Re: Clang/LLVM version requirements

[Paolo Bonzini <pbonzini@redhat.com>]

Il 13/09/2012 13:52, Jan Beulich ha scritto:
> 
> So the patch is fine by me if it covers that misalignment case.
> But it seems a little heavy handed - I'd think that instead of the
> sub-section, we could just create an arbitrary other section, or
> even allow uninitialized variable (it's unclear to me why Paolo
> wrote the comment - in c/s 25479:61dfb3da56b - regarding BSS
> the way it is now) - after all we only need to make sure that
> - the space gets properly allocated in trampoline.S, i.e. also in
>   reloc.bin
> - all accesses are PC-relative
> Neither has anything to do with use of uninitialized variables.

We cannot use BSS because it doesn't appear in reloc.S.

Apart from BSS, there would be no benefit for reloc to move itself to
BOOT_TRAMPOLINE, because BOOT_TRAMPOLINE is not a known location
anymore.  So it was easier to just run it in place from where we put it,
in the middle of head.S, but this means BSS disappears after extracting
reloc.bin.

So it's not really because the code must be relocatable, but more
because of the way we extract the binary data and put it in the middle
of head.S.

Initialized data works as long as you pass -fno-zero-initialized-in-bss
to the compiler or it is eaten.  I used assembly to declare the alloc
variable, because I wasn't sure of which GCC versions need the option,
and whether older versions are supported for compiling Xen.

Paolo