From nobody Wed Feb 11 06:31:19 2026 Received: from out30-99.freemail.mail.aliyun.com (out30-99.freemail.mail.aliyun.com [115.124.30.99]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id EEEC43D97A for ; Wed, 21 Feb 2024 09:28:13 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=115.124.30.99 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1708507696; cv=none; b=dwwe7Lnp4EZI+JxJF1WJchKKcpuM+dVELrIryUtyrnkVyYFkfTrv4lfY9rvXRlPVQjzFcjVRd7A8ADv+9ute+ZbXO4GxfdNn9ohRXPu/lS7sP+ktXwqBaFdkyluSp1P8dXrjGUAFAAXBwJ1AHVMu3Ba/53u4v/xTKImF19I0kAY= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1708507696; c=relaxed/simple; bh=XYc4TdVvWWJquVpA+tfkNKBtu2+qh+Pphd2uWChX58M=; h=From:To:Cc:Subject:Date:Message-Id:In-Reply-To:References: MIME-Version; b=sXHZj6nWkz8WYXC4PnHl3P8LXLkXhkZnOTyOgaUbFUUsSljfK+IQKwaB4Pogf4Hkpdj8TG77ZYn+3YgBvIveZePZFFrnTJtsuqk1kqaQXJ1024Q7VTY/cdJYBN/i6M5ufIDkbimchaIlMo+wut3W8/H0EYB7hRQsHMViIAomXIk= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com; spf=pass smtp.mailfrom=linux.alibaba.com; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b=N8dnYYs4; arc=none smtp.client-ip=115.124.30.99 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b="N8dnYYs4" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.alibaba.com; s=default; t=1708507686; h=From:To:Subject:Date:Message-Id:MIME-Version; bh=ImWD/LFam8uPK216oOawCtCWE3yiiD0csaC4SgFveA8=; b=N8dnYYs4/RxkmL95EU9cddIr52GZltuM1MHI6/tbV4RPoxPXk9YywsZeJpvCq+yb+h+GRpBqbYGz0eOfRhOoOK4DZoYPDa3SlNqtFT0iQgtTOpcc2tV/1WQy7vIwK1QQ+b4NrsPOQ3rv4egKD2VaQ+Kdtn+92jFwoAlKI90t/iU= X-Alimail-AntiSpam: AC=PASS;BC=-1|-1;BR=01201311R951e4;CH=green;DM=||false|;DS=||;FP=0|-1|-1|-1|0|-1|-1|-1;HT=ay29a033018046056;MF=baolin.wang@linux.alibaba.com;NM=1;PH=DS;RN=10;SR=0;TI=SMTPD_---0W0zUu6x_1708507684; Received: from localhost(mailfrom:baolin.wang@linux.alibaba.com fp:SMTPD_---0W0zUu6x_1708507684) by smtp.aliyun-inc.com; Wed, 21 Feb 2024 17:28:05 +0800 From: Baolin Wang To: akpm@linux-foundation.org Cc: muchun.song@linux.dev, osalvador@suse.de, david@redhat.com, linmiaohe@huawei.com, naoya.horiguchi@nec.com, mhocko@kernel.org, baolin.wang@linux.alibaba.com, linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: [RFC PATCH 2/3] mm: hugetlb: make the hugetlb migration strategy consistent Date: Wed, 21 Feb 2024 17:27:54 +0800 Message-Id: <0514e5139b17ecf3cd9e09d86c93e586c56688dc.1708507022.git.baolin.wang@linux.alibaba.com> X-Mailer: git-send-email 2.39.3 In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" As discussed in previous thread [1], there is an inconsistency when handing hugetlb migration. When handling the migration of freed hugetlb, it prevents fallback to other NUMA nodes in alloc_and_dissolve_hugetlb_folio(). However, when dealing with in-use hugetlb, it allows fallback to other NUMA nodes in alloc_hugetlb_folio_nodemask(), which can break the per-node hugetlb pool and might result in unexpected failures when node bound workloads doesn't g= et what is asssumed available. To make hugetlb migration strategy more clear, we should list all the scena= rios of hugetlb migration and analyze whether allocation fallback is permitted: 1) Memory offline: will call dissolve_free_huge_pages() to free the freed h= ugetlb, and call do_migrate_range() to migrate the in-use hugetlb. Both can break t= he per-node hugetlb pool, but as this is an explicit offlining operation, no b= etter choice. So should allow the hugetlb allocation fallback. 2) Memory failure: same as memory offline. Should allow fallback to a diffe= rent node might be the only option to handle it, otherwise the impact of poisoned mem= ory can be amplified. 3) Longterm pinning: will call migrate_longterm_unpinnable_pages() to migra= te in-use and not-longterm-pinnable hugetlb, which can break the per-node pool. But w= e should fail to longterm pinning if can not allocate on current node to avoid break= ing the per-node pool. 4) Syscalls (mbind, migrate_pages, move_pages): these are explicit users op= eration to move pages to other nodes, so fallback to other nodes should not be proh= ibited. 5) alloc_contig_range: used by CMA allocation and virtio-mem fake-offline t= o allocate given range of pages. Now the freed hugetlb migration is not allowed to fal= lback, to keep consistency, the in-use hugetlb migration should be also not allowed t= o fallback. 6) alloc_contig_pages: used by kfence, pgtable_debug etc. The strategy shou= ld be consistent with that of alloc_contig_range(). Based on the analysis of the various scenarios above, determine whether fal= lback is permitted according to the migration reason in alloc_hugetlb_folio_nodemask= (). [1] https://lore.kernel.org/all/6f26ce22d2fcd523418a085f2c588fe0776d46e7.17= 06794035.git.baolin.wang@linux.alibaba.com/ Signed-off-by: Baolin Wang --- include/linux/hugetlb.h | 4 ++-- mm/hugetlb.c | 28 ++++++++++++++++++++++++++-- mm/mempolicy.c | 2 +- mm/migrate.c | 2 +- 4 files changed, 30 insertions(+), 6 deletions(-) diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h index 77b30a8c6076..fa122dc509cf 100644 --- a/include/linux/hugetlb.h +++ b/include/linux/hugetlb.h @@ -747,7 +747,7 @@ int isolate_or_dissolve_huge_page(struct page *page, st= ruct list_head *list); struct folio *alloc_hugetlb_folio(struct vm_area_struct *vma, unsigned long addr, int avoid_reserve); struct folio *alloc_hugetlb_folio_nodemask(struct hstate *h, int preferred= _nid, - nodemask_t *nmask, gfp_t gfp_mask); + nodemask_t *nmask, gfp_t gfp_mask, int reason); int hugetlb_add_to_page_cache(struct folio *folio, struct address_space *m= apping, pgoff_t idx); void restore_reserve_on_error(struct hstate *h, struct vm_area_struct *vma, @@ -1065,7 +1065,7 @@ static inline struct folio *alloc_hugetlb_folio(struc= t vm_area_struct *vma, =20 static inline struct folio * alloc_hugetlb_folio_nodemask(struct hstate *h, int preferred_nid, - nodemask_t *nmask, gfp_t gfp_mask) + nodemask_t *nmask, gfp_t gfp_mask, int reason) { return NULL; } diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 68283e54c899..a55cfc7844bc 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -2621,8 +2621,10 @@ struct folio *alloc_buddy_hugetlb_folio_with_mpol(st= ruct hstate *h, =20 /* folio migration callback function */ struct folio *alloc_hugetlb_folio_nodemask(struct hstate *h, int preferred= _nid, - nodemask_t *nmask, gfp_t gfp_mask) + nodemask_t *nmask, gfp_t gfp_mask, int reason) { + bool allowed_fallback =3D false; + spin_lock_irq(&hugetlb_lock); if (available_huge_pages(h)) { struct folio *folio; @@ -2636,6 +2638,28 @@ struct folio *alloc_hugetlb_folio_nodemask(struct hs= tate *h, int preferred_nid, } spin_unlock_irq(&hugetlb_lock); =20 + if (gfp_mask & __GFP_THISNODE) + goto alloc_new; + + /* + * Note: the memory offline, memory failure and migration syscalls can br= eak + * the per-node hugetlb pool. Other cases can not allocate new hugetlb on + * other nodes. + */ + switch (reason) { + case MR_MEMORY_HOTPLUG: + case MR_MEMORY_FAILURE: + case MR_SYSCALL: + case MR_MEMPOLICY_MBIND: + allowed_fallback =3D true; + break; + default: + break; + } + + if (!allowed_fallback) + gfp_mask |=3D __GFP_THISNODE; +alloc_new: return alloc_migrate_hugetlb_folio(h, gfp_mask, preferred_nid, nmask); } =20 @@ -6666,7 +6690,7 @@ static struct folio *alloc_hugetlb_folio_vma(struct h= state *h, =20 gfp_mask =3D htlb_alloc_mask(h); node =3D huge_node(vma, address, gfp_mask, &mpol, &nodemask); - folio =3D alloc_hugetlb_folio_nodemask(h, node, nodemask, gfp_mask); + folio =3D alloc_hugetlb_folio_nodemask(h, node, nodemask, gfp_mask, -1); mpol_cond_put(mpol); =20 return folio; diff --git a/mm/mempolicy.c b/mm/mempolicy.c index 98ceb12e0e17..436e817eeaeb 100644 --- a/mm/mempolicy.c +++ b/mm/mempolicy.c @@ -1228,7 +1228,7 @@ static struct folio *alloc_migration_target_by_mpol(s= truct folio *src, h =3D folio_hstate(src); gfp =3D htlb_alloc_mask(h); nodemask =3D policy_nodemask(gfp, pol, ilx, &nid); - return alloc_hugetlb_folio_nodemask(h, nid, nodemask, gfp); + return alloc_hugetlb_folio_nodemask(h, nid, nodemask, gfp, MR_MEMPOLICY_= MBIND); } =20 if (folio_test_large(src)) diff --git a/mm/migrate.c b/mm/migrate.c index bde63010a3cf..0c2b70800da3 100644 --- a/mm/migrate.c +++ b/mm/migrate.c @@ -2022,7 +2022,7 @@ struct folio *alloc_migration_target(struct folio *sr= c, unsigned long private) =20 gfp_mask =3D htlb_modify_alloc_mask(h, gfp_mask); return alloc_hugetlb_folio_nodemask(h, nid, - mtc->nmask, gfp_mask); + mtc->nmask, gfp_mask, mtc->reason); } =20 if (folio_test_large(src)) { --=20 2.39.3