Linux Virtualization - Sep 2020 - [PATCH v6 42/76] x86/sev-es: Setup early #VC handler

On Mon, Aug 24, 2020 at 10:54:37AM +0200, Joerg Roedel
wrote:
> +#ifdef CONFIG_AMD_MEM_ENCRYPT
> +static void set_early_idt_handler(gate_desc *idt, int n, void *handler)
> +{
> +	struct idt_data data;
> +	gate_desc desc;
> +
> +	init_idt_data(&data, n, handler);
> +	idt_init_desc(&desc, &data);
> +	native_write_idt_entry(idt, n, &desc);
> +}
> +#endif
> +
> +static struct desc_ptr early_idt_descr __initdata = {
> +	.size		= IDT_TABLE_SIZE - 1,
> +	.address	= 0 /* Needs physical address of idt_table - initialized at
runtime. */,
> +};
> +
> +void __init early_idt_setup(unsigned long physbase)
> +{
> +	void __maybe_unused *handler;
> +	gate_desc *idt;
> +
> +	idt = fixup_pointer(idt_table, physbase);
> +
> +#ifdef CONFIG_AMD_MEM_ENCRYPT
> +	/* VMM Communication Exception */
> +	handler = fixup_pointer(vc_no_ghcb, physbase);
> +	set_early_idt_handler(idt, X86_TRAP_VC, handler);
This function is used only once AFAICT - you might just as well add its
three-lined body here and save yourself the function definition and
ifdeffery above...
> +#endif
> +
> +	/* Initialize IDT descriptor and load IDT */
> +	early_idt_descr.address = (unsigned long)idt;
> +	native_load_idt(&early_idt_descr);
> +}
> -- 
> 2.28.0
> 
-- 
Regards/Gruss,
    Boris.

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

On Mon, Aug 31, 2020 at 11:45:41AM +0200, Borislav Petkov
wrote:
> On Mon, Aug 24, 2020 at 10:54:37AM +0200, Joerg Roedel wrote:
> > +#ifdef CONFIG_AMD_MEM_ENCRYPT
> > +	/* VMM Communication Exception */
> > +	handler = fixup_pointer(vc_no_ghcb, physbase);
> > +	set_early_idt_handler(idt, X86_TRAP_VC, handler);
> 
> This function is used only once AFAICT - you might just as well add its
> three-lined body here and save yourself the function definition and
> ifdeffery above...
True, but having a separate function might be handy when support for #VE
and #HV is developed. Those might also need to setup their early
handlers here, no?

Regards,

	Joerg

On Tue, Sep 01, 2020 at 02:59:22PM +0200, Joerg Roedel
wrote:
> True, but having a separate function might be handy when support for #VE
> and #HV is developed. Those might also need to setup their early
> handlers here, no?
Ok.

-- 
Regards/Gruss,
    Boris.

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