Re: [PATCH] x86/sev-es: Do not use copy_from_kernel_nofault in early #VC handler

[Dave Hansen]

On 9/6/23 15:45, Adam Dunlap wrote:
>  static int vc_fetch_insn_kernel(struct es_em_ctxt *ctxt,
> -				unsigned char *buffer)
> +				unsigned char *buffer, bool is_early)
>  {
> -	return copy_from_kernel_nofault(buffer, (unsigned char *)ctxt->regs->ip, MAX_INSN_SIZE);
> +	if (is_early) {
> +		memcpy(buffer, (unsigned char *)ctxt->regs->ip, MAX_INSN_SIZE);
> +		return 0;
> +	} else {
> +		return copy_from_kernel_nofault(buffer, (unsigned char *)ctxt->regs->ip,
> +			MAX_INSN_SIZE);
> +	}
>  }

This isn't the normal way we do these kinds of things.

If we go with this solution, they next guy who tries
copy_from_kernel_nofault() will hit the same issue, and start plumbing
their own 'is_early' through _their_ little chunk of arch/x86.

Usually, we'll add some gunk in arch/x86/boot/compressed/misc.h to
override the troublesome implementation.  In this case, it would make a
lot of sense to somehow avoid touching boot_cpu_data.x86_virt_bits in
the first place.