Merge tag 'erofs-for-6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs

Pull erofs updates from Gao Xiang:
 "No outstanding new feature for this cycle.

  Most of these commits are decompression cleanups which are part of the
  ongoing development for subpage/folio compression support as well as
  xattr cleanups for the upcoming xattr bloom filter optimization [1].

  In addition, there are bugfixes to address some corner cases of
  compressed images due to global data de-duplication and arm64 16k
  pages.

  Summary:

   - Fix rare I/O hang on deduplicated compressed images due to loop
     hooked chains

   - Fix compact compression layout of 16k blocks on arm64 devices

   - Fix atomic context detection of async decompression

   - Decompression/Xattr code cleanups"

Link: https://lore.kernel.org/r/20230621083209.116024-1-jefflexu@linux.alibaba.com [1]

* tag 'erofs-for-6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
  erofs: clean up zmap.c
  erofs: remove unnecessary goto
  erofs: Fix detection of atomic context
  erofs: use separate xattr parsers for listxattr/getxattr
  erofs: unify inline/shared xattr iterators for listxattr/getxattr
  erofs: make the size of read data stored in buffer_ofs
  erofs: unify xattr_iter structures
  erofs: use absolute position in xattr iterator
  erofs: fix compact 4B support for 16k block size
  erofs: convert erofs_read_metabuf() to erofs_bread() for xattr
  erofs: use poison pointer to replace the hard-coded address
  erofs: use struct lockref to replace handcrafted approach
  erofs: adapt managed inode operations into folios
  erofs: kill hooked chains to avoid loops on deduplicated compressed images
  erofs: avoid on-stack pagepool directly passed by arguments
  erofs: allocate extra bvec pages directly instead of retrying
  erofs: clean up z_erofs_pcluster_readmore()
  erofs: remove the member readahead from struct z_erofs_decompress_frontend
  erofs: fold in z_erofs_decompress()

fs/erofs/compress.h

@@ -89,8 +89,7 @@ static inline bool erofs_page_is_managed(const struct erofs_sb_info *sbi,
 
 int z_erofs_fixup_insize(struct z_erofs_decompress_req *rq, const char *padbuf,
 			 unsigned int padbufsize);
-int z_erofs_decompress(struct z_erofs_decompress_req *rq,
-		       struct page **pagepool);
+extern const struct z_erofs_decompressor erofs_decompressors[];
 
 /* prototypes for specific algorithms */
 int z_erofs_lzma_decompress(struct z_erofs_decompress_req *rq,

fs/erofs/decompressor.c

@@ -363,7 +363,7 @@ static int z_erofs_transform_plain(struct z_erofs_decompress_req *rq,
 	return 0;
 }
 
-static struct z_erofs_decompressor decompressors[] = {
+const struct z_erofs_decompressor erofs_decompressors[] = {
 	[Z_EROFS_COMPRESSION_SHIFTED] = {
 		.decompress = z_erofs_transform_plain,
 		.name = "shifted"
@@ -383,9 +383,3 @@ static struct z_erofs_decompressor decompressors[] = {
 	},
 #endif
 };
-
-int z_erofs_decompress(struct z_erofs_decompress_req *rq,
-		       struct page **pagepool)
-{
-	return decompressors[rq->alg].decompress(rq, pagepool);
-}

fs/erofs/internal.h

@@ -208,46 +208,12 @@ enum {
 	EROFS_ZIP_CACHE_READAROUND
 };
 
-#define EROFS_LOCKED_MAGIC     (INT_MIN | 0xE0F510CCL)
-
 /* basic unit of the workstation of a super_block */
 struct erofs_workgroup {
-	/* the workgroup index in the workstation */
 	pgoff_t index;
-
-	/* overall workgroup reference count */
-	atomic_t refcount;
+	struct lockref lockref;
 };
 
-static inline bool erofs_workgroup_try_to_freeze(struct erofs_workgroup *grp,
-						 int val)
-{
-	preempt_disable();
-	if (val != atomic_cmpxchg(&grp->refcount, val, EROFS_LOCKED_MAGIC)) {
-		preempt_enable();
-		return false;
-	}
-	return true;
-}
-
-static inline void erofs_workgroup_unfreeze(struct erofs_workgroup *grp,
-					    int orig_val)
-{
-	/*
-	 * other observers should notice all modifications
-	 * in the freezing period.
-	 */
-	smp_mb();
-	atomic_set(&grp->refcount, orig_val);
-	preempt_enable();
-}
-
-static inline int erofs_wait_on_workgroup_freezed(struct erofs_workgroup *grp)
-{
-	return atomic_cond_read_relaxed(&grp->refcount,
-					VAL != EROFS_LOCKED_MAGIC);
-}
-
 enum erofs_kmap_type {
 	EROFS_NO_KMAP,		/* don't map the buffer */
 	EROFS_KMAP,		/* use kmap_local_page() to map the buffer */
@@ -486,7 +452,7 @@ static inline void erofs_pagepool_add(struct page **pagepool, struct page *page)
 void erofs_release_pages(struct page **pagepool);
 
 #ifdef CONFIG_EROFS_FS_ZIP
-int erofs_workgroup_put(struct erofs_workgroup *grp);
+void erofs_workgroup_put(struct erofs_workgroup *grp);
 struct erofs_workgroup *erofs_find_workgroup(struct super_block *sb,
 					     pgoff_t index);
 struct erofs_workgroup *erofs_insert_workgroup(struct super_block *sb,
@@ -500,7 +466,6 @@ int __init z_erofs_init_zip_subsystem(void);
 void z_erofs_exit_zip_subsystem(void);
 int erofs_try_to_free_all_cached_pages(struct erofs_sb_info *sbi,
 				       struct erofs_workgroup *egrp);
-int erofs_try_to_free_cached_page(struct page *page);
 int z_erofs_load_lz4_config(struct super_block *sb,
 			    struct erofs_super_block *dsb,
 			    struct z_erofs_lz4_cfgs *lz4, int len);
@@ -511,6 +476,7 @@ void erofs_put_pcpubuf(void *ptr);
 int erofs_pcpubuf_growsize(unsigned int nrpages);
 void __init erofs_pcpubuf_init(void);
 void erofs_pcpubuf_exit(void);
+int erofs_init_managed_cache(struct super_block *sb);
 #else
 static inline void erofs_shrinker_register(struct super_block *sb) {}
 static inline void erofs_shrinker_unregister(struct super_block *sb) {}
@@ -530,6 +496,7 @@ static inline int z_erofs_load_lz4_config(struct super_block *sb,
 }
 static inline void erofs_pcpubuf_init(void) {}
 static inline void erofs_pcpubuf_exit(void) {}
+static inline int erofs_init_managed_cache(struct super_block *sb) { return 0; }
 #endif	/* !CONFIG_EROFS_FS_ZIP */
 
 #ifdef CONFIG_EROFS_FS_ZIP_LZMA

fs/erofs/super.c

@@ -599,68 +599,6 @@ static int erofs_fc_parse_param(struct fs_context *fc,
 	return 0;
 }
 
-#ifdef CONFIG_EROFS_FS_ZIP
-static const struct address_space_operations managed_cache_aops;
-
-static bool erofs_managed_cache_release_folio(struct folio *folio, gfp_t gfp)
-{
-	bool ret = true;
-	struct address_space *const mapping = folio->mapping;
-
-	DBG_BUGON(!folio_test_locked(folio));
-	DBG_BUGON(mapping->a_ops != &managed_cache_aops);
-
-	if (folio_test_private(folio))
-		ret = erofs_try_to_free_cached_page(&folio->page);
-
-	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_invalidate_folio(struct folio *folio,
-					       size_t offset, size_t length)
-{
-	const size_t stop = length + offset;
-
-	DBG_BUGON(!folio_test_locked(folio));
-
-	/* Check for potential overflow in debug mode */
-	DBG_BUGON(stop > folio_size(folio) || stop < length);
-
-	if (offset == 0 && stop == folio_size(folio))
-		while (!erofs_managed_cache_release_folio(folio, GFP_NOFS))
-			cond_resched();
-}
-
-static const struct address_space_operations managed_cache_aops = {
-	.release_folio = erofs_managed_cache_release_folio,
-	.invalidate_folio = erofs_managed_cache_invalidate_folio,
-};
-
-static int erofs_init_managed_cache(struct super_block *sb)
-{
-	struct erofs_sb_info *const sbi = EROFS_SB(sb);
-	struct inode *const inode = new_inode(sb);
-
-	if (!inode)
-		return -ENOMEM;
-
-	set_nlink(inode, 1);
-	inode->i_size = OFFSET_MAX;
-
-	inode->i_mapping->a_ops = &managed_cache_aops;
-	mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
-	sbi->managed_cache = inode;
-	return 0;
-}
-#else
-static int erofs_init_managed_cache(struct super_block *sb) { return 0; }
-#endif
-
 static struct inode *erofs_nfs_get_inode(struct super_block *sb,
 					 u64 ino, u32 generation)
 {
@@ -1016,10 +954,8 @@ static int __init erofs_module_init(void)
 					       sizeof(struct erofs_inode), 0,
 					       SLAB_RECLAIM_ACCOUNT,
 					       erofs_inode_init_once);
-	if (!erofs_inode_cachep) {
-		err = -ENOMEM;
-		goto icache_err;
-	}
+	if (!erofs_inode_cachep)
+		return -ENOMEM;
 
 	err = erofs_init_shrinker();
 	if (err)
@@ -1054,7 +990,6 @@ static int __init erofs_module_init(void)
 	erofs_exit_shrinker();
 shrinker_err:
 	kmem_cache_destroy(erofs_inode_cachep);
-icache_err:
 	return err;
 }
 

fs/erofs/utils.c

@@ -4,7 +4,6 @@
  *             https://www.huawei.com/
  */
 #include "internal.h"
-#include <linux/pagevec.h>
 
 struct page *erofs_allocpage(struct page **pagepool, gfp_t gfp)
 {
@@ -33,22 +32,21 @@ void erofs_release_pages(struct page **pagepool)
 /* global shrink count (for all mounted EROFS instances) */
 static atomic_long_t erofs_global_shrink_cnt;
 
-static int erofs_workgroup_get(struct erofs_workgroup *grp)
+static bool erofs_workgroup_get(struct erofs_workgroup *grp)
 {
-	int o;
+	if (lockref_get_not_zero(&grp->lockref))
+		return true;
 
-repeat:
-	o = erofs_wait_on_workgroup_freezed(grp);
-	if (o <= 0)
-		return -1;
-
-	if (atomic_cmpxchg(&grp->refcount, o, o + 1) != o)
-		goto repeat;
+	spin_lock(&grp->lockref.lock);
+	if (__lockref_is_dead(&grp->lockref)) {
+		spin_unlock(&grp->lockref.lock);
+		return false;
+	}
 
-	/* decrease refcount paired by erofs_workgroup_put */
-	if (o == 1)
+	if (!grp->lockref.count++)
 		atomic_long_dec(&erofs_global_shrink_cnt);
-	return 0;
+	spin_unlock(&grp->lockref.lock);
+	return true;
 }
 
 struct erofs_workgroup *erofs_find_workgroup(struct super_block *sb,
@@ -61,7 +59,7 @@ struct erofs_workgroup *erofs_find_workgroup(struct super_block *sb,
 	rcu_read_lock();
 	grp = xa_load(&sbi->managed_pslots, index);
 	if (grp) {
-		if (erofs_workgroup_get(grp)) {
+		if (!erofs_workgroup_get(grp)) {
 			/* prefer to relax rcu read side */
 			rcu_read_unlock();
 			goto repeat;
@@ -80,11 +78,10 @@ struct erofs_workgroup *erofs_insert_workgroup(struct super_block *sb,
 	struct erofs_workgroup *pre;
 
 	/*
-	 * Bump up a reference count before making this visible
-	 * to others for the XArray in order to avoid potential
-	 * UAF without serialized by xa_lock.
+	 * Bump up before making this visible to others for the XArray in order
+	 * to avoid potential UAF without serialized by xa_lock.
 	 */
-	atomic_inc(&grp->refcount);
+	lockref_get(&grp->lockref);
 
 repeat:
 	xa_lock(&sbi->managed_pslots);
@@ -93,13 +90,13 @@ struct erofs_workgroup *erofs_insert_workgroup(struct super_block *sb,
 	if (pre) {
 		if (xa_is_err(pre)) {
 			pre = ERR_PTR(xa_err(pre));
-		} else if (erofs_workgroup_get(pre)) {
+		} else if (!erofs_workgroup_get(pre)) {
 			/* try to legitimize the current in-tree one */
 			xa_unlock(&sbi->managed_pslots);
 			cond_resched();
 			goto repeat;
 		}
-		atomic_dec(&grp->refcount);
+		lockref_put_return(&grp->lockref);
 		grp = pre;
 	}
 	xa_unlock(&sbi->managed_pslots);
@@ -112,38 +109,34 @@ static void  __erofs_workgroup_free(struct erofs_workgroup *grp)
 	erofs_workgroup_free_rcu(grp);
 }
 
-int erofs_workgroup_put(struct erofs_workgroup *grp)
+void erofs_workgroup_put(struct erofs_workgroup *grp)
 {
-	int count = atomic_dec_return(&grp->refcount);
+	if (lockref_put_or_lock(&grp->lockref))
+		return;
 
-	if (count == 1)
+	DBG_BUGON(__lockref_is_dead(&grp->lockref));
+	if (grp->lockref.count == 1)
 		atomic_long_inc(&erofs_global_shrink_cnt);
-	else if (!count)
-		__erofs_workgroup_free(grp);
-	return count;
+	--grp->lockref.count;
+	spin_unlock(&grp->lockref.lock);
 }
 
 static bool erofs_try_to_release_workgroup(struct erofs_sb_info *sbi,
 					   struct erofs_workgroup *grp)
 {
-	/*
-	 * If managed cache is on, refcount of workgroups
-	 * themselves could be < 0 (freezed). In other words,
-	 * there is no guarantee that all refcounts > 0.
-	 */
-	if (!erofs_workgroup_try_to_freeze(grp, 1))
-		return false;
+	int free = false;
+
+	spin_lock(&grp->lockref.lock);
+	if (grp->lockref.count)
+		goto out;
 
 	/*
-	 * Note that all cached pages should be unattached
-	 * before deleted from the XArray. Otherwise some
-	 * cached pages could be still attached to the orphan
-	 * old workgroup when the new one is available in the tree.
+	 * Note that all cached pages should be detached before deleted from
+	 * the XArray. Otherwise some cached pages could be still attached to
+	 * the orphan old workgroup when the new one is available in the tree.
 	 */
-	if (erofs_try_to_free_all_cached_pages(sbi, grp)) {
-		erofs_workgroup_unfreeze(grp, 1);
-		return false;
-	}
+	if (erofs_try_to_free_all_cached_pages(sbi, grp))
+		goto out;
 
 	/*
 	 * It's impossible to fail after the workgroup is freezed,
@@ -152,10 +145,13 @@ static bool erofs_try_to_release_workgroup(struct erofs_sb_info *sbi,
 	 */
 	DBG_BUGON(__xa_erase(&sbi->managed_pslots, grp->index) != grp);
 
-	/* last refcount should be connected with its managed pslot.  */
-	erofs_workgroup_unfreeze(grp, 0);
-	__erofs_workgroup_free(grp);
-	return true;
+	lockref_mark_dead(&grp->lockref);
+	free = true;
+out:
+	spin_unlock(&grp->lockref.lock);
+	if (free)
+		__erofs_workgroup_free(grp);
+	return free;
 }
 
 static unsigned long erofs_shrink_workstation(struct erofs_sb_info *sbi,

fs/erofs/zmap.c

@@ -22,8 +22,8 @@ struct z_erofs_maprecorder {
 	bool partialref;
 };
 
-static int legacy_load_cluster_from_disk(struct z_erofs_maprecorder *m,
-					 unsigned long lcn)
+static int z_erofs_load_full_lcluster(struct z_erofs_maprecorder *m,
+				      unsigned long lcn)
 {
 	struct inode *const inode = m->inode;
 	struct erofs_inode *const vi = EROFS_I(inode);
@@ -129,7 +129,7 @@ static int unpack_compacted_index(struct z_erofs_maprecorder *m,
 	u8 *in, type;
 	bool big_pcluster;
 
-	if (1 << amortizedshift == 4)
+	if (1 << amortizedshift == 4 && lclusterbits <= 14)
 		vcnt = 2;
 	else if (1 << amortizedshift == 2 && lclusterbits == 12)
 		vcnt = 16;
@@ -226,12 +226,11 @@ static int unpack_compacted_index(struct z_erofs_maprecorder *m,
 	return 0;
 }
 
-static int compacted_load_cluster_from_disk(struct z_erofs_maprecorder *m,
-					    unsigned long lcn, bool lookahead)
+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 unsigned int lclusterbits = vi->z_logical_clusterbits;
 	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);
@@ -239,9 +238,6 @@ static int compacted_load_cluster_from_disk(struct z_erofs_maprecorder *m,
 	unsigned int amortizedshift;
 	erofs_off_t pos;
 
-	if (lclusterbits != 12)
-		return -EOPNOTSUPP;
-
 	if (lcn >= totalidx)
 		return -EINVAL;
 
@@ -281,23 +277,23 @@ static int compacted_load_cluster_from_disk(struct z_erofs_maprecorder *m,
 	return unpack_compacted_index(m, amortizedshift, pos, lookahead);
 }
 
-static int z_erofs_load_cluster_from_disk(struct z_erofs_maprecorder *m,
-					  unsigned int lcn, bool lookahead)
+static int z_erofs_load_lcluster_from_disk(struct z_erofs_maprecorder *m,
+					   unsigned int lcn, bool lookahead)
 {
-	const unsigned int datamode = EROFS_I(m->inode)->datalayout;
-
-	if (datamode == EROFS_INODE_COMPRESSED_FULL)
-		return legacy_load_cluster_from_disk(m, lcn);
-
-	if (datamode == EROFS_INODE_COMPRESSED_COMPACT)
-		return compacted_load_cluster_from_disk(m, lcn, lookahead);
-
-	return -EINVAL;
+	switch (EROFS_I(m->inode)->datalayout) {
+	case EROFS_INODE_COMPRESSED_FULL:
+		return z_erofs_load_full_lcluster(m, lcn);
+	case EROFS_INODE_COMPRESSED_COMPACT:
+		return z_erofs_load_compact_lcluster(m, lcn, lookahead);
+	default:
+		return -EINVAL;
+	}
 }
 
 static int z_erofs_extent_lookback(struct z_erofs_maprecorder *m,
 				   unsigned int lookback_distance)
 {
+	struct super_block *sb = m->inode->i_sb;
 	struct erofs_inode *const vi = EROFS_I(m->inode);
 	const unsigned int lclusterbits = vi->z_logical_clusterbits;
 
@@ -305,21 +301,15 @@ static int z_erofs_extent_lookback(struct z_erofs_maprecorder *m,
 		unsigned long lcn = m->lcn - lookback_distance;
 		int err;
 
-		/* load extent head logical cluster if needed */
-		err = z_erofs_load_cluster_from_disk(m, lcn, false);
+		err = z_erofs_load_lcluster_from_disk(m, lcn, false);
 		if (err)
 			return err;
 
 		switch (m->type) {
 		case Z_EROFS_LCLUSTER_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];
+			if (!lookback_distance)
+				goto err_bogus;
 			continue;
 		case Z_EROFS_LCLUSTER_TYPE_PLAIN:
 		case Z_EROFS_LCLUSTER_TYPE_HEAD1:
@@ -328,16 +318,15 @@ static int z_erofs_extent_lookback(struct z_erofs_maprecorder *m,
 			m->map->m_la = (lcn << lclusterbits) | m->clusterofs;
 			return 0;
 		default:
-			erofs_err(m->inode->i_sb,
-				  "unknown type %u @ lcn %lu of nid %llu",
+			erofs_err(sb, "unknown type %u @ lcn %lu of nid %llu",
 				  m->type, lcn, vi->nid);
 			DBG_BUGON(1);
 			return -EOPNOTSUPP;
 		}
 	}
-
-	erofs_err(m->inode->i_sb, "bogus lookback distance @ nid %llu",
-		  vi->nid);
+err_bogus:
+	erofs_err(sb, "bogus lookback distance %u @ lcn %lu of nid %llu",
+		  lookback_distance, m->lcn, vi->nid);
 	DBG_BUGON(1);
 	return -EFSCORRUPTED;
 }
@@ -369,7 +358,7 @@ static int z_erofs_get_extent_compressedlen(struct z_erofs_maprecorder *m,
 	if (m->compressedblks)
 		goto out;
 
-	err = z_erofs_load_cluster_from_disk(m, lcn, false);
+	err = z_erofs_load_lcluster_from_disk(m, lcn, false);
 	if (err)
 		return err;
 
@@ -401,9 +390,8 @@ static int z_erofs_get_extent_compressedlen(struct z_erofs_maprecorder *m,
 			break;
 		fallthrough;
 	default:
-		erofs_err(m->inode->i_sb,
-			  "cannot found CBLKCNT @ lcn %lu of nid %llu",
-			  lcn, vi->nid);
+		erofs_err(sb, "cannot found CBLKCNT @ lcn %lu of nid %llu", lcn,
+			  vi->nid);
 		DBG_BUGON(1);
 		return -EFSCORRUPTED;
 	}
@@ -411,9 +399,7 @@ static int z_erofs_get_extent_compressedlen(struct z_erofs_maprecorder *m,
 	map->m_plen = erofs_pos(sb, m->compressedblks);
 	return 0;
 err_bonus_cblkcnt:
-	erofs_err(m->inode->i_sb,
-		  "bogus CBLKCNT @ lcn %lu of nid %llu",
-		  lcn, vi->nid);
+	erofs_err(sb, "bogus CBLKCNT @ lcn %lu of nid %llu", lcn, vi->nid);
 	DBG_BUGON(1);
 	return -EFSCORRUPTED;
 }
@@ -434,7 +420,7 @@ static int z_erofs_get_extent_decompressedlen(struct z_erofs_maprecorder *m)
 			return 0;
 		}
 
-		err = z_erofs_load_cluster_from_disk(m, lcn, true);
+		err = z_erofs_load_lcluster_from_disk(m, lcn, true);
 		if (err)
 			return err;
 
@@ -481,7 +467,7 @@ static int z_erofs_do_map_blocks(struct inode *inode,
 	initial_lcn = ofs >> lclusterbits;
 	endoff = ofs & ((1 << lclusterbits) - 1);
 
-	err = z_erofs_load_cluster_from_disk(&m, initial_lcn, false);
+	err = z_erofs_load_lcluster_from_disk(&m, initial_lcn, false);
 	if (err)
 		goto unmap_out;
 
@@ -539,8 +525,7 @@ static int z_erofs_do_map_blocks(struct inode *inode,
 	if (flags & EROFS_GET_BLOCKS_FINDTAIL) {
 		vi->z_tailextent_headlcn = m.lcn;
 		/* for non-compact indexes, fragmentoff is 64 bits */
-		if (fragment &&
-		    vi->datalayout == EROFS_INODE_COMPRESSED_FULL)
+		if (fragment && vi->datalayout == EROFS_INODE_COMPRESSED_FULL)
 			vi->z_fragmentoff |= (u64)m.pblk << 32;
 	}
 	if (ztailpacking && m.lcn == vi->z_tailextent_headlcn) {