On 17.02.2022 17:06, Andrew Cooper wrote: > On 17/02/2022 14:48, Jan Beulich wrote: >> On 17.02.2022 13:06, Andrew Cooper wrote: >>> On 17/02/2022 10:42, Jan Beulich wrote: >>>> On 17.02.2022 11:01, Andrew Cooper wrote: >>>>> --- a/xen/arch/x86/xen.lds.S >>>>> +++ b/xen/arch/x86/xen.lds.S >>>>> @@ -87,6 +87,7 @@ SECTIONS >>>>> *(.text.unlikely) >>>>> *(.fixup) >>>>> *(.text.kexec) >>>>> + kexec_reloc_end = .; >>>> Does this maybe want aligning on a 4- or even 8-byte boundary? If >>>> so, imo preferably not here, but by adding a trailing .align in the >>>> .S file. >>> There's no special need for it to be aligned, and it is anyway as the >>> stack is the last object in it. >> You mean it anyway would be, if the stack was aligned? Or am I to imply >> that you've amended the patch to add alignment there? > > I have aligned reloc_stack stack because that's a no-brainer.
With this ... > With that suitably aligned, kexec_reloc_end becomes aligned naturally > (because reloc_stack is the final object), and I don't think there's > much point putting anything explicit in the linker script. ... I certainly agree with this. Jan