erofs: tidy up synchronous decompression

[ Upstream commit cc831ab336 ]

 - Get rid of `sbi->opt.max_sync_decompress_pages` since it's fixed as
   3 all the time;

 - Add Z_EROFS_MAX_SYNC_DECOMPRESS_BYTES in bytes instead of in pages,
   since for non-4K pages, 3-page limitation makes no sense;

 - Move `sync_decompress` to sbi to avoid unexpected remount impact;

 - Fold z_erofs_is_sync_decompress() into its caller;

 - Better description of sysfs entry `sync_decompress`.

Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Stable-dep-of: 1aee05e814 ("erofs: fix use-after-free on sbi->sync_decompress")
Signed-off-by: Sasha Levin <sashal@kernel.org>

Documentation/ABI/testing/sysfs-fs-erofs

@@ -10,12 +10,16 @@ Description:	Shows all enabled kernel features.
 What:		/sys/fs/erofs/<disk>/sync_decompress
 Date:		November 2021
 Contact:	"Huang Jianan" <huangjianan@oppo.com>
-Description:	Control strategy of sync decompression:
+Description:	Control strategy of synchronous decompression. Synchronous
+		decompression tries to decompress in the reader thread for
+		synchronous reads and small asynchronous reads (<= 12 KiB):
 
-		- 0 (default, auto): enable for readpage, and enable for
-		  readahead on atomic contexts only.
-		- 1 (force on): enable for readpage and readahead.
-		- 2 (force off): disable for all situations.
+		- 0 (auto, default): apply to synchronous reads only, but will
+		                     switch to 1 (force on) if any decompression
+		                     request is detected in atomic contexts;
+		- 1 (force on): apply to synchronous reads and small
+		                asynchronous reads;
+		- 2 (force off): disable synchronous decompression completely.
 
 What:		/sys/fs/erofs/<disk>/drop_caches
 Date:		November 2024

fs/erofs/internal.h

@@ -59,10 +59,6 @@ enum {
 struct erofs_mount_opts {
 	/* current strategy of how to use managed cache */
 	unsigned char cache_strategy;
-	/* strategy of sync decompression (0 - auto, 1 - force on, 2 - force off) */
-	unsigned int sync_decompress;
-	/* threshold for decompression synchronously */
-	unsigned int max_sync_decompress_pages;
 	unsigned int mount_opt;
 };
 
@@ -116,6 +112,7 @@ struct erofs_sb_info {
 	/* managed XArray arranged in physical block number */
 	struct xarray managed_pslots;
 
+	unsigned int sync_decompress;	/* strategy for sync decompression */
 	unsigned int shrinker_run_no;
 	u16 available_compr_algs;
 

fs/erofs/super.c

@@ -379,8 +379,7 @@ static void erofs_default_options(struct erofs_sb_info *sbi)
 {
 #ifdef CONFIG_EROFS_FS_ZIP
 	sbi->opt.cache_strategy = EROFS_ZIP_CACHE_READAROUND;
-	sbi->opt.max_sync_decompress_pages = 3;
-	sbi->opt.sync_decompress = EROFS_SYNC_DECOMPRESS_AUTO;
+	sbi->sync_decompress = EROFS_SYNC_DECOMPRESS_AUTO;
 #endif
 #ifdef CONFIG_EROFS_FS_XATTR
 	set_opt(&sbi->opt, XATTR_USER);

fs/erofs/sysfs.c

@@ -59,7 +59,7 @@ static struct erofs_attr erofs_attr_##_name = {			\
 #define ATTR_LIST(name) (&erofs_attr_##name.attr)
 
 #ifdef CONFIG_EROFS_FS_ZIP
-EROFS_ATTR_RW_UI(sync_decompress, erofs_mount_opts);
+EROFS_ATTR_RW_UI(sync_decompress, erofs_sb_info);
 EROFS_ATTR_FUNC(drop_caches, 0200);
 #endif
 #ifdef CONFIG_EROFS_FS_ZIP_ACCEL

fs/erofs/zdata.c

@@ -9,6 +9,7 @@
 #include <linux/cpuhotplug.h>
 #include <trace/events/erofs.h>
 
+#define Z_EROFS_MAX_SYNC_DECOMPRESS_BYTES	12288
 #define Z_EROFS_PCLUSTER_MAX_PAGES	(Z_EROFS_PCLUSTER_MAX_SIZE / PAGE_SIZE)
 #define Z_EROFS_INLINE_BVECS		2
 
@@ -1097,21 +1098,6 @@ static int z_erofs_scan_folio(struct z_erofs_frontend *f,
 	return err;
 }
 
-static bool z_erofs_is_sync_decompress(struct erofs_sb_info *sbi,
-				       unsigned int readahead_pages)
-{
-	/* auto: enable for read_folio, disable for readahead */
-	if ((sbi->opt.sync_decompress == EROFS_SYNC_DECOMPRESS_AUTO) &&
-	    !readahead_pages)
-		return true;
-
-	if ((sbi->opt.sync_decompress == EROFS_SYNC_DECOMPRESS_FORCE_ON) &&
-	    (readahead_pages <= sbi->opt.max_sync_decompress_pages))
-		return true;
-
-	return false;
-}
-
 static bool z_erofs_page_is_invalidated(struct page *page)
 {
 	return !page_folio(page)->mapping && !z_erofs_is_shortlived_page(page);
@@ -1473,9 +1459,9 @@ static void z_erofs_decompress_kickoff(struct z_erofs_decompressqueue *io,
 #else
 		queue_work(z_erofs_workqueue, &io->u.work);
 #endif
-		/* enable sync decompression for readahead */
-		if (sbi->opt.sync_decompress == EROFS_SYNC_DECOMPRESS_AUTO)
-			sbi->opt.sync_decompress = EROFS_SYNC_DECOMPRESS_FORCE_ON;
+		/* See `sync_decompress` in sysfs-fs-erofs for more details */
+		if (sbi->sync_decompress == EROFS_SYNC_DECOMPRESS_AUTO)
+			sbi->sync_decompress = EROFS_SYNC_DECOMPRESS_FORCE_ON;
 		return;
 	}
 	gfp_flag = memalloc_noio_save();
@@ -1795,16 +1781,21 @@ drain_io:
 	z_erofs_decompress_kickoff(q[JQ_SUBMIT], nr_bios);
 }
 
-static int z_erofs_runqueue(struct z_erofs_frontend *f, unsigned int rapages)
+static int z_erofs_runqueue(struct z_erofs_frontend *f, unsigned int rabytes)
 {
 	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, rapages);
+	int syncmode = sbi->sync_decompress;
+	bool force_fg;
 	int err;
 
+	force_fg = (syncmode == EROFS_SYNC_DECOMPRESS_AUTO && !rabytes) ||
+		(syncmode == EROFS_SYNC_DECOMPRESS_FORCE_ON &&
+			(rabytes <= Z_EROFS_MAX_SYNC_DECOMPRESS_BYTES));
+
 	if (f->head == Z_EROFS_PCLUSTER_TAIL)
 		return 0;
-	z_erofs_submit_queue(f, io, &force_fg, !!rapages);
+	z_erofs_submit_queue(f, io, &force_fg, !!rabytes);
 
 	/* handle bypass queue (no i/o pclusters) immediately */
 	err = z_erofs_decompress_queue(&io[JQ_BYPASS], &f->pagepool);
@@ -1925,7 +1916,7 @@ 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, nrpages);
+	(void)z_erofs_runqueue(&f, nrpages << PAGE_SHIFT);
 	erofs_put_metabuf(&f.map.buf);
 	erofs_release_pages(&f.pagepool);
 }