On Tue, Jul 14, 2020 at 06:34:24PM -0700, Kees Cook wrote: > On Tue, Jul 14, 2020 at 02:08:36PM +0200, Joerg Roedel wrote: > > +# make sure head64.c is built without stack protector > > +nostackp := $(call cc-option, -fno-stack-protector) > > +CFLAGS_head64.o := $(nostackp) > > Recent refactoring[1] for stack protector suggests this should just > unconditionally be: > > CFLAGS_head64.o += -fno-stack-protector > > But otherwise, yeah, this should be fine here -- it's all early init > stuff. > > Reviewed-by: Kees Cook <keescook@xxxxxxxxxxxx> Thanks, I am not sure this patch will be needed in the next version, as I am currently rebasing to tip/master, which also made idt_descr static in kernel/idt.c. So with that I think I have to move the early IDT init functions to kernel/idt.c too and setup %gs earlier in head_64.S to make stack-protector happy. The %gs setup actually needs to happen two times, one time when the kernel still runs identity mapped and then again when it switched to virtual addresses. Regards, Joerg
