, Dec 31, 2015 at 11:20 PM, Fabio Estevam festevam@gmail.com wrote:

Hi Bin,

On Thu, Dec 31, 2015 at 7:32 AM, Bin Meng bmeng.cn@gmail.com wrote:

...

I also see this behavior on ls1021atwr board. I agree with Simon, the correct fix should fix the PCIe driver to return 0 instead of -EINVAL.

Do you mean like this for imx?

Yes

--- a/drivers/pci/pcie_imx.c +++ b/drivers/pci/pcie_imx.c @@ -381,7 +381,7 @@ static int imx_pcie_read_config(struct pci_controller *hose, ret = imx_pcie_addr_valid(d); if (ret) { *val = 0xffffffff;

•           return ret;
  

•           return 0;
    }
  
    va_address = get_bus_address(d, where);
  
  

and like this for layerscape:

Yes

--- a/drivers/pci/pcie_layerscape.c +++ b/drivers/pci/pcie_layerscape.c @@ -314,7 +314,7 @@ static int ls_pcie_read_config(struct pci_controller *hose,

    if (ls_pcie_addr_valid(hose, d)) {
            *val = 0xffffffff;

•           return -EINVAL;
  

•           return 0;
    }
  
    if (PCI_BUS(d) == hose->first_busno)
  
  

Again, I was wondering why we created two drivers for the same (or similar) PCIe IPs.

Regards, Bin