Re: [RFC][PATCH 15/22] x86,vmx: Remove .fixup usage

[Peter Zijlstra]

On Sat, Nov 06, 2021 at 08:05:51AM +0100, Paolo Bonzini wrote:
> On 11/5/21 19:17, Sean Christopherson wrote:
> > On Thu, Nov 04, 2021, Peter Zijlstra wrote:
> > > In the vmread exceptin path, use the, thus far, unused output register
> > > to push the @fault argument onto the stack. This, in turn, enables the
> > > exception handler to not do pushes and only modify that register when
> > > an exception does occur.
> > > 
> > > Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> > > ---
> > >   arch/x86/kvm/vmx/vmx_ops.h |   14 ++++++--------
> > >   1 file changed, 6 insertions(+), 8 deletions(-)
> > > 
> > > --- a/arch/x86/kvm/vmx/vmx_ops.h
> > > +++ b/arch/x86/kvm/vmx/vmx_ops.h
> > > @@ -80,9 +80,11 @@ static __always_inline unsigned long __v
> > >   		      * @field, and bounce through the trampoline to preserve
> > >   		      * volatile registers.
> > >   		      */
> > > -		     "push $0\n\t"
> > > +		     "xorl %k1, %k1\n\t"
> > > +		     "2:\n\t"
> > > +		     "push %1\n\t"
> > >   		     "push %2\n\t"
> > 
> > This trick doesn't work if the compiler selects the same GPR for %1 and %2, as
> > the "field" will get lost.
> 
> Ouch, good catch.  It should be actually very simple to fix it, just mark
> "value" as an "early clobber" output:
> 
>       : ASM_CALL_CONSTRAINT, "=&r"(value) : "r"(field) : "cc");
> 
> That's an output which is written before the instruction is finished using
> the input operands.  The manual even says "this operand may not lie in a
> register that is read by the instruction or as part of any memory address",
> which is exactly what you caught with %1 and %2 both being the same GPR.

Yes, but as Sean points out, that will negatively affect code-gen on the
happy path. But perhaps that's acceptable if we add the asm-goto-output
variant?