On Wednesday 05 March 2014, Liviu Dudau wrote:

+
+	pr_debug("Parsing ranges property...\n");
+	for_each_of_pci_range(&parser, &range) {
+		/* Read next ranges element */
+		pr_debug("pci_space: 0x%08x pci_addr:0x%016llx ",
+				range.pci_space, range.pci_addr);
+		pr_debug("cpu_addr:0x%016llx size:0x%016llx\n",
+					range.cpu_addr, range.size);
+
+		/*
+		 * If we failed translation or got a zero-sized region
+		 * then skip this range
+		 */
+		if (range.cpu_addr == OF_BAD_ADDR || range.size == 0)
+			continue;
+
+		res = kzalloc(sizeof(struct resource), GFP_KERNEL);
+		if (!res)
+			return -ENOMEM;
+
+		of_pci_range_to_resource(&range, dev, res);
+
+		if (resource_type(res) == IORESOURCE_IO)
+			*io_base = range.cpu_addr;
+
+		pci_add_resource_offset(resources, res,
+				res->start - range.pci_addr);
+	}

As mentioned regarding the pci_register_io_range() helper, x86
would not enter the 'resource_type(res) == IORESOURCE_IO' code path,
which on the one hand is fine so we can return an error from
pci_register_io_range() there, but I think it will lead to
io_base getting an uninitialized content.

There could also be other reasons why pci_register_io_range() fails,
e.g. because the space is exhausted, and I think we should try to
catch that here and skip the pci_add_resource_offset() and io_base
assignment then.

	Arnd