erofs: add sysfs node to control sync decompression strategy

Although readpage is a synchronous path, there will be no additional
kworker scheduling overhead in non-atomic contexts together with
dm-verity.

Let's add a sysfs node to disable sync decompression as an option.

Link: https://lore.kernel.org/r/20211206143552.8384-1-huangjianan@oppo.com
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Huang Jianan <huangjianan@oppo.com>
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>

Documentation/ABI/testing/sysfs-fs-erofs

@@ -5,3 +5,12 @@ Description:	Shows all enabled kernel features.
 		Supported features:
 		zero_padding, compr_cfgs, big_pcluster, chunked_file,
 		device_table, compr_head2, sb_chksum.
+
+What:		/sys/fs/erofs/<disk>/sync_decompress
+Date:		November 2021
+Contact:	"Huang Jianan" <huangjianan@oppo.com>
+Description:	Control strategy of sync decompression
+		- 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.

fs/erofs/internal.h

@@ -56,12 +56,18 @@ struct erofs_device_info {
 	u32 mapped_blkaddr;
 };
 
+enum {
+	EROFS_SYNC_DECOMPRESS_AUTO,
+	EROFS_SYNC_DECOMPRESS_FORCE_ON,
+	EROFS_SYNC_DECOMPRESS_FORCE_OFF
+};
+
 struct erofs_mount_opts {
 #ifdef CONFIG_EROFS_FS_ZIP
 	/* current strategy of how to use managed cache */
 	unsigned char cache_strategy;
-	/* strategy of sync decompression (false - auto, true - force on) */
-	bool readahead_sync_decompress;
+	/* 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;

fs/erofs/super.c

@@ -423,7 +423,7 @@ static void erofs_default_options(struct erofs_fs_context *ctx)
 #ifdef CONFIG_EROFS_FS_ZIP
 	ctx->opt.cache_strategy = EROFS_ZIP_CACHE_READAROUND;
 	ctx->opt.max_sync_decompress_pages = 3;
-	ctx->opt.readahead_sync_decompress = false;
+	ctx->opt.sync_decompress = EROFS_SYNC_DECOMPRESS_AUTO;
 #endif
 #ifdef CONFIG_EROFS_FS_XATTR
 	set_opt(&ctx->opt, XATTR_USER);

fs/erofs/sysfs.c

@@ -16,6 +16,7 @@ enum {
 
 enum {
 	struct_erofs_sb_info,
+	struct_erofs_mount_opts,
 };
 
 struct erofs_attr {
@@ -54,7 +55,14 @@ 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);
+#endif
+
 static struct attribute *erofs_attrs[] = {
+#ifdef CONFIG_EROFS_FS_ZIP
+	ATTR_LIST(sync_decompress),
+#endif
 	NULL,
 };
 ATTRIBUTE_GROUPS(erofs);
@@ -85,6 +93,8 @@ static unsigned char *__struct_ptr(struct erofs_sb_info *sbi,
 {
 	if (struct_type == struct_erofs_sb_info)
 		return (unsigned char *)sbi + offset;
+	if (struct_type == struct_erofs_mount_opts)
+		return (unsigned char *)&sbi->opt + offset;
 	return NULL;
 }
 
@@ -130,6 +140,11 @@ static ssize_t erofs_attr_store(struct kobject *kobj, struct attribute *attr,
 			return ret;
 		if (t != (unsigned int)t)
 			return -ERANGE;
+#ifdef CONFIG_EROFS_FS_ZIP
+		if (!strcmp(a->attr.name, "sync_decompress") &&
+		    (t > EROFS_SYNC_DECOMPRESS_FORCE_OFF))
+			return -EINVAL;
+#endif
 		*(unsigned int *)ptr = t;
 		return len;
 	case attr_pointer_bool:

fs/erofs/zdata.c

@@ -762,6 +762,21 @@ err_out:
 	goto out;
 }
 
+static bool z_erofs_get_sync_decompress_policy(struct erofs_sb_info *sbi,
+				       unsigned int readahead_pages)
+{
+	/* auto: enable for readpage, 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 void z_erofs_decompressqueue_work(struct work_struct *work);
 static void z_erofs_decompress_kickoff(struct z_erofs_decompressqueue *io,
 				       bool sync, int bios)
@@ -784,7 +799,9 @@ static void z_erofs_decompress_kickoff(struct z_erofs_decompressqueue *io,
 	/* Use workqueue and sync decompression for atomic contexts only */
 	if (in_atomic() || irqs_disabled()) {
 		queue_work(z_erofs_workqueue, &io->u.work);
-		sbi->opt.readahead_sync_decompress = true;
+		/* enable sync decompression for readahead */
+		if (sbi->opt.sync_decompress == EROFS_SYNC_DECOMPRESS_AUTO)
+			sbi->opt.sync_decompress = EROFS_SYNC_DECOMPRESS_FORCE_ON;
 		return;
 	}
 	z_erofs_decompressqueue_work(&io->u.work);
@@ -1435,6 +1452,7 @@ skip:
 static int z_erofs_readpage(struct file *file, struct page *page)
 {
 	struct inode *const inode = page->mapping->host;
+	struct erofs_sb_info *const sbi = EROFS_I_SB(inode);
 	struct z_erofs_decompress_frontend f = DECOMPRESS_FRONTEND_INIT(inode);
 	struct page *pagepool = NULL;
 	int err;
@@ -1450,7 +1468,8 @@ static int z_erofs_readpage(struct file *file, struct page *page)
 	(void)z_erofs_collector_end(&f.clt);
 
 	/* if some compressed cluster ready, need submit them anyway */
-	z_erofs_runqueue(inode->i_sb, &f, &pagepool, true);
+	z_erofs_runqueue(inode->i_sb, &f, &pagepool,
+			 z_erofs_get_sync_decompress_policy(sbi, 0));
 
 	if (err)
 		erofs_err(inode->i_sb, "failed to read, err [%d]", err);
@@ -1501,8 +1520,7 @@ static void z_erofs_readahead(struct readahead_control *rac)
 	(void)z_erofs_collector_end(&f.clt);
 
 	z_erofs_runqueue(inode->i_sb, &f, &pagepool,
-			 sbi->opt.readahead_sync_decompress &&
-			 nr_pages <= sbi->opt.max_sync_decompress_pages);
+			 z_erofs_get_sync_decompress_policy(sbi, nr_pages));
 	if (f.map.mpage)
 		put_page(f.map.mpage);
 	erofs_release_pages(&pagepool);