[PATCH v7 3/5] mm/hugetlb: add support for mempolicy MPOL_PREFERRED_MANY
From: Feng Tang <hidden>
Date: 2021-08-03 05:59:43
Also in:
linux-api, lkml
Subsystem:
hugetlb subsystem, memory management, the rest · Maintainers:
Muchun Song, Oscar Salvador, Andrew Morton, Linus Torvalds
From: Ben Widawsky <redacted> Implement the missing huge page allocation functionality while obeying the preferred node semantics. This is similar to the implementation for general page allocation, as it uses a fallback mechanism to try multiple preferred nodes first, and then all other nodes. [akpm: fix compling issue when merging with other hugetlb patch] [Thanks to 0day bot for catching the missing #ifdef CONFIG_NUMA issue] Link: https://lore.kernel.org/r/20200630212517.308045-12-ben.widawsky@intel.com (local) Suggested-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Ben Widawsky <redacted> Co-developed-by: Feng Tang <redacted> Signed-off-by: Feng Tang <redacted> --- mm/hugetlb.c | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+)
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 95714fb28150..9279f6d478d9 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c@@ -1166,7 +1166,20 @@ static struct page *dequeue_huge_page_vma(struct hstate *h, gfp_mask = htlb_alloc_mask(h); nid = huge_node(vma, address, gfp_mask, &mpol, &nodemask); +#ifdef CONFIG_NUMA + if (mpol->mode == MPOL_PREFERRED_MANY) { + page = dequeue_huge_page_nodemask(h, gfp_mask, nid, nodemask); + if (page) + goto check_reserve; + /* Fallback to all nodes */ + nodemask = NULL; + } +#endif page = dequeue_huge_page_nodemask(h, gfp_mask, nid, nodemask); + +#ifdef CONFIG_NUMA +check_reserve: +#endif if (page && !avoid_reserve && vma_has_reserves(vma, chg)) { SetHPageRestoreReserve(page); h->resv_huge_pages--;
@@ -2147,6 +2160,21 @@ struct page *alloc_buddy_huge_page_with_mpol(struct hstate *h, nodemask_t *nodemask; nid = huge_node(vma, addr, gfp_mask, &mpol, &nodemask); +#ifdef CONFIG_NUMA + if (mpol->mode == MPOL_PREFERRED_MANY) { + gfp_t gfp = gfp_mask | __GFP_NOWARN; + + gfp &= ~(__GFP_DIRECT_RECLAIM | __GFP_NOFAIL); + page = alloc_surplus_huge_page(h, gfp, nid, nodemask, false); + if (page) { + mpol_cond_put(mpol); + return page; + } + + /* Fallback to all nodes */ + nodemask = NULL; + } +#endif page = alloc_surplus_huge_page(h, gfp_mask, nid, nodemask, false); mpol_cond_put(mpol);
--
2.14.1