On 31.05.2012, at 23:58, Scott Wood wrote: > On 05/30/2012 06:00 AM, Alexander Graf wrote: >> @@ -179,6 +182,20 @@ static int mpc8544_load_device_tree(CPUPPCState *env, >> } >> } >> >> + /* XXX These should go into their respective devices' code */ >> + sprintf(soc, "/soc8544@%x", MPC8544_CCSRBAR_BASE); > > This should just be "/soc@%x" (something like "ccsr@%x" or "immr@%x" > would be better, but hasn't really caught on yet). "soc8544"-type names > have been discouraged for a while now. > >> + qemu_devtree_setprop_string(fdt, soc, "compatible", "simple-bus"); > > It should also have a compatible of "fsl,mpc8544-immr". > >> + qemu_devtree_setprop_cell2(fdt, soc, "reg", MPC8544_CCSRBAR_BASE, >> + MPC8544_CCSRBAR_REGSIZE); > > We don't need reg here anymore. > > I realize this was probably meant to produce the same output as the > current device tree, but if we're introducing new code to generate this > stuff we should follow current best practices.
Good point. I guess we should clean that up as a patch at the end of the series though. First convert the state 1:1 over, then improve it. Alex
