Hello. I rewrote my earlier patch to romcc_io.h (where I banned some device ranges from pci_locate_device) because it was too aggressive. With this patch to sb700_early_setup.c I achieve the same on my board (serial output, and it hangs in the same place), but hopefully with less breakage for other boards. I first tried to hardcode PCI bus, device and function in sb700_lpc_init() only but it did hang elsewhere, so I finally thought it more coherent to hardcode all devices in the file than just some functions, and it works for me. I still have the doubt of why was there a pci_locate_device in the first place, since from my more than limited knowledge it does not look like the devices in sb700 may appear in other bus,device,function addresses. But the original even had an error message at one place for when the device was not found.
Please review and/or test and commit if you think appropiate (I guess it's not urgent, since my board still does not boot).
This patch applies to svn 5739
Signed-off-by: Xavi Drudis Ferran xdrudis@tinet.cat
---
Thank you for commiting my previous patches.
Now I plan to have a look at fidvid.c and AMD's Bios and Kernel Developer Guide. I'm not sure I'll be able to fix much, and it may take me a while. I guess it's better not to send patches or anything until I get something that works for me.