Merge tag 'erofs-for-6.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs

Pull erofs updates from Gao Xiang:
 "Updates for folio conversions for compressed inodes: While large folio
  support for compressed data could work now, it remains disabled since
  the stress test could hang due to page migration in a few hours after
  enabling it. I need more time to investigate further before enabling
  this feature.

  Additionally, clean up stream decompressors and tracepoints for
  simplicity.

  Summary:

   - More folio conversions for compressed inodes

   - Stream decompressor (LZMA/DEFLATE/ZSTD) cleanups

   - Minor tracepoint cleanup"

* tag 'erofs-for-6.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
  erofs: silence uninitialized variable warning in z_erofs_scan_folio()
  erofs: avoid refcounting short-lived pages
  erofs: get rid of z_erofs_map_blocks_iter_* tracepoints
  erofs: tidy up stream decompressors
  erofs: refine z_erofs_{init,exit}_subsystem()
  erofs: move each decompressor to its own source file
  erofs: tidy up `struct z_erofs_bvec`
  erofs: teach z_erofs_scan_folios() to handle multi-page folios
  erofs: convert z_erofs_read_fragment() to folios
  erofs: convert z_erofs_pcluster_readmore() to folios

fs/erofs/compress.h

@@ -24,6 +24,8 @@ struct z_erofs_decompressor {
 		      void *data, int size);
 	int (*decompress)(struct z_erofs_decompress_req *rq,
 			  struct page **pagepool);
+	int (*init)(void);
+	void (*exit)(void);
 	char *name;
 };
 
@@ -52,17 +54,14 @@ struct z_erofs_decompressor {
  */
 
 /*
- * short-lived pages are pages directly from buddy system with specific
- * page->private (no need to set PagePrivate since these are non-LRU /
- * non-movable pages and bypass reclaim / migration code).
+ * Currently, short-lived pages are pages directly from buddy system
+ * with specific page->private (Z_EROFS_SHORTLIVED_PAGE).
+ * In the future world of Memdescs, it should be type 0 (Misc) memory
+ * which type can be checked with a new helper.
  */
 static inline bool z_erofs_is_shortlived_page(struct page *page)
 {
-	if (page->private != Z_EROFS_SHORTLIVED_PAGE)
-		return false;
-
-	DBG_BUGON(page->mapping);
-	return true;
+	return page->private == Z_EROFS_SHORTLIVED_PAGE;
 }
 
 static inline bool z_erofs_put_shortlivedpage(struct page **pagepool,
@@ -70,32 +69,32 @@ static inline bool z_erofs_put_shortlivedpage(struct page **pagepool,
 {
 	if (!z_erofs_is_shortlived_page(page))
 		return false;
-
-	/* short-lived pages should not be used by others at the same time */
-	if (page_ref_count(page) > 1) {
-		put_page(page);
-	} else {
-		/* follow the pcluster rule above. */
-		erofs_pagepool_add(pagepool, page);
-	}
+	erofs_pagepool_add(pagepool, page);
 	return true;
 }
 
+extern const struct z_erofs_decompressor z_erofs_lzma_decomp;
+extern const struct z_erofs_decompressor z_erofs_deflate_decomp;
+extern const struct z_erofs_decompressor z_erofs_zstd_decomp;
+extern const struct z_erofs_decompressor *z_erofs_decomp[];
+
+struct z_erofs_stream_dctx {
+	struct z_erofs_decompress_req *rq;
+	unsigned int inpages, outpages;	/* # of {en,de}coded pages */
+	int no, ni;			/* the current {en,de}coded page # */
+
+	unsigned int avail_out;		/* remaining bytes in the decoded buffer */
+	unsigned int inbuf_pos, inbuf_sz;
+					/* current status of the encoded buffer */
+	u8 *kin, *kout;			/* buffer mapped pointers */
+	void *bounce;			/* bounce buffer for inplace I/Os */
+	bool bounced;			/* is the bounce buffer used now? */
+};
+
+int z_erofs_stream_switch_bufs(struct z_erofs_stream_dctx *dctx, void **dst,
+			       void **src, struct page **pgpl);
 int z_erofs_fixup_insize(struct z_erofs_decompress_req *rq, const char *padbuf,
 			 unsigned int padbufsize);
-extern const struct z_erofs_decompressor erofs_decompressors[];
-
-/* prototypes for specific algorithms */
-int z_erofs_load_lzma_config(struct super_block *sb,
-			struct erofs_super_block *dsb, void *data, int size);
-int z_erofs_load_deflate_config(struct super_block *sb,
-			struct erofs_super_block *dsb, void *data, int size);
-int z_erofs_load_zstd_config(struct super_block *sb,
-			struct erofs_super_block *dsb, void *data, int size);
-int z_erofs_lzma_decompress(struct z_erofs_decompress_req *rq,
-			    struct page **pagepool);
-int z_erofs_deflate_decompress(struct z_erofs_decompress_req *rq,
-			       struct page **pagepool);
-int z_erofs_zstd_decompress(struct z_erofs_decompress_req *rq,
-			    struct page **pgpl);
+int __init z_erofs_init_decompressor(void);
+void z_erofs_exit_decompressor(void);
 #endif

fs/erofs/decompressor.c

@@ -2,6 +2,7 @@
 /*
  * Copyright (C) 2019 HUAWEI, Inc.
  *             https://www.huawei.com/
+ * Copyright (C) 2024 Alibaba Cloud
  */
 #include "compress.h"
 #include <linux/lz4.h>
@@ -109,7 +110,6 @@ static int z_erofs_lz4_prepare_dstpages(struct z_erofs_lz4_decompress_ctx *ctx,
 
 		if (top) {
 			victim = availables[--top];
-			get_page(victim);
 		} else {
 			victim = __erofs_allocpage(pagepool, rq->gfp, true);
 			if (!victim)
@@ -371,40 +371,113 @@ static int z_erofs_transform_plain(struct z_erofs_decompress_req *rq,
 	return 0;
 }
 
-const struct z_erofs_decompressor erofs_decompressors[] = {
-	[Z_EROFS_COMPRESSION_SHIFTED] = {
+int z_erofs_stream_switch_bufs(struct z_erofs_stream_dctx *dctx, void **dst,
+			       void **src, struct page **pgpl)
+{
+	struct z_erofs_decompress_req *rq = dctx->rq;
+	struct super_block *sb = rq->sb;
+	struct page **pgo, *tmppage;
+	unsigned int j;
+
+	if (!dctx->avail_out) {
+		if (++dctx->no >= dctx->outpages || !rq->outputsize) {
+			erofs_err(sb, "insufficient space for decompressed data");
+			return -EFSCORRUPTED;
+		}
+
+		if (dctx->kout)
+			kunmap_local(dctx->kout);
+		dctx->avail_out = min(rq->outputsize, PAGE_SIZE - rq->pageofs_out);
+		rq->outputsize -= dctx->avail_out;
+		pgo = &rq->out[dctx->no];
+		if (!*pgo && rq->fillgaps) {		/* deduped */
+			*pgo = erofs_allocpage(pgpl, rq->gfp);
+			if (!*pgo) {
+				dctx->kout = NULL;
+				return -ENOMEM;
+			}
+			set_page_private(*pgo, Z_EROFS_SHORTLIVED_PAGE);
+		}
+		if (*pgo) {
+			dctx->kout = kmap_local_page(*pgo);
+			*dst = dctx->kout + rq->pageofs_out;
+		} else {
+			*dst = dctx->kout = NULL;
+		}
+		rq->pageofs_out = 0;
+	}
+
+	if (dctx->inbuf_pos == dctx->inbuf_sz && rq->inputsize) {
+		if (++dctx->ni >= dctx->inpages) {
+			erofs_err(sb, "invalid compressed data");
+			return -EFSCORRUPTED;
+		}
+		if (dctx->kout) /* unlike kmap(), take care of the orders */
+			kunmap_local(dctx->kout);
+		kunmap_local(dctx->kin);
+
+		dctx->inbuf_sz = min_t(u32, rq->inputsize, PAGE_SIZE);
+		rq->inputsize -= dctx->inbuf_sz;
+		dctx->kin = kmap_local_page(rq->in[dctx->ni]);
+		*src = dctx->kin;
+		dctx->bounced = false;
+		if (dctx->kout) {
+			j = (u8 *)*dst - dctx->kout;
+			dctx->kout = kmap_local_page(rq->out[dctx->no]);
+			*dst = dctx->kout + j;
+		}
+		dctx->inbuf_pos = 0;
+	}
+
+	/*
+	 * Handle overlapping: Use the given bounce buffer if the input data is
+	 * under processing; Or utilize short-lived pages from the on-stack page
+	 * pool, where pages are shared among the same request.  Note that only
+	 * a few inplace I/O pages need to be doubled.
+	 */
+	if (!dctx->bounced && rq->out[dctx->no] == rq->in[dctx->ni]) {
+		memcpy(dctx->bounce, *src, dctx->inbuf_sz);
+		*src = dctx->bounce;
+		dctx->bounced = true;
+	}
+
+	for (j = dctx->ni + 1; j < dctx->inpages; ++j) {
+		if (rq->out[dctx->no] != rq->in[j])
+			continue;
+		tmppage = erofs_allocpage(pgpl, rq->gfp);
+		if (!tmppage)
+			return -ENOMEM;
+		set_page_private(tmppage, Z_EROFS_SHORTLIVED_PAGE);
+		copy_highpage(tmppage, rq->in[j]);
+		rq->in[j] = tmppage;
+	}
+	return 0;
+}
+
+const struct z_erofs_decompressor *z_erofs_decomp[] = {
+	[Z_EROFS_COMPRESSION_SHIFTED] = &(const struct z_erofs_decompressor) {
 		.decompress = z_erofs_transform_plain,
 		.name = "shifted"
 	},
-	[Z_EROFS_COMPRESSION_INTERLACED] = {
+	[Z_EROFS_COMPRESSION_INTERLACED] = &(const struct z_erofs_decompressor) {
 		.decompress = z_erofs_transform_plain,
 		.name = "interlaced"
 	},
-	[Z_EROFS_COMPRESSION_LZ4] = {
+	[Z_EROFS_COMPRESSION_LZ4] = &(const struct z_erofs_decompressor) {
 		.config = z_erofs_load_lz4_config,
 		.decompress = z_erofs_lz4_decompress,
+		.init = z_erofs_gbuf_init,
+		.exit = z_erofs_gbuf_exit,
 		.name = "lz4"
 	},
 #ifdef CONFIG_EROFS_FS_ZIP_LZMA
-	[Z_EROFS_COMPRESSION_LZMA] = {
-		.config = z_erofs_load_lzma_config,
-		.decompress = z_erofs_lzma_decompress,
-		.name = "lzma"
-	},
+	[Z_EROFS_COMPRESSION_LZMA] = &z_erofs_lzma_decomp,
 #endif
 #ifdef CONFIG_EROFS_FS_ZIP_DEFLATE
-	[Z_EROFS_COMPRESSION_DEFLATE] = {
-		.config = z_erofs_load_deflate_config,
-		.decompress = z_erofs_deflate_decompress,
-		.name = "deflate"
-	},
+	[Z_EROFS_COMPRESSION_DEFLATE] = &z_erofs_deflate_decomp,
 #endif
 #ifdef CONFIG_EROFS_FS_ZIP_ZSTD
-	[Z_EROFS_COMPRESSION_ZSTD] = {
-		.config = z_erofs_load_zstd_config,
-		.decompress = z_erofs_zstd_decompress,
-		.name = "zstd"
-	},
+	[Z_EROFS_COMPRESSION_ZSTD] = &z_erofs_zstd_decomp,
 #endif
 };
 
@@ -432,6 +505,7 @@ int z_erofs_parse_cfgs(struct super_block *sb, struct erofs_super_block *dsb)
 	offset = EROFS_SUPER_OFFSET + sbi->sb_size;
 	alg = 0;
 	for (algs = sbi->available_compr_algs; algs; algs >>= 1, ++alg) {
+		const struct z_erofs_decompressor *dec = z_erofs_decomp[alg];
 		void *data;
 
 		if (!(algs & 1))
@@ -443,16 +517,13 @@ int z_erofs_parse_cfgs(struct super_block *sb, struct erofs_super_block *dsb)
 			break;
 		}
 
-		if (alg >= ARRAY_SIZE(erofs_decompressors) ||
-		    !erofs_decompressors[alg].config) {
+		if (alg < Z_EROFS_COMPRESSION_MAX && dec && dec->config) {
+			ret = dec->config(sb, dsb, data, size);
+		} else {
 			erofs_err(sb, "algorithm %d isn't enabled on this kernel",
 				  alg);
 			ret = -EOPNOTSUPP;
-		} else {
-			ret = erofs_decompressors[alg].config(sb,
-					dsb, data, size);
 		}
-
 		kfree(data);
 		if (ret)
 			break;
@@ -460,3 +531,28 @@ int z_erofs_parse_cfgs(struct super_block *sb, struct erofs_super_block *dsb)
 	erofs_put_metabuf(&buf);
 	return ret;
 }
+
+int __init z_erofs_init_decompressor(void)
+{
+	int i, err;
+
+	for (i = 0; i < Z_EROFS_COMPRESSION_MAX; ++i) {
+		err = z_erofs_decomp[i] ? z_erofs_decomp[i]->init() : 0;
+		if (err) {
+			while (--i)
+				if (z_erofs_decomp[i])
+					z_erofs_decomp[i]->exit();
+			return err;
+		}
+	}
+	return 0;
+}
+
+void z_erofs_exit_decompressor(void)
+{
+	int i;
+
+	for (i = 0; i < Z_EROFS_COMPRESSION_MAX; ++i)
+		if (z_erofs_decomp[i])
+			z_erofs_decomp[i]->exit();
+}

fs/erofs/decompressor_deflate.c

@@ -15,7 +15,7 @@ static DECLARE_WAIT_QUEUE_HEAD(z_erofs_deflate_wq);
 
 module_param_named(deflate_streams, z_erofs_deflate_nstrms, uint, 0444);
 
-void z_erofs_deflate_exit(void)
+static void z_erofs_deflate_exit(void)
 {
 	/* there should be no running fs instance */
 	while (z_erofs_deflate_avail_strms) {
@@ -41,7 +41,7 @@ void z_erofs_deflate_exit(void)
 	}
 }
 
-int __init z_erofs_deflate_init(void)
+static int __init z_erofs_deflate_init(void)
 {
 	/* by default, use # of possible CPUs instead */
 	if (!z_erofs_deflate_nstrms)
@@ -49,7 +49,7 @@ int __init z_erofs_deflate_init(void)
 	return 0;
 }
 
-int z_erofs_load_deflate_config(struct super_block *sb,
+static int z_erofs_load_deflate_config(struct super_block *sb,
 			struct erofs_super_block *dsb, void *data, int size)
 {
 	struct z_erofs_deflate_cfgs *dfl = data;
@@ -97,27 +97,26 @@ int z_erofs_load_deflate_config(struct super_block *sb,
 	return -ENOMEM;
 }
 
-int z_erofs_deflate_decompress(struct z_erofs_decompress_req *rq,
-			       struct page **pgpl)
+static int z_erofs_deflate_decompress(struct z_erofs_decompress_req *rq,
+				      struct page **pgpl)
 {
-	const unsigned int nrpages_out =
-		PAGE_ALIGN(rq->pageofs_out + rq->outputsize) >> PAGE_SHIFT;
-	const unsigned int nrpages_in =
-		PAGE_ALIGN(rq->inputsize) >> PAGE_SHIFT;
 	struct super_block *sb = rq->sb;
-	unsigned int insz, outsz, pofs;
+	struct z_erofs_stream_dctx dctx = {
+		.rq = rq,
+		.inpages = PAGE_ALIGN(rq->inputsize) >> PAGE_SHIFT,
+		.outpages = PAGE_ALIGN(rq->pageofs_out + rq->outputsize)
+				>> PAGE_SHIFT,
+		.no = -1, .ni = 0,
+	};
 	struct z_erofs_deflate *strm;
-	u8 *kin, *kout = NULL;
-	bool bounced = false;
-	int no = -1, ni = 0, j = 0, zerr, err;
+	int zerr, err;
 
 	/* 1. get the exact DEFLATE compressed size */
-	kin = kmap_local_page(*rq->in);
-	err = z_erofs_fixup_insize(rq, kin + rq->pageofs_in,
-			min_t(unsigned int, rq->inputsize,
-			      sb->s_blocksize - rq->pageofs_in));
+	dctx.kin = kmap_local_page(*rq->in);
+	err = z_erofs_fixup_insize(rq, dctx.kin + rq->pageofs_in,
+			min(rq->inputsize, sb->s_blocksize - rq->pageofs_in));
 	if (err) {
-		kunmap_local(kin);
+		kunmap_local(dctx.kin);
 		return err;
 	}
 
@@ -134,102 +133,35 @@ int z_erofs_deflate_decompress(struct z_erofs_decompress_req *rq,
 	spin_unlock(&z_erofs_deflate_lock);
 
 	/* 3. multi-call decompress */
-	insz = rq->inputsize;
-	outsz = rq->outputsize;
 	zerr = zlib_inflateInit2(&strm->z, -MAX_WBITS);
 	if (zerr != Z_OK) {
 		err = -EIO;
 		goto failed_zinit;
 	}
 
-	pofs = rq->pageofs_out;
-	strm->z.avail_in = min_t(u32, insz, PAGE_SIZE - rq->pageofs_in);
-	insz -= strm->z.avail_in;
-	strm->z.next_in = kin + rq->pageofs_in;
+	rq->fillgaps = true;	/* DEFLATE doesn't support NULL output buffer */
+	strm->z.avail_in = min(rq->inputsize, PAGE_SIZE - rq->pageofs_in);
+	rq->inputsize -= strm->z.avail_in;
+	strm->z.next_in = dctx.kin + rq->pageofs_in;
 	strm->z.avail_out = 0;
+	dctx.bounce = strm->bounce;
 
 	while (1) {
-		if (!strm->z.avail_out) {
-			if (++no >= nrpages_out || !outsz) {
-				erofs_err(sb, "insufficient space for decompressed data");
-				err = -EFSCORRUPTED;
-				break;
-			}
-
-			if (kout)
-				kunmap_local(kout);
-			strm->z.avail_out = min_t(u32, outsz, PAGE_SIZE - pofs);
-			outsz -= strm->z.avail_out;
-			if (!rq->out[no]) {
-				rq->out[no] = erofs_allocpage(pgpl, rq->gfp);
-				if (!rq->out[no]) {
-					kout = NULL;
-					err = -ENOMEM;
-					break;
-				}
-				set_page_private(rq->out[no],
-						 Z_EROFS_SHORTLIVED_PAGE);
-			}
-			kout = kmap_local_page(rq->out[no]);
-			strm->z.next_out = kout + pofs;
-			pofs = 0;
-		}
-
-		if (!strm->z.avail_in && insz) {
-			if (++ni >= nrpages_in) {
-				erofs_err(sb, "invalid compressed data");
-				err = -EFSCORRUPTED;
-				break;
-			}
-
-			if (kout) { /* unlike kmap(), take care of the orders */
-				j = strm->z.next_out - kout;
-				kunmap_local(kout);
-			}
-			kunmap_local(kin);
-			strm->z.avail_in = min_t(u32, insz, PAGE_SIZE);
-			insz -= strm->z.avail_in;
-			kin = kmap_local_page(rq->in[ni]);
-			strm->z.next_in = kin;
-			bounced = false;
-			if (kout) {
-				kout = kmap_local_page(rq->out[no]);
-				strm->z.next_out = kout + j;
-			}
-		}
-
-		/*
-		 * Handle overlapping: Use bounced buffer if the compressed
-		 * data is under processing; Or use short-lived pages from the
-		 * on-stack pagepool where pages share among the same request
-		 * and not _all_ inplace I/O pages are needed to be doubled.
-		 */
-		if (!bounced && rq->out[no] == rq->in[ni]) {
-			memcpy(strm->bounce, strm->z.next_in, strm->z.avail_in);
-			strm->z.next_in = strm->bounce;
-			bounced = true;
-		}
-
-		for (j = ni + 1; j < nrpages_in; ++j) {
-			struct page *tmppage;
-
-			if (rq->out[no] != rq->in[j])
-				continue;
-			tmppage = erofs_allocpage(pgpl, rq->gfp);
-			if (!tmppage) {
-				err = -ENOMEM;
-				goto failed;
-			}
-			set_page_private(tmppage, Z_EROFS_SHORTLIVED_PAGE);
-			copy_highpage(tmppage, rq->in[j]);
-			rq->in[j] = tmppage;
-		}
+		dctx.avail_out = strm->z.avail_out;
+		dctx.inbuf_sz = strm->z.avail_in;
+		err = z_erofs_stream_switch_bufs(&dctx,
+					(void **)&strm->z.next_out,
+					(void **)&strm->z.next_in, pgpl);
+		if (err)
+			break;
+		strm->z.avail_out = dctx.avail_out;
+		strm->z.avail_in = dctx.inbuf_sz;
 
 		zerr = zlib_inflate(&strm->z, Z_SYNC_FLUSH);
-		if (zerr != Z_OK || !(outsz + strm->z.avail_out)) {
+		if (zerr != Z_OK || !(rq->outputsize + strm->z.avail_out)) {
 			if (zerr == Z_OK && rq->partial_decoding)
 				break;
-			if (zerr == Z_STREAM_END && !outsz)
+			if (zerr == Z_STREAM_END && !rq->outputsize)
 				break;
 			erofs_err(sb, "failed to decompress %d in[%u] out[%u]",
 				  zerr, rq->inputsize, rq->outputsize);
@@ -237,13 +169,12 @@ int z_erofs_deflate_decompress(struct z_erofs_decompress_req *rq,
 			break;
 		}
 	}
-failed:
 	if (zlib_inflateEnd(&strm->z) != Z_OK && !err)
 		err = -EIO;
-	if (kout)
-		kunmap_local(kout);
+	if (dctx.kout)
+		kunmap_local(dctx.kout);
 failed_zinit:
-	kunmap_local(kin);
+	kunmap_local(dctx.kin);
 	/* 4. push back DEFLATE stream context to the global list */
 	spin_lock(&z_erofs_deflate_lock);
 	strm->next = z_erofs_deflate_head;
@@ -252,3 +183,11 @@ int z_erofs_deflate_decompress(struct z_erofs_decompress_req *rq,
 	wake_up(&z_erofs_deflate_wq);
 	return err;
 }
+
+const struct z_erofs_decompressor z_erofs_deflate_decomp = {
+	.config = z_erofs_load_deflate_config,
+	.decompress = z_erofs_deflate_decompress,
+	.init = z_erofs_deflate_init,
+	.exit = z_erofs_deflate_exit,
+	.name = "deflate",
+};

fs/erofs/decompressor_lzma.c

@@ -5,7 +5,6 @@
 struct z_erofs_lzma {
 	struct z_erofs_lzma *next;
 	struct xz_dec_microlzma *state;
-	struct xz_buf buf;
 	u8 bounce[PAGE_SIZE];
 };
 
@@ -18,7 +17,7 @@ static DECLARE_WAIT_QUEUE_HEAD(z_erofs_lzma_wq);
 
 module_param_named(lzma_streams, z_erofs_lzma_nstrms, uint, 0444);
 
-void z_erofs_lzma_exit(void)
+static void z_erofs_lzma_exit(void)
 {
 	/* there should be no running fs instance */
 	while (z_erofs_lzma_avail_strms) {
@@ -46,7 +45,7 @@ void z_erofs_lzma_exit(void)
 	}
 }
 
-int __init z_erofs_lzma_init(void)
+static int __init z_erofs_lzma_init(void)
 {
 	unsigned int i;
 
@@ -70,7 +69,7 @@ int __init z_erofs_lzma_init(void)
 	return 0;
 }
 
-int z_erofs_load_lzma_config(struct super_block *sb,
+static int z_erofs_load_lzma_config(struct super_block *sb,
 			struct erofs_super_block *dsb, void *data, int size)
 {
 	static DEFINE_MUTEX(lzma_resize_mutex);
@@ -147,26 +146,28 @@ int z_erofs_load_lzma_config(struct super_block *sb,
 	return err;
 }
 
-int z_erofs_lzma_decompress(struct z_erofs_decompress_req *rq,
-			    struct page **pgpl)
+static int z_erofs_lzma_decompress(struct z_erofs_decompress_req *rq,
+				   struct page **pgpl)
 {
-	const unsigned int nrpages_out =
-		PAGE_ALIGN(rq->pageofs_out + rq->outputsize) >> PAGE_SHIFT;
-	const unsigned int nrpages_in =
-		PAGE_ALIGN(rq->inputsize) >> PAGE_SHIFT;
-	unsigned int inlen, outlen, pageofs;
+	struct super_block *sb = rq->sb;
+	struct z_erofs_stream_dctx dctx = {
+		.rq = rq,
+		.inpages = PAGE_ALIGN(rq->inputsize) >> PAGE_SHIFT,
+		.outpages = PAGE_ALIGN(rq->pageofs_out + rq->outputsize)
+				>> PAGE_SHIFT,
+		.no = -1, .ni = 0,
+	};
+	struct xz_buf buf = {};
 	struct z_erofs_lzma *strm;
-	u8 *kin;
-	bool bounced = false;
-	int no, ni, j, err = 0;
+	enum xz_ret xz_err;
+	int err;
 
 	/* 1. get the exact LZMA compressed size */
-	kin = kmap(*rq->in);
-	err = z_erofs_fixup_insize(rq, kin + rq->pageofs_in,
-			min_t(unsigned int, rq->inputsize,
-			      rq->sb->s_blocksize - rq->pageofs_in));
+	dctx.kin = kmap_local_page(*rq->in);
+	err = z_erofs_fixup_insize(rq, dctx.kin + rq->pageofs_in,
+			min(rq->inputsize, sb->s_blocksize - rq->pageofs_in));
 	if (err) {
-		kunmap(*rq->in);
+		kunmap_local(dctx.kin);
 		return err;
 	}
 
@@ -183,108 +184,45 @@ int z_erofs_lzma_decompress(struct z_erofs_decompress_req *rq,
 	spin_unlock(&z_erofs_lzma_lock);
 
 	/* 3. multi-call decompress */
-	inlen = rq->inputsize;
-	outlen = rq->outputsize;
-	xz_dec_microlzma_reset(strm->state, inlen, outlen,
+	xz_dec_microlzma_reset(strm->state, rq->inputsize, rq->outputsize,
 			       !rq->partial_decoding);
-	pageofs = rq->pageofs_out;
-	strm->buf.in = kin + rq->pageofs_in;
-	strm->buf.in_pos = 0;
-	strm->buf.in_size = min_t(u32, inlen, PAGE_SIZE - rq->pageofs_in);
-	inlen -= strm->buf.in_size;
-	strm->buf.out = NULL;
-	strm->buf.out_pos = 0;
-	strm->buf.out_size = 0;
-
-	for (ni = 0, no = -1;;) {
-		enum xz_ret xz_err;
-
-		if (strm->buf.out_pos == strm->buf.out_size) {
-			if (strm->buf.out) {
-				kunmap(rq->out[no]);
-				strm->buf.out = NULL;
-			}
-
-			if (++no >= nrpages_out || !outlen) {
-				erofs_err(rq->sb, "decompressed buf out of bound");
-				err = -EFSCORRUPTED;
-				break;
-			}
-			strm->buf.out_pos = 0;
-			strm->buf.out_size = min_t(u32, outlen,
-						   PAGE_SIZE - pageofs);
-			outlen -= strm->buf.out_size;
-			if (!rq->out[no] && rq->fillgaps) {	/* deduped */
-				rq->out[no] = erofs_allocpage(pgpl, rq->gfp);
-				if (!rq->out[no]) {
-					err = -ENOMEM;
-					break;
-				}
-				set_page_private(rq->out[no],
-						 Z_EROFS_SHORTLIVED_PAGE);
-			}
-			if (rq->out[no])
-				strm->buf.out = kmap(rq->out[no]) + pageofs;
-			pageofs = 0;
-		} else if (strm->buf.in_pos == strm->buf.in_size) {
-			kunmap(rq->in[ni]);
-
-			if (++ni >= nrpages_in || !inlen) {
-				erofs_err(rq->sb, "compressed buf out of bound");
-				err = -EFSCORRUPTED;
-				break;
-			}
-			strm->buf.in_pos = 0;
-			strm->buf.in_size = min_t(u32, inlen, PAGE_SIZE);
-			inlen -= strm->buf.in_size;
-			kin = kmap(rq->in[ni]);
-			strm->buf.in = kin;
-			bounced = false;
-		}
+	buf.in_size = min(rq->inputsize, PAGE_SIZE - rq->pageofs_in);
+	rq->inputsize -= buf.in_size;
+	buf.in = dctx.kin + rq->pageofs_in,
+	dctx.bounce = strm->bounce;
+	do {
+		dctx.avail_out = buf.out_size - buf.out_pos;
+		dctx.inbuf_sz = buf.in_size;
+		dctx.inbuf_pos = buf.in_pos;
+		err = z_erofs_stream_switch_bufs(&dctx, (void **)&buf.out,
+						 (void **)&buf.in, pgpl);
+		if (err)
+			break;
 
-		/*
-		 * Handle overlapping: Use bounced buffer if the compressed
-		 * data is under processing; Otherwise, Use short-lived pages
-		 * from the on-stack pagepool where pages share with the same
-		 * request.
-		 */
-		if (!bounced && rq->out[no] == rq->in[ni]) {
-			memcpy(strm->bounce, strm->buf.in, strm->buf.in_size);
-			strm->buf.in = strm->bounce;
-			bounced = true;
+		if (buf.out_size == buf.out_pos) {
+			buf.out_size = dctx.avail_out;
+			buf.out_pos = 0;
 		}
-		for (j = ni + 1; j < nrpages_in; ++j) {
-			struct page *tmppage;
+		buf.in_size = dctx.inbuf_sz;
+		buf.in_pos = dctx.inbuf_pos;
 
-			if (rq->out[no] != rq->in[j])
-				continue;
-			tmppage = erofs_allocpage(pgpl, rq->gfp);
-			if (!tmppage) {
-				err = -ENOMEM;
-				goto failed;
-			}
-			set_page_private(tmppage, Z_EROFS_SHORTLIVED_PAGE);
-			copy_highpage(tmppage, rq->in[j]);
-			rq->in[j] = tmppage;
-		}
-		xz_err = xz_dec_microlzma_run(strm->state, &strm->buf);
-		DBG_BUGON(strm->buf.out_pos > strm->buf.out_size);
-		DBG_BUGON(strm->buf.in_pos > strm->buf.in_size);
+		xz_err = xz_dec_microlzma_run(strm->state, &buf);
+		DBG_BUGON(buf.out_pos > buf.out_size);
+		DBG_BUGON(buf.in_pos > buf.in_size);
 
 		if (xz_err != XZ_OK) {
-			if (xz_err == XZ_STREAM_END && !outlen)
+			if (xz_err == XZ_STREAM_END && !rq->outputsize)
 				break;
-			erofs_err(rq->sb, "failed to decompress %d in[%u] out[%u]",
+			erofs_err(sb, "failed to decompress %d in[%u] out[%u]",
 				  xz_err, rq->inputsize, rq->outputsize);
 			err = -EFSCORRUPTED;
 			break;
 		}
-	}
-failed:
-	if (no < nrpages_out && strm->buf.out)
-		kunmap(rq->out[no]);
-	if (ni < nrpages_in)
-		kunmap(rq->in[ni]);
+	} while (1);
+
+	if (dctx.kout)
+		kunmap_local(dctx.kout);
+	kunmap_local(dctx.kin);
 	/* 4. push back LZMA stream context to the global list */
 	spin_lock(&z_erofs_lzma_lock);
 	strm->next = z_erofs_lzma_head;
@@ -293,3 +231,11 @@ int z_erofs_lzma_decompress(struct z_erofs_decompress_req *rq,
 	wake_up(&z_erofs_lzma_wq);
 	return err;
 }
+
+const struct z_erofs_decompressor z_erofs_lzma_decomp = {
+	.config = z_erofs_load_lzma_config,
+	.decompress = z_erofs_lzma_decompress,
+	.init = z_erofs_lzma_init,
+	.exit = z_erofs_lzma_exit,
+	.name = "lzma"
+};

fs/erofs/decompressor_zstd.c

@@ -34,7 +34,7 @@ static struct z_erofs_zstd *z_erofs_isolate_strms(bool all)
 	return strm;
 }
 
-void z_erofs_zstd_exit(void)
+static void z_erofs_zstd_exit(void)
 {
 	while (z_erofs_zstd_avail_strms) {
 		struct z_erofs_zstd *strm, *n;
@@ -49,7 +49,7 @@ void z_erofs_zstd_exit(void)
 	}
 }
 
-int __init z_erofs_zstd_init(void)
+static int __init z_erofs_zstd_init(void)
 {
 	/* by default, use # of possible CPUs instead */
 	if (!z_erofs_zstd_nstrms)
@@ -72,7 +72,7 @@ int __init z_erofs_zstd_init(void)
 	return 0;
 }
 
-int z_erofs_load_zstd_config(struct super_block *sb,
+static int z_erofs_load_zstd_config(struct super_block *sb,
 			struct erofs_super_block *dsb, void *data, int size)
 {
 	static DEFINE_MUTEX(zstd_resize_mutex);
@@ -135,30 +135,29 @@ int z_erofs_load_zstd_config(struct super_block *sb,
 	return strm ? -ENOMEM : 0;
 }
 
-int z_erofs_zstd_decompress(struct z_erofs_decompress_req *rq,
-			    struct page **pgpl)
+static int z_erofs_zstd_decompress(struct z_erofs_decompress_req *rq,
+				   struct page **pgpl)
 {
-	const unsigned int nrpages_out =
-		PAGE_ALIGN(rq->pageofs_out + rq->outputsize) >> PAGE_SHIFT;
-	const unsigned int nrpages_in =
-		PAGE_ALIGN(rq->inputsize) >> PAGE_SHIFT;
-	zstd_dstream *stream;
 	struct super_block *sb = rq->sb;
-	unsigned int insz, outsz, pofs;
-	struct z_erofs_zstd *strm;
+	struct z_erofs_stream_dctx dctx = {
+		.rq = rq,
+		.inpages = PAGE_ALIGN(rq->inputsize) >> PAGE_SHIFT,
+		.outpages = PAGE_ALIGN(rq->pageofs_out + rq->outputsize)
+				>> PAGE_SHIFT,
+		.no = -1, .ni = 0,
+	};
 	zstd_in_buffer in_buf = { NULL, 0, 0 };
 	zstd_out_buffer out_buf = { NULL, 0, 0 };
-	u8 *kin, *kout = NULL;
-	bool bounced = false;
-	int no = -1, ni = 0, j = 0, zerr, err;
+	struct z_erofs_zstd *strm;
+	zstd_dstream *stream;
+	int zerr, err;
 
 	/* 1. get the exact compressed size */
-	kin = kmap_local_page(*rq->in);
-	err = z_erofs_fixup_insize(rq, kin + rq->pageofs_in,
-			min_t(unsigned int, rq->inputsize,
-			      sb->s_blocksize - rq->pageofs_in));
+	dctx.kin = kmap_local_page(*rq->in);
+	err = z_erofs_fixup_insize(rq, dctx.kin + rq->pageofs_in,
+			min(rq->inputsize, sb->s_blocksize - rq->pageofs_in));
 	if (err) {
-		kunmap_local(kin);
+		kunmap_local(dctx.kin);
 		return err;
 	}
 
@@ -166,109 +165,48 @@ int z_erofs_zstd_decompress(struct z_erofs_decompress_req *rq,
 	strm = z_erofs_isolate_strms(false);
 
 	/* 3. multi-call decompress */
-	insz = rq->inputsize;
-	outsz = rq->outputsize;
 	stream = zstd_init_dstream(z_erofs_zstd_max_dictsize, strm->wksp, strm->wkspsz);
 	if (!stream) {
 		err = -EIO;
 		goto failed_zinit;
 	}
 
-	pofs = rq->pageofs_out;
-	in_buf.size = min_t(u32, insz, PAGE_SIZE - rq->pageofs_in);
-	insz -= in_buf.size;
-	in_buf.src = kin + rq->pageofs_in;
+	rq->fillgaps = true;	/* ZSTD doesn't support NULL output buffer */
+	in_buf.size = min_t(u32, rq->inputsize, PAGE_SIZE - rq->pageofs_in);
+	rq->inputsize -= in_buf.size;
+	in_buf.src = dctx.kin + rq->pageofs_in;
+	dctx.bounce = strm->bounce;
+
 	do {
-		if (out_buf.size == out_buf.pos) {
-			if (++no >= nrpages_out || !outsz) {
-				erofs_err(sb, "insufficient space for decompressed data");
-				err = -EFSCORRUPTED;
-				break;
-			}
+		dctx.avail_out = out_buf.size - out_buf.pos;
+		dctx.inbuf_sz = in_buf.size;
+		dctx.inbuf_pos = in_buf.pos;
+		err = z_erofs_stream_switch_bufs(&dctx, &out_buf.dst,
+						 (void **)&in_buf.src, pgpl);
+		if (err)
+			break;
 
-			if (kout)
-				kunmap_local(kout);
-			out_buf.size = min_t(u32, outsz, PAGE_SIZE - pofs);
-			outsz -= out_buf.size;
-			if (!rq->out[no]) {
-				rq->out[no] = erofs_allocpage(pgpl, rq->gfp);
-				if (!rq->out[no]) {
-					kout = NULL;
-					err = -ENOMEM;
-					break;
-				}
-				set_page_private(rq->out[no],
-						 Z_EROFS_SHORTLIVED_PAGE);
-			}
-			kout = kmap_local_page(rq->out[no]);
-			out_buf.dst = kout + pofs;
+		if (out_buf.size == out_buf.pos) {
+			out_buf.size = dctx.avail_out;
 			out_buf.pos = 0;
-			pofs = 0;
-		}
-
-		if (in_buf.size == in_buf.pos && insz) {
-			if (++ni >= nrpages_in) {
-				erofs_err(sb, "invalid compressed data");
-				err = -EFSCORRUPTED;
-				break;
-			}
-
-			if (kout) /* unlike kmap(), take care of the orders */
-				kunmap_local(kout);
-			kunmap_local(kin);
-			in_buf.size = min_t(u32, insz, PAGE_SIZE);
-			insz -= in_buf.size;
-			kin = kmap_local_page(rq->in[ni]);
-			in_buf.src = kin;
-			in_buf.pos = 0;
-			bounced = false;
-			if (kout) {
-				j = (u8 *)out_buf.dst - kout;
-				kout = kmap_local_page(rq->out[no]);
-				out_buf.dst = kout + j;
-			}
 		}
+		in_buf.size = dctx.inbuf_sz;
+		in_buf.pos = dctx.inbuf_pos;
 
-		/*
-		 * Handle overlapping: Use bounced buffer if the compressed
-		 * data is under processing; Or use short-lived pages from the
-		 * on-stack pagepool where pages share among the same request
-		 * and not _all_ inplace I/O pages are needed to be doubled.
-		 */
-		if (!bounced && rq->out[no] == rq->in[ni]) {
-			memcpy(strm->bounce, in_buf.src, in_buf.size);
-			in_buf.src = strm->bounce;
-			bounced = true;
-		}
-
-		for (j = ni + 1; j < nrpages_in; ++j) {
-			struct page *tmppage;
-
-			if (rq->out[no] != rq->in[j])
-				continue;
-			tmppage = erofs_allocpage(pgpl, rq->gfp);
-			if (!tmppage) {
-				err = -ENOMEM;
-				goto failed;
-			}
-			set_page_private(tmppage, Z_EROFS_SHORTLIVED_PAGE);
-			copy_highpage(tmppage, rq->in[j]);
-			rq->in[j] = tmppage;
-		}
 		zerr = zstd_decompress_stream(stream, &out_buf, &in_buf);
-		if (zstd_is_error(zerr) || (!zerr && outsz)) {
+		if (zstd_is_error(zerr) || (!zerr && rq->outputsize)) {
 			erofs_err(sb, "failed to decompress in[%u] out[%u]: %s",
 				  rq->inputsize, rq->outputsize,
 				  zerr ? zstd_get_error_name(zerr) : "unexpected end of stream");
 			err = -EFSCORRUPTED;
 			break;
 		}
-	} while (outsz || out_buf.pos < out_buf.size);
-failed:
-	if (kout)
-		kunmap_local(kout);
+	} while (rq->outputsize || out_buf.pos < out_buf.size);
+
+	if (dctx.kout)
+		kunmap_local(dctx.kout);
 failed_zinit:
-	kunmap_local(kin);
+	kunmap_local(dctx.kin);
 	/* 4. push back ZSTD stream context to the global list */
 	spin_lock(&z_erofs_zstd_lock);
 	strm->next = z_erofs_zstd_head;
@@ -277,3 +215,11 @@ int z_erofs_zstd_decompress(struct z_erofs_decompress_req *rq,
 	wake_up(&z_erofs_zstd_wq);
 	return err;
 }
+
+const struct z_erofs_decompressor z_erofs_zstd_decomp = {
+	.config = z_erofs_load_zstd_config,
+	.decompress = z_erofs_zstd_decompress,
+	.init = z_erofs_zstd_init,
+	.exit = z_erofs_zstd_exit,
+	.name = "zstd",
+};

fs/erofs/internal.h

@@ -312,17 +312,13 @@ static inline unsigned int erofs_inode_datalayout(unsigned int ifmt)
 	return (ifmt >> EROFS_I_DATALAYOUT_BIT) & EROFS_I_DATALAYOUT_MASK;
 }
 
-/*
- * Different from grab_cache_page_nowait(), reclaiming is never triggered
- * when allocating new pages.
- */
-static inline
-struct page *erofs_grab_cache_page_nowait(struct address_space *mapping,
-					  pgoff_t index)
+/* reclaiming is never triggered when allocating new folios. */
+static inline struct folio *erofs_grab_folio_nowait(struct address_space *as,
+						    pgoff_t index)
 {
-	return pagecache_get_page(mapping, index,
+	return __filemap_get_folio(as, index,
 			FGP_LOCK|FGP_CREAT|FGP_NOFS|FGP_NOWAIT,
-			readahead_gfp_mask(mapping) & ~__GFP_RECLAIM);
+			readahead_gfp_mask(as) & ~__GFP_RECLAIM);
 }
 
 /* Has a disk mapping */
@@ -458,8 +454,8 @@ void erofs_shrinker_register(struct super_block *sb);
 void erofs_shrinker_unregister(struct super_block *sb);
 int __init erofs_init_shrinker(void);
 void erofs_exit_shrinker(void);
-int __init z_erofs_init_zip_subsystem(void);
-void z_erofs_exit_zip_subsystem(void);
+int __init z_erofs_init_subsystem(void);
+void z_erofs_exit_subsystem(void);
 int erofs_try_to_free_all_cached_folios(struct erofs_sb_info *sbi,
 					struct erofs_workgroup *egrp);
 int z_erofs_map_blocks_iter(struct inode *inode, struct erofs_map_blocks *map,
@@ -476,37 +472,11 @@ static inline void erofs_shrinker_register(struct super_block *sb) {}
 static inline void erofs_shrinker_unregister(struct super_block *sb) {}
 static inline int erofs_init_shrinker(void) { return 0; }
 static inline void erofs_exit_shrinker(void) {}
-static inline int z_erofs_init_zip_subsystem(void) { return 0; }
-static inline void z_erofs_exit_zip_subsystem(void) {}
-static inline int z_erofs_gbuf_init(void) { return 0; }
-static inline void z_erofs_gbuf_exit(void) {}
+static inline int z_erofs_init_subsystem(void) { return 0; }
+static inline void z_erofs_exit_subsystem(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
-int __init z_erofs_lzma_init(void);
-void z_erofs_lzma_exit(void);
-#else
-static inline int z_erofs_lzma_init(void) { return 0; }
-static inline int z_erofs_lzma_exit(void) { return 0; }
-#endif	/* !CONFIG_EROFS_FS_ZIP_LZMA */
-
-#ifdef CONFIG_EROFS_FS_ZIP_DEFLATE
-int __init z_erofs_deflate_init(void);
-void z_erofs_deflate_exit(void);
-#else
-static inline int z_erofs_deflate_init(void) { return 0; }
-static inline int z_erofs_deflate_exit(void) { return 0; }
-#endif	/* !CONFIG_EROFS_FS_ZIP_DEFLATE */
-
-#ifdef CONFIG_EROFS_FS_ZIP_ZSTD
-int __init z_erofs_zstd_init(void);
-void z_erofs_zstd_exit(void);
-#else
-static inline int z_erofs_zstd_init(void) { return 0; }
-static inline int z_erofs_zstd_exit(void) { return 0; }
-#endif	/* !CONFIG_EROFS_FS_ZIP_ZSTD */
-
 #ifdef CONFIG_EROFS_FS_ONDEMAND
 int erofs_fscache_register_fs(struct super_block *sb);
 void erofs_fscache_unregister_fs(struct super_block *sb);

fs/erofs/super.c

@@ -849,23 +849,7 @@ static int __init erofs_module_init(void)
 	if (err)
 		goto shrinker_err;
 
-	err = z_erofs_lzma_init();
-	if (err)
-		goto lzma_err;
-
-	err = z_erofs_deflate_init();
-	if (err)
-		goto deflate_err;
-
-	err = z_erofs_zstd_init();
-	if (err)
-		goto zstd_err;
-
-	err = z_erofs_gbuf_init();
-	if (err)
-		goto gbuf_err;
-
-	err = z_erofs_init_zip_subsystem();
+	err = z_erofs_init_subsystem();
 	if (err)
 		goto zip_err;
 
@@ -882,16 +866,8 @@ static int __init erofs_module_init(void)
 fs_err:
 	erofs_exit_sysfs();
 sysfs_err:
-	z_erofs_exit_zip_subsystem();
+	z_erofs_exit_subsystem();
 zip_err:
-	z_erofs_gbuf_exit();
-gbuf_err:
-	z_erofs_zstd_exit();
-zstd_err:
-	z_erofs_deflate_exit();
-deflate_err:
-	z_erofs_lzma_exit();
-lzma_err:
 	erofs_exit_shrinker();
 shrinker_err:
 	kmem_cache_destroy(erofs_inode_cachep);
@@ -906,13 +882,9 @@ static void __exit erofs_module_exit(void)
 	rcu_barrier();
 
 	erofs_exit_sysfs();
-	z_erofs_exit_zip_subsystem();
-	z_erofs_zstd_exit();
-	z_erofs_deflate_exit();
-	z_erofs_lzma_exit();
+	z_erofs_exit_subsystem();
 	erofs_exit_shrinker();
 	kmem_cache_destroy(erofs_inode_cachep);
-	z_erofs_gbuf_exit();
 }
 
 static int erofs_statfs(struct dentry *dentry, struct kstatfs *buf)

fs/erofs/zdata.c

@@ -19,10 +19,7 @@
 typedef void *z_erofs_next_pcluster_t;
 
 struct z_erofs_bvec {
-	union {
-		struct page *page;
-		struct folio *folio;
-	};
+	struct page *page;
 	int offset;
 	unsigned int end;
 };
@@ -449,44 +446,51 @@ static inline int erofs_cpu_hotplug_init(void) { return 0; }
 static inline void erofs_cpu_hotplug_destroy(void) {}
 #endif
 
-void z_erofs_exit_zip_subsystem(void)
+void z_erofs_exit_subsystem(void)
 {
 	erofs_cpu_hotplug_destroy();
 	erofs_destroy_percpu_workers();
 	destroy_workqueue(z_erofs_workqueue);
 	z_erofs_destroy_pcluster_pool();
+	z_erofs_exit_decompressor();
 }
 
-int __init z_erofs_init_zip_subsystem(void)
+int __init z_erofs_init_subsystem(void)
 {
-	int err = z_erofs_create_pcluster_pool();
+	int err = z_erofs_init_decompressor();
 
 	if (err)
-		goto out_error_pcluster_pool;
+		goto err_decompressor;
+
+	err = z_erofs_create_pcluster_pool();
+	if (err)
+		goto err_pcluster_pool;
 
 	z_erofs_workqueue = alloc_workqueue("erofs_worker",
 			WQ_UNBOUND | WQ_HIGHPRI, num_possible_cpus());
 	if (!z_erofs_workqueue) {
 		err = -ENOMEM;
-		goto out_error_workqueue_init;
+		goto err_workqueue_init;
 	}
 
 	err = erofs_init_percpu_workers();
 	if (err)
-		goto out_error_pcpu_worker;
+		goto err_pcpu_worker;
 
 	err = erofs_cpu_hotplug_init();
 	if (err < 0)
-		goto out_error_cpuhp_init;
+		goto err_cpuhp_init;
 	return err;
 
-out_error_cpuhp_init:
+err_cpuhp_init:
 	erofs_destroy_percpu_workers();
-out_error_pcpu_worker:
+err_pcpu_worker:
 	destroy_workqueue(z_erofs_workqueue);
-out_error_workqueue_init:
+err_workqueue_init:
 	z_erofs_destroy_pcluster_pool();
-out_error_pcluster_pool:
+err_pcluster_pool:
+	z_erofs_exit_decompressor();
+err_decompressor:
 	return err;
 }
 
@@ -617,32 +621,31 @@ static void z_erofs_bind_cache(struct z_erofs_decompress_frontend *fe)
 		fe->mode = Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE;
 }
 
-/* called by erofs_shrinker to get rid of all cached compressed bvecs */
+/* (erofs_shrinker) disconnect cached encoded data with pclusters */
 int erofs_try_to_free_all_cached_folios(struct erofs_sb_info *sbi,
 					struct erofs_workgroup *grp)
 {
 	struct z_erofs_pcluster *const pcl =
 		container_of(grp, struct z_erofs_pcluster, obj);
 	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
+	struct folio *folio;
 	int i;
 
 	DBG_BUGON(z_erofs_is_inline_pcluster(pcl));
-	/* There is no actice user since the pcluster is now freezed */
+	/* Each cached folio contains one page unless bs > ps is supported */
 	for (i = 0; i < pclusterpages; ++i) {
-		struct folio *folio = pcl->compressed_bvecs[i].folio;
-
-		if (!folio)
-			continue;
+		if (pcl->compressed_bvecs[i].page) {
+			folio = page_folio(pcl->compressed_bvecs[i].page);
+			/* Avoid reclaiming or migrating this folio */
+			if (!folio_trylock(folio))
+				return -EBUSY;
 
-		/* Avoid reclaiming or migrating this folio */
-		if (!folio_trylock(folio))
-			return -EBUSY;
-
-		if (!erofs_folio_is_managed(sbi, folio))
-			continue;
-		pcl->compressed_bvecs[i].folio = NULL;
-		folio_detach_private(folio);
-		folio_unlock(folio);
+			if (!erofs_folio_is_managed(sbi, folio))
+				continue;
+			pcl->compressed_bvecs[i].page = NULL;
+			folio_detach_private(folio);
+			folio_unlock(folio);
+		}
 	}
 	return 0;
 }
@@ -650,9 +653,9 @@ int erofs_try_to_free_all_cached_folios(struct erofs_sb_info *sbi,
 static bool z_erofs_cache_release_folio(struct folio *folio, gfp_t gfp)
 {
 	struct z_erofs_pcluster *pcl = folio_get_private(folio);
-	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
+	struct z_erofs_bvec *bvec = pcl->compressed_bvecs;
+	struct z_erofs_bvec *end = bvec + z_erofs_pclusterpages(pcl);
 	bool ret;
-	int i;
 
 	if (!folio_test_private(folio))
 		return true;
@@ -661,9 +664,9 @@ static bool z_erofs_cache_release_folio(struct folio *folio, gfp_t gfp)
 	spin_lock(&pcl->obj.lockref.lock);
 	if (pcl->obj.lockref.count <= 0) {
 		DBG_BUGON(z_erofs_is_inline_pcluster(pcl));
-		for (i = 0; i < pclusterpages; ++i) {
-			if (pcl->compressed_bvecs[i].folio == folio) {
-				pcl->compressed_bvecs[i].folio = NULL;
+		for (; bvec < end; ++bvec) {
+			if (bvec->page && page_folio(bvec->page) == folio) {
+				bvec->page = NULL;
 				folio_detach_private(folio);
 				ret = true;
 				break;
@@ -925,7 +928,7 @@ static void z_erofs_pcluster_end(struct z_erofs_decompress_frontend *fe)
 	fe->pcl = NULL;
 }
 
-static int z_erofs_read_fragment(struct super_block *sb, struct page *page,
+static int z_erofs_read_fragment(struct super_block *sb, struct folio *folio,
 			unsigned int cur, unsigned int end, erofs_off_t pos)
 {
 	struct inode *packed_inode = EROFS_SB(sb)->packed_inode;
@@ -938,113 +941,109 @@ static int z_erofs_read_fragment(struct super_block *sb, struct page *page,
 
 	buf.mapping = packed_inode->i_mapping;
 	for (; cur < end; cur += cnt, pos += cnt) {
-		cnt = min_t(unsigned int, end - cur,
-			    sb->s_blocksize - erofs_blkoff(sb, pos));
+		cnt = min(end - cur, sb->s_blocksize - erofs_blkoff(sb, pos));
 		src = erofs_bread(&buf, pos, EROFS_KMAP);
 		if (IS_ERR(src)) {
 			erofs_put_metabuf(&buf);
 			return PTR_ERR(src);
 		}
-		memcpy_to_page(page, cur, src, cnt);
+		memcpy_to_folio(folio, cur, src, cnt);
 	}
 	erofs_put_metabuf(&buf);
 	return 0;
 }
 
-static int z_erofs_scan_folio(struct z_erofs_decompress_frontend *fe,
+static int z_erofs_scan_folio(struct z_erofs_decompress_frontend *f,
 			      struct folio *folio, bool ra)
 {
-	struct inode *const inode = fe->inode;
-	struct erofs_map_blocks *const map = &fe->map;
+	struct inode *const inode = f->inode;
+	struct erofs_map_blocks *const map = &f->map;
 	const loff_t offset = folio_pos(folio);
-	const unsigned int bs = i_blocksize(inode), fs = folio_size(folio);
-	bool tight = true, exclusive;
-	unsigned int cur, end, len, split;
+	const unsigned int bs = i_blocksize(inode);
+	unsigned int end = folio_size(folio), split = 0, cur, pgs;
+	bool tight, excl;
 	int err = 0;
 
+	tight = (bs == PAGE_SIZE);
 	z_erofs_onlinefolio_init(folio);
-	split = 0;
-	end = fs;
-repeat:
-	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;
-	}
-
-	cur = offset > map->m_la ? 0 : map->m_la - offset;
-	/* bump split parts first to avoid several separate cases */
-	++split;
-
-	if (!(map->m_flags & EROFS_MAP_MAPPED)) {
-		folio_zero_segment(folio, cur, end);
-		tight = false;
-		goto next_part;
-	}
-
-	if (map->m_flags & EROFS_MAP_FRAGMENT) {
-		erofs_off_t fpos = offset + cur - map->m_la;
+	do {
+		if (offset + end - 1 < map->m_la ||
+		    offset + end - 1 >= map->m_la + map->m_llen) {
+			z_erofs_pcluster_end(f);
+			map->m_la = offset + end - 1;
+			map->m_llen = 0;
+			err = z_erofs_map_blocks_iter(inode, map, 0);
+			if (err)
+				break;
+		}
 
-		len = min_t(unsigned int, map->m_llen - fpos, end - cur);
-		err = z_erofs_read_fragment(inode->i_sb, &folio->page, cur,
-			cur + len, EROFS_I(inode)->z_fragmentoff + fpos);
-		if (err)
-			goto out;
-		tight = false;
-		goto next_part;
-	}
+		cur = offset > map->m_la ? 0 : map->m_la - offset;
+		pgs = round_down(cur, PAGE_SIZE);
+		/* bump split parts first to avoid several separate cases */
+		++split;
+
+		if (!(map->m_flags & EROFS_MAP_MAPPED)) {
+			folio_zero_segment(folio, cur, end);
+			tight = false;
+		} else if (map->m_flags & EROFS_MAP_FRAGMENT) {
+			erofs_off_t fpos = offset + cur - map->m_la;
+
+			err = z_erofs_read_fragment(inode->i_sb, folio, cur,
+					cur + min(map->m_llen - fpos, end - cur),
+					EROFS_I(inode)->z_fragmentoff + fpos);
+			if (err)
+				break;
+			tight = false;
+		} else {
+			if (!f->pcl) {
+				err = z_erofs_pcluster_begin(f);
+				if (err)
+					break;
+				f->pcl->besteffort |= !ra;
+			}
 
-	if (!fe->pcl) {
-		err = z_erofs_pcluster_begin(fe);
-		if (err)
-			goto out;
-		fe->pcl->besteffort |= !ra;
-	}
+			pgs = round_down(end - 1, PAGE_SIZE);
+			/*
+			 * Ensure this partial page belongs to this submit chain
+			 * rather than other concurrent submit chains or
+			 * noio(bypass) chains since those chains are handled
+			 * asynchronously thus it cannot be used for inplace I/O
+			 * or bvpage (should be processed in the strict order.)
+			 */
+			tight &= (f->mode >= Z_EROFS_PCLUSTER_FOLLOWED);
+			excl = false;
+			if (cur <= pgs) {
+				excl = (split <= 1) || tight;
+				cur = pgs;
+			}
 
-	/*
-	 * Ensure the current partial folio belongs to this submit chain rather
-	 * than other concurrent submit chains or the noio(bypass) chain since
-	 * those chains are handled asynchronously thus the folio 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 && bs == fs)));
-	if (cur)
-		tight &= (fe->mode >= Z_EROFS_PCLUSTER_FOLLOWED);
-
-	err = z_erofs_attach_page(fe, &((struct z_erofs_bvec) {
-					.page = &folio->page,
-					.offset = offset - map->m_la,
-					.end = end,
-				  }), exclusive);
-	if (err)
-		goto out;
-
-	z_erofs_onlinefolio_split(folio);
-	if (fe->pcl->pageofs_out != (map->m_la & ~PAGE_MASK))
-		fe->pcl->multibases = true;
-	if (fe->pcl->length < offset + end - map->m_la) {
-		fe->pcl->length = offset + end - map->m_la;
-		fe->pcl->pageofs_out = map->m_la & ~PAGE_MASK;
-	}
-	if ((map->m_flags & EROFS_MAP_FULL_MAPPED) &&
-	    !(map->m_flags & EROFS_MAP_PARTIAL_REF) &&
-	    fe->pcl->length == map->m_llen)
-		fe->pcl->partial = false;
-next_part:
-	/* shorten the remaining extent to update progress */
-	map->m_llen = offset + cur - map->m_la;
-	map->m_flags &= ~EROFS_MAP_FULL_MAPPED;
-
-	end = cur;
-	if (end > 0)
-		goto repeat;
+			err = z_erofs_attach_page(f, &((struct z_erofs_bvec) {
+				.page = folio_page(folio, pgs >> PAGE_SHIFT),
+				.offset = offset + pgs - map->m_la,
+				.end = end - pgs, }), excl);
+			if (err)
+				break;
 
-out:
+			z_erofs_onlinefolio_split(folio);
+			if (f->pcl->pageofs_out != (map->m_la & ~PAGE_MASK))
+				f->pcl->multibases = true;
+			if (f->pcl->length < offset + end - map->m_la) {
+				f->pcl->length = offset + end - map->m_la;
+				f->pcl->pageofs_out = map->m_la & ~PAGE_MASK;
+			}
+			if ((map->m_flags & EROFS_MAP_FULL_MAPPED) &&
+			    !(map->m_flags & EROFS_MAP_PARTIAL_REF) &&
+			    f->pcl->length == map->m_llen)
+				f->pcl->partial = false;
+		}
+		/* shorten the remaining extent to update progress */
+		map->m_llen = offset + cur - map->m_la;
+		map->m_flags &= ~EROFS_MAP_FULL_MAPPED;
+		if (cur <= pgs) {
+			split = cur < pgs;
+			tight = (bs == PAGE_SIZE);
+		}
+	} while ((end = cur) > 0);
 	z_erofs_onlinefolio_end(folio, err);
 	return err;
 }
@@ -1066,7 +1065,7 @@ static bool z_erofs_is_sync_decompress(struct erofs_sb_info *sbi,
 
 static bool z_erofs_page_is_invalidated(struct page *page)
 {
-	return !page->mapping && !z_erofs_is_shortlived_page(page);
+	return !page_folio(page)->mapping && !z_erofs_is_shortlived_page(page);
 }
 
 struct z_erofs_decompress_backend {
@@ -1221,8 +1220,8 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be,
 	struct z_erofs_pcluster *pcl = be->pcl;
 	unsigned int pclusterpages = z_erofs_pclusterpages(pcl);
 	const struct z_erofs_decompressor *decomp =
-				&erofs_decompressors[pcl->algorithmformat];
-	int i, err2;
+				z_erofs_decomp[pcl->algorithmformat];
+	int i, j, jtop, err2;
 	struct page *page;
 	bool overlapped;
 
@@ -1280,10 +1279,9 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be,
 		WRITE_ONCE(pcl->compressed_bvecs[0].page, NULL);
 		put_page(page);
 	} else {
+		/* managed folios are still left in compressed_bvecs[] */
 		for (i = 0; i < pclusterpages; ++i) {
-			/* consider shortlived pages added when decompressing */
 			page = be->compressed_pages[i];
-
 			if (!page ||
 			    erofs_folio_is_managed(sbi, page_folio(page)))
 				continue;
@@ -1294,21 +1292,31 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be,
 	if (be->compressed_pages < be->onstack_pages ||
 	    be->compressed_pages >= be->onstack_pages + Z_EROFS_ONSTACK_PAGES)
 		kvfree(be->compressed_pages);
-	z_erofs_fill_other_copies(be, err);
 
+	jtop = 0;
+	z_erofs_fill_other_copies(be, err);
 	for (i = 0; i < be->nr_pages; ++i) {
 		page = be->decompressed_pages[i];
 		if (!page)
 			continue;
 
 		DBG_BUGON(z_erofs_page_is_invalidated(page));
-
-		/* recycle all individual short-lived pages */
-		if (z_erofs_put_shortlivedpage(be->pagepool, page))
+		if (!z_erofs_is_shortlived_page(page)) {
+			z_erofs_onlinefolio_end(page_folio(page), err);
 			continue;
-		z_erofs_onlinefolio_end(page_folio(page), err);
+		}
+		if (pcl->algorithmformat != Z_EROFS_COMPRESSION_LZ4) {
+			erofs_pagepool_add(be->pagepool, page);
+			continue;
+		}
+		for (j = 0; j < jtop && be->decompressed_pages[j] != page; ++j)
+			;
+		if (j >= jtop)	/* this bounce page is newly detected */
+			be->decompressed_pages[jtop++] = page;
 	}
-
+	while (jtop)
+		erofs_pagepool_add(be->pagepool,
+				   be->decompressed_pages[--jtop]);
 	if (be->decompressed_pages != be->onstack_pages)
 		kvfree(be->decompressed_pages);
 
@@ -1419,7 +1427,7 @@ static void z_erofs_fill_bio_vec(struct bio_vec *bvec,
 	bool tocache = false;
 	struct z_erofs_bvec zbv;
 	struct address_space *mapping;
-	struct page *page;
+	struct folio *folio;
 	int bs = i_blocksize(f->inode);
 
 	/* Except for inplace folios, the entire folio can be used for I/Os */
@@ -1429,23 +1437,25 @@ static void z_erofs_fill_bio_vec(struct bio_vec *bvec,
 	spin_lock(&pcl->obj.lockref.lock);
 	zbv = pcl->compressed_bvecs[nr];
 	spin_unlock(&pcl->obj.lockref.lock);
-	if (!zbv.folio)
+	if (!zbv.page)
 		goto out_allocfolio;
 
-	bvec->bv_page = &zbv.folio->page;
+	bvec->bv_page = zbv.page;
 	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 (zbv.folio->private == (void *)Z_EROFS_PREALLOCATED_PAGE) {
-		zbv.folio->private = 0;
+	if (folio->private == (void *)Z_EROFS_PREALLOCATED_PAGE) {
+		folio->private = 0;
 		tocache = true;
 		goto out_tocache;
 	}
 
-	mapping = READ_ONCE(zbv.folio->mapping);
+	mapping = READ_ONCE(folio->mapping);
 	/*
 	 * File-backed folios for inplace I/Os are all locked steady,
 	 * therefore it is impossible for `mapping` to be NULL.
@@ -1457,21 +1467,21 @@ static void z_erofs_fill_bio_vec(struct bio_vec *bvec,
 		return;
 	}
 
-	folio_lock(zbv.folio);
-	if (zbv.folio->mapping == mc) {
+	folio_lock(folio);
+	if (folio->mapping == mc) {
 		/*
 		 * The cached folio is still in managed cache but without
 		 * a valid `->private` pcluster hint.  Let's reconnect them.
 		 */
-		if (!folio_test_private(zbv.folio)) {
-			folio_attach_private(zbv.folio, pcl);
+		if (!folio_test_private(folio)) {
+			folio_attach_private(folio, pcl);
 			/* compressed_bvecs[] already takes a ref before */
-			folio_put(zbv.folio);
+			folio_put(folio);
 		}
 
 		/* no need to submit if it is already up-to-date */
-		if (folio_test_uptodate(zbv.folio)) {
-			folio_unlock(zbv.folio);
+		if (folio_test_uptodate(folio)) {
+			folio_unlock(folio);
 			bvec->bv_page = NULL;
 		}
 		return;
@@ -1481,32 +1491,31 @@ static void z_erofs_fill_bio_vec(struct bio_vec *bvec,
 	 * It has been truncated, so it's unsafe to reuse this one. Let's
 	 * allocate a new page for compressed data.
 	 */
-	DBG_BUGON(zbv.folio->mapping);
+	DBG_BUGON(folio->mapping);
 	tocache = true;
-	folio_unlock(zbv.folio);
-	folio_put(zbv.folio);
+	folio_unlock(folio);
+	folio_put(folio);
 out_allocfolio:
-	page = erofs_allocpage(&f->pagepool, gfp | __GFP_NOFAIL);
+	zbv.page = erofs_allocpage(&f->pagepool, gfp | __GFP_NOFAIL);
 	spin_lock(&pcl->obj.lockref.lock);
-	if (pcl->compressed_bvecs[nr].folio) {
-		erofs_pagepool_add(&f->pagepool, page);
+	if (pcl->compressed_bvecs[nr].page) {
+		erofs_pagepool_add(&f->pagepool, zbv.page);
 		spin_unlock(&pcl->obj.lockref.lock);
 		cond_resched();
 		goto repeat;
 	}
-	pcl->compressed_bvecs[nr].folio = zbv.folio = page_folio(page);
+	bvec->bv_page = pcl->compressed_bvecs[nr].page = zbv.page;
+	folio = page_folio(zbv.page);
+	/* first mark it as a temporary shortlived folio (now 1 ref) */
+	folio->private = (void *)Z_EROFS_SHORTLIVED_PAGE;
 	spin_unlock(&pcl->obj.lockref.lock);
-	bvec->bv_page = page;
 out_tocache:
 	if (!tocache || bs != PAGE_SIZE ||
-	    filemap_add_folio(mc, zbv.folio, pcl->obj.index + nr, gfp)) {
-		/* turn into a temporary shortlived folio (1 ref) */
-		zbv.folio->private = (void *)Z_EROFS_SHORTLIVED_PAGE;
+	    filemap_add_folio(mc, folio, pcl->obj.index + nr, gfp))
 		return;
-	}
-	folio_attach_private(zbv.folio, pcl);
+	folio_attach_private(folio, pcl);
 	/* drop a refcount added by allocpage (then 2 refs in total here) */
-	folio_put(zbv.folio);
+	folio_put(folio);
 }
 
 static struct z_erofs_decompressqueue *jobqueue_init(struct super_block *sb,
@@ -1767,7 +1776,6 @@ static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f,
 		end = round_up(end, PAGE_SIZE);
 	} else {
 		end = round_up(map->m_la, PAGE_SIZE);
-
 		if (!map->m_llen)
 			return;
 	}
@@ -1775,15 +1783,15 @@ static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f,
 	cur = map->m_la + map->m_llen - 1;
 	while ((cur >= end) && (cur < i_size_read(inode))) {
 		pgoff_t index = cur >> PAGE_SHIFT;
-		struct page *page;
+		struct folio *folio;
 
-		page = erofs_grab_cache_page_nowait(inode->i_mapping, index);
-		if (page) {
-			if (PageUptodate(page))
-				unlock_page(page);
+		folio = erofs_grab_folio_nowait(inode->i_mapping, index);
+		if (!IS_ERR_OR_NULL(folio)) {
+			if (folio_test_uptodate(folio))
+				folio_unlock(folio);
 			else
-				z_erofs_scan_folio(f, page_folio(page), !!rac);
-			put_page(page);
+				z_erofs_scan_folio(f, folio, !!rac);
+			folio_put(folio);
 		}
 
 		if (cur < PAGE_SIZE)

fs/erofs/zmap.c

@@ -686,7 +686,7 @@ int z_erofs_map_blocks_iter(struct inode *inode, struct erofs_map_blocks *map,
 	struct erofs_inode *const vi = EROFS_I(inode);
 	int err = 0;
 
-	trace_z_erofs_map_blocks_iter_enter(inode, map, flags);
+	trace_erofs_map_blocks_enter(inode, map, flags);
 
 	/* when trying to read beyond EOF, leave it unmapped */
 	if (map->m_la >= inode->i_size) {
@@ -713,7 +713,7 @@ int z_erofs_map_blocks_iter(struct inode *inode, struct erofs_map_blocks *map,
 out:
 	if (err)
 		map->m_llen = 0;
-	trace_z_erofs_map_blocks_iter_exit(inode, map, flags, err);
+	trace_erofs_map_blocks_exit(inode, map, flags, err);
 	return err;
 }
 

include/trace/events/erofs.h

@@ -143,7 +143,8 @@ TRACE_EVENT(erofs_readpages,
 		__entry->raw)
 );
 
-DECLARE_EVENT_CLASS(erofs__map_blocks_enter,
+TRACE_EVENT(erofs_map_blocks_enter,
+
 	TP_PROTO(struct inode *inode, struct erofs_map_blocks *map,
 		 unsigned int flags),
 
@@ -171,21 +172,8 @@ DECLARE_EVENT_CLASS(erofs__map_blocks_enter,
 		  __entry->flags ? show_map_flags(__entry->flags) : "NULL")
 );
 
-DEFINE_EVENT(erofs__map_blocks_enter, erofs_map_blocks_enter,
-	TP_PROTO(struct inode *inode, struct erofs_map_blocks *map,
-		 unsigned flags),
-
-	TP_ARGS(inode, map, flags)
-);
-
-DEFINE_EVENT(erofs__map_blocks_enter, z_erofs_map_blocks_iter_enter,
-	TP_PROTO(struct inode *inode, struct erofs_map_blocks *map,
-		 unsigned int flags),
-
-	TP_ARGS(inode, map, flags)
-);
+TRACE_EVENT(erofs_map_blocks_exit,
 
-DECLARE_EVENT_CLASS(erofs__map_blocks_exit,
 	TP_PROTO(struct inode *inode, struct erofs_map_blocks *map,
 		 unsigned int flags, int ret),
 
@@ -223,20 +211,6 @@ DECLARE_EVENT_CLASS(erofs__map_blocks_exit,
 		  show_mflags(__entry->mflags), __entry->ret)
 );
 
-DEFINE_EVENT(erofs__map_blocks_exit, erofs_map_blocks_exit,
-	TP_PROTO(struct inode *inode, struct erofs_map_blocks *map,
-		 unsigned flags, int ret),
-
-	TP_ARGS(inode, map, flags, ret)
-);
-
-DEFINE_EVENT(erofs__map_blocks_exit, z_erofs_map_blocks_iter_exit,
-	TP_PROTO(struct inode *inode, struct erofs_map_blocks *map,
-		 unsigned int flags, int ret),
-
-	TP_ARGS(inode, map, flags, ret)
-);
-
 TRACE_EVENT(erofs_destroy_inode,
 	TP_PROTO(struct inode *inode),