On Fri, Oct 25, 2013 at 12:18 PM, Karicheri, Muralidharan <m-karicheri2@xxxxxx> wrote: >>> -----Original Message----- >>> From: Andrew Murray [mailto:amurray@xxxxxxxxxxxxxxxxxxx] >>> Sent: Friday, October 25, 2013 12:27 PM >>> To: Karicheri, Muralidharan >>> Cc: linux-pci@xxxxxxxxxxxxxxx >>> Subject: Re: AER driver on ARM platform >>> >>> On 25 October 2013 16:56, Karicheri, Muralidharan <m-karicheri2@xxxxxx> wrote: >>> > Did anyone successfully tested PCIe AER driver on an ARM platform? I saw this enabled in >>> a tegra platform. Could someone from nVidia respond? >>> >>> I've previously used this on a custom ARM platform - it works. >>> >>> You need to ensure that PCIe port services are enabled, I believe at one point this required >>> 'pcie_ports=native' to be set on the kernel command line. >>> >>> Andrew Murray >>> > > Yes. That was it. It probes the driver now. You should not have to boot with "pcie_ports=native" unless you are debugging or working around a firmware issue. It looks like get_port_device_capability() expects the arch to implement a pcie_port_platform_notify() that sets a bitmask of PCIe services that Linux should take control of. But that's only implemented for ACPI systems (x86 and ia64), so probably every other arch is stuck with the "pcie_ports=native" hack. This is a bug, and somebody should fix it. Consider yourself empowered to implement whatever you need on ARM to negotiate for control of these PCIe services. Bjorn -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html