[PATCH 1/4] erofs: simplify z_erofs_load_compact_lcluster()

[PATCH 1/4] erofs: simplify z_erofs_load_compact_lcluster()

[Gao Xiang]

 - Get rid of unpack_compacted_index() and fold it into
   z_erofs_load_compact_lcluster();

 - Avoid a goto.

Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
---
 fs/erofs/zmap.c | 89 ++++++++++++++++++++-----------------------------
 1 file changed, 36 insertions(+), 53 deletions(-)

diff --git a/fs/erofs/zmap.c b/fs/erofs/zmap.c
index 4535f2f0a014..b9e35089c9b8 100644
--- a/fs/erofs/zmap.c
+++ b/fs/erofs/zmap.c
@@ -97,17 +97,48 @@ static int get_compacted_la_distance(unsigned int lobits,
 	return d1;
 }
 
-static int unpack_compacted_index(struct z_erofs_maprecorder *m,
-				  unsigned int amortizedshift,
-				  erofs_off_t pos, bool lookahead)
+static int z_erofs_load_compact_lcluster(struct z_erofs_maprecorder *m,
+					 unsigned long lcn, bool lookahead)
 {
-	struct erofs_inode *const vi = EROFS_I(m->inode);
+	struct inode *const inode = m->inode;
+	struct erofs_inode *const vi = EROFS_I(inode);
+	const erofs_off_t ebase = sizeof(struct z_erofs_map_header) +
+		ALIGN(erofs_iloc(inode) + vi->inode_isize + vi->xattr_isize, 8);
 	const unsigned int lclusterbits = vi->z_logical_clusterbits;
+	const unsigned int totalidx = erofs_iblks(inode);
+	unsigned int compacted_4b_initial, compacted_2b, amortizedshift;
 	unsigned int vcnt, lo, lobits, encodebits, nblk, bytes;
-	bool big_pcluster;
+	bool big_pcluster = vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_1;
+	erofs_off_t pos;
 	u8 *in, type;
 	int i;
 
+	if (lcn >= totalidx || lclusterbits > 14)
+		return -EINVAL;
+
+	m->lcn = lcn;
+	/* used to align to 32-byte (compacted_2b) alignment */
+	compacted_4b_initial = ((32 - ebase % 32) / 4) & 7;
+	compacted_2b = 0;
+	if ((vi->z_advise & Z_EROFS_ADVISE_COMPACTED_2B) &&
+	    compacted_4b_initial < totalidx)
+		compacted_2b = rounddown(totalidx - compacted_4b_initial, 16);
+
+	pos = ebase;
+	amortizedshift = 2;	/* compact_4b */
+	if (lcn >= compacted_4b_initial) {
+		pos += compacted_4b_initial * 4;
+		lcn -= compacted_4b_initial;
+		if (lcn < compacted_2b) {
+			amortizedshift = 1;
+		} else {
+			pos += compacted_2b * 2;
+			lcn -= compacted_2b;
+		}
+	}
+	pos += lcn * (1 << amortizedshift);
+
+	/* figure out the lcluster count in this pack */
 	if (1 << amortizedshift == 4 && lclusterbits <= 14)
 		vcnt = 2;
 	else if (1 << amortizedshift == 2 && lclusterbits <= 12)
@@ -122,7 +153,6 @@ static int unpack_compacted_index(struct z_erofs_maprecorder *m,
 	/* it doesn't equal to round_up(..) */
 	m->nextpackoff = round_down(pos, vcnt << amortizedshift) +
 			 (vcnt << amortizedshift);
-	big_pcluster = vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_1;
 	lobits = max(lclusterbits, ilog2(Z_EROFS_LI_D0_CBLKCNT) + 1U);
 	encodebits = ((vcnt << amortizedshift) - sizeof(__le32)) * 8 / vcnt;
 	bytes = pos & ((vcnt << amortizedshift) - 1);
@@ -207,53 +237,6 @@ static int unpack_compacted_index(struct z_erofs_maprecorder *m,
 	return 0;
 }
 
-static int z_erofs_load_compact_lcluster(struct z_erofs_maprecorder *m,
-					 unsigned long lcn, bool lookahead)
-{
-	struct inode *const inode = m->inode;
-	struct erofs_inode *const vi = EROFS_I(inode);
-	const erofs_off_t ebase = sizeof(struct z_erofs_map_header) +
-		ALIGN(erofs_iloc(inode) + vi->inode_isize + vi->xattr_isize, 8);
-	unsigned int totalidx = erofs_iblks(inode);
-	unsigned int compacted_4b_initial, compacted_2b;
-	unsigned int amortizedshift;
-	erofs_off_t pos;
-
-	if (lcn >= totalidx || vi->z_logical_clusterbits > 14)
-		return -EINVAL;
-
-	m->lcn = lcn;
-	/* used to align to 32-byte (compacted_2b) alignment */
-	compacted_4b_initial = (32 - ebase % 32) / 4;
-	if (compacted_4b_initial == 32 / 4)
-		compacted_4b_initial = 0;
-
-	if ((vi->z_advise & Z_EROFS_ADVISE_COMPACTED_2B) &&
-	    compacted_4b_initial < totalidx)
-		compacted_2b = rounddown(totalidx - compacted_4b_initial, 16);
-	else
-		compacted_2b = 0;
-
-	pos = ebase;
-	if (lcn < compacted_4b_initial) {
-		amortizedshift = 2;
-		goto out;
-	}
-	pos += compacted_4b_initial * 4;
-	lcn -= compacted_4b_initial;
-
-	if (lcn < compacted_2b) {
-		amortizedshift = 1;
-		goto out;
-	}
-	pos += compacted_2b * 2;
-	lcn -= compacted_2b;
-	amortizedshift = 2;
-out:
-	pos += lcn * (1 << amortizedshift);
-	return unpack_compacted_index(m, amortizedshift, pos, lookahead);
-}
-
 static int z_erofs_load_lcluster_from_disk(struct z_erofs_maprecorder *m,
 					   unsigned int lcn, bool lookahead)
 {
-- 
2.43.5

[PATCH 2/4] erofs: get rid of `z_erofs_next_pcluster_t`

[Gao Xiang]

It was originally intended for tagged pointer reservation.

Now all encoded data can be represented uniformally with
`struct z_erofs_pcluster` as described in commit bf1aa03980f4
("erofs: sunset `struct erofs_workgroup`"), let's drop it too.

Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
---
 fs/erofs/zdata.c | 80 ++++++++++++++++++------------------------------
 1 file changed, 29 insertions(+), 51 deletions(-)

diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index 19ef4ff2a134..8bafc4d9edbe 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -12,12 +12,6 @@
 #define Z_EROFS_PCLUSTER_MAX_PAGES	(Z_EROFS_PCLUSTER_MAX_SIZE / PAGE_SIZE)
 #define Z_EROFS_INLINE_BVECS		2
 
-/*
- * let's leave a type here in case of introducing
- * another tagged pointer later.
- */
-typedef void *z_erofs_next_pcluster_t;
-
 struct z_erofs_bvec {
 	struct page *page;
 	int offset;
@@ -48,7 +42,7 @@ struct z_erofs_pcluster {
 	struct lockref lockref;
 
 	/* A: point to next chained pcluster or TAILs */
-	z_erofs_next_pcluster_t next;
+	struct z_erofs_pcluster *next;
 
 	/* I: start block address of this pcluster */
 	erofs_off_t index;
@@ -94,12 +88,11 @@ struct z_erofs_pcluster {
 
 /* the end of a chain of pclusters */
 #define Z_EROFS_PCLUSTER_TAIL           ((void *) 0x700 + POISON_POINTER_DELTA)
-#define Z_EROFS_PCLUSTER_NIL            (NULL)
 
 struct z_erofs_decompressqueue {
 	struct super_block *sb;
+	struct z_erofs_pcluster *head;
 	atomic_t pending_bios;
-	z_erofs_next_pcluster_t head;
 
 	union {
 		struct completion done;
@@ -493,8 +486,7 @@ struct z_erofs_decompress_frontend {
 
 	struct page *pagepool;
 	struct page *candidate_bvpage;
-	struct z_erofs_pcluster *pcl;
-	z_erofs_next_pcluster_t owned_head;
+	struct z_erofs_pcluster *pcl, *head;
 	enum z_erofs_pclustermode mode;
 
 	erofs_off_t headoffset;
@@ -504,7 +496,7 @@ struct z_erofs_decompress_frontend {
 };
 
 #define DECOMPRESS_FRONTEND_INIT(__i) { \
-	.inode = __i, .owned_head = Z_EROFS_PCLUSTER_TAIL, \
+	.inode = __i, .head = Z_EROFS_PCLUSTER_TAIL, \
 	.mode = Z_EROFS_PCLUSTER_FOLLOWED }
 
 static bool z_erofs_should_alloc_cache(struct z_erofs_decompress_frontend *fe)
@@ -752,9 +744,7 @@ static int z_erofs_register_pcluster(struct z_erofs_decompress_frontend *fe)
 	pcl->algorithmformat = map->m_algorithmformat;
 	pcl->length = 0;
 	pcl->partial = true;
-
-	/* new pclusters should be claimed as type 1, primary and followed */
-	pcl->next = fe->owned_head;
+	pcl->next = fe->head;
 	pcl->pageofs_out = map->m_la & ~PAGE_MASK;
 	fe->mode = Z_EROFS_PCLUSTER_FOLLOWED;
 
@@ -790,8 +780,7 @@ static int z_erofs_register_pcluster(struct z_erofs_decompress_frontend *fe)
 			goto err_out;
 		}
 	}
-	fe->owned_head = &pcl->next;
-	fe->pcl = pcl;
+	fe->head = fe->pcl = pcl;
 	return 0;
 
 err_out:
@@ -810,7 +799,7 @@ static int z_erofs_pcluster_begin(struct z_erofs_decompress_frontend *fe)
 
 	DBG_BUGON(fe->pcl);
 	/* must be Z_EROFS_PCLUSTER_TAIL or pointed to previous pcluster */
-	DBG_BUGON(fe->owned_head == Z_EROFS_PCLUSTER_NIL);
+	DBG_BUGON(!fe->head);
 
 	if (!(map->m_flags & EROFS_MAP_META)) {
 		while (1) {
@@ -838,10 +827,9 @@ static int z_erofs_pcluster_begin(struct z_erofs_decompress_frontend *fe)
 	if (ret == -EEXIST) {
 		mutex_lock(&fe->pcl->lock);
 		/* check if this pcluster hasn't been linked into any chain. */
-		if (cmpxchg(&fe->pcl->next, Z_EROFS_PCLUSTER_NIL,
-			    fe->owned_head) == Z_EROFS_PCLUSTER_NIL) {
+		if (!cmpxchg(&fe->pcl->next, NULL, fe->head)) {
 			/* .. so it can be attached to our submission chain */
-			fe->owned_head = &fe->pcl->next;
+			fe->head = fe->pcl;
 			fe->mode = Z_EROFS_PCLUSTER_FOLLOWED;
 		} else {	/* otherwise, it belongs to an inflight chain */
 			fe->mode = Z_EROFS_PCLUSTER_INFLIGHT;
@@ -1394,7 +1382,7 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be,
 	pcl->vcnt = 0;
 
 	/* pcluster lock MUST be taken before the following line */
-	WRITE_ONCE(pcl->next, Z_EROFS_PCLUSTER_NIL);
+	WRITE_ONCE(pcl->next, NULL);
 	mutex_unlock(&pcl->lock);
 
 	if (z_erofs_is_inline_pcluster(pcl))
@@ -1412,16 +1400,14 @@ static int z_erofs_decompress_queue(const struct z_erofs_decompressqueue *io,
 		.pagepool = pagepool,
 		.decompressed_secondary_bvecs =
 			LIST_HEAD_INIT(be.decompressed_secondary_bvecs),
+		.pcl = io->head,
 	};
-	z_erofs_next_pcluster_t owned = io->head;
+	struct z_erofs_pcluster *next;
 	int err = io->eio ? -EIO : 0;
 
-	while (owned != Z_EROFS_PCLUSTER_TAIL) {
-		DBG_BUGON(owned == Z_EROFS_PCLUSTER_NIL);
-
-		be.pcl = container_of(owned, struct z_erofs_pcluster, next);
-		owned = READ_ONCE(be.pcl->next);
-
+	for (; be.pcl != Z_EROFS_PCLUSTER_TAIL; be.pcl = next) {
+		DBG_BUGON(!be.pcl);
+		next = READ_ONCE(be.pcl->next);
 		err = z_erofs_decompress_pcluster(&be, err) ?: err;
 	}
 	return err;
@@ -1631,18 +1617,13 @@ enum {
 	NR_JOBQUEUES,
 };
 
-static void move_to_bypass_jobqueue(struct z_erofs_pcluster *pcl,
-				    z_erofs_next_pcluster_t qtail[],
-				    z_erofs_next_pcluster_t owned_head)
+static void z_erofs_move_to_bypass_queue(struct z_erofs_pcluster *pcl,
+					 struct z_erofs_pcluster *next,
+					 struct z_erofs_pcluster **qtail[])
 {
-	z_erofs_next_pcluster_t *const submit_qtail = qtail[JQ_SUBMIT];
-	z_erofs_next_pcluster_t *const bypass_qtail = qtail[JQ_BYPASS];
-
 	WRITE_ONCE(pcl->next, Z_EROFS_PCLUSTER_TAIL);
-
-	WRITE_ONCE(*submit_qtail, owned_head);
-	WRITE_ONCE(*bypass_qtail, &pcl->next);
-
+	WRITE_ONCE(*qtail[JQ_SUBMIT], next);
+	WRITE_ONCE(*qtail[JQ_BYPASS], pcl);
 	qtail[JQ_BYPASS] = &pcl->next;
 }
 
@@ -1677,9 +1658,9 @@ static void z_erofs_submit_queue(struct z_erofs_decompress_frontend *f,
 {
 	struct super_block *sb = f->inode->i_sb;
 	struct address_space *mc = MNGD_MAPPING(EROFS_SB(sb));
-	z_erofs_next_pcluster_t qtail[NR_JOBQUEUES];
+	struct z_erofs_pcluster **qtail[NR_JOBQUEUES];
 	struct z_erofs_decompressqueue *q[NR_JOBQUEUES];
-	z_erofs_next_pcluster_t owned_head = f->owned_head;
+	struct z_erofs_pcluster *pcl, *next;
 	/* bio is NULL initially, so no need to initialize last_{index,bdev} */
 	erofs_off_t last_pa;
 	unsigned int nr_bios = 0;
@@ -1695,22 +1676,19 @@ static void z_erofs_submit_queue(struct z_erofs_decompress_frontend *f,
 	qtail[JQ_SUBMIT] = &q[JQ_SUBMIT]->head;
 
 	/* by default, all need io submission */
-	q[JQ_SUBMIT]->head = owned_head;
+	q[JQ_SUBMIT]->head = next = f->head;
 
 	do {
 		struct erofs_map_dev mdev;
-		struct z_erofs_pcluster *pcl;
 		erofs_off_t cur, end;
 		struct bio_vec bvec;
 		unsigned int i = 0;
 		bool bypass = true;
 
-		DBG_BUGON(owned_head == Z_EROFS_PCLUSTER_NIL);
-		pcl = container_of(owned_head, struct z_erofs_pcluster, next);
-		owned_head = READ_ONCE(pcl->next);
-
+		pcl = next;
+		next = READ_ONCE(pcl->next);
 		if (z_erofs_is_inline_pcluster(pcl)) {
-			move_to_bypass_jobqueue(pcl, qtail, owned_head);
+			z_erofs_move_to_bypass_queue(pcl, next, qtail);
 			continue;
 		}
 
@@ -1782,8 +1760,8 @@ static void z_erofs_submit_queue(struct z_erofs_decompress_frontend *f,
 		if (!bypass)
 			qtail[JQ_SUBMIT] = &pcl->next;
 		else
-			move_to_bypass_jobqueue(pcl, qtail, owned_head);
-	} while (owned_head != Z_EROFS_PCLUSTER_TAIL);
+			z_erofs_move_to_bypass_queue(pcl, next, qtail);
+	} while (next != Z_EROFS_PCLUSTER_TAIL);
 
 	if (bio) {
 		if (erofs_is_fileio_mode(EROFS_SB(sb)))
@@ -1815,7 +1793,7 @@ static int z_erofs_runqueue(struct z_erofs_decompress_frontend *f,
 	bool force_fg = z_erofs_is_sync_decompress(sbi, ra_folios);
 	int err;
 
-	if (f->owned_head == Z_EROFS_PCLUSTER_TAIL)
+	if (f->head == Z_EROFS_PCLUSTER_TAIL)
 		return 0;
 	z_erofs_submit_queue(f, io, &force_fg, !!ra_folios);
 
-- 
2.43.5

[PATCH 3/4] erofs: tidy up zdata.c

[Gao Xiang]

All small code style adjustments, no logic changes:

 - z_erofs_decompress_frontend => z_erofs_frontend;
 - z_erofs_decompress_backend => z_erofs_backend;
 - Use Z_EROFS_DEFINE_FRONTEND() to replace DECOMPRESS_FRONTEND_INIT();
 - `nr_folios` should be `nrpages` in z_erofs_readahead();
 - Refine in-line comments.

Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
---
 fs/erofs/zdata.c | 111 ++++++++++++++++++-----------------------------
 1 file changed, 43 insertions(+), 68 deletions(-)

diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index 8bafc4d9edbe..dcbb692ac058 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -455,31 +455,25 @@ int __init z_erofs_init_subsystem(void)
 }
 
 enum z_erofs_pclustermode {
+	/* It has previously been linked into another processing chain */
 	Z_EROFS_PCLUSTER_INFLIGHT,
 	/*
-	 * a weak form of Z_EROFS_PCLUSTER_FOLLOWED, the difference is that it
-	 * could be dispatched into bypass queue later due to uptodated managed
-	 * pages. All related online pages cannot be reused for inplace I/O (or
-	 * bvpage) since it can be directly decoded without I/O submission.
+	 * A weaker form of Z_EROFS_PCLUSTER_FOLLOWED; the difference is that it
+	 * may be dispatched to the bypass queue later due to uptodated managed
+	 * folios.  All file-backed folios related to this pcluster cannot be
+	 * reused for in-place I/O (or bvpage) since the pcluster may be decoded
+	 * in a separate queue (and thus out of order).
 	 */
 	Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE,
 	/*
-	 * The pcluster was just linked to a decompression chain by us.  It can
-	 * also be linked with the remaining pclusters, which means if the
-	 * processing page is the tail page of a pcluster, this pcluster can
-	 * safely use the whole page (since the previous pcluster is within the
-	 * same chain) for in-place I/O, as illustrated below:
-	 *  ___________________________________________________
-	 * |  tail (partial) page  |    head (partial) page    |
-	 * |  (of the current pcl) |   (of the previous pcl)   |
-	 * |___PCLUSTER_FOLLOWED___|_____PCLUSTER_FOLLOWED_____|
-	 *
-	 * [  (*) the page above can be used as inplace I/O.   ]
+	 * The pcluster has just been linked to our processing chain.
+	 * File-backed folios (except for the head page) related to it can be
+	 * used for in-place I/O (or bvpage).
 	 */
 	Z_EROFS_PCLUSTER_FOLLOWED,
 };
 
-struct z_erofs_decompress_frontend {
+struct z_erofs_frontend {
 	struct inode *const inode;
 	struct erofs_map_blocks map;
 	struct z_erofs_bvec_iter biter;
@@ -495,11 +489,11 @@ struct z_erofs_decompress_frontend {
 	unsigned int icur;
 };
 
-#define DECOMPRESS_FRONTEND_INIT(__i) { \
-	.inode = __i, .head = Z_EROFS_PCLUSTER_TAIL, \
-	.mode = Z_EROFS_PCLUSTER_FOLLOWED }
+#define Z_EROFS_DEFINE_FRONTEND(fe, i, ho) struct z_erofs_frontend fe = { \
+	.inode = i, .head = Z_EROFS_PCLUSTER_TAIL, \
+	.mode = Z_EROFS_PCLUSTER_FOLLOWED, .headoffset = ho }
 
-static bool z_erofs_should_alloc_cache(struct z_erofs_decompress_frontend *fe)
+static bool z_erofs_should_alloc_cache(struct z_erofs_frontend *fe)
 {
 	unsigned int cachestrategy = EROFS_I_SB(fe->inode)->opt.cache_strategy;
 
@@ -516,7 +510,7 @@ static bool z_erofs_should_alloc_cache(struct z_erofs_decompress_frontend *fe)
 	return false;
 }
 
-static void z_erofs_bind_cache(struct z_erofs_decompress_frontend *fe)
+static void z_erofs_bind_cache(struct z_erofs_frontend *fe)
 {
 	struct address_space *mc = MNGD_MAPPING(EROFS_I_SB(fe->inode));
 	struct z_erofs_pcluster *pcl = fe->pcl;
@@ -673,7 +667,7 @@ int erofs_init_managed_cache(struct super_block *sb)
 }
 
 /* callers must be with pcluster lock held */
-static int z_erofs_attach_page(struct z_erofs_decompress_frontend *fe,
+static int z_erofs_attach_page(struct z_erofs_frontend *fe,
 			       struct z_erofs_bvec *bvec, bool exclusive)
 {
 	struct z_erofs_pcluster *pcl = fe->pcl;
@@ -719,7 +713,7 @@ static bool z_erofs_get_pcluster(struct z_erofs_pcluster *pcl)
 	return true;
 }
 
-static int z_erofs_register_pcluster(struct z_erofs_decompress_frontend *fe)
+static int z_erofs_register_pcluster(struct z_erofs_frontend *fe)
 {
 	struct erofs_map_blocks *map = &fe->map;
 	struct super_block *sb = fe->inode->i_sb;
@@ -789,7 +783,7 @@ static int z_erofs_register_pcluster(struct z_erofs_decompress_frontend *fe)
 	return err;
 }
 
-static int z_erofs_pcluster_begin(struct z_erofs_decompress_frontend *fe)
+static int z_erofs_pcluster_begin(struct z_erofs_frontend *fe)
 {
 	struct erofs_map_blocks *map = &fe->map;
 	struct super_block *sb = fe->inode->i_sb;
@@ -862,14 +856,9 @@ static int z_erofs_pcluster_begin(struct z_erofs_decompress_frontend *fe)
 	return 0;
 }
 
-/*
- * keep in mind that no referenced pclusters will be freed
- * only after a RCU grace period.
- */
 static void z_erofs_rcu_callback(struct rcu_head *head)
 {
-	z_erofs_free_pcluster(container_of(head,
-			struct z_erofs_pcluster, rcu));
+	z_erofs_free_pcluster(container_of(head, struct z_erofs_pcluster, rcu));
 }
 
 static bool __erofs_try_to_release_pcluster(struct erofs_sb_info *sbi,
@@ -911,8 +900,7 @@ static bool erofs_try_to_release_pcluster(struct erofs_sb_info *sbi,
 	return free;
 }
 
-unsigned long z_erofs_shrink_scan(struct erofs_sb_info *sbi,
-				  unsigned long nr_shrink)
+unsigned long z_erofs_shrink_scan(struct erofs_sb_info *sbi, unsigned long nr)
 {
 	struct z_erofs_pcluster *pcl;
 	unsigned int freed = 0;
@@ -926,7 +914,7 @@ unsigned long z_erofs_shrink_scan(struct erofs_sb_info *sbi,
 		xa_unlock(&sbi->managed_pslots);
 
 		++freed;
-		if (!--nr_shrink)
+		if (!--nr)
 			return freed;
 		xa_lock(&sbi->managed_pslots);
 	}
@@ -955,7 +943,7 @@ static void z_erofs_put_pcluster(struct erofs_sb_info *sbi,
 		call_rcu(&pcl->rcu, z_erofs_rcu_callback);
 }
 
-static void z_erofs_pcluster_end(struct z_erofs_decompress_frontend *fe)
+static void z_erofs_pcluster_end(struct z_erofs_frontend *fe)
 {
 	struct z_erofs_pcluster *pcl = fe->pcl;
 
@@ -968,13 +956,9 @@ static void z_erofs_pcluster_end(struct z_erofs_decompress_frontend *fe)
 	if (fe->candidate_bvpage)
 		fe->candidate_bvpage = NULL;
 
-	/*
-	 * if all pending pages are added, don't hold its reference
-	 * any longer if the pcluster isn't hosted by ourselves.
-	 */
+	/* Drop refcount if it doesn't belong to our processing chain */
 	if (fe->mode < Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE)
 		z_erofs_put_pcluster(EROFS_I_SB(fe->inode), pcl, false);
-
 	fe->pcl = NULL;
 }
 
@@ -1003,7 +987,7 @@ static int z_erofs_read_fragment(struct super_block *sb, struct folio *folio,
 	return 0;
 }
 
-static int z_erofs_scan_folio(struct z_erofs_decompress_frontend *f,
+static int z_erofs_scan_folio(struct z_erofs_frontend *f,
 			      struct folio *folio, bool ra)
 {
 	struct inode *const inode = f->inode;
@@ -1118,7 +1102,7 @@ static bool z_erofs_page_is_invalidated(struct page *page)
 	return !page_folio(page)->mapping && !z_erofs_is_shortlived_page(page);
 }
 
-struct z_erofs_decompress_backend {
+struct z_erofs_backend {
 	struct page *onstack_pages[Z_EROFS_ONSTACK_PAGES];
 	struct super_block *sb;
 	struct z_erofs_pcluster *pcl;
@@ -1138,7 +1122,7 @@ struct z_erofs_bvec_item {
 	struct list_head list;
 };
 
-static void z_erofs_do_decompressed_bvec(struct z_erofs_decompress_backend *be,
+static void z_erofs_do_decompressed_bvec(struct z_erofs_backend *be,
 					 struct z_erofs_bvec *bvec)
 {
 	struct z_erofs_bvec_item *item;
@@ -1161,8 +1145,7 @@ static void z_erofs_do_decompressed_bvec(struct z_erofs_decompress_backend *be,
 	list_add(&item->list, &be->decompressed_secondary_bvecs);
 }
 
-static void z_erofs_fill_other_copies(struct z_erofs_decompress_backend *be,
-				      int err)
+static void z_erofs_fill_other_copies(struct z_erofs_backend *be, int err)
 {
 	unsigned int off0 = be->pcl->pageofs_out;
 	struct list_head *p, *n;
@@ -1203,7 +1186,7 @@ static void z_erofs_fill_other_copies(struct z_erofs_decompress_backend *be,
 	}
 }
 
-static void z_erofs_parse_out_bvecs(struct z_erofs_decompress_backend *be)
+static void z_erofs_parse_out_bvecs(struct z_erofs_backend *be)
 {
 	struct z_erofs_pcluster *pcl = be->pcl;
 	struct z_erofs_bvec_iter biter;
@@ -1228,8 +1211,7 @@ static void z_erofs_parse_out_bvecs(struct z_erofs_decompress_backend *be)
 		z_erofs_put_shortlivedpage(be->pagepool, old_bvpage);
 }
 
-static int z_erofs_parse_in_bvecs(struct z_erofs_decompress_backend *be,
-				  bool *overlapped)
+static int z_erofs_parse_in_bvecs(struct z_erofs_backend *be, bool *overlapped)
 {
 	struct z_erofs_pcluster *pcl = be->pcl;
 	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
@@ -1264,8 +1246,7 @@ static int z_erofs_parse_in_bvecs(struct z_erofs_decompress_backend *be,
 	return err;
 }
 
-static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be,
-				       int err)
+static int z_erofs_decompress_pcluster(struct z_erofs_backend *be, int err)
 {
 	struct erofs_sb_info *const sbi = EROFS_SB(be->sb);
 	struct z_erofs_pcluster *pcl = be->pcl;
@@ -1395,7 +1376,7 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be,
 static int z_erofs_decompress_queue(const struct z_erofs_decompressqueue *io,
 				    struct page **pagepool)
 {
-	struct z_erofs_decompress_backend be = {
+	struct z_erofs_backend be = {
 		.sb = io->sb,
 		.pagepool = pagepool,
 		.decompressed_secondary_bvecs =
@@ -1473,7 +1454,7 @@ static void z_erofs_decompress_kickoff(struct z_erofs_decompressqueue *io,
 }
 
 static void z_erofs_fill_bio_vec(struct bio_vec *bvec,
-				 struct z_erofs_decompress_frontend *f,
+				 struct z_erofs_frontend *f,
 				 struct z_erofs_pcluster *pcl,
 				 unsigned int nr,
 				 struct address_space *mc)
@@ -1652,7 +1633,7 @@ static void z_erofs_endio(struct bio *bio)
 		bio_put(bio);
 }
 
-static void z_erofs_submit_queue(struct z_erofs_decompress_frontend *f,
+static void z_erofs_submit_queue(struct z_erofs_frontend *f,
 				 struct z_erofs_decompressqueue *fgq,
 				 bool *force_fg, bool readahead)
 {
@@ -1785,17 +1766,16 @@ static void z_erofs_submit_queue(struct z_erofs_decompress_frontend *f,
 	z_erofs_decompress_kickoff(q[JQ_SUBMIT], nr_bios);
 }
 
-static int z_erofs_runqueue(struct z_erofs_decompress_frontend *f,
-			    unsigned int ra_folios)
+static int z_erofs_runqueue(struct z_erofs_frontend *f, unsigned int rapages)
 {
 	struct z_erofs_decompressqueue io[NR_JOBQUEUES];
 	struct erofs_sb_info *sbi = EROFS_I_SB(f->inode);
-	bool force_fg = z_erofs_is_sync_decompress(sbi, ra_folios);
+	bool force_fg = z_erofs_is_sync_decompress(sbi, rapages);
 	int err;
 
 	if (f->head == Z_EROFS_PCLUSTER_TAIL)
 		return 0;
-	z_erofs_submit_queue(f, io, &force_fg, !!ra_folios);
+	z_erofs_submit_queue(f, io, &force_fg, !!rapages);
 
 	/* handle bypass queue (no i/o pclusters) immediately */
 	err = z_erofs_decompress_queue(&io[JQ_BYPASS], &f->pagepool);
@@ -1813,7 +1793,7 @@ static int z_erofs_runqueue(struct z_erofs_decompress_frontend *f,
  * Since partial uptodate is still unimplemented for now, we have to use
  * approximate readmore strategies as a start.
  */
-static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f,
+static void z_erofs_pcluster_readmore(struct z_erofs_frontend *f,
 		struct readahead_control *rac, bool backmost)
 {
 	struct inode *inode = f->inode;
@@ -1868,12 +1848,10 @@ static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f,
 static int z_erofs_read_folio(struct file *file, struct folio *folio)
 {
 	struct inode *const inode = folio->mapping->host;
-	struct z_erofs_decompress_frontend f = DECOMPRESS_FRONTEND_INIT(inode);
+	Z_EROFS_DEFINE_FRONTEND(f, inode, folio_pos(folio));
 	int err;
 
 	trace_erofs_read_folio(folio, false);
-	f.headoffset = (erofs_off_t)folio->index << PAGE_SHIFT;
-
 	z_erofs_pcluster_readmore(&f, NULL, true);
 	err = z_erofs_scan_folio(&f, folio, false);
 	z_erofs_pcluster_readmore(&f, NULL, false);
@@ -1893,17 +1871,14 @@ static int z_erofs_read_folio(struct file *file, struct folio *folio)
 static void z_erofs_readahead(struct readahead_control *rac)
 {
 	struct inode *const inode = rac->mapping->host;
-	struct z_erofs_decompress_frontend f = DECOMPRESS_FRONTEND_INIT(inode);
+	Z_EROFS_DEFINE_FRONTEND(f, inode, readahead_pos(rac));
 	struct folio *head = NULL, *folio;
-	unsigned int nr_folios;
+	unsigned int nrpages = readahead_count(rac);
 	int err;
 
-	f.headoffset = readahead_pos(rac);
-
 	z_erofs_pcluster_readmore(&f, rac, true);
-	nr_folios = readahead_count(rac);
-	trace_erofs_readpages(inode, readahead_index(rac), nr_folios, false);
-
+	nrpages = readahead_count(rac);
+	trace_erofs_readpages(inode, readahead_index(rac), nrpages, false);
 	while ((folio = readahead_folio(rac))) {
 		folio->private = head;
 		head = folio;
@@ -1922,7 +1897,7 @@ static void z_erofs_readahead(struct readahead_control *rac)
 	z_erofs_pcluster_readmore(&f, rac, false);
 	z_erofs_pcluster_end(&f);
 
-	(void)z_erofs_runqueue(&f, nr_folios);
+	(void)z_erofs_runqueue(&f, nrpages);
 	erofs_put_metabuf(&f.map.buf);
 	erofs_release_pages(&f.pagepool);
 }
-- 
2.43.5

[PATCH 4/4] erofs: convert z_erofs_bind_cache() to folios

[Gao Xiang]

The managed cache uses a pseudo inode to keep (necessary) compressed
data.

Currently, it still uses zero-order folios, so this is just a trivial
conversion, except that the use of the pagepool is temporarily dropped.

Drop some obsoleted comments too.

Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
---
 fs/erofs/compress.h | 23 +-------------------
 fs/erofs/zdata.c    | 51 ++++++++++++++++++---------------------------
 2 files changed, 21 insertions(+), 53 deletions(-)

diff --git a/fs/erofs/compress.h b/fs/erofs/compress.h
index 7bfe251680ec..65ff39401020 100644
--- a/fs/erofs/compress.h
+++ b/fs/erofs/compress.h
@@ -29,29 +29,8 @@ struct z_erofs_decompressor {
 	char *name;
 };
 
-/* some special page->private (unsigned long, see below) */
 #define Z_EROFS_SHORTLIVED_PAGE		(-1UL << 2)
-#define Z_EROFS_PREALLOCATED_PAGE	(-2UL << 2)
-
-/*
- * For all pages in a pcluster, page->private should be one of
- * Type                         Last 2bits      page->private
- * short-lived page             00              Z_EROFS_SHORTLIVED_PAGE
- * preallocated page (tryalloc) 00              Z_EROFS_PREALLOCATED_PAGE
- * cached/managed page          00              pointer to z_erofs_pcluster
- * online page (file-backed,    01/10/11        sub-index << 2 | count
- *              some pages can be used for inplace I/O)
- *
- * page->mapping should be one of
- * Type                 page->mapping
- * short-lived page     NULL
- * preallocated page    NULL
- * cached/managed page  non-NULL or NULL (invalidated/truncated page)
- * online page          non-NULL
- *
- * For all managed pages, PG_private should be set with 1 extra refcount,
- * which is used for page reclaim / migration.
- */
+#define Z_EROFS_PREALLOCATED_FOLIO	((void *)(-2UL << 2))
 
 /*
  * Currently, short-lived pages are pages directly from buddy system
diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index dcbb692ac058..b3fb18d1a9c8 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -516,13 +516,11 @@ static void z_erofs_bind_cache(struct z_erofs_frontend *fe)
 	struct z_erofs_pcluster *pcl = fe->pcl;
 	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
 	bool shouldalloc = z_erofs_should_alloc_cache(fe);
-	bool standalone = true;
-	/*
-	 * optimistic allocation without direct reclaim since inplace I/O
-	 * can be used if low memory otherwise.
-	 */
+	bool may_bypass = true;
+	/* Optimistic allocation, as in-place I/O can be used as a fallback */
 	gfp_t gfp = (mapping_gfp_mask(mc) & ~__GFP_DIRECT_RECLAIM) |
 			__GFP_NOMEMALLOC | __GFP_NORETRY | __GFP_NOWARN;
+	struct folio *folio, *newfolio;
 	unsigned int i;
 
 	if (i_blocksize(fe->inode) != PAGE_SIZE ||
@@ -530,47 +528,42 @@ static void z_erofs_bind_cache(struct z_erofs_frontend *fe)
 		return;
 
 	for (i = 0; i < pclusterpages; ++i) {
-		struct page *page, *newpage;
-
 		/* Inaccurate check w/o locking to avoid unneeded lookups */
 		if (READ_ONCE(pcl->compressed_bvecs[i].page))
 			continue;
 
-		page = find_get_page(mc, pcl->index + i);
-		if (!page) {
-			/* I/O is needed, no possible to decompress directly */
-			standalone = false;
+		folio = filemap_get_folio(mc, pcl->index + i);
+		if (IS_ERR(folio)) {
+			may_bypass = false;
 			if (!shouldalloc)
 				continue;
 
 			/*
-			 * Try cached I/O if allocation succeeds or fallback to
-			 * in-place I/O instead to avoid any direct reclaim.
+			 * Allocate a managed folio for cached I/O, or it may be
+			 * then filled with a file-backed folio for in-place I/O
 			 */
-			newpage = erofs_allocpage(&fe->pagepool, gfp);
-			if (!newpage)
+			newfolio = filemap_alloc_folio(gfp, 0);
+			if (!newfolio)
 				continue;
-			set_page_private(newpage, Z_EROFS_PREALLOCATED_PAGE);
+			newfolio->private = Z_EROFS_PREALLOCATED_FOLIO;
+			folio = NULL;
 		}
 		spin_lock(&pcl->lockref.lock);
 		if (!pcl->compressed_bvecs[i].page) {
-			pcl->compressed_bvecs[i].page = page ? page : newpage;
+			pcl->compressed_bvecs[i].page =
+				folio_page(folio ?: newfolio, 0);
 			spin_unlock(&pcl->lockref.lock);
 			continue;
 		}
 		spin_unlock(&pcl->lockref.lock);
-
-		if (page)
-			put_page(page);
-		else if (newpage)
-			erofs_pagepool_add(&fe->pagepool, newpage);
+		folio_put(folio ?: newfolio);
 	}
 
 	/*
-	 * don't do inplace I/O if all compressed pages are available in
-	 * managed cache since it can be moved to the bypass queue instead.
+	 * Don't perform in-place I/O if all compressed pages are available in
+	 * the managed cache, as the pcluster can be moved to the bypass queue.
 	 */
-	if (standalone)
+	if (may_bypass)
 		fe->mode = Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE;
 }
 
@@ -1481,12 +1474,8 @@ static void z_erofs_fill_bio_vec(struct bio_vec *bvec,
 	DBG_BUGON(z_erofs_is_shortlived_page(bvec->bv_page));
 
 	folio = page_folio(zbv.page);
-	/*
-	 * Handle preallocated cached folios.  We tried to allocate such folios
-	 * without triggering direct reclaim.  If allocation failed, inplace
-	 * file-backed folios will be used instead.
-	 */
-	if (folio->private == (void *)Z_EROFS_PREALLOCATED_PAGE) {
+	/* For preallocated managed folios, add them to page cache here */
+	if (folio->private == Z_EROFS_PREALLOCATED_FOLIO) {
 		tocache = true;
 		goto out_tocache;
 	}
-- 
2.43.5

Re: [PATCH 1/4] erofs: simplify z_erofs_load_compact_lcluster()

[Chao Yu]

On 1/14/25 11:44, Gao Xiang wrote:
>   - Get rid of unpack_compacted_index() and fold it into
>     z_erofs_load_compact_lcluster();
> 
>   - Avoid a goto.
> 
> Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>

Reviewed-by: Chao Yu <chao@kernel.org>

Thanks,

Re: [PATCH 2/4] erofs: get rid of `z_erofs_next_pcluster_t`

[Chao Yu]

On 1/14/25 11:44, Gao Xiang wrote:
> It was originally intended for tagged pointer reservation.
> 
> Now all encoded data can be represented uniformally with
> `struct z_erofs_pcluster` as described in commit bf1aa03980f4
> ("erofs: sunset `struct erofs_workgroup`"), let's drop it too.
> 
> Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>

Reviewed-by: Chao Yu <chao@kernel.org>

Thanks,

Re: [PATCH 3/4] erofs: tidy up zdata.c

[Chao Yu]

On 1/14/25 11:44, Gao Xiang wrote:
> All small code style adjustments, no logic changes:
> 
>   - z_erofs_decompress_frontend => z_erofs_frontend;
>   - z_erofs_decompress_backend => z_erofs_backend;
>   - Use Z_EROFS_DEFINE_FRONTEND() to replace DECOMPRESS_FRONTEND_INIT();
>   - `nr_folios` should be `nrpages` in z_erofs_readahead();
>   - Refine in-line comments.
> 
> Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>

Reviewed-by: Chao Yu <chao@kernel.org>

Thanks,

Re: [PATCH 4/4] erofs: convert z_erofs_bind_cache() to folios

[Chao Yu]

On 1/14/25 11:44, Gao Xiang wrote:
> The managed cache uses a pseudo inode to keep (necessary) compressed
> data.
> 
> Currently, it still uses zero-order folios, so this is just a trivial
> conversion, except that the use of the pagepool is temporarily dropped.
> 
> Drop some obsoleted comments too.
> 
> Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>

Reviewed-by: Chao Yu <chao@kernel.org>

Thanks,