Re: [PATCH 02/12] x86/mm/hotplug: Remove pgd_list use from the memory hotplug code

[Ingo Molnar <mingo@kernel.org>]

* Oleg Nesterov <oleg@redhat.com> wrote:

> On 06/11, Ingo Molnar wrote:
> >
> >  void sync_global_pgds(unsigned long start, unsigned long end, int removed)
> >  {
> > @@ -169,29 +169,33 @@ void sync_global_pgds(unsigned long start, unsigned long end, int removed)
> >
> >  	for (address = start; address <= end; address += PGDIR_SIZE) {
> >  		const pgd_t *pgd_ref = pgd_offset_k(address);
> > -		struct page *page;
> > +		struct task_struct *g, *p;
> >
> >  		/*
> > -		 * When it is called after memory hot remove, pgd_none()
> > -		 * returns true. In this case (removed == 1), we must clear
> > -		 * the PGD entries in the local PGD level page.
> > +		 * When this function is called after memory hot remove,
> > +		 * pgd_none() already returns true, but only the reference
> > +		 * kernel PGD has been cleared, not the process PGDs.
> > +		 *
> > +		 * So clear the affected entries in every process PGD as well:
> >  		 */
> >  		if (pgd_none(*pgd_ref) && !removed)
> >  			continue;
> >
> >  		spin_lock(&pgd_lock);
> > -		list_for_each_entry(page, &pgd_list, lru) {
> > -			pgd_t *pgd;
> > +
> > +		for_each_process_thread(g, p) {
> > +			pgd_t *pgd = p->mm->pgd;
> 
> and we use the same pgd later: set_pgd(pgd, *pgd_ref).
> looks like this should be pgd_offset(p->mm, address) ?
> 
> And obviously you need to check ->mm != NULL first?

Yes.

> And perhaps it makes sense to swap "for (address)" and for_each_process() 
> loops...

Yes, but I'd flip around the logic in a later, separate patch.

Note that PGDIR_SIZE is 512 GB here, so the outer loop will most likely execute at 
most twice in practice.

Thanks,

	Ingo