1 files changed, 119 insertions, 164 deletions

diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index 5f1890e309c6..036f610e044b 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -143,22 +143,17 @@ static inline void z_erofs_onlinepage_split(struct page *page)
 	atomic_inc((atomic_t *)&page->private);
 }
 
-static inline void z_erofs_page_mark_eio(struct page *page)
+static void z_erofs_onlinepage_endio(struct page *page, int err)
 {
-	int orig;
+	int orig, v;
+
+	DBG_BUGON(!PagePrivate(page));
 
 	do {
 		orig = atomic_read((atomic_t *)&page->private);
-	} while (atomic_cmpxchg((atomic_t *)&page->private, orig,
-				orig | Z_EROFS_PAGE_EIO) != orig);
-}
-
-static inline void z_erofs_onlinepage_endio(struct page *page)
-{
-	unsigned int v;
+		v = (orig - 1) | (err ? Z_EROFS_PAGE_EIO : 0);
+	} while (atomic_cmpxchg((atomic_t *)&page->private, orig, v) != orig);
 
-	DBG_BUGON(!PagePrivate(page));
-	v = atomic_dec_return((atomic_t *)&page->private);
 	if (!(v & ~Z_EROFS_PAGE_EIO)) {
 		set_page_private(page, 0);
 		ClearPagePrivate(page);
@@ -507,19 +502,17 @@ enum z_erofs_pclustermode {
 	 */
 	Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE,
 	/*
-	 * The current collection has been linked with the owned chain, and
-	 * could also be linked with the remaining collections, which means
-	 * if the processing page is the tail page of the collection, thus
-	 * the current collection can safely use the whole page (since
-	 * the previous collection is under control) for in-place I/O, as
-	 * illustrated below:
-	 *  ________________________________________________________________
-	 * |  tail (partial) page |          head (partial) page           |
-	 * |  (of the current cl) |      (of the previous collection)      |
-	 * |                      |                                        |
-	 * |__PCLUSTER_FOLLOWED___|___________PCLUSTER_FOLLOWED____________|
+	 * 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 above page can be used as inplace I/O.               ]
+	 * [  (*) the page above can be used as inplace I/O.   ]
 	 */
 	Z_EROFS_PCLUSTER_FOLLOWED,
 };
@@ -535,8 +528,6 @@ struct z_erofs_decompress_frontend {
 	z_erofs_next_pcluster_t owned_head;
 	enum z_erofs_pclustermode mode;
 
-	/* used for applying cache strategy on the fly */
-	bool backmost;
 	erofs_off_t headoffset;
 
 	/* a pointer used to pick up inplace I/O pages */
@@ -545,7 +536,7 @@ struct z_erofs_decompress_frontend {
 
 #define DECOMPRESS_FRONTEND_INIT(__i) { \
 	.inode = __i, .owned_head = Z_EROFS_PCLUSTER_TAIL, \
-	.mode = Z_EROFS_PCLUSTER_FOLLOWED, .backmost = true }
+	.mode = Z_EROFS_PCLUSTER_FOLLOWED }
 
 static bool z_erofs_should_alloc_cache(struct z_erofs_decompress_frontend *fe)
 {
@@ -554,7 +545,7 @@ static bool z_erofs_should_alloc_cache(struct z_erofs_decompress_frontend *fe)
 	if (cachestrategy <= EROFS_ZIP_CACHE_DISABLED)
 		return false;
 
-	if (fe->backmost)
+	if (!(fe->map.m_flags & EROFS_MAP_FULL_MAPPED))
 		return true;
 
 	if (cachestrategy >= EROFS_ZIP_CACHE_READAROUND &&
@@ -851,9 +842,11 @@ err_out:
 	return err;
 }
 
-static int z_erofs_collector_begin(struct z_erofs_decompress_frontend *fe)
+static int z_erofs_pcluster_begin(struct z_erofs_decompress_frontend *fe)
 {
 	struct erofs_map_blocks *map = &fe->map;
+	struct super_block *sb = fe->inode->i_sb;
+	erofs_blk_t blknr = erofs_blknr(sb, map->m_pa);
 	struct erofs_workgroup *grp = NULL;
 	int ret;
 
@@ -863,8 +856,7 @@ static int z_erofs_collector_begin(struct z_erofs_decompress_frontend *fe)
 	DBG_BUGON(fe->owned_head == Z_EROFS_PCLUSTER_NIL);
 
 	if (!(map->m_flags & EROFS_MAP_META)) {
-		grp = erofs_find_workgroup(fe->inode->i_sb,
-					   map->m_pa >> PAGE_SHIFT);
+		grp = erofs_find_workgroup(sb, blknr);
 	} else if ((map->m_pa & ~PAGE_MASK) + map->m_plen > PAGE_SIZE) {
 		DBG_BUGON(1);
 		return -EFSCORRUPTED;
@@ -883,9 +875,26 @@ static int z_erofs_collector_begin(struct z_erofs_decompress_frontend *fe)
 	} else if (ret) {
 		return ret;
 	}
+
 	z_erofs_bvec_iter_begin(&fe->biter, &fe->pcl->bvset,
 				Z_EROFS_INLINE_BVECS, fe->pcl->vcnt);
-	/* since file-backed online pages are traversed in reverse order */
+	if (!z_erofs_is_inline_pcluster(fe->pcl)) {
+		/* bind cache first when cached decompression is preferred */
+		z_erofs_bind_cache(fe);
+	} else {
+		void *mptr;
+
+		mptr = erofs_read_metabuf(&map->buf, sb, blknr, EROFS_NO_KMAP);
+		if (IS_ERR(mptr)) {
+			ret = PTR_ERR(mptr);
+			erofs_err(sb, "failed to get inline data %d", ret);
+			return ret;
+		}
+		get_page(map->buf.page);
+		WRITE_ONCE(fe->pcl->compressed_bvecs[0].page, map->buf.page);
+		fe->mode = Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE;
+	}
+	/* file-backed inplace I/O pages are traversed in reverse order */
 	fe->icur = z_erofs_pclusterpages(fe->pcl);
 	return 0;
 }
@@ -908,12 +917,12 @@ void erofs_workgroup_free_rcu(struct erofs_workgroup *grp)
 	call_rcu(&pcl->rcu, z_erofs_rcu_callback);
 }
 
-static bool z_erofs_collector_end(struct z_erofs_decompress_frontend *fe)
+static void z_erofs_pcluster_end(struct z_erofs_decompress_frontend *fe)
 {
 	struct z_erofs_pcluster *pcl = fe->pcl;
 
 	if (!pcl)
-		return false;
+		return;
 
 	z_erofs_bvec_iter_end(&fe->biter);
 	mutex_unlock(&pcl->lock);
@@ -929,37 +938,29 @@ static bool z_erofs_collector_end(struct z_erofs_decompress_frontend *fe)
 		erofs_workgroup_put(&pcl->obj);
 
 	fe->pcl = NULL;
-	return true;
 }
 
-static int z_erofs_read_fragment(struct inode *inode, erofs_off_t pos,
-				 struct page *page, unsigned int pageofs,
-				 unsigned int len)
+static int z_erofs_read_fragment(struct super_block *sb, struct page *page,
+			unsigned int cur, unsigned int end, erofs_off_t pos)
 {
-	struct super_block *sb = inode->i_sb;
-	struct inode *packed_inode = EROFS_I_SB(inode)->packed_inode;
+	struct inode *packed_inode = EROFS_SB(sb)->packed_inode;
 	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
-	u8 *src, *dst;
-	unsigned int i, cnt;
+	unsigned int cnt;
+	u8 *src;
 
 	if (!packed_inode)
 		return -EFSCORRUPTED;
 
 	buf.inode = packed_inode;
-	pos += EROFS_I(inode)->z_fragmentoff;
-	for (i = 0; i < len; i += cnt) {
-		cnt = min_t(unsigned int, len - i,
+	for (; cur < end; cur += cnt, pos += cnt) {
+		cnt = min_t(unsigned int, end - cur,
 			    sb->s_blocksize - erofs_blkoff(sb, pos));
 		src = erofs_bread(&buf, erofs_blknr(sb, pos), EROFS_KMAP);
 		if (IS_ERR(src)) {
 			erofs_put_metabuf(&buf);
 			return PTR_ERR(src);
 		}
-
-		dst = kmap_local_page(page);
-		memcpy(dst + pageofs + i, src + erofs_blkoff(sb, pos), cnt);
-		kunmap_local(dst);
-		pos += cnt;
+		memcpy_to_page(page, cur, src + erofs_blkoff(sb, pos), cnt);
 	}
 	erofs_put_metabuf(&buf);
 	return 0;
@@ -972,94 +973,60 @@ static int z_erofs_do_read_page(struct z_erofs_decompress_frontend *fe,
 	struct erofs_map_blocks *const map = &fe->map;
 	const loff_t offset = page_offset(page);
 	bool tight = true, exclusive;
-	unsigned int cur, end, spiltted;
+	unsigned int cur, end, len, split;
 	int err = 0;
 
-	/* register locked file pages as online pages in pack */
 	z_erofs_onlinepage_init(page);
 
-	spiltted = 0;
+	split = 0;
 	end = PAGE_SIZE;
 repeat:
-	cur = end - 1;
-
-	if (offset + cur < map->m_la ||
-	    offset + cur >= map->m_la + map->m_llen) {
-		if (z_erofs_collector_end(fe))
-			fe->backmost = false;
-		map->m_la = offset + cur;
+	if (offset + end - 1 < map->m_la ||
+	    offset + end - 1 >= map->m_la + map->m_llen) {
+		z_erofs_pcluster_end(fe);
+		map->m_la = offset + end - 1;
 		map->m_llen = 0;
 		err = z_erofs_map_blocks_iter(inode, map, 0);
 		if (err)
 			goto out;
-	} else {
-		if (fe->pcl)
-			goto hitted;
-		/* didn't get a valid pcluster previously (very rare) */
 	}
 
-	if (!(map->m_flags & EROFS_MAP_MAPPED) ||
-	    map->m_flags & EROFS_MAP_FRAGMENT)
-		goto hitted;
-
-	err = z_erofs_collector_begin(fe);
-	if (err)
-		goto out;
+	cur = offset > map->m_la ? 0 : map->m_la - offset;
+	/* bump split parts first to avoid several separate cases */
+	++split;
 
-	if (z_erofs_is_inline_pcluster(fe->pcl)) {
-		void *mp;
-
-		mp = erofs_read_metabuf(&fe->map.buf, inode->i_sb,
-					erofs_blknr(inode->i_sb, map->m_pa),
-					EROFS_NO_KMAP);
-		if (IS_ERR(mp)) {
-			err = PTR_ERR(mp);
-			erofs_err(inode->i_sb,
-				  "failed to get inline page, err %d", err);
-			goto out;
-		}
-		get_page(fe->map.buf.page);
-		WRITE_ONCE(fe->pcl->compressed_bvecs[0].page,
-			   fe->map.buf.page);
-		fe->mode = Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE;
-	} else {
-		/* bind cache first when cached decompression is preferred */
-		z_erofs_bind_cache(fe);
-	}
-hitted:
-	/*
-	 * Ensure the current partial page belongs to this submit chain rather
-	 * than other concurrent submit chains or the noio(bypass) chain since
-	 * those chains are handled asynchronously thus the page cannot be used
-	 * for inplace I/O or bvpage (should be processed in a strict order.)
-	 */
-	tight &= (fe->mode > Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE);
-
-	cur = end - min_t(unsigned int, offset + end - map->m_la, end);
 	if (!(map->m_flags & EROFS_MAP_MAPPED)) {
 		zero_user_segment(page, cur, end);
+		tight = false;
 		goto next_part;
 	}
+
 	if (map->m_flags & EROFS_MAP_FRAGMENT) {
-		unsigned int pageofs, skip, len;
+		erofs_off_t fpos = offset + cur - map->m_la;
 
-		if (offset > map->m_la) {
-			pageofs = 0;
-			skip = offset - map->m_la;
-		} else {
-			pageofs = map->m_la & ~PAGE_MASK;
-			skip = 0;
-		}
-		len = min_t(unsigned int, map->m_llen - skip, end - cur);
-		err = z_erofs_read_fragment(inode, skip, page, pageofs, len);
+		len = min_t(unsigned int, map->m_llen - fpos, end - cur);
+		err = z_erofs_read_fragment(inode->i_sb, page, cur, cur + len,
+				EROFS_I(inode)->z_fragmentoff + fpos);
 		if (err)
 			goto out;
-		++spiltted;
 		tight = false;
 		goto next_part;
 	}
 
-	exclusive = (!cur && (!spiltted || tight));
+	if (!fe->pcl) {
+		err = z_erofs_pcluster_begin(fe);
+		if (err)
+			goto out;
+	}
+
+	/*
+	 * Ensure the current partial page belongs to this submit chain rather
+	 * than other concurrent submit chains or the noio(bypass) chain since
+	 * those chains are handled asynchronously thus the page cannot be used
+	 * for inplace I/O or bvpage (should be processed in a strict order.)
+	 */
+	tight &= (fe->mode > Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE);
+	exclusive = (!cur && ((split <= 1) || tight));
 	if (cur)
 		tight &= (fe->mode >= Z_EROFS_PCLUSTER_FOLLOWED);
 
@@ -1072,8 +1039,6 @@ hitted:
 		goto out;
 
 	z_erofs_onlinepage_split(page);
-	/* bump up the number of spiltted parts of a page */
-	++spiltted;
 	if (fe->pcl->pageofs_out != (map->m_la & ~PAGE_MASK))
 		fe->pcl->multibases = true;
 	if (fe->pcl->length < offset + end - map->m_la) {
@@ -1094,9 +1059,7 @@ next_part:
 		goto repeat;
 
 out:
-	if (err)
-		z_erofs_page_mark_eio(page);
-	z_erofs_onlinepage_endio(page);
+	z_erofs_onlinepage_endio(page, err);
 	return err;
 }
 
@@ -1144,10 +1107,11 @@ static void z_erofs_do_decompressed_bvec(struct z_erofs_decompress_backend *be,
 					 struct z_erofs_bvec *bvec)
 {
 	struct z_erofs_bvec_item *item;
+	unsigned int pgnr;
 
-	if (!((bvec->offset + be->pcl->pageofs_out) & ~PAGE_MASK)) {
-		unsigned int pgnr;
-
+	if (!((bvec->offset + be->pcl->pageofs_out) & ~PAGE_MASK) &&
+	    (bvec->end == PAGE_SIZE ||
+	     bvec->offset + bvec->end == be->pcl->length)) {
 		pgnr = (bvec->offset + be->pcl->pageofs_out) >> PAGE_SHIFT;
 		DBG_BUGON(pgnr >= be->nr_pages);
 		if (!be->decompressed_pages[pgnr]) {
@@ -1198,9 +1162,7 @@ static void z_erofs_fill_other_copies(struct z_erofs_decompress_backend *be,
 			cur += len;
 		}
 		kunmap_local(dst);
-		if (err)
-			z_erofs_page_mark_eio(bvi->bvec.page);
-		z_erofs_onlinepage_endio(bvi->bvec.page);
+		z_erofs_onlinepage_endio(bvi->bvec.page, err);
 		list_del(p);
 		kfree(bvi);
 	}
@@ -1371,9 +1333,7 @@ out:
 		/* recycle all individual short-lived pages */
 		if (z_erofs_put_shortlivedpage(be->pagepool, page))
 			continue;
-		if (err)
-			z_erofs_page_mark_eio(page);
-		z_erofs_onlinepage_endio(page);
+		z_erofs_onlinepage_endio(page, err);
 	}
 
 	if (be->decompressed_pages != be->onstack_pages)
@@ -1409,7 +1369,10 @@ static void z_erofs_decompress_queue(const struct z_erofs_decompressqueue *io,
 		owned = READ_ONCE(be.pcl->next);
 
 		z_erofs_decompress_pcluster(&be, io->eio ? -EIO : 0);
-		erofs_workgroup_put(&be.pcl->obj);
+		if (z_erofs_is_inline_pcluster(be.pcl))
+			z_erofs_free_pcluster(be.pcl);
+		else
+			erofs_workgroup_put(&be.pcl->obj);
 	}
 }
 
@@ -1841,21 +1804,16 @@ static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f,
 	}
 
 	cur = map->m_la + map->m_llen - 1;
-	while (cur >= end) {
+	while ((cur >= end) && (cur < i_size_read(inode))) {
 		pgoff_t index = cur >> PAGE_SHIFT;
 		struct page *page;
 
 		page = erofs_grab_cache_page_nowait(inode->i_mapping, index);
 		if (page) {
-			if (PageUptodate(page)) {
+			if (PageUptodate(page))
 				unlock_page(page);
-			} else {
-				err = z_erofs_do_read_page(f, page);
-				if (err)
-					erofs_err(inode->i_sb,
-						  "readmore error at page %lu @ nid %llu",
-						  index, EROFS_I(inode)->nid);
-			}
+			else
+				(void)z_erofs_do_read_page(f, page);
 			put_page(page);
 		}
 
@@ -1867,25 +1825,25 @@ 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 page *page = &folio->page;
-	struct inode *const inode = page->mapping->host;
+	struct inode *const inode = folio->mapping->host;
 	struct erofs_sb_info *const sbi = EROFS_I_SB(inode);
 	struct z_erofs_decompress_frontend f = DECOMPRESS_FRONTEND_INIT(inode);
 	int err;
 
-	trace_erofs_readpage(page, false);
-	f.headoffset = (erofs_off_t)page->index << PAGE_SHIFT;
+	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_do_read_page(&f, page);
+	err = z_erofs_do_read_page(&f, &folio->page);
 	z_erofs_pcluster_readmore(&f, NULL, false);
-	(void)z_erofs_collector_end(&f);
+	z_erofs_pcluster_end(&f);
 
 	/* if some compressed cluster ready, need submit them anyway */
 	z_erofs_runqueue(&f, z_erofs_is_sync_decompress(sbi, 0), false);
 
-	if (err)
-		erofs_err(inode->i_sb, "failed to read, err [%d]", err);
+	if (err && err != -EINTR)
+		erofs_err(inode->i_sb, "read error %d @ %lu of nid %llu",
+			  err, folio->index, EROFS_I(inode)->nid);
 
 	erofs_put_metabuf(&f.map.buf);
 	erofs_release_pages(&f.pagepool);
@@ -1897,38 +1855,35 @@ static void z_erofs_readahead(struct readahead_control *rac)
 	struct inode *const inode = rac->mapping->host;
 	struct erofs_sb_info *const sbi = EROFS_I_SB(inode);
 	struct z_erofs_decompress_frontend f = DECOMPRESS_FRONTEND_INIT(inode);
-	struct page *head = NULL, *page;
-	unsigned int nr_pages;
+	struct folio *head = NULL, *folio;
+	unsigned int nr_folios;
+	int err;
 
 	f.headoffset = readahead_pos(rac);
 
 	z_erofs_pcluster_readmore(&f, rac, true);
-	nr_pages = readahead_count(rac);
-	trace_erofs_readpages(inode, readahead_index(rac), nr_pages, false);
+	nr_folios = readahead_count(rac);
+	trace_erofs_readpages(inode, readahead_index(rac), nr_folios, false);
 
-	while ((page = readahead_page(rac))) {
-		set_page_private(page, (unsigned long)head);
-		head = page;
+	while ((folio = readahead_folio(rac))) {
+		folio->private = head;
+		head = folio;
 	}
 
+	/* traverse in reverse order for best metadata I/O performance */
 	while (head) {
-		struct page *page = head;
-		int err;
-
-		/* traversal in reverse order */
-		head = (void *)page_private(page);
+		folio = head;
+		head = folio_get_private(folio);
 
-		err = z_erofs_do_read_page(&f, page);
-		if (err)
-			erofs_err(inode->i_sb,
-				  "readahead error at page %lu @ nid %llu",
-				  page->index, EROFS_I(inode)->nid);
-		put_page(page);
+		err = z_erofs_do_read_page(&f, &folio->page);
+		if (err && err != -EINTR)
+			erofs_err(inode->i_sb, "readahead error at folio %lu @ nid %llu",
+				  folio->index, EROFS_I(inode)->nid);
 	}
 	z_erofs_pcluster_readmore(&f, rac, false);
-	(void)z_erofs_collector_end(&f);
+	z_erofs_pcluster_end(&f);
 
-	z_erofs_runqueue(&f, z_erofs_is_sync_decompress(sbi, nr_pages), true);
+	z_erofs_runqueue(&f, z_erofs_is_sync_decompress(sbi, nr_folios), true);
 	erofs_put_metabuf(&f.map.buf);
 	erofs_release_pages(&f.pagepool);
 }