Re: [PATCH] x86/mm/encrypt: Use %a asm operand modifier to obtain %rip-relative address

[Uros Bizjak]

On Tue, Nov 21, 2023 at 10:05 AM Kirill A. Shutemov
<kirill@xxxxxxxxxxxxx> wrote:
>
> On Mon, Nov 20, 2023 at 08:58:29PM +0100, Uros Bizjak wrote:
> > On Mon, Nov 20, 2023 at 6:15 PM Borislav Petkov <bp@xxxxxxxxx> wrote:
> > >
> > > On Mon, Nov 20, 2023 at 04:33:50PM +0100, Uros Bizjak wrote:
> > > > The "a" asm operand modifier substitutes a memory reference, with the
> > > > actual operand treated as address.  For x86_64, when a symbol is
> > > > provided, the "a" modifier emits "sym(%rip)" instead of "sym".
> > > >
> > > > Clang allows only "i" and "r" operand constraints with an "a" modifier,
> > > > so the patch normalizes the modifier/constraint pair to "a"/"i"
> > >
> > > s/the patch normalizes/normalize/
> > >
> > > > which is consistent between both compilers.
> > > >
> > > > No functional change intended.
> > > >
> > > > Suggested-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> > > > Cc: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
> > > > Cc: Andy Lutomirski <luto@xxxxxxxxxx>
> > > > Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > > > Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> > > > Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> > > > Cc: Borislav Petkov <bp@xxxxxxxxx>
> > > > Cc: "H. Peter Anvin" <hpa@xxxxxxxxx>
> > > > Signed-off-by: Uros Bizjak <ubizjak@xxxxxxxxx>
> > > > ---
> > > >  arch/x86/mm/mem_encrypt_identity.c | 16 ++++------------
> > > >  1 file changed, 4 insertions(+), 12 deletions(-)
> > > >
> > > > diff --git a/arch/x86/mm/mem_encrypt_identity.c b/arch/x86/mm/mem_encrypt_identity.c
> > > > index d73aeb16417f..6a351fdd1b39 100644
> > > > --- a/arch/x86/mm/mem_encrypt_identity.c
> > > > +++ b/arch/x86/mm/mem_encrypt_identity.c
> > > > @@ -346,9 +346,7 @@ void __init sme_encrypt_kernel(struct boot_params *bp)
> > > >        * We're running identity mapped, so we must obtain the address to the
> > > >        * SME encryption workarea using rip-relative addressing.
> > > >        */
> > > > -     asm ("lea sme_workarea(%%rip), %0"
> > > > -          : "=r" (workarea_start)
> > > > -          : "p" (sme_workarea));
> > > > +     asm ("lea %a1, %0" : "=r" (workarea_start) : "i" (sme_workarea));
> > >
> > > Yeah, I saw that particular subthread today.
> > >
> > > Are you sure this "a" modifier DTRT with all gcc version we support?
> > >
> > > I.e., from 5.1 onwards...
> >
> > Yes [1].
> >
> > [1] https://godbolt.org/z/aWe7b16rT
>
> But Clang generates RIP-relative load only starting from version 15.
> Clange 11 claimed to be supported.

Huh, what a bummer...

The patch is retracted then.

Thanks,
Uros.