Tomasz Maciej Nowak <tome...@o2.pl> writes: > This commit fixes regression on Linksys WRT1900 (Mamba) where this device > doesn't have USB 3.0 controller integrated in SoC, instead it has Etron > EJ168 connected to PCIe lane. Previously enabled in kernel 4.4 and 4.9, > was lost in transition to 4.14.
This is not correct. CONFIG_USB_XHCI_PCI defaults to y when CONFIG_USB_XHCI_HCD and CONFIG_USB_PCI are enabled. The EJ168 on Mamba works just fine on v4.14 without this patch. The xhci-pci driver is also enabled and binds to the EJ168 on v4.19. And it almost works - as shown by my usbmon traces. There would obviously not be any port or bus to trace if the EJ168 driver was disabled... So this fix has no effect. The v4.14 -> v4.19 regression on Mamba must be caused by something else. But I am starting to wonder if it might be unrelated to OpenWrt, EJ168 and Mamba at all. There are some suspicious-looking changes to the serial driver (io_edgeport) I am using. I'll do some testing with other devices and platforms. Bjørn _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel
