Re: [PATCH v2 04/12] x86/x86: Add early_is_tdx_guest() interface

[Kuppuswamy, Sathyanarayanan]

On 6/17/21 10:05 AM, Borislav Petkov wrote:
> On Sat, Jun 12, 2021 at 02:04:45PM -0700, Kuppuswamy Sathyanarayanan wrote:
>
> > Subject: Re: [PATCH v2 04/12] x86/x86: Add early_is_tdx_guest() interface
>
> Subject prefix should be "x86/tdx:" ofc.

I will fix this in next version.

> > diff --git a/arch/x86/boot/compressed/tdx.c b/arch/x86/boot/compressed/tdx.c
> > new file mode 100644
> > index 000000000000..ddfa4a6d1939
> > --- /dev/null
> > +++ b/arch/x86/boot/compressed/tdx.c
> > @@ -0,0 +1,28 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * tdx.c - Early boot code for TDX
> > + */
> > +
> > +#include <asm/tdx.h>
>
> Please no kernel proper includes in the decompressor stage - that thing
> is an include hell mess and needs cleaning up. Use cpuid_count() from
> arch/x86/boot/cpuflags.c by exporting it properly and add the other
> defines here instead of using kernel proper facilities.
>
> I know, I know, there is other misuse but it has to stop.

I will re-use cpuid_count() from cpuflags.c. But it is missing definition
for cpuid_eax(0). So, I will have to define a new function to handle it.
Hope its alright with you.

> > +static int __ro_after_init tdx_guest = -1;
> > +
> > +static inline bool native_cpuid_has_tdx_guest(void)
>
> Why is this function prefixed with "native_"?
>
> > +{
> > +	u32 eax = TDX_CPUID_LEAF_ID, sig[3] = {0};
> > +
> > +	if (native_cpuid_eax(0) < TDX_CPUID_LEAF_ID)
> > +		return false;
> > +
> > +	native_cpuid(&eax, &sig[0], &sig[1], &sig[2]);
> > +
> > +	return !memcmp("IntelTDX    ", sig, 12);
> > +}
> > +
> > +bool early_is_tdx_guest(void)
>
> So I guess this is going to be used somewhere, because I don't see it.
> Or is it going away in favor of prot_guest_has(PR_GUEST_TDX) ?

It is used for handling TDX specific I/O support in decompressor code.

You can find its usage in https://lore.kernel.org/patchwork/patch/1444186/

May be I should move this patch next to that patch for easy reading.

> This is the problem with sending new versions of patches as a reply to
> the old ones in the patchset: I get confused. Why don't you wait until
> the whole thing has been reviewed and then send a new revision which has
> all the changes and I can find my way in the context?
>
> And with the amount of changes so far, you should probably send a new
> revision of the initial support set now-ish instead of me reviewing this
> one 'til the end.

I will send the new series with review fixes. Also, since this patch is only
used by I/O support code, I will move this patch to another TDX series which
adds TDX I/O support.

> Thx.

--
Sathyanarayanan Kuppuswamy
Linux Kernel Developer