Re: [PATCH v9 10/43] x86/sev: Check SEV-SNP features support

[Borislav Petkov]

On Fri, Jan 28, 2022 at 11:17:31AM -0600, Brijesh Singh wrote:
> diff --git a/arch/x86/boot/compressed/idt_64.c b/arch/x86/boot/compressed/idt_64.c
> index 9b93567d663a..63e9044ab1d6 100644
> --- a/arch/x86/boot/compressed/idt_64.c
> +++ b/arch/x86/boot/compressed/idt_64.c
> @@ -39,7 +39,15 @@ void load_stage1_idt(void)
>  	load_boot_idt(&boot_idt_desc);
>  }
>  
> -/* Setup IDT after kernel jumping to  .Lrelocated */
> +/*
> + * Setup IDT after kernel jumping to  .Lrelocated
> + *
> + * initialize_identity_maps() needs a PF handler setup. The PF handler setup
> + * needs to happen in load_stage2_idt() where the IDT is loaded and there the
> + * VC IDT entry gets setup too in order to handle VCs, one needs a GHCB which
> + * gets setup with an already setup table which is done in
> + * initialize_identity_maps() and this is where the circle is complete.
> + */

I've beefed it up more, please use this one instead:

/*
 * Setup IDT after kernel jumping to  .Lrelocated.
 *
 * initialize_identity_maps() needs a #PF handler to be setup
 * in order to be able to fault-in identity mapping ranges; see
 * do_boot_page_fault().
 *
 * This #PF handler setup needs to happen in load_stage2_idt() where the
 * IDT is loaded and there the #VC IDT entry gets setup too.
 *
 * In order to be able to handle #VCs, one needs a GHCB which
 * gets setup with an already set up pagetable, which is done in
 * initialize_identity_maps(). And there's the catch 22: the boot #VC
 * handler do_boot_stage2_vc() needs to call early_setup_ghcb() itself
 * (and, especially set_page_decrypted()) because the SEV-ES setup code
 * cannot initialize a GHCB as there's no #PF handler yet...
 */

> diff --git a/arch/x86/kernel/sev.c b/arch/x86/kernel/sev.c
> index 19ad09712902..24df739c9c05 100644
> --- a/arch/x86/kernel/sev.c
> +++ b/arch/x86/kernel/sev.c
> @@ -43,6 +43,9 @@ static struct ghcb boot_ghcb_page __bss_decrypted __aligned(PAGE_SIZE);
>   */
>  static struct ghcb __initdata *boot_ghcb;
>  
> +/* Bitmap of SEV features supported by the hypervisor */
> +static u64 sev_hv_features __ro_after_init;
> +
>  /* #VC handler runtime per-CPU data */
>  struct sev_es_runtime_data {
>  	struct ghcb ghcb_page;
> @@ -766,6 +769,18 @@ void __init sev_es_init_vc_handling(void)
>  	if (!sev_es_check_cpu_features())
>  		panic("SEV-ES CPU Features missing");
>  
> +	/*
> +	 * SEV-SNP is supported in v2 of the GHCB spec which mandates support for HV
> +	 * features. If SEV-SNP is enabled, then check if the hypervisor supports
> +	 * the SEV-SNP features.

You guys have been completely brainwashed by marketing. I say:

"s/SEV-SNP/SNP/g

And please do that everywhere in sev-specific files."

and you go and slap that "SEV-" thing everywhere instead. Why? That file
is already called sev.c so it must be SEV-something. Lemme simplify that
comment for ya:

	/*
	 * SNP is supported in v2 of the GHCB spec which mandates support for HV
	 * features.
	 */

That's it, no more needed - the rest should be visible from the code.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette