Hello u-boot-users,
This is my attempt at adding CONFIG_PCISLAVE support to the MPC8349EMDS board. I noticed that there was some shared infrastructure for the PCI controller in the 83xx-series CPU's, so I removed all of the old code, and used the infrastructure instead.
I have boot tested the board in host mode (using the 83XX_GENERIC_PCI infrastructure) but I do not have the IO Expander board, so I cannot test that PCI cards still work. Linux happily initializes the PCI interface, and lspci -vvv is identical before and after my patch.
I copied the configuration bit unlocking routine from other boards support for CONFIG_PCISLAVE. I added it to the shared infrastructure so it would have access to the static pci_controller that is set up and used by the other infrastructure code.
I then went ahead and added support for CONFIG_PCISLAVE to the MPC8349EMDS board. I have boot tested the board in a desktop pc. I have booted Linux on the board in this mode (with CONFIG_PCI=n so it does not try to re-initialize other PCI devices on the bus. Linux is only set up for Host-mode PCI), and have successfully developed both U-Boot and Linux drivers that implement a virtual serial port and virtual ethernet device (between the host pc and the MPC8349EMDS board) using only the PCI bus and the MPC8349EMDS message registers. I'm happy to post them if there is any interest.
I'm looking for suggestions and improvements so that these can be accepted into the U-Boot source.
Thanks, Ira