[PATCH] mempolicy: alloc_pages_mpol() for NUMA policy without vma: fix

From: Hugh Dickins <hughd@google.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>,
	domenico cerasuolo <mimmocerasuolo@gmail.com>,
	Andi Kleen <ak@linux.intel.com>, Christoph Lameter <cl@linux.co>,
	Matthew Wilcox <willy@infradead.org>,
	Mike Kravetz <mike.kravetz@oracle.com>,
	David Hildenbrand <david@redhat.com>,
	Suren Baghdasaryan <surenb@google.com>,
	Yang Shi <shy828301@gmail.com>,
	Sidhartha Kumar <sidhartha.kumar@oracle.com>,
	Vishal Moola <vishal.moola@gmamil.com>,
	Kefeng Wang <wangkefeng.wang@huawei.com>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Tejun Heo <tj@kernel.org>,
	Mel Gorman <mgorman@techsingularity.net>,
	Michal Hocko <mhocko@suse.com>,
	"Huang, Ying" <ying.huang@intel.com>,
	Nhat Pham <nphamcs@gmail.com>,
	Yosry Ahmed <yosryahmed@google.com>,
	linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: [PATCH] mempolicy: alloc_pages_mpol() for NUMA policy without vma: fix
Date: Mon, 23 Oct 2023 23:44:18 -0700 (PDT)	[thread overview]
Message-ID: <00dc4f56-e623-7c85-29ea-4211e93063f6@google.com> (raw)
In-Reply-To: <20231023190555.GA1964810@cmpxchg.org>

mm-unstable commit 48a7bd12d57f ("mempolicy: alloc_pages_mpol() for NUMA
policy without vma") ended read_swap_cache_async() supporting NULL vma -
okay; but missed the NULL mpol being passed to __read_swap_cache_async()
by zswap_writeback_entry() - oops!

Since its other callers all give good mpol, add get_task_policy(current)
there in mm/zswap.c, to produce the same good-enough behaviour as before
(and task policy, acted on in current task, does not require the refcount
to be dup'ed).

But if that policy is (quite reasonably) MPOL_INTERLEAVE, then ilx must
be NO_INTERLEAVE_INDEX rather than 0, to provide the same distribution
as before: move that definition from mempolicy.c to mempolicy.h.

Reported-by: Domenico Cerasuolo <mimmocerasuolo@gmail.com>
Closes: https://lore.kernel.org/linux-mm/74e34633-6060-f5e3-aee-7040d43f2e93@google.com/T/#mf08c877d1884fc7867f9e328cdf02257ff3b3ae9
Suggested-by: Johannes Weiner <hannes@cmpxchg.org>
Fixes: 48a7bd12d57f ("mempolicy: alloc_pages_mpol() for NUMA policy without vma")
Signed-off-by: Hugh Dickins <hughd@google.com>
---
 include/linux/mempolicy.h | 2 ++
 mm/mempolicy.c            | 2 --
 mm/zswap.c                | 7 +++++--
 3 files changed, 7 insertions(+), 4 deletions(-)

diff --git a/include/linux/mempolicy.h b/include/linux/mempolicy.h
index 2801d5b0a4e9..dd9ed2ce7fd5 100644
--- a/include/linux/mempolicy.h
+++ b/include/linux/mempolicy.h
@@ -17,6 +17,8 @@
 
 struct mm_struct;
 
+#define NO_INTERLEAVE_INDEX (-1UL)	/* use task il_prev for interleaving */
+
 #ifdef CONFIG_NUMA
 
 /*
diff --git a/mm/mempolicy.c b/mm/mempolicy.c
index 898ee2e3c85b..989293180eb6 100644
--- a/mm/mempolicy.c
+++ b/mm/mempolicy.c
@@ -114,8 +114,6 @@
 #define MPOL_MF_INVERT       (MPOL_MF_INTERNAL << 1)	/* Invert check for nodemask */
 #define MPOL_MF_WRLOCK       (MPOL_MF_INTERNAL << 2)	/* Write-lock walked vmas */
 
-#define NO_INTERLEAVE_INDEX (-1UL)
-
 static struct kmem_cache *policy_cache;
 static struct kmem_cache *sn_cache;
 
diff --git a/mm/zswap.c b/mm/zswap.c
index 37d2b1cb2ecb..060857adca76 100644
--- a/mm/zswap.c
+++ b/mm/zswap.c
@@ -24,6 +24,7 @@
 #include <linux/swap.h>
 #include <linux/crypto.h>
 #include <linux/scatterlist.h>
+#include <linux/mempolicy.h>
 #include <linux/mempool.h>
 #include <linux/zpool.h>
 #include <crypto/acompress.h>
@@ -1057,6 +1058,7 @@ static int zswap_writeback_entry(struct zswap_entry *entry,
 {
 	swp_entry_t swpentry = entry->swpentry;
 	struct page *page;
+	struct mempolicy *mpol;
 	struct scatterlist input, output;
 	struct crypto_acomp_ctx *acomp_ctx;
 	struct zpool *pool = zswap_find_zpool(entry);
@@ -1075,8 +1077,9 @@ static int zswap_writeback_entry(struct zswap_entry *entry,
 	}
 
 	/* try to allocate swap cache page */
-	page = __read_swap_cache_async(swpentry, GFP_KERNEL, NULL, 0,
-				       &page_was_allocated);
+	mpol = get_task_policy(current);
+	page = __read_swap_cache_async(swpentry, GFP_KERNEL, mpol,
+				NO_INTERLEAVE_INDEX, &page_was_allocated);
 	if (!page) {
 		ret = -ENOMEM;
 		goto fail;
-- 
2.35.3

