On di, 2015-07-28 at 18:45 +0800, Ley Foon Tan wrote:

+config PCIE_ALTERA
+	bool "Altera PCIe controller"
+	depends on ARCH_SOCFPGA
+	depends on OF
+	select PCI_MSI_IRQ_DOMAIN if PCI_MSI
+	help
+	  Say Y here if you want to enable PCIe controller support for Altera
+	  SoCFPGA family of SoCs.

+obj-$(CONFIG_PCIE_ALTERA) += pcie-altera.o

+#include <linux/module.h>

+static const struct of_device_id altera_pcie_of_match[] = {
+	{ .compatible = "altr,pcie-root-port-1.0", },
+	{},
+};
+MODULE_DEVICE_TABLE(of, altera_pcie_of_match);

+static struct platform_driver altera_pcie_driver = {
+	[...]
+		.owner	= THIS_MODULE,
+	]...]
+};
+
+module_platform_driver(altera_pcie_driver);
+
+MODULE_AUTHOR("Ley Foon Tan [off-list ref]");
+MODULE_DESCRIPTION("Altera PCIe host controller driver");
+MODULE_LICENSE("GPL v2");

PCIE_ALTERA is a bool symbol. So pcie-altera.o is built-in only. Yet
pcie-altera.c uses a number of module-specific constructs. Should
PCIE_ALTERA perhaps be tristate?

Likewise for PCIE_ALTERA_MSI in 4/6.

Thanks,


Paul Bolle