Re: [PATCH RESEND v9 2/2] x86/tdx: Support vmalloc() for tdx_enc_status_changed()

[Dave Hansen]

On 8/10/23 19:12, Dexuan Cui wrote:
> +	if (!is_vmalloc_addr((void *)start))
> +		return tdx_enc_status_changed_phys(__pa(start), __pa(end), enc);
> +
> +	while (start < end) {
> +		phys_addr_t start_pa = slow_virt_to_phys((void *)start);
> +		phys_addr_t end_pa = start_pa + PAGE_SIZE;
> +
> +		if (!tdx_enc_status_changed_phys(start_pa, end_pa, enc))
> +			return false;
> +
> +		start += PAGE_SIZE;
> +	}

This creates two different paths for vmalloc() and the direct map.
There are two different ways to do va=>pa conversion, for instance.
Here's a single loop that works for both cases:

	unsigned long step = end - start;
	unsigned long addr;

	/* Step through page-by-page for vmalloc() mappings: */
	if (is_vmalloc_addr((void *)vaddr))
		step = PAGE_SIZE;

	for (addr = start; addr < end; addr += step) {
		phys_addr_t start_pa = slow_virt_to_phys(addr);
		phys_addr_t end_pa   = start_pa + step;

		if (!tdx_enc_status_changed_phys(start_pa, end_pa, enc))
			return false;
	}

Note that this also doesn't abuse 'start' by making it a loop variable.
It also, uh, uses a for() loop.

The only downside is that it costs a page table walk for direct map
virt=>phys conversion.  I can live with that.