From nobody Fri Dec 19 07:55:05 2025 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 5475BC10F04 for ; Wed, 6 Dec 2023 09:46:45 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1377287AbjLFJqf (ORCPT ); Wed, 6 Dec 2023 04:46:35 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:57134 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1377267AbjLFJqb (ORCPT ); Wed, 6 Dec 2023 04:46:31 -0500 Received: from out-178.mta1.migadu.com (out-178.mta1.migadu.com [95.215.58.178]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 4CA4EB9 for ; Wed, 6 Dec 2023 01:46:37 -0800 (PST) X-Report-Abuse: Please report any abuse attempt to abuse@migadu.com and include these headers. From: Chengming Zhou Date: Wed, 06 Dec 2023 09:46:25 +0000 Subject: [PATCH 2/7] mm/zswap: split zswap rb-tree MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable Message-Id: <20231206-zswap-lock-optimize-v1-2-e25b059f9c3a@bytedance.com> References: <20231206-zswap-lock-optimize-v1-0-e25b059f9c3a@bytedance.com> In-Reply-To: <20231206-zswap-lock-optimize-v1-0-e25b059f9c3a@bytedance.com> To: Vitaly Wool , Nhat Pham , Johannes Weiner , Michal Hocko , Seth Jennings , Dan Streetman , Andrew Morton , Yosry Ahmed Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org, Chengming Zhou X-Developer-Signature: v=1; a=ed25519-sha256; t=1701855988; l=6878; i=zhouchengming@bytedance.com; s=20231204; h=from:subject:message-id; bh=91S6pF6My/bUROdLpQALha8Vf8LlqYuhabIDQmuzuWI=; b=wPMG7SuyQz02Gmf6/zoPkNXypnlp9pzAs3LaGP2c2Co5uxHZeHfDwL1yFMzUvPUEcTV7ChSyQ ajsHFI83VQLCQGcg29JkEtY4nc6utk5FaXQG6gU7Y5SrzXyxOn7RFyY X-Developer-Key: i=zhouchengming@bytedance.com; a=ed25519; pk=xFTmRtMG3vELGJBUiml7OYNdM393WOMv0iWWeQEVVdA= X-Migadu-Flow: FLOW_OUT Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Each swapfile has one rb-tree to search the mapping of swp_entry_t to zswap_entry, that use a spinlock to protect, which can cause heavy lock contention if multiple tasks zswap_store/load concurrently. Optimize the scalability problem by splitting the zswap rb-tree into multiple rb-trees, each corresponds to SWAP_ADDRESS_SPACE_PAGES (64M), just like we did in the swap cache address_space splitting. Signed-off-by: Chengming Zhou --- include/linux/zswap.h | 4 +-- mm/swapfile.c | 2 +- mm/zswap.c | 69 ++++++++++++++++++++++++++++++++---------------= ---- 3 files changed, 47 insertions(+), 28 deletions(-) diff --git a/include/linux/zswap.h b/include/linux/zswap.h index 7cccc02cb9e9..d3a8bc300b70 100644 --- a/include/linux/zswap.h +++ b/include/linux/zswap.h @@ -30,7 +30,7 @@ struct zswap_lruvec_state { bool zswap_store(struct folio *folio); bool zswap_load(struct folio *folio); void zswap_invalidate(int type, pgoff_t offset); -int zswap_swapon(int type); +int zswap_swapon(int type, unsigned long nr_pages); void zswap_swapoff(int type); void zswap_memcg_offline_cleanup(struct mem_cgroup *memcg); void zswap_lruvec_state_init(struct lruvec *lruvec); @@ -50,7 +50,7 @@ static inline bool zswap_load(struct folio *folio) } =20 static inline void zswap_invalidate(int type, pgoff_t offset) {} -static inline int zswap_swapon(int type) {} +static inline int zswap_swapon(int type, unsigned long nr_pages) {} static inline void zswap_swapoff(int type) {} static inline void zswap_memcg_offline_cleanup(struct mem_cgroup *memcg) {} static inline void zswap_lruvec_state_init(struct lruvec *lruvec) {} diff --git a/mm/swapfile.c b/mm/swapfile.c index 939e7590feda..da8367a3e076 100644 --- a/mm/swapfile.c +++ b/mm/swapfile.c @@ -3163,7 +3163,7 @@ SYSCALL_DEFINE2(swapon, const char __user *, specialf= ile, int, swap_flags) if (error) goto bad_swap_unlock_inode; =20 - error =3D zswap_swapon(p->type); + error =3D zswap_swapon(p->type, maxpages); if (error) goto free_swap_address_space; =20 diff --git a/mm/zswap.c b/mm/zswap.c index 5e2b8d5ee33b..a6b4859a0164 100644 --- a/mm/zswap.c +++ b/mm/zswap.c @@ -234,6 +234,7 @@ struct zswap_tree { }; =20 static struct zswap_tree *zswap_trees[MAX_SWAPFILES]; +static unsigned int nr_zswap_trees[MAX_SWAPFILES]; =20 /* RCU-protected iteration */ static LIST_HEAD(zswap_pools); @@ -260,6 +261,10 @@ static bool zswap_has_pool; * helpers and fwd declarations **********************************/ =20 +#define swap_zswap_tree(entry) \ + (&zswap_trees[swp_type(entry)][swp_offset(entry) \ + >> SWAP_ADDRESS_SPACE_SHIFT]) + #define zswap_pool_debug(msg, p) \ pr_debug("%s pool %s/%s\n", msg, (p)->tfm_name, \ zpool_get_type((p)->zpools[0])) @@ -885,7 +890,7 @@ static enum lru_status shrink_memcg_cb(struct list_head= *item, struct list_lru_o * until the entry is verified to still be alive in the tree. */ swpoffset =3D swp_offset(entry->swpentry); - tree =3D zswap_trees[swp_type(entry->swpentry)]; + tree =3D swap_zswap_tree(entry->swpentry); list_lru_isolate(l, item); /* * It's safe to drop the lock here because we return either @@ -1535,10 +1540,9 @@ static void zswap_fill_page(void *ptr, unsigned long= value) bool zswap_store(struct folio *folio) { swp_entry_t swp =3D folio->swap; - int type =3D swp_type(swp); pgoff_t offset =3D swp_offset(swp); struct page *page =3D &folio->page; - struct zswap_tree *tree =3D zswap_trees[type]; + struct zswap_tree *tree =3D swap_zswap_tree(swp); struct zswap_entry *entry, *dupentry; struct scatterlist input, output; struct crypto_acomp_ctx *acomp_ctx; @@ -1610,7 +1614,7 @@ bool zswap_store(struct folio *folio) src =3D kmap_local_page(page); if (zswap_is_page_same_filled(src, &value)) { kunmap_local(src); - entry->swpentry =3D swp_entry(type, offset); + entry->swpentry =3D swp; entry->length =3D 0; entry->value =3D value; atomic_inc(&zswap_same_filled_pages); @@ -1688,7 +1692,7 @@ bool zswap_store(struct folio *folio) mutex_unlock(acomp_ctx->mutex); =20 /* populate entry */ - entry->swpentry =3D swp_entry(type, offset); + entry->swpentry =3D swp; entry->handle =3D handle; entry->length =3D dlen; =20 @@ -1748,10 +1752,9 @@ bool zswap_store(struct folio *folio) bool zswap_load(struct folio *folio) { swp_entry_t swp =3D folio->swap; - int type =3D swp_type(swp); pgoff_t offset =3D swp_offset(swp); struct page *page =3D &folio->page; - struct zswap_tree *tree =3D zswap_trees[type]; + struct zswap_tree *tree =3D swap_zswap_tree(swp); struct zswap_entry *entry; struct scatterlist input, output; struct crypto_acomp_ctx *acomp_ctx; @@ -1835,7 +1838,7 @@ bool zswap_load(struct folio *folio) =20 void zswap_invalidate(int type, pgoff_t offset) { - struct zswap_tree *tree =3D zswap_trees[type]; + struct zswap_tree *tree =3D swap_zswap_tree(swp_entry(type, offset)); struct zswap_entry *entry; =20 /* find */ @@ -1850,37 +1853,53 @@ void zswap_invalidate(int type, pgoff_t offset) spin_unlock(&tree->lock); } =20 -int zswap_swapon(int type) +int zswap_swapon(int type, unsigned long nr_pages) { - struct zswap_tree *tree; + struct zswap_tree *trees, *tree; + unsigned int nr, i; =20 - tree =3D kzalloc(sizeof(*tree), GFP_KERNEL); - if (!tree) { + nr =3D DIV_ROUND_UP(nr_pages, SWAP_ADDRESS_SPACE_PAGES); + trees =3D kvcalloc(nr, sizeof(*tree), GFP_KERNEL); + if (!trees) { pr_err("alloc failed, zswap disabled for swap type %d\n", type); return -ENOMEM; } =20 - tree->rbroot =3D RB_ROOT; - spin_lock_init(&tree->lock); - zswap_trees[type] =3D tree; + for (i =3D 0; i < nr; i++) { + tree =3D trees + i; + tree->rbroot =3D RB_ROOT; + spin_lock_init(&tree->lock); + } + + nr_zswap_trees[type] =3D nr; + zswap_trees[type] =3D trees; return 0; } =20 void zswap_swapoff(int type) { - struct zswap_tree *tree =3D zswap_trees[type]; - struct zswap_entry *entry, *n; + struct zswap_tree *trees =3D zswap_trees[type]; + unsigned int i; =20 - if (!tree) + if (!trees) return; =20 - /* walk the tree and free everything */ - spin_lock(&tree->lock); - rbtree_postorder_for_each_entry_safe(entry, n, &tree->rbroot, rbnode) - zswap_free_entry(entry); - tree->rbroot =3D RB_ROOT; - spin_unlock(&tree->lock); - kfree(tree); + for (i =3D 0; i < nr_zswap_trees[type]; i++) { + struct zswap_tree *tree =3D trees + i; + struct zswap_entry *entry, *n; + + /* walk the tree and free everything */ + spin_lock(&tree->lock); + rbtree_postorder_for_each_entry_safe(entry, n, + &tree->rbroot, + rbnode) + zswap_free_entry(entry); + tree->rbroot =3D RB_ROOT; + spin_unlock(&tree->lock); + } + + kvfree(trees); + nr_zswap_trees[type] =3D 0; zswap_trees[type] =3D NULL; } =20 --=20 b4 0.10.1