On Mon, Mar 25, 2013 at 01:31:28PM +0100, Michal Hocko wrote:
> On Fri 22-03-13 16:23:48, Naoya Horiguchi wrote:
> > Currently migrate_huge_page() takes a pointer to a hugepage to be
> > migrated as an argument, instead of taking a pointer to the list of
> > hugepages to be migrated. This behavior was introduced in commit
> > 189ebff28 ("hugetlb: simplify migrate_huge_page()"), and was OK
> > because until now hugepage migration is enabled only for soft-offlining
> > which takes only one hugepage in a single call.
> >
> > But the situation will change in the later patches in this series
> > which enable other users of page migration to support hugepage migration.
> > They can kick migration for both of normal pages and hugepages
> > in a single call, so we need to go back to original implementation
> > of using linked lists to collect the hugepages to be migrated.
>
> If the purpose of this patch is to reduce code duplication then you
> should remove migrate_huge_page as it doesn't have any caller anymore.
Yes, that makes sense. I'll do this.
> [...]
> > @@ -1482,12 +1483,20 @@ static int soft_offline_huge_page(struct page *page, int flags)
> > unlock_page(hpage);
> >
> > /* Keep page count to indicate a given hugepage is isolated. */
> > - ret = migrate_huge_page(hpage, new_page, MPOL_MF_MOVE_ALL,
> > - MIGRATE_SYNC);
> > - put_page(hpage);
> > + list_move(&hpage->lru, &pagelist);
> > + ret = migrate_pages(&pagelist, new_page, MPOL_MF_MOVE_ALL,
> > + MIGRATE_SYNC, MR_MEMORY_FAILURE);
> > if (ret) {
> > pr_info("soft offline: %#lx: migration failed %d, type %lx\n",
> > pfn, ret, page->flags);
> > + /*
> > + * We know that soft_offline_huge_page() tries to migrate
> > + * only one hugepage pointed to by hpage, so we need not
> > + * run through the pagelist here.
> > + */
> > + putback_active_hugepage(hpage);
>
> Maybe I am missing something but why we didn't need to call this before
> when using migrate_huge_page?
migrate_huge_page() does not need list handling before/after the call,
because it's defined to migrate only one hugepage, and it has a page as
an argument, not list_head.
> > + if (ret > 0)
> > + ret = -EIO;
> > } else {
> > set_page_hwpoison_huge_page(hpage);
> > dequeue_hwpoisoned_huge_page(hpage);
> > diff --git v3.9-rc3.orig/mm/migrate.c v3.9-rc3/mm/migrate.c
> > index f69f354..66030b6 100644
> > --- v3.9-rc3.orig/mm/migrate.c
> > +++ v3.9-rc3/mm/migrate.c
> > @@ -981,6 +981,8 @@ static int unmap_and_move_huge_page(new_page_t get_new_page,
> >
> > unlock_page(hpage);
> > out:
> > + if (rc != -EAGAIN)
> > + putback_active_hugepage(hpage);
>
> And why do you put it here? If it is called from migrate_pages then the
> caller already does the clean-up (putback_lru_pages).
What the caller of migrate_pages() cleans up is the (huge)pages which failed
to be migrated. And what the above code cleans up is the source hugepage
after the migration succeeds.
The latter clean-up code originally existed, but removed in 189ebff28
("hugetlb: simplify migrate_huge_page()").
This commit cleans up the code based on that there was only one user
of hugepage migration, but that's not true any more.
So the above hunk is a part of revert of the commit.
But it's not a simple revert, because there's one difference between
now and before 189ebff28 that we link hugepages in-use to hugepage_activelist.
Then we finally come to the above change.
Thanks,
Naoya
> > put_page(new_hpage);
> > if (result) {
> > if (rc)
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>