[RFC PATCH v3 0/2] erofs: support compressed fragments data

[RFC PATCH v3 0/2] erofs: support compressed fragments data

[Yue Hu]

From: Yue Hu <huyue2@coolpad.com>

This feature can merge tail of per-file or the whole files into a
special inode to achieve greater compression ratio.

Meanwhile, also add a interlaced uncompressed data layout support for
compressed files since fragments feature (and later) can use it.

mkfs v5: https://lore.kernel.org/all/cover.1661687617.git.huyue2@coolpad.com/

changes from v2:
 - enhance the condition to check if pcluster is interlaced or not;
 - no typo.

changes from v1:
 - fix a compiling error without CONFIG_EROFS_FS_ZIP, reported by kernel test
   robot <lkp@intel.com>;
 - introduce the term 'interlaced' for patch 1/2 suggested by Xiang;
 - fix packed inode failure path when read super pointed out by Xiang;
 - use kmap_local_page instead of kmap_atomic pointed out by Xiang;
 - use a simpler way to avoid call read fragment data twice suggested by Xiang;
 - update commit message change.

Yue Hu (2):
  erofs: support interlaced uncompressed data for compressed files
  erofs: support on-disk compressed fragments data

 fs/erofs/compress.h     |  3 +++
 fs/erofs/decompressor.c | 12 +++++----
 fs/erofs/erofs_fs.h     | 28 ++++++++++++++-----
 fs/erofs/internal.h     | 16 ++++++++---
 fs/erofs/super.c        | 15 +++++++++++
 fs/erofs/sysfs.c        |  2 ++
 fs/erofs/zdata.c        | 60 +++++++++++++++++++++++++++++++++++++++--
 fs/erofs/zdata.h        |  3 +++
 fs/erofs/zmap.c         | 33 ++++++++++++++++++++++-
 9 files changed, 155 insertions(+), 17 deletions(-)

-- 
2.17.1

[RFC PATCH v3 1/2] erofs: support interlaced uncompressed data for compressed files

[Yue Hu]

From: Yue Hu <huyue2@coolpad.com>

Currently, there is no start offset when writing uncompressed data to
disk blocks for compressed files. However, we are using in-place I/O
which will decrease the number of memory copies a lot if we write it
just from an offset of 'pageofs_out'. So, let's support it.

Signed-off-by: Yue Hu <huyue2@coolpad.com>
---
 fs/erofs/compress.h     |  3 +++
 fs/erofs/decompressor.c | 12 +++++++-----
 fs/erofs/erofs_fs.h     |  2 ++
 fs/erofs/zdata.c        | 12 +++++++++++-
 fs/erofs/zdata.h        |  3 +++
 5 files changed, 26 insertions(+), 6 deletions(-)

diff --git a/fs/erofs/compress.h b/fs/erofs/compress.h
index 26fa170090b8..cef26c63d6b1 100644
--- a/fs/erofs/compress.h
+++ b/fs/erofs/compress.h
@@ -15,6 +15,9 @@ struct z_erofs_decompress_req {
 	unsigned short pageofs_in, pageofs_out;
 	unsigned int inputsize, outputsize;
 
+	/* cut point of interlaced uncompressed data */
+	unsigned int interlaced_offset;
+
 	/* indicate the algorithm will be used for decompression */
 	unsigned int alg;
 	bool inplace_io, partial_decoding, fillgaps;
diff --git a/fs/erofs/decompressor.c b/fs/erofs/decompressor.c
index 2d55569f96ac..e5dc8eb992b1 100644
--- a/fs/erofs/decompressor.c
+++ b/fs/erofs/decompressor.c
@@ -340,18 +340,20 @@ static int z_erofs_shifted_transform(struct z_erofs_decompress_req *rq,
 	src = kmap_atomic(*rq->in) + rq->pageofs_in;
 	if (rq->out[0]) {
 		dst = kmap_atomic(rq->out[0]);
-		memcpy(dst + rq->pageofs_out, src, righthalf);
+		memcpy(dst + rq->pageofs_out, src + rq->interlaced_offset,
+		       righthalf);
 		kunmap_atomic(dst);
 	}
 
 	if (nrpages_out == 2) {
 		DBG_BUGON(!rq->out[1]);
-		if (rq->out[1] == *rq->in) {
-			memmove(src, src + righthalf, lefthalf);
-		} else {
+		if (rq->out[1] != *rq->in) {
 			dst = kmap_atomic(rq->out[1]);
-			memcpy(dst, src + righthalf, lefthalf);
+			memcpy(dst, rq->interlaced_offset ? src :
+						(src + righthalf), lefthalf);
 			kunmap_atomic(dst);
+		} else if (!rq->interlaced_offset) {
+			memmove(src, src + righthalf, lefthalf);
 		}
 	}
 	kunmap_atomic(src);
diff --git a/fs/erofs/erofs_fs.h b/fs/erofs/erofs_fs.h
index 2b48373f690b..5c1de6d7ad71 100644
--- a/fs/erofs/erofs_fs.h
+++ b/fs/erofs/erofs_fs.h
@@ -295,11 +295,13 @@ struct z_erofs_lzma_cfgs {
  * bit 1 : HEAD1 big pcluster (0 - off; 1 - on)
  * bit 2 : HEAD2 big pcluster (0 - off; 1 - on)
  * bit 3 : tailpacking inline pcluster (0 - off; 1 - on)
+ * bit 4 : interlaced plain pcluster (0 - off; 1 - on)
  */
 #define Z_EROFS_ADVISE_COMPACTED_2B		0x0001
 #define Z_EROFS_ADVISE_BIG_PCLUSTER_1		0x0002
 #define Z_EROFS_ADVISE_BIG_PCLUSTER_2		0x0004
 #define Z_EROFS_ADVISE_INLINE_PCLUSTER		0x0008
+#define Z_EROFS_ADVISE_INTERLACED_PCLUSTER	0x0010
 
 struct z_erofs_map_header {
 	__le16	h_reserved1;
diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index 5792ca9e0d5e..bbaa3a924852 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -481,6 +481,7 @@ static void z_erofs_try_to_claim_pcluster(struct z_erofs_decompress_frontend *f)
 
 static int z_erofs_register_pcluster(struct z_erofs_decompress_frontend *fe)
 {
+	struct erofs_inode *const vi = EROFS_I(fe->inode);
 	struct erofs_map_blocks *map = &fe->map;
 	bool ztailpacking = map->m_flags & EROFS_MAP_META;
 	struct z_erofs_pcluster *pcl;
@@ -508,6 +509,12 @@ static int z_erofs_register_pcluster(struct z_erofs_decompress_frontend *fe)
 	pcl->pageofs_out = map->m_la & ~PAGE_MASK;
 	fe->mode = Z_EROFS_PCLUSTER_FOLLOWED;
 
+	pcl->interlaced = false;
+	if ((vi->z_advise & Z_EROFS_ADVISE_INTERLACED_PCLUSTER) &&
+	    pcl->algorithmformat == Z_EROFS_COMPRESSION_SHIFTED &&
+	    pcl->pageofs_out)
+		pcl->interlaced = true;
+
 	/*
 	 * lock all primary followed works before visible to others
 	 * and mutex_trylock *never* fails for a new pcluster.
@@ -972,7 +979,7 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be,
 	struct erofs_sb_info *const sbi = EROFS_SB(be->sb);
 	struct z_erofs_pcluster *pcl = be->pcl;
 	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
-	unsigned int i, inputsize;
+	unsigned int i, inputsize, interlaced_offset;
 	int err2;
 	struct page *page;
 	bool overlapped;
@@ -1015,6 +1022,8 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be,
 	else
 		inputsize = pclusterpages * PAGE_SIZE;
 
+	interlaced_offset = pcl->interlaced ? pcl->pageofs_out : 0;
+
 	err = z_erofs_decompress(&(struct z_erofs_decompress_req) {
 					.sb = be->sb,
 					.in = be->compressed_pages,
@@ -1023,6 +1032,7 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be,
 					.pageofs_out = pcl->pageofs_out,
 					.inputsize = inputsize,
 					.outputsize = pcl->length,
+					.interlaced_offset = interlaced_offset,
 					.alg = pcl->algorithmformat,
 					.inplace_io = overlapped,
 					.partial_decoding = pcl->partial,
diff --git a/fs/erofs/zdata.h b/fs/erofs/zdata.h
index e7f04c4fbb81..75f3a52fc66f 100644
--- a/fs/erofs/zdata.h
+++ b/fs/erofs/zdata.h
@@ -87,6 +87,9 @@ struct z_erofs_pcluster {
 	/* L: indicate several pageofs_outs or not */
 	bool multibases;
 
+	/* I: whether interlaced uncompressed data or not */
+	bool interlaced;
+
 	/* A: compressed bvecs (can be cached or inplaced pages) */
 	struct z_erofs_bvec compressed_bvecs[];
 };
-- 
2.17.1

[RFC PATCH v3 2/2] erofs: support on-disk compressed fragments data

[Yue Hu]

From: Yue Hu <huyue2@coolpad.com>

Introduce on-disk compressed fragments data feature.

This approach adds a new field called `h_fragmentoff' in the per-file
compression header to indicate the fragment offset of each tail pcluster
or the whole file in the special packed inode.

Like ztailpacking, it will also find and record the 'headlcn' of the
tail pcluster when initializing per-inode zmap for making follow-on
requests more easy.

Signed-off-by: Yue Hu <huyue2@coolpad.com>
---
 fs/erofs/erofs_fs.h | 26 ++++++++++++++++++------
 fs/erofs/internal.h | 16 ++++++++++++---
 fs/erofs/super.c    | 15 ++++++++++++++
 fs/erofs/sysfs.c    |  2 ++
 fs/erofs/zdata.c    | 48 ++++++++++++++++++++++++++++++++++++++++++++-
 fs/erofs/zmap.c     | 33 ++++++++++++++++++++++++++++++-
 6 files changed, 129 insertions(+), 11 deletions(-)

diff --git a/fs/erofs/erofs_fs.h b/fs/erofs/erofs_fs.h
index 5c1de6d7ad71..340fb980fe03 100644
--- a/fs/erofs/erofs_fs.h
+++ b/fs/erofs/erofs_fs.h
@@ -25,6 +25,7 @@
 #define EROFS_FEATURE_INCOMPAT_DEVICE_TABLE	0x00000008
 #define EROFS_FEATURE_INCOMPAT_COMPR_HEAD2	0x00000008
 #define EROFS_FEATURE_INCOMPAT_ZTAILPACKING	0x00000010
+#define EROFS_FEATURE_INCOMPAT_FRAGMENTS	0x00000020
 #define EROFS_ALL_FEATURE_INCOMPAT		\
 	(EROFS_FEATURE_INCOMPAT_ZERO_PADDING | \
 	 EROFS_FEATURE_INCOMPAT_COMPR_CFGS | \
@@ -32,7 +33,8 @@
 	 EROFS_FEATURE_INCOMPAT_CHUNKED_FILE | \
 	 EROFS_FEATURE_INCOMPAT_DEVICE_TABLE | \
 	 EROFS_FEATURE_INCOMPAT_COMPR_HEAD2 | \
-	 EROFS_FEATURE_INCOMPAT_ZTAILPACKING)
+	 EROFS_FEATURE_INCOMPAT_ZTAILPACKING | \
+	 EROFS_FEATURE_INCOMPAT_FRAGMENTS)
 
 #define EROFS_SB_EXTSLOT_SIZE	16
 
@@ -71,7 +73,9 @@ struct erofs_super_block {
 	} __packed u1;
 	__le16 extra_devices;	/* # of devices besides the primary device */
 	__le16 devt_slotoff;	/* startoff = devt_slotoff * devt_slotsize */
-	__u8 reserved2[38];
+	__u8 reserved[6];
+	__le64 packed_nid;	/* nid of the special packed inode */
+	__u8 reserved2[24];
 };
 
 /*
@@ -296,17 +300,26 @@ struct z_erofs_lzma_cfgs {
  * bit 2 : HEAD2 big pcluster (0 - off; 1 - on)
  * bit 3 : tailpacking inline pcluster (0 - off; 1 - on)
  * bit 4 : interlaced plain pcluster (0 - off; 1 - on)
+ * bit 5 : fragment pcluster (0 - off; 1 - on)
  */
 #define Z_EROFS_ADVISE_COMPACTED_2B		0x0001
 #define Z_EROFS_ADVISE_BIG_PCLUSTER_1		0x0002
 #define Z_EROFS_ADVISE_BIG_PCLUSTER_2		0x0004
 #define Z_EROFS_ADVISE_INLINE_PCLUSTER		0x0008
 #define Z_EROFS_ADVISE_INTERLACED_PCLUSTER	0x0010
+#define Z_EROFS_ADVISE_FRAGMENT_PCLUSTER	0x0020
 
+#define Z_EROFS_FRAGMENT_INODE_BIT              7
 struct z_erofs_map_header {
-	__le16	h_reserved1;
-	/* indicates the encoded size of tailpacking data */
-	__le16  h_idata_size;
+	union {
+		/* direct addressing for fragment offset */
+		__le32  h_fragmentoff;
+		struct {
+			__le16  h_reserved1;
+			/* indicates the encoded size of tailpacking data */
+			__le16  h_idata_size;
+		};
+	};
 	__le16	h_advise;
 	/*
 	 * bit 0-3 : algorithm type of head 1 (logical cluster type 01);
@@ -315,7 +328,8 @@ struct z_erofs_map_header {
 	__u8	h_algorithmtype;
 	/*
 	 * bit 0-2 : logical cluster bits - 12, e.g. 0 for 4096;
-	 * bit 3-7 : reserved.
+	 * bit 3-6 : reserved;
+	 * bit 7   : move the whole file into packed inode or not.
 	 */
 	__u8	h_clusterbits;
 };
diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
index cfee49d33b95..7b9d31bab928 100644
--- a/fs/erofs/internal.h
+++ b/fs/erofs/internal.h
@@ -120,6 +120,7 @@ struct erofs_sb_info {
 	struct inode *managed_cache;
 
 	struct erofs_sb_lz4_info lz4;
+	struct inode *packed_inode;
 #endif	/* CONFIG_EROFS_FS_ZIP */
 	struct erofs_dev_context *devs;
 	struct dax_device *dax_dev;
@@ -306,6 +307,7 @@ EROFS_FEATURE_FUNCS(chunked_file, incompat, INCOMPAT_CHUNKED_FILE)
 EROFS_FEATURE_FUNCS(device_table, incompat, INCOMPAT_DEVICE_TABLE)
 EROFS_FEATURE_FUNCS(compr_head2, incompat, INCOMPAT_COMPR_HEAD2)
 EROFS_FEATURE_FUNCS(ztailpacking, incompat, INCOMPAT_ZTAILPACKING)
+EROFS_FEATURE_FUNCS(fragments, incompat, INCOMPAT_FRAGMENTS)
 EROFS_FEATURE_FUNCS(sb_chksum, compat, COMPAT_SB_CHKSUM)
 
 /* atomic flag definitions */
@@ -341,8 +343,13 @@ struct erofs_inode {
 			unsigned char  z_algorithmtype[2];
 			unsigned char  z_logical_clusterbits;
 			unsigned long  z_tailextent_headlcn;
-			erofs_off_t    z_idataoff;
-			unsigned short z_idata_size;
+			union {
+				struct {
+					erofs_off_t    z_idataoff;
+					unsigned short z_idata_size;
+				};
+				erofs_off_t z_fragmentoff;
+			};
 		};
 #endif	/* CONFIG_EROFS_FS_ZIP */
 	};
@@ -400,6 +407,7 @@ extern const struct address_space_operations z_erofs_aops;
 enum {
 	BH_Encoded = BH_PrivateStart,
 	BH_FullMapped,
+	BH_Fragment,
 };
 
 /* Has a disk mapping */
@@ -410,6 +418,8 @@ enum {
 #define EROFS_MAP_ENCODED	(1 << BH_Encoded)
 /* The length of extent is full */
 #define EROFS_MAP_FULL_MAPPED	(1 << BH_FullMapped)
+/* Located in the special packed inode */
+#define EROFS_MAP_FRAGMENT	(1 << BH_Fragment)
 
 struct erofs_map_blocks {
 	struct erofs_buf buf;
@@ -431,7 +441,7 @@ struct erofs_map_blocks {
 #define EROFS_GET_BLOCKS_FIEMAP	0x0002
 /* Used to map the whole extent if non-negligible data is requested for LZMA */
 #define EROFS_GET_BLOCKS_READMORE	0x0004
-/* Used to map tail extent for tailpacking inline pcluster */
+/* Used to map tail extent for tailpacking inline or fragment pcluster */
 #define EROFS_GET_BLOCKS_FINDTAIL	0x0008
 
 enum {
diff --git a/fs/erofs/super.c b/fs/erofs/super.c
index 3173debeaa5a..8170c0d8ab92 100644
--- a/fs/erofs/super.c
+++ b/fs/erofs/super.c
@@ -381,6 +381,17 @@ static int erofs_read_superblock(struct super_block *sb)
 #endif
 	sbi->islotbits = ilog2(sizeof(struct erofs_inode_compact));
 	sbi->root_nid = le16_to_cpu(dsb->root_nid);
+#ifdef CONFIG_EROFS_FS_ZIP
+	sbi->packed_inode = NULL;
+	if (erofs_sb_has_fragments(sbi)) {
+		sbi->packed_inode =
+			erofs_iget(sb, le64_to_cpu(dsb->packed_nid), false);
+		if (IS_ERR(sbi->packed_inode)) {
+			ret = PTR_ERR(sbi->packed_inode);
+			goto out;
+		}
+	}
+#endif
 	sbi->inos = le64_to_cpu(dsb->inos);
 
 	sbi->build_time = le64_to_cpu(dsb->build_time);
@@ -411,6 +422,8 @@ static int erofs_read_superblock(struct super_block *sb)
 		erofs_info(sb, "EXPERIMENTAL compressed inline data feature in use. Use at your own risk!");
 	if (erofs_is_fscache_mode(sb))
 		erofs_info(sb, "EXPERIMENTAL fscache-based on-demand read feature in use. Use at your own risk!");
+	if (erofs_sb_has_fragments(sbi))
+		erofs_info(sb, "EXPERIMENTAL compressed fragments feature in use. Use at your own risk!");
 out:
 	erofs_put_metabuf(&buf);
 	return ret;
@@ -908,6 +921,8 @@ static void erofs_put_super(struct super_block *sb)
 #ifdef CONFIG_EROFS_FS_ZIP
 	iput(sbi->managed_cache);
 	sbi->managed_cache = NULL;
+	iput(sbi->packed_inode);
+	sbi->packed_inode = NULL;
 #endif
 	erofs_fscache_unregister_cookie(&sbi->s_fscache);
 }
diff --git a/fs/erofs/sysfs.c b/fs/erofs/sysfs.c
index c1383e508bbe..1b52395be82a 100644
--- a/fs/erofs/sysfs.c
+++ b/fs/erofs/sysfs.c
@@ -76,6 +76,7 @@ EROFS_ATTR_FEATURE(device_table);
 EROFS_ATTR_FEATURE(compr_head2);
 EROFS_ATTR_FEATURE(sb_chksum);
 EROFS_ATTR_FEATURE(ztailpacking);
+EROFS_ATTR_FEATURE(fragments);
 
 static struct attribute *erofs_feat_attrs[] = {
 	ATTR_LIST(zero_padding),
@@ -86,6 +87,7 @@ static struct attribute *erofs_feat_attrs[] = {
 	ATTR_LIST(compr_head2),
 	ATTR_LIST(sb_chksum),
 	ATTR_LIST(ztailpacking),
+	ATTR_LIST(fragments),
 	NULL,
 };
 ATTRIBUTE_GROUPS(erofs_feat);
diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index bbaa3a924852..ecafe85c2763 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -657,6 +657,33 @@ static bool should_alloc_managed_pages(struct z_erofs_decompress_frontend *fe,
 		la < fe->headoffset;
 }
 
+static int z_erofs_read_fragment_data(struct page *page, unsigned int pageofs,
+				      loff_t start, unsigned int len)
+{
+	struct inode *const inode = page->mapping->host;
+	erofs_off_t offset = EROFS_I(inode)->z_fragmentoff + start;
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	u8 *src, *dst;
+	unsigned int i, cnt;
+
+	for (i = 0; i < len; i += cnt) {
+		cnt = min(EROFS_BLKSIZ - (unsigned int)erofs_blkoff(offset),
+			  len - i);
+		src = erofs_bread(&buf, EROFS_I_SB(inode)->packed_inode,
+				  erofs_blknr(offset), EROFS_KMAP);
+		if (IS_ERR(src))
+			return PTR_ERR(src);
+
+		dst = kmap_local_page(page);
+		memcpy(dst + pageofs + i, src + erofs_blkoff(offset), cnt);
+		kunmap_local(dst);
+
+		offset += cnt;
+	}
+	erofs_put_metabuf(&buf);
+	return 0;
+}
+
 static int z_erofs_do_read_page(struct z_erofs_decompress_frontend *fe,
 				struct page *page, struct page **pagepool)
 {
@@ -695,7 +722,8 @@ static int z_erofs_do_read_page(struct z_erofs_decompress_frontend *fe,
 		/* didn't get a valid pcluster previously (very rare) */
 	}
 
-	if (!(map->m_flags & EROFS_MAP_MAPPED))
+	if (!(map->m_flags & EROFS_MAP_MAPPED) ||
+	    map->m_flags & EROFS_MAP_FRAGMENT)
 		goto hitted;
 
 	err = z_erofs_collector_begin(fe);
@@ -742,6 +770,24 @@ static int z_erofs_do_read_page(struct z_erofs_decompress_frontend *fe,
 		zero_user_segment(page, cur, end);
 		goto next_part;
 	}
+	if (map->m_flags & EROFS_MAP_FRAGMENT) {
+		unsigned int pageofs, skip, len;
+
+		if (map->m_la < offset) {
+			pageofs = 0;
+			skip = offset - map->m_la;
+		} else {
+			pageofs = map->m_la & ~PAGE_MASK;
+			skip = 0;
+		}
+		len = min((unsigned int)(map->m_llen - skip), end - cur);
+		err = z_erofs_read_fragment_data(page, pageofs, skip, len);
+		if (err)
+			goto out;
+		++spiltted;
+		tight = false;
+		goto next_part;
+	}
 
 	exclusive = (!cur && (!spiltted || tight));
 	if (cur)
diff --git a/fs/erofs/zmap.c b/fs/erofs/zmap.c
index d58549ca1df9..2863ed6c6df9 100644
--- a/fs/erofs/zmap.c
+++ b/fs/erofs/zmap.c
@@ -123,6 +123,23 @@ static int z_erofs_fill_inode_lazy(struct inode *inode)
 		if (err < 0)
 			goto out_unlock;
 	}
+	if (vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER) {
+		vi->z_fragmentoff = le32_to_cpu(h->h_fragmentoff);
+
+		if (h->h_clusterbits >> Z_EROFS_FRAGMENT_INODE_BIT) {
+			vi->z_tailextent_headlcn = 0;
+		} else {
+			struct erofs_map_blocks map = {
+				.buf = __EROFS_BUF_INITIALIZER
+			};
+
+			err = z_erofs_do_map_blocks(inode, &map,
+						    EROFS_GET_BLOCKS_FINDTAIL);
+			erofs_put_metabuf(&map.buf);
+			if (err < 0)
+				goto out_unlock;
+		}
+	}
 	/* paired with smp_mb() at the beginning of the function */
 	smp_mb();
 	set_bit(EROFS_I_Z_INITED_BIT, &vi->flags);
@@ -598,6 +615,7 @@ static int z_erofs_do_map_blocks(struct inode *inode,
 {
 	struct erofs_inode *const vi = EROFS_I(inode);
 	bool ztailpacking = vi->z_advise & Z_EROFS_ADVISE_INLINE_PCLUSTER;
+	bool inpacked = vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER;
 	struct z_erofs_maprecorder m = {
 		.inode = inode,
 		.map = map,
@@ -672,6 +690,9 @@ static int z_erofs_do_map_blocks(struct inode *inode,
 		map->m_flags |= EROFS_MAP_META;
 		map->m_pa = vi->z_idataoff;
 		map->m_plen = vi->z_idata_size;
+	} else if (inpacked && m.lcn == vi->z_tailextent_headlcn) {
+		map->m_flags |= EROFS_MAP_FRAGMENT;
+		DBG_BUGON(!map->m_la);
 	} else {
 		map->m_pa = blknr_to_addr(m.pblk);
 		err = z_erofs_get_extent_compressedlen(&m, initial_lcn);
@@ -709,6 +730,7 @@ int z_erofs_map_blocks_iter(struct inode *inode,
 			    struct erofs_map_blocks *map,
 			    int flags)
 {
+	struct erofs_inode *const vi = EROFS_I(inode);
 	int err = 0;
 
 	trace_z_erofs_map_blocks_iter_enter(inode, map, flags);
@@ -725,6 +747,14 @@ int z_erofs_map_blocks_iter(struct inode *inode,
 	if (err)
 		goto out;
 
+	if ((vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER) &&
+	    !vi->z_tailextent_headlcn) {
+		map->m_llen = map->m_la + 1;
+		map->m_la = 0;
+		map->m_flags = EROFS_MAP_MAPPED | EROFS_MAP_FRAGMENT;
+		goto out;
+	}
+
 	err = z_erofs_do_map_blocks(inode, map, flags);
 out:
 	trace_z_erofs_map_blocks_iter_exit(inode, map, flags, err);
@@ -751,7 +781,8 @@ static int z_erofs_iomap_begin_report(struct inode *inode, loff_t offset,
 	iomap->length = map.m_llen;
 	if (map.m_flags & EROFS_MAP_MAPPED) {
 		iomap->type = IOMAP_MAPPED;
-		iomap->addr = map.m_pa;
+		iomap->addr = map.m_flags & EROFS_MAP_FRAGMENT ?
+			      IOMAP_NULL_ADDR : map.m_pa;
 	} else {
 		iomap->type = IOMAP_HOLE;
 		iomap->addr = IOMAP_NULL_ADDR;
-- 
2.17.1