On Mon, Jul 04, 2016 at 01:46:24AM +0900, Yoshinori Sato wrote: > FDT address is P1SEG. So not virtual address. > > Signed-off-by: Yoshinori Sato <ys...@users.sourceforge.jp> > --- > arch/sh/kernel/setup.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/arch/sh/kernel/setup.c b/arch/sh/kernel/setup.c > index 86f2792..8e3b099 100644 > --- a/arch/sh/kernel/setup.c > +++ b/arch/sh/kernel/setup.c > @@ -254,7 +254,7 @@ void __ref sh_fdt_init(phys_addr_t dt_phys) > #ifdef CONFIG_USE_BUILTIN_DTB > dt_virt = __dtb_start; > #else > - dt_virt = phys_to_virt(dt_phys); > + dt_virt = (void *)P1SEGADDR(dt_phys); > #endif > > if (!dt_virt || !early_init_dt_scan(dt_virt)) { > --
I don't think this change is correct, and I'm not sure what the motivation is. It certainly can't work with !CONFIG_29BIT, and likely can't work on nommu either (it won't work on J2). Maybe we have different ideas about the sort of physical address the boot loader is expected to pass; I would expect it to be something that, when passed to phys_to_virt, yields an address the kernel can use to access the memory. This does not necessarily mean it's MMU-mapped memory; it could be (and in practice will be, I think) an address in the P1 segment obtained by adding PAGE_OFFSET (see asm/page.h). Rich