On Fri, 2023-11-03 at 19:34 +0100, David Hildenbrand wrote:
> On 03.11.23 18:30, Nina Schoetterl-Glausch wrote:
> > The length of the facility list accessed when interpretively executing
> > STFLE is the same as the hosts facility list (in case of format-0)
> > When shadowing, copy only those bytes.
> > The memory following the facility list need not be accessible, in which
> > case we'd wrongly inject a validity intercept.
> >
> > Fixes: 66b630d5b7f2 ("KVM: s390: vsie: support STFLE interpretation")
> > Signed-off-by: Nina Schoetterl-Glausch <nsg@xxxxxxxxxxxxx>
> > ---
> > arch/s390/include/asm/facility.h | 6 ++++++
> > arch/s390/kernel/Makefile | 2 +-
> > arch/s390/kernel/facility.c | 18 ++++++++++++++++++
> > arch/s390/kvm/vsie.c | 12 +++++++++++-
> > 4 files changed, 36 insertions(+), 2 deletions(-)
> > create mode 100644 arch/s390/kernel/facility.c
[...]
> > diff --git a/arch/s390/kernel/facility.c b/arch/s390/kernel/facility.c
[...]
> > +#include <asm/facility.h>
> > +
> > +unsigned int stfle_size(void)
> > +{
> > + static unsigned int size = 0;
> > + u64 dummy;
> > +
> > + if (!size) {
> > + size = __stfle_asm(&dummy, 1) + 1;
> > + }
> > + return size;
> > +}
> > +EXPORT_SYMBOL(stfle_size);
>
> Possible races? Should have to use an atomic?
Good point. Calling __stfle_asm multiple times is fine
and AFAIK torn reads/writes aren't possible. I don't see a way
for the compiler to break things either.
But it might indeed be nicer to use an atomic, without
any downsides.
>
> No access to documentation, but sounds plausible.
>
> Acked-by: David Hildenbrand <david@xxxxxxxxxx>
Thanks!
