Re: [PATCHv11 11/19] x86/tdx: Convert shared memory back to private on kexec

[Kirill A. Shutemov]

On Fri, May 31, 2024 at 05:14:42PM +0200, Borislav Petkov wrote:
> On Tue, May 28, 2024 at 12:55:14PM +0300, Kirill A. Shutemov wrote:
> > +static void tdx_kexec_finish(void)
> > +{
> > +	unsigned long addr, end;
> > +	long found = 0, shared;
> > +
> > +	lockdep_assert_irqs_disabled();
> > +
> > +	addr = PAGE_OFFSET;
> > +	end  = PAGE_OFFSET + get_max_mapped();
> > +
> > +	while (addr < end) {
> > +		unsigned long size;
> > +		unsigned int level;
> > +		pte_t *pte;
> > +
> > +		pte = lookup_address(addr, &level);
> > +		size = page_level_size(level);
> > +
> > +		if (pte && pte_decrypted(*pte)) {
> > +			int pages = size / PAGE_SIZE;
> > +
> > +			/*
> > +			 * Touching memory with shared bit set triggers implicit
> > +			 * conversion to shared.
> > +			 *
> > +			 * Make sure nobody touches the shared range from
> > +			 * now on.
> > +			 */
> > +			set_pte(pte, __pte(0));
> > +
> 
> Format the below into a comment here:
> 
> /* 
> 
> The only thing one can do at this point on failure is panic. It is
> reasonable to proceed, especially for the crash case because the
> kexec-ed kernel is using a different page table so there won't be
> a mismatch between shared/private marking of the page so it doesn't
> matter.

Page tables would not make a difference here. We will switch to identity
mappings soon. And kexec-ed kernel will build new page tables from
scratch.

I will drop the part after "It is reasonable to proceed".


-- 
  Kiryl Shutsemau / Kirill A. Shutemov