On Tue, Apr 10, 2018 at 06:17:04PM +0200, Marek Vasut wrote: > On 04/10/2018 05:28 PM, Geert Uytterhoeven wrote: ... > >>> rcar_pcie_get_resources() is called while the device is > >>> runtime-enabled/resumed, > >>> pci_free_resource_list() is called while the device is runtime-disabled. > > rcar_pcie_get_resources() is NOT a pair function for > pci_free_resource_list() . rcar_pcie_parse_request_of_pci_ranges() is a > pair function for pci_free_resource_list(). > > rcar_pcie_parse_request_of_pci_ranges() calls > of_pci_get_host_bridge_resources() internally, so every single function > called after successful call of rcar_pcie_parse_request_of_pci_ranges() > must call pci_free_resource_list(). > > Both of_pci_get_host_bridge_resources() and pci_free_resource_list() are > called with runtime PM disabled. > > The naming of the functions is confusing though. Hi, thanks everyone for their efforts in preparing/reviewing this patch. It seems there are some differences of opinion on how best to handle the error paths but unlike earlier versions this one seems correct to me. If that turns out to be false we can address it. But I don't think its likely things will be enhanced by continuing this review. Lorenzo, please consider taking this patch in its current form. Reviewed-by: Simon Horman <horms+renesas@xxxxxxxxxxxx>
