Re: [PATCH v7 05/10] x86/tdx: Handle port I/O

[Sathyanarayanan Kuppuswamy]

On 10/17/21 12:58 PM, Thomas Gleixner wrote:
> On Tue, Oct 05 2021 at 13:41, Kuppuswamy Sathyanarayanan wrote:
> >   									\
> >   static inline void outs##bwl(int port, const void *addr, unsigned long count) \
> >   {									\
> > -	if (sev_key_active()) {						\
> > +	if (sev_key_active() ||						\
> > +	    cc_platform_has(CC_ATTR_GUEST_UNROLL_STRING_IO)) {		\
> Instead of adding an extra check, can you please replace that
> sev_key_active() with cc_platform_has() completely?

Yes. sev_key_active() can be removed and replaced with
cc_platform_has().

Thomas Lendacky also proposed to introduce as common
static key which can be set by both AMD SEV and TDX code.

@Thomas Lendacky, any comments?

> > +/*
> > + * tdx_handle_early_io() cannot be re-used in #VE handler for handling
> > + * I/O because the way of handling string I/O is different between
> > + * normal and early I/O case. Also, once trace support is enabled,
> > + * tdx_handle_io() will be extended to use trace calls which is also
> > + * not valid for early I/O cases.
> > + */
> > +static void tdx_handle_io(struct pt_regs *regs, u32 exit_qual)
> > +{
> > +	struct tdx_hypercall_output outh;
> > +	int out, size, port, ret;
> > +	bool string;
> > +	u64 mask;
> > +
> > +	string = VE_IS_IO_STRING(exit_qual);
> > +
> > +	/* I/O strings ops are unrolled at build time. */
> Fancy. The compiler can evaluate sev_key_active() and
> cc_platform_has() at build time?

It is incorrect. I will fix this in next version. How about following
change?

I/O strings are replaced with in/out instructions. If string is reported,
report it with BUG.

> Thanks,
>
>          tglx

--
Sathyanarayanan Kuppuswamy
Linux Kernel Developer