Merge tag 'erofs-for-5.18-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs

Pull erofs updates from Gao Xiang:
 "In this cycle, we continue converting to use meta buffers for all
  remaining uncompressed paths to prepare for the upcoming subpage,
  folio and fscache features.

  We also fixed a double-free issue when sysfs initialization fails,
  which was reported by syzbot.

  Besides, in order for the userspace to control per-file timestamp
  easier, we now switch to record mtime instead of ctime with a
  compatible feature marked. And there are also some code cleanups and
  documentation update as usual.

  Summary:

   - Avoid using page structure directly for all uncompressed paths

   - Fix a double-free issue when sysfs initialization fails

   - Complete DAX description for erofs

   - Use mtime instead since there's no (easy) way for users to control
     ctime

   - Several code cleanups"

* tag 'erofs-for-5.18-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
  erofs: rename ctime to mtime
  erofs: use meta buffers for inode lookup
  erofs: use meta buffers for reading directories
  fs: erofs: add sanity check for kobject in erofs_unregister_sysfs
  erofs: refine managed inode stuffs
  erofs: clean up z_erofs_extent_lookback
  erofs: silence warnings related to impossible m_plen
  Documentation/filesystem/dax: update DAX description on erofs
  erofs: clean up preload_compressed_pages()
  erofs: get rid of `struct z_erofs_collector'
  erofs: use meta buffers for erofs_read_superblock()

10 files changed, 185 insertions, 197 deletions

diff --git a/fs/erofs/data.c b/fs/erofs/data.c
index 226a57c57ee6..780db1e5f4b7 100644
--- a/fs/erofs/data.c
+++ b/fs/erofs/data.c
@@ -28,10 +28,10 @@ void erofs_put_metabuf(struct erofs_buf *buf)
 	buf->page = NULL;
 }
 
-void *erofs_read_metabuf(struct erofs_buf *buf, struct super_block *sb,
-			erofs_blk_t blkaddr, enum erofs_kmap_type type)
+void *erofs_bread(struct erofs_buf *buf, struct inode *inode,
+		  erofs_blk_t blkaddr, enum erofs_kmap_type type)
 {
-	struct address_space *const mapping = sb->s_bdev->bd_inode->i_mapping;
+	struct address_space *const mapping = inode->i_mapping;
 	erofs_off_t offset = blknr_to_addr(blkaddr);
 	pgoff_t index = offset >> PAGE_SHIFT;
 	struct page *page = buf->page;
@@ -60,6 +60,12 @@ void *erofs_read_metabuf(struct erofs_buf *buf, struct super_block *sb,
 	return buf->base + (offset & ~PAGE_MASK);
 }
 
+void *erofs_read_metabuf(struct erofs_buf *buf, struct super_block *sb,
+			 erofs_blk_t blkaddr, enum erofs_kmap_type type)
+{
+	return erofs_bread(buf, sb->s_bdev->bd_inode, blkaddr, type);
+}
+
 static int erofs_map_blocks_flatmode(struct inode *inode,
 				     struct erofs_map_blocks *map,
 				     int flags)
diff --git a/fs/erofs/dir.c b/fs/erofs/dir.c
index eee9b0b31b63..18e59821c597 100644
--- a/fs/erofs/dir.c
+++ b/fs/erofs/dir.c
@@ -2,6 +2,7 @@
 /*
  * Copyright (C) 2017-2018 HUAWEI, Inc.
  *             https://www.huawei.com/
+ * Copyright (C) 2022, Alibaba Cloud
  */
 #include "internal.h"
 
@@ -67,7 +68,7 @@ static int erofs_fill_dentries(struct inode *dir, struct dir_context *ctx,
 static int erofs_readdir(struct file *f, struct dir_context *ctx)
 {
 	struct inode *dir = file_inode(f);
-	struct address_space *mapping = dir->i_mapping;
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
 	const size_t dirsize = i_size_read(dir);
 	unsigned int i = ctx->pos / EROFS_BLKSIZ;
 	unsigned int ofs = ctx->pos % EROFS_BLKSIZ;
@@ -75,26 +76,19 @@ static int erofs_readdir(struct file *f, struct dir_context *ctx)
 	bool initial = true;
 
 	while (ctx->pos < dirsize) {
-		struct page *dentry_page;
 		struct erofs_dirent *de;
 		unsigned int nameoff, maxsize;
 
-		dentry_page = read_mapping_page(mapping, i, NULL);
-		if (dentry_page == ERR_PTR(-ENOMEM)) {
-			err = -ENOMEM;
-			break;
-		} else if (IS_ERR(dentry_page)) {
+		de = erofs_bread(&buf, dir, i, EROFS_KMAP);
+		if (IS_ERR(de)) {
 			erofs_err(dir->i_sb,
 				  "fail to readdir of logical block %u of nid %llu",
 				  i, EROFS_I(dir)->nid);
-			err = -EFSCORRUPTED;
+			err = PTR_ERR(de);
 			break;
 		}
 
-		de = (struct erofs_dirent *)kmap(dentry_page);
-
 		nameoff = le16_to_cpu(de->nameoff);
-
 		if (nameoff < sizeof(struct erofs_dirent) ||
 		    nameoff >= PAGE_SIZE) {
 			erofs_err(dir->i_sb,
@@ -119,10 +113,6 @@ static int erofs_readdir(struct file *f, struct dir_context *ctx)
 		err = erofs_fill_dentries(dir, ctx, de, &ofs,
 					  nameoff, maxsize);
 skip_this:
-		kunmap(dentry_page);
-
-		put_page(dentry_page);
-
 		ctx->pos = blknr_to_addr(i) + ofs;
 
 		if (err)
@@ -130,6 +120,7 @@ skip_this:
 		++i;
 		ofs = 0;
 	}
+	erofs_put_metabuf(&buf);
 	return err < 0 ? err : 0;
 }
 
diff --git a/fs/erofs/erofs_fs.h b/fs/erofs/erofs_fs.h
index 3ea62c6fb00a..1238ca104f09 100644
--- a/fs/erofs/erofs_fs.h
+++ b/fs/erofs/erofs_fs.h
@@ -12,6 +12,7 @@
 #define EROFS_SUPER_OFFSET      1024
 
 #define EROFS_FEATURE_COMPAT_SB_CHKSUM          0x00000001
+#define EROFS_FEATURE_COMPAT_MTIME              0x00000002
 
 /*
  * Any bits that aren't in EROFS_ALL_FEATURE_INCOMPAT should
@@ -186,8 +187,8 @@ struct erofs_inode_extended {
 
 	__le32 i_uid;
 	__le32 i_gid;
-	__le64 i_ctime;
-	__le32 i_ctime_nsec;
+	__le64 i_mtime;
+	__le32 i_mtime_nsec;
 	__le32 i_nlink;
 	__u8   i_reserved2[16];
 };
diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
index ff62f84f47d3..e8b37ba5e9ad 100644
--- a/fs/erofs/inode.c
+++ b/fs/erofs/inode.c
@@ -113,8 +113,8 @@ static void *erofs_read_inode(struct erofs_buf *buf,
 		set_nlink(inode, le32_to_cpu(die->i_nlink));
 
 		/* extended inode has its own timestamp */
-		inode->i_ctime.tv_sec = le64_to_cpu(die->i_ctime);
-		inode->i_ctime.tv_nsec = le32_to_cpu(die->i_ctime_nsec);
+		inode->i_ctime.tv_sec = le64_to_cpu(die->i_mtime);
+		inode->i_ctime.tv_nsec = le32_to_cpu(die->i_mtime_nsec);
 
 		inode->i_size = le64_to_cpu(die->i_size);
 
diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
index 5aa2cf2c2f80..5298c4ee277d 100644
--- a/fs/erofs/internal.h
+++ b/fs/erofs/internal.h
@@ -479,6 +479,8 @@ struct erofs_map_dev {
 extern const struct file_operations erofs_file_fops;
 void erofs_unmap_metabuf(struct erofs_buf *buf);
 void erofs_put_metabuf(struct erofs_buf *buf);
+void *erofs_bread(struct erofs_buf *buf, struct inode *inode,
+		  erofs_blk_t blkaddr, enum erofs_kmap_type type);
 void *erofs_read_metabuf(struct erofs_buf *buf, struct super_block *sb,
 			 erofs_blk_t blkaddr, enum erofs_kmap_type type);
 int erofs_map_dev(struct super_block *sb, struct erofs_map_dev *dev);
diff --git a/fs/erofs/namei.c b/fs/erofs/namei.c
index 8629e616028c..554efa363317 100644
--- a/fs/erofs/namei.c
+++ b/fs/erofs/namei.c
@@ -2,6 +2,7 @@
 /*
  * Copyright (C) 2017-2018 HUAWEI, Inc.
  *             https://www.huawei.com/
+ * Copyright (C) 2022, Alibaba Cloud
  */
 #include "xattr.h"
 
@@ -86,14 +87,14 @@ static struct erofs_dirent *find_target_dirent(struct erofs_qstr *name,
 	return ERR_PTR(-ENOENT);
 }
 
-static struct page *find_target_block_classic(struct inode *dir,
-					      struct erofs_qstr *name,
-					      int *_ndirents)
+static void *find_target_block_classic(struct erofs_buf *target,
+				       struct inode *dir,
+				       struct erofs_qstr *name,
+				       int *_ndirents)
 {
 	unsigned int startprfx, endprfx;
 	int head, back;
-	struct address_space *const mapping = dir->i_mapping;
-	struct page *candidate = ERR_PTR(-ENOENT);
+	void *candidate = ERR_PTR(-ENOENT);
 
 	startprfx = endprfx = 0;
 	head = 0;
@@ -101,10 +102,11 @@ static struct page *find_target_block_classic(struct inode *dir,
 
 	while (head <= back) {
 		const int mid = head + (back - head) / 2;
-		struct page *page = read_mapping_page(mapping, mid, NULL);
+		struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+		struct erofs_dirent *de;
 
-		if (!IS_ERR(page)) {
-			struct erofs_dirent *de = kmap_atomic(page);
+		de = erofs_bread(&buf, dir, mid, EROFS_KMAP);
+		if (!IS_ERR(de)) {
 			const int nameoff = nameoff_from_disk(de->nameoff,
 							      EROFS_BLKSIZ);
 			const int ndirents = nameoff / sizeof(*de);
@@ -113,13 +115,12 @@ static struct page *find_target_block_classic(struct inode *dir,
 			struct erofs_qstr dname;
 
 			if (!ndirents) {
-				kunmap_atomic(de);
-				put_page(page);
+				erofs_put_metabuf(&buf);
 				erofs_err(dir->i_sb,
 					  "corrupted dir block %d @ nid %llu",
 					  mid, EROFS_I(dir)->nid);
 				DBG_BUGON(1);
-				page = ERR_PTR(-EFSCORRUPTED);
+				de = ERR_PTR(-EFSCORRUPTED);
 				goto out;
 			}
 
@@ -135,7 +136,6 @@ static struct page *find_target_block_classic(struct inode *dir,
 
 			/* string comparison without already matched prefix */
 			diff = erofs_dirnamecmp(name, &dname, &matched);
-			kunmap_atomic(de);
 
 			if (!diff) {
 				*_ndirents = 0;
@@ -145,11 +145,12 @@ static struct page *find_target_block_classic(struct inode *dir,
 				startprfx = matched;
 
 				if (!IS_ERR(candidate))
-					put_page(candidate);
-				candidate = page;
+					erofs_put_metabuf(target);
+				*target = buf;
+				candidate = de;
 				*_ndirents = ndirents;
 			} else {
-				put_page(page);
+				erofs_put_metabuf(&buf);
 
 				back = mid - 1;
 				endprfx = matched;
@@ -158,8 +159,8 @@ static struct page *find_target_block_classic(struct inode *dir,
 		}
 out:		/* free if the candidate is valid */
 		if (!IS_ERR(candidate))
-			put_page(candidate);
-		return page;
+			erofs_put_metabuf(target);
+		return de;
 	}
 	return candidate;
 }
@@ -169,8 +170,7 @@ int erofs_namei(struct inode *dir,
 		erofs_nid_t *nid, unsigned int *d_type)
 {
 	int ndirents;
-	struct page *page;
-	void *data;
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
 	struct erofs_dirent *de;
 	struct erofs_qstr qn;
 
@@ -181,26 +181,20 @@ int erofs_namei(struct inode *dir,
 	qn.end = name->name + name->len;
 
 	ndirents = 0;
-	page = find_target_block_classic(dir, &qn, &ndirents);
 
-	if (IS_ERR(page))
-		return PTR_ERR(page);
+	de = find_target_block_classic(&buf, dir, &qn, &ndirents);
+	if (IS_ERR(de))
+		return PTR_ERR(de);
 
-	data = kmap_atomic(page);
 	/* the target page has been mapped */
 	if (ndirents)
-		de = find_target_dirent(&qn, data, EROFS_BLKSIZ, ndirents);
-	else
-		de = (struct erofs_dirent *)data;
+		de = find_target_dirent(&qn, (u8 *)de, EROFS_BLKSIZ, ndirents);
 
 	if (!IS_ERR(de)) {
 		*nid = le64_to_cpu(de->nid);
 		*d_type = de->file_type;
 	}
-
-	kunmap_atomic(data);
-	put_page(page);
-
+	erofs_put_metabuf(&buf);
 	return PTR_ERR_OR_ZERO(de);
 }
 
diff --git a/fs/erofs/super.c b/fs/erofs/super.c
index 915eefe0d7e2..e178100c162a 100644
--- a/fs/erofs/super.c
+++ b/fs/erofs/super.c
@@ -281,21 +281,19 @@ static int erofs_init_devices(struct super_block *sb,
 static int erofs_read_superblock(struct super_block *sb)
 {
 	struct erofs_sb_info *sbi;
-	struct page *page;
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
 	struct erofs_super_block *dsb;
 	unsigned int blkszbits;
 	void *data;
 	int ret;
 
-	page = read_mapping_page(sb->s_bdev->bd_inode->i_mapping, 0, NULL);
-	if (IS_ERR(page)) {
+	data = erofs_read_metabuf(&buf, sb, 0, EROFS_KMAP);
+	if (IS_ERR(data)) {
 		erofs_err(sb, "cannot read erofs superblock");
-		return PTR_ERR(page);
+		return PTR_ERR(data);
 	}
 
 	sbi = EROFS_SB(sb);
-
-	data = kmap(page);
 	dsb = (struct erofs_super_block *)(data + EROFS_SUPER_OFFSET);
 
 	ret = -EINVAL;
@@ -365,8 +363,7 @@ static int erofs_read_superblock(struct super_block *sb)
 	if (erofs_sb_has_ztailpacking(sbi))
 		erofs_info(sb, "EXPERIMENTAL compressed inline data feature in use. Use at your own risk!");
 out:
-	kunmap(page);
-	put_page(page);
+	erofs_put_metabuf(&buf);
 	return ret;
 }
 
@@ -535,6 +532,11 @@ static int erofs_managed_cache_releasepage(struct page *page, gfp_t gfp_mask)
 	return ret;
 }
 
+/*
+ * It will be called only on inode eviction. In case that there are still some
+ * decompression requests in progress, wait with rescheduling for a bit here.
+ * We could introduce an extra locking instead but it seems unnecessary.
+ */
 static void erofs_managed_cache_invalidatepage(struct page *page,
 					       unsigned int offset,
 					       unsigned int length)
@@ -568,8 +570,7 @@ static int erofs_init_managed_cache(struct super_block *sb)
 	inode->i_size = OFFSET_MAX;
 
 	inode->i_mapping->a_ops = &managed_cache_aops;
-	mapping_set_gfp_mask(inode->i_mapping,
-			     GFP_NOFS | __GFP_HIGHMEM | __GFP_MOVABLE);
+	mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
 	sbi->managed_cache = inode;
 	return 0;
 }
diff --git a/fs/erofs/sysfs.c b/fs/erofs/sysfs.c
index dac252bc9228..f3babf1e6608 100644
--- a/fs/erofs/sysfs.c
+++ b/fs/erofs/sysfs.c
@@ -221,9 +221,11 @@ void erofs_unregister_sysfs(struct super_block *sb)
 {
 	struct erofs_sb_info *sbi = EROFS_SB(sb);
 
-	kobject_del(&sbi->s_kobj);
-	kobject_put(&sbi->s_kobj);
-	wait_for_completion(&sbi->s_kobj_unregister);
+	if (sbi->s_kobj.state_in_sysfs) {
+		kobject_del(&sbi->s_kobj);
+		kobject_put(&sbi->s_kobj);
+		wait_for_completion(&sbi->s_kobj_unregister);
+	}
 }
 
 int __init erofs_init_sysfs(void)
diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index 27d42ffdafd2..0ed880f42525 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -192,7 +192,10 @@ enum z_erofs_collectmode {
 	COLLECT_PRIMARY_FOLLOWED,
 };
 
-struct z_erofs_collector {
+struct z_erofs_decompress_frontend {
+	struct inode *const inode;
+	struct erofs_map_blocks map;
+
 	struct z_erofs_pagevec_ctor vector;
 
 	struct z_erofs_pcluster *pcl, *tailpcl;
@@ -202,13 +205,6 @@ struct z_erofs_collector {
 	z_erofs_next_pcluster_t owned_head;
 
 	enum z_erofs_collectmode mode;
-};
-
-struct z_erofs_decompress_frontend {
-	struct inode *const inode;
-
-	struct z_erofs_collector clt;
-	struct erofs_map_blocks map;
 
 	bool readahead;
 	/* used for applying cache strategy on the fly */
@@ -216,30 +212,30 @@ struct z_erofs_decompress_frontend {
 	erofs_off_t headoffset;
 };
 
-#define COLLECTOR_INIT() { \
-	.owned_head = Z_EROFS_PCLUSTER_TAIL, \
-	.mode = COLLECT_PRIMARY_FOLLOWED }
-
 #define DECOMPRESS_FRONTEND_INIT(__i) { \
-	.inode = __i, .clt = COLLECTOR_INIT(), \
-	.backmost = true, }
+	.inode = __i, .owned_head = Z_EROFS_PCLUSTER_TAIL, \
+	.mode = COLLECT_PRIMARY_FOLLOWED }
 
 static struct page *z_pagemap_global[Z_EROFS_VMAP_GLOBAL_PAGES];
 static DEFINE_MUTEX(z_pagemap_global_lock);
 
-static void preload_compressed_pages(struct z_erofs_collector *clt,
-				     struct address_space *mc,
-				     enum z_erofs_cache_alloctype type,
-				     struct page **pagepool)
+static void z_erofs_bind_cache(struct z_erofs_decompress_frontend *fe,
+			       enum z_erofs_cache_alloctype type,
+			       struct page **pagepool)
 {
-	struct z_erofs_pcluster *pcl = clt->pcl;
+	struct address_space *mc = MNGD_MAPPING(EROFS_I_SB(fe->inode));
+	struct z_erofs_pcluster *pcl = fe->pcl;
 	bool standalone = true;
+	/*
+	 * optimistic allocation without direct reclaim since inplace I/O
+	 * can be used if low memory otherwise.
+	 */
 	gfp_t gfp = (mapping_gfp_mask(mc) & ~__GFP_DIRECT_RECLAIM) |
 			__GFP_NOMEMALLOC | __GFP_NORETRY | __GFP_NOWARN;
 	struct page **pages;
 	pgoff_t index;
 
-	if (clt->mode < COLLECT_PRIMARY_FOLLOWED)
+	if (fe->mode < COLLECT_PRIMARY_FOLLOWED)
 		return;
 
 	pages = pcl->compressed_pages;
@@ -288,7 +284,7 @@ static void preload_compressed_pages(struct z_erofs_collector *clt,
 	 * managed cache since it can be moved to the bypass queue instead.
 	 */
 	if (standalone)
-		clt->mode = COLLECT_PRIMARY_FOLLOWED_NOINPLACE;
+		fe->mode = COLLECT_PRIMARY_FOLLOWED_NOINPLACE;
 }
 
 /* called by erofs_shrinker to get rid of all compressed_pages */
@@ -350,47 +346,47 @@ int erofs_try_to_free_cached_page(struct page *page)
 }
 
 /* page_type must be Z_EROFS_PAGE_TYPE_EXCLUSIVE */
-static bool z_erofs_try_inplace_io(struct z_erofs_collector *clt,
+static bool z_erofs_try_inplace_io(struct z_erofs_decompress_frontend *fe,
 				   struct page *page)
 {
-	struct z_erofs_pcluster *const pcl = clt->pcl;
+	struct z_erofs_pcluster *const pcl = fe->pcl;
 
-	while (clt->icpage_ptr > pcl->compressed_pages)
-		if (!cmpxchg(--clt->icpage_ptr, NULL, page))
+	while (fe->icpage_ptr > pcl->compressed_pages)
+		if (!cmpxchg(--fe->icpage_ptr, NULL, page))
 			return true;
 	return false;
 }
 
 /* callers must be with collection lock held */
-static int z_erofs_attach_page(struct z_erofs_collector *clt,
+static int z_erofs_attach_page(struct z_erofs_decompress_frontend *fe,
 			       struct page *page, enum z_erofs_page_type type,
 			       bool pvec_safereuse)
 {
 	int ret;
 
 	/* give priority for inplaceio */
-	if (clt->mode >= COLLECT_PRIMARY &&
+	if (fe->mode >= COLLECT_PRIMARY &&
 	    type == Z_EROFS_PAGE_TYPE_EXCLUSIVE &&
-	    z_erofs_try_inplace_io(clt, page))
+	    z_erofs_try_inplace_io(fe, page))
 		return 0;
 
-	ret = z_erofs_pagevec_enqueue(&clt->vector, page, type,
+	ret = z_erofs_pagevec_enqueue(&fe->vector, page, type,
 				      pvec_safereuse);
-	clt->cl->vcnt += (unsigned int)ret;
+	fe->cl->vcnt += (unsigned int)ret;
 	return ret ? 0 : -EAGAIN;
 }
 
-static void z_erofs_try_to_claim_pcluster(struct z_erofs_collector *clt)
+static void z_erofs_try_to_claim_pcluster(struct z_erofs_decompress_frontend *f)
 {
-	struct z_erofs_pcluster *pcl = clt->pcl;
-	z_erofs_next_pcluster_t *owned_head = &clt->owned_head;
+	struct z_erofs_pcluster *pcl = f->pcl;
+	z_erofs_next_pcluster_t *owned_head = &f->owned_head;
 
 	/* type 1, nil pcluster (this pcluster doesn't belong to any chain.) */
 	if (cmpxchg(&pcl->next, Z_EROFS_PCLUSTER_NIL,
 		    *owned_head) == Z_EROFS_PCLUSTER_NIL) {
 		*owned_head = &pcl->next;
 		/* so we can attach this pcluster to our submission chain. */
-		clt->mode = COLLECT_PRIMARY_FOLLOWED;
+		f->mode = COLLECT_PRIMARY_FOLLOWED;
 		return;
 	}
 
@@ -401,24 +397,24 @@ static void z_erofs_try_to_claim_pcluster(struct z_erofs_collector *clt)
 	if (cmpxchg(&pcl->next, Z_EROFS_PCLUSTER_TAIL,
 		    *owned_head) == Z_EROFS_PCLUSTER_TAIL) {
 		*owned_head = Z_EROFS_PCLUSTER_TAIL;
-		clt->mode = COLLECT_PRIMARY_HOOKED;
-		clt->tailpcl = NULL;
+		f->mode = COLLECT_PRIMARY_HOOKED;
+		f->tailpcl = NULL;
 		return;
 	}
 	/* type 3, it belongs to a chain, but it isn't the end of the chain */
-	clt->mode = COLLECT_PRIMARY;
+	f->mode = COLLECT_PRIMARY;
 }
 
-static int z_erofs_lookup_collection(struct z_erofs_collector *clt,
+static int z_erofs_lookup_collection(struct z_erofs_decompress_frontend *fe,
 				     struct inode *inode,
 				     struct erofs_map_blocks *map)
 {
-	struct z_erofs_pcluster *pcl = clt->pcl;
+	struct z_erofs_pcluster *pcl = fe->pcl;
 	struct z_erofs_collection *cl;
 	unsigned int length;
 
 	/* to avoid unexpected loop formed by corrupted images */
-	if (clt->owned_head == &pcl->next || pcl == clt->tailpcl) {
+	if (fe->owned_head == &pcl->next || pcl == fe->tailpcl) {
 		DBG_BUGON(1);
 		return -EFSCORRUPTED;
 	}
@@ -449,15 +445,15 @@ static int z_erofs_lookup_collection(struct z_erofs_collector *clt,
 	}
 	mutex_lock(&cl->lock);
 	/* used to check tail merging loop due to corrupted images */
-	if (clt->owned_head == Z_EROFS_PCLUSTER_TAIL)
-		clt->tailpcl = pcl;
+	if (fe->owned_head == Z_EROFS_PCLUSTER_TAIL)
+		fe->tailpcl = pcl;
 
-	z_erofs_try_to_claim_pcluster(clt);
-	clt->cl = cl;
+	z_erofs_try_to_claim_pcluster(fe);
+	fe->cl = cl;
 	return 0;
 }
 
-static int z_erofs_register_collection(struct z_erofs_collector *clt,
+static int z_erofs_register_collection(struct z_erofs_decompress_frontend *fe,
 				       struct inode *inode,
 				       struct erofs_map_blocks *map)
 {
@@ -485,8 +481,8 @@ static int z_erofs_register_collection(struct z_erofs_collector *clt,
 			Z_EROFS_PCLUSTER_FULL_LENGTH : 0);
 
 	/* new pclusters should be claimed as type 1, primary and followed */
-	pcl->next = clt->owned_head;
-	clt->mode = COLLECT_PRIMARY_FOLLOWED;
+	pcl->next = fe->owned_head;
+	fe->mode = COLLECT_PRIMARY_FOLLOWED;
 
 	cl = z_erofs_primarycollection(pcl);
 	cl->pageofs = map->m_la & ~PAGE_MASK;
@@ -512,18 +508,18 @@ static int z_erofs_register_collection(struct z_erofs_collector *clt,
 		}
 
 		if (grp != &pcl->obj) {
-			clt->pcl = container_of(grp,
+			fe->pcl = container_of(grp,
 					struct z_erofs_pcluster, obj);
 			err = -EEXIST;
 			goto err_out;
 		}
 	}
 	/* used to check tail merging loop due to corrupted images */
-	if (clt->owned_head == Z_EROFS_PCLUSTER_TAIL)
-		clt->tailpcl = pcl;
-	clt->owned_head = &pcl->next;
-	clt->pcl = pcl;
-	clt->cl = cl;
+	if (fe->owned_head == Z_EROFS_PCLUSTER_TAIL)
+		fe->tailpcl = pcl;
+	fe->owned_head = &pcl->next;
+	fe->pcl = pcl;
+	fe->cl = cl;
 	return 0;
 
 err_out:
@@ -532,18 +528,18 @@ err_out:
 	return err;
 }
 
-static int z_erofs_collector_begin(struct z_erofs_collector *clt,
+static int z_erofs_collector_begin(struct z_erofs_decompress_frontend *fe,
 				   struct inode *inode,
 				   struct erofs_map_blocks *map)
 {
 	struct erofs_workgroup *grp;
 	int ret;
 
-	DBG_BUGON(clt->cl);
+	DBG_BUGON(fe->cl);
 
 	/* must be Z_EROFS_PCLUSTER_TAIL or pointed to previous collection */
-	DBG_BUGON(clt->owned_head == Z_EROFS_PCLUSTER_NIL);
-	DBG_BUGON(clt->owned_head == Z_EROFS_PCLUSTER_TAIL_CLOSED);
+	DBG_BUGON(fe->owned_head == Z_EROFS_PCLUSTER_NIL);
+	DBG_BUGON(fe->owned_head == Z_EROFS_PCLUSTER_TAIL_CLOSED);
 
 	if (map->m_flags & EROFS_MAP_META) {
 		if ((map->m_pa & ~PAGE_MASK) + map->m_plen > PAGE_SIZE) {
@@ -555,28 +551,28 @@ static int z_erofs_collector_begin(struct z_erofs_collector *clt,
 
 	grp = erofs_find_workgroup(inode->i_sb, map->m_pa >> PAGE_SHIFT);
 	if (grp) {
-		clt->pcl = container_of(grp, struct z_erofs_pcluster, obj);
+		fe->pcl = container_of(grp, struct z_erofs_pcluster, obj);
 	} else {
 tailpacking:
-		ret = z_erofs_register_collection(clt, inode, map);
+		ret = z_erofs_register_collection(fe, inode, map);
 		if (!ret)
 			goto out;
 		if (ret != -EEXIST)
 			return ret;
 	}
 
-	ret = z_erofs_lookup_collection(clt, inode, map);
+	ret = z_erofs_lookup_collection(fe, inode, map);
 	if (ret) {
-		erofs_workgroup_put(&clt->pcl->obj);
+		erofs_workgroup_put(&fe->pcl->obj);
 		return ret;
 	}
 
 out:
-	z_erofs_pagevec_ctor_init(&clt->vector, Z_EROFS_NR_INLINE_PAGEVECS,
-				  clt->cl->pagevec, clt->cl->vcnt);
+	z_erofs_pagevec_ctor_init(&fe->vector, Z_EROFS_NR_INLINE_PAGEVECS,
+				  fe->cl->pagevec, fe->cl->vcnt);
 	/* since file-backed online pages are traversed in reverse order */
-	clt->icpage_ptr = clt->pcl->compressed_pages +
-			z_erofs_pclusterpages(clt->pcl);
+	fe->icpage_ptr = fe->pcl->compressed_pages +
+			z_erofs_pclusterpages(fe->pcl);
 	return 0;
 }
 
@@ -610,24 +606,24 @@ static void z_erofs_collection_put(struct z_erofs_collection *cl)
 	erofs_workgroup_put(&pcl->obj);
 }
 
-static bool z_erofs_collector_end(struct z_erofs_collector *clt)
+static bool z_erofs_collector_end(struct z_erofs_decompress_frontend *fe)
 {
-	struct z_erofs_collection *cl = clt->cl;
+	struct z_erofs_collection *cl = fe->cl;
 
 	if (!cl)
 		return false;
 
-	z_erofs_pagevec_ctor_exit(&clt->vector, false);
+	z_erofs_pagevec_ctor_exit(&fe->vector, false);
 	mutex_unlock(&cl->lock);
 
 	/*
 	 * if all pending pages are added, don't hold its reference
 	 * any longer if the pcluster isn't hosted by ourselves.
 	 */
-	if (clt->mode < COLLECT_PRIMARY_FOLLOWED_NOINPLACE)
+	if (fe->mode < COLLECT_PRIMARY_FOLLOWED_NOINPLACE)
 		z_erofs_collection_put(cl);
 
-	clt->cl = NULL;
+	fe->cl = NULL;
 	return true;
 }
 
@@ -651,7 +647,6 @@ static int z_erofs_do_read_page(struct z_erofs_decompress_frontend *fe,
 	struct inode *const inode = fe->inode;
 	struct erofs_sb_info *const sbi = EROFS_I_SB(inode);
 	struct erofs_map_blocks *const map = &fe->map;
-	struct z_erofs_collector *const clt = &fe->clt;
 	const loff_t offset = page_offset(page);
 	bool tight = true;
 
@@ -672,7 +667,7 @@ repeat:
 	if (offset + cur >= map->m_la &&
 	    offset + cur < map->m_la + map->m_llen) {
 		/* didn't get a valid collection previously (very rare) */
-		if (!clt->cl)
+		if (!fe->cl)
 			goto restart_now;
 		goto hitted;
 	}
@@ -680,7 +675,7 @@ repeat:
 	/* go ahead the next map_blocks */
 	erofs_dbg("%s: [out-of-range] pos %llu", __func__, offset + cur);
 
-	if (z_erofs_collector_end(clt))
+	if (z_erofs_collector_end(fe))
 		fe->backmost = false;
 
 	map->m_la = offset + cur;
@@ -693,11 +688,11 @@ restart_now:
 	if (!(map->m_flags & EROFS_MAP_MAPPED))
 		goto hitted;
 
-	err = z_erofs_collector_begin(clt, inode, map);
+	err = z_erofs_collector_begin(fe, inode, map);
 	if (err)
 		goto err_out;
 
-	if (z_erofs_is_inline_pcluster(clt->pcl)) {
+	if (z_erofs_is_inline_pcluster(fe->pcl)) {
 		void *mp;
 
 		mp = erofs_read_metabuf(&fe->map.buf, inode->i_sb,
@@ -709,20 +704,18 @@ restart_now:
 			goto err_out;
 		}
 		get_page(fe->map.buf.page);
-		WRITE_ONCE(clt->pcl->compressed_pages[0], fe->map.buf.page);
-		clt->mode = COLLECT_PRIMARY_FOLLOWED_NOINPLACE;
+		WRITE_ONCE(fe->pcl->compressed_pages[0], fe->map.buf.page);
+		fe->mode = COLLECT_PRIMARY_FOLLOWED_NOINPLACE;
 	} else {
-		/* preload all compressed pages (can change mode if needed) */
+		/* bind cache first when cached decompression is preferred */
 		if (should_alloc_managed_pages(fe, sbi->opt.cache_strategy,
 					       map->m_la))
 			cache_strategy = TRYALLOC;
 		else
 			cache_strategy = DONTALLOC;
 
-		preload_compressed_pages(clt, MNGD_MAPPING(sbi),
-					 cache_strategy, pagepool);
+		z_erofs_bind_cache(fe, cache_strategy, pagepool);
 	}
-
 hitted:
 	/*
 	 * Ensure the current partial page belongs to this submit chain rather
@@ -730,8 +723,8 @@ hitted:
 	 * those chains are handled asynchronously thus the page cannot be used
 	 * for inplace I/O or pagevec (should be processed in strict order.)
 	 */
-	tight &= (clt->mode >= COLLECT_PRIMARY_HOOKED &&
-		  clt->mode != COLLECT_PRIMARY_FOLLOWED_NOINPLACE);
+	tight &= (fe->mode >= COLLECT_PRIMARY_HOOKED &&
+		  fe->mode != COLLECT_PRIMARY_FOLLOWED_NOINPLACE);
 
 	cur = end - min_t(unsigned int, offset + end - map->m_la, end);
 	if (!(map->m_flags & EROFS_MAP_MAPPED)) {
@@ -746,18 +739,18 @@ hitted:
 				Z_EROFS_VLE_PAGE_TYPE_TAIL_SHARED));
 
 	if (cur)
-		tight &= (clt->mode >= COLLECT_PRIMARY_FOLLOWED);
+		tight &= (fe->mode >= COLLECT_PRIMARY_FOLLOWED);
 
 retry:
-	err = z_erofs_attach_page(clt, page, page_type,
-				  clt->mode >= COLLECT_PRIMARY_FOLLOWED);
+	err = z_erofs_attach_page(fe, page, page_type,
+				  fe->mode >= COLLECT_PRIMARY_FOLLOWED);
 	/* should allocate an additional short-lived page for pagevec */
 	if (err == -EAGAIN) {
 		struct page *const newpage =
 				alloc_page(GFP_NOFS | __GFP_NOFAIL);
 
 		set_page_private(newpage, Z_EROFS_SHORTLIVED_PAGE);
-		err = z_erofs_attach_page(clt, newpage,
+		err = z_erofs_attach_page(fe, newpage,
 					  Z_EROFS_PAGE_TYPE_EXCLUSIVE, true);
 		if (!err)
 			goto retry;
@@ -773,7 +766,7 @@ retry:
 	/* bump up the number of spiltted parts of a page */
 	++spiltted;
 	/* also update nr_pages */
-	clt->cl->nr_pages = max_t(pgoff_t, clt->cl->nr_pages, index + 1);
+	fe->cl->nr_pages = max_t(pgoff_t, fe->cl->nr_pages, index + 1);
 next_part:
 	/* can be used for verification */
 	map->m_llen = offset + cur - map->m_la;
@@ -1098,10 +1091,10 @@ static void z_erofs_decompress_kickoff(struct z_erofs_decompressqueue *io,
 static struct page *pickup_page_for_submission(struct z_erofs_pcluster *pcl,
 					       unsigned int nr,
 					       struct page **pagepool,
-					       struct address_space *mc,
-					       gfp_t gfp)
+					       struct address_space *mc)
 {
 	const pgoff_t index = pcl->obj.index;
+	gfp_t gfp = mapping_gfp_mask(mc);
 	bool tocache = false;
 
 	struct address_space *mapping;
@@ -1309,7 +1302,7 @@ static void z_erofs_submit_queue(struct super_block *sb,
 	z_erofs_next_pcluster_t qtail[NR_JOBQUEUES];
 	struct z_erofs_decompressqueue *q[NR_JOBQUEUES];
 	void *bi_private;
-	z_erofs_next_pcluster_t owned_head = f->clt.owned_head;
+	z_erofs_next_pcluster_t owned_head = f->owned_head;
 	/* bio is NULL initially, so no need to initialize last_{index,bdev} */
 	pgoff_t last_index;
 	struct block_device *last_bdev;
@@ -1357,8 +1350,7 @@ static void z_erofs_submit_queue(struct super_block *sb,
 			struct page *page;
 
 			page = pickup_page_for_submission(pcl, i++, pagepool,
-							  MNGD_MAPPING(sbi),
-							  GFP_NOFS);
+							  MNGD_MAPPING(sbi));
 			if (!page)
 				continue;
 
@@ -1416,7 +1408,7 @@ static void z_erofs_runqueue(struct super_block *sb,
 {
 	struct z_erofs_decompressqueue io[NR_JOBQUEUES];
 
-	if (f->clt.owned_head == Z_EROFS_PCLUSTER_TAIL)
+	if (f->owned_head == Z_EROFS_PCLUSTER_TAIL)
 		return;
 	z_erofs_submit_queue(sb, f, pagepool, io, &force_fg);
 
@@ -1516,7 +1508,7 @@ static int z_erofs_readpage(struct file *file, struct page *page)
 	err = z_erofs_do_read_page(&f, page, &pagepool);
 	z_erofs_pcluster_readmore(&f, NULL, 0, &pagepool, false);
 
-	(void)z_erofs_collector_end(&f.clt);
+	(void)z_erofs_collector_end(&f);
 
 	/* if some compressed cluster ready, need submit them anyway */
 	z_erofs_runqueue(inode->i_sb, &f, &pagepool,
@@ -1566,7 +1558,7 @@ static void z_erofs_readahead(struct readahead_control *rac)
 		put_page(page);
 	}
 	z_erofs_pcluster_readmore(&f, rac, 0, &pagepool, false);
-	(void)z_erofs_collector_end(&f.clt);
+	(void)z_erofs_collector_end(&f);
 
 	z_erofs_runqueue(inode->i_sb, &f, &pagepool,
 			 z_erofs_get_sync_decompress_policy(sbi, nr_pages));
diff --git a/fs/erofs/zmap.c b/fs/erofs/zmap.c
index 361b1d6e4bf9..572f0b8151ba 100644
--- a/fs/erofs/zmap.c
+++ b/fs/erofs/zmap.c
@@ -431,48 +431,47 @@ static int z_erofs_extent_lookback(struct z_erofs_maprecorder *m,
 				   unsigned int lookback_distance)
 {
 	struct erofs_inode *const vi = EROFS_I(m->inode);
-	struct erofs_map_blocks *const map = m->map;
 	const unsigned int lclusterbits = vi->z_logical_clusterbits;
-	unsigned long lcn = m->lcn;
-	int err;
 
-	if (lcn < lookback_distance) {
-		erofs_err(m->inode->i_sb,
-			  "bogus lookback distance @ nid %llu", vi->nid);
-		DBG_BUGON(1);
-		return -EFSCORRUPTED;
-	}
+	while (m->lcn >= lookback_distance) {
+		unsigned long lcn = m->lcn - lookback_distance;
+		int err;
 
-	/* load extent head logical cluster if needed */
-	lcn -= lookback_distance;
-	err = z_erofs_load_cluster_from_disk(m, lcn, false);
-	if (err)
-		return err;
+		/* load extent head logical cluster if needed */
+		err = z_erofs_load_cluster_from_disk(m, lcn, false);
+		if (err)
+			return err;
 
-	switch (m->type) {
-	case Z_EROFS_VLE_CLUSTER_TYPE_NONHEAD:
-		if (!m->delta[0]) {
+		switch (m->type) {
+		case Z_EROFS_VLE_CLUSTER_TYPE_NONHEAD:
+			if (!m->delta[0]) {
+				erofs_err(m->inode->i_sb,
+					  "invalid lookback distance 0 @ nid %llu",
+					  vi->nid);
+				DBG_BUGON(1);
+				return -EFSCORRUPTED;
+			}
+			lookback_distance = m->delta[0];
+			continue;
+		case Z_EROFS_VLE_CLUSTER_TYPE_PLAIN:
+		case Z_EROFS_VLE_CLUSTER_TYPE_HEAD1:
+		case Z_EROFS_VLE_CLUSTER_TYPE_HEAD2:
+			m->headtype = m->type;
+			m->map->m_la = (lcn << lclusterbits) | m->clusterofs;
+			return 0;
+		default:
 			erofs_err(m->inode->i_sb,
-				  "invalid lookback distance 0 @ nid %llu",
-				  vi->nid);
+				  "unknown type %u @ lcn %lu of nid %llu",
+				  m->type, lcn, vi->nid);
 			DBG_BUGON(1);
-			return -EFSCORRUPTED;
+			return -EOPNOTSUPP;
 		}
-		return z_erofs_extent_lookback(m, m->delta[0]);
-	case Z_EROFS_VLE_CLUSTER_TYPE_PLAIN:
-	case Z_EROFS_VLE_CLUSTER_TYPE_HEAD1:
-	case Z_EROFS_VLE_CLUSTER_TYPE_HEAD2:
-		m->headtype = m->type;
-		map->m_la = (lcn << lclusterbits) | m->clusterofs;
-		break;
-	default:
-		erofs_err(m->inode->i_sb,
-			  "unknown type %u @ lcn %lu of nid %llu",
-			  m->type, lcn, vi->nid);
-		DBG_BUGON(1);
-		return -EOPNOTSUPP;
 	}
-	return 0;
+
+	erofs_err(m->inode->i_sb, "bogus lookback distance @ nid %llu",
+		  vi->nid);
+	DBG_BUGON(1);
+	return -EFSCORRUPTED;
 }
 
 static int z_erofs_get_extent_compressedlen(struct z_erofs_maprecorder *m,
@@ -494,7 +493,7 @@ static int z_erofs_get_extent_compressedlen(struct z_erofs_maprecorder *m,
 	     !(vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_1)) ||
 	    ((m->headtype == Z_EROFS_VLE_CLUSTER_TYPE_HEAD2) &&
 	     !(vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_2))) {
-		map->m_plen = 1 << lclusterbits;
+		map->m_plen = 1ULL << lclusterbits;
 		return 0;
 	}
 	lcn = m->lcn + 1;
@@ -540,7 +539,7 @@ static int z_erofs_get_extent_compressedlen(struct z_erofs_maprecorder *m,
 		return -EFSCORRUPTED;
 	}
 out:
-	map->m_plen = m->compressedlcs << lclusterbits;
+	map->m_plen = (u64)m->compressedlcs << lclusterbits;
 	return 0;
 err_bonus_cblkcnt:
 	erofs_err(m->inode->i_sb,