* Glauber de Oliveira Costa (gcosta at redhat.com) wrote: > I've been working on a port of paravirt_ops for the x86_64 architectures > for a couple of days. Here's the result so far. I'm posting it soon for > early cooments. The more, the better. Thanks a lot for getting this started Glauber. As Rusty mentioned, we'll want to follow the no export paravirt_ops motion upstream for x86_64 as well. > Index: linux-2.6.19-quilt/arch/x86_64/kernel/entry.S > =================================================================== > --- linux-2.6.19-quilt.orig/arch/x86_64/kernel/entry.S > +++ linux-2.6.19-quilt/arch/x86_64/kernel/entry.S > @@ -1064,6 +1064,22 @@ KPROBE_ENTRY(int3) > CFI_ENDPROC > KPROBE_END(int3) > > +#ifdef CONFIG_PARAVIRT > +ENTRY(native_iret) > +1: iretq > +.section __ex_table,"a" > + .align 8 > + .quad 1b, bad_iret > +.previous > +ENDPROC(native_iret) > + > +ENTRY(native_irq_enable_sysexit) > + sti > + sysretq > +ENDPROC(native_irq_enable_sysexit) sysexit is legacy mode only and I think we'll need a story for swapgs
