A new fscache-based shared domain mode is going to be introduced for
erofs. In which case, same data blobs in same domain will be shared
and reused to reduce on-disk space usage.

The implementation of sharing blobs will be introduced in subsequent
patches.

Signed-off-by: Jia Zhu <zhujia.zj@bytedance.com>
---
 fs/erofs/fscache.c  | 130 ++++++++++++++++++++++++++++++++++++++------
 fs/erofs/internal.h |   9 +++
 2 files changed, 121 insertions(+), 18 deletions(-)

diff --git a/fs/erofs/fscache.c b/fs/erofs/fscache.c
index d72e2a7ea6ab..10d3f0511f15 100644
--- a/fs/erofs/fscache.c
+++ b/fs/erofs/fscache.c
@@ -1,10 +1,14 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 /*
  * Copyright (C) 2022, Alibaba Cloud
+ * Copyright (C) 2022, Bytedance Inc. All rights reserved.
  */
 #include <linux/fscache.h>
 #include "internal.h"
 
+static DEFINE_MUTEX(erofs_domain_list_lock);
+static LIST_HEAD(erofs_domain_list);
+
 static struct netfs_io_request *erofs_fscache_alloc_request(struct address_space *mapping,
 					     loff_t start, size_t len)
 {
@@ -417,6 +421,99 @@ const struct address_space_operations erofs_fscache_access_aops = {
 	.readahead = erofs_fscache_readahead,
 };
 
+static void erofs_fscache_domain_put(struct erofs_domain *domain)
+{
+	if (!domain)
+		return;
+	mutex_lock(&erofs_domain_list_lock);
+	if (refcount_dec_and_test(&domain->ref)) {
+		list_del(&domain->list);
+		fscache_relinquish_volume(domain->volume, NULL, false);
+		mutex_unlock(&erofs_domain_list_lock);
+		kfree(domain->domain_id);
+		kfree(domain);
+		return;
+	}
+	mutex_unlock(&erofs_domain_list_lock);
+}
+
+static int erofs_fscache_register_volume(struct super_block *sb)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+	char *domain_id = sbi->opt.domain_id;
+	struct fscache_volume *volume;
+	char *name;
+	int ret = 0;
+
+	name = kasprintf(GFP_KERNEL, "erofs,%s",
+			 domain_id ? domain_id : sbi->opt.fsid);
+	if (!name)
+		return -ENOMEM;
+
+	volume = fscache_acquire_volume(name, NULL, NULL, 0);
+	if (IS_ERR_OR_NULL(volume)) {
+		erofs_err(sb, "failed to register volume for %s", name);
+		ret = volume ? PTR_ERR(volume) : -EOPNOTSUPP;
+		volume = NULL;
+	}
+
+	sbi->volume = volume;
+	kfree(name);
+	return ret;
+}
+
+static int erofs_fscache_init_domain(struct super_block *sb)
+{
+	int err;
+	struct erofs_domain *domain;
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+
+	domain = kzalloc(sizeof(struct erofs_domain), GFP_KERNEL);
+	if (!domain)
+		return -ENOMEM;
+
+	domain->domain_id = kstrdup(sbi->opt.domain_id, GFP_KERNEL);
+	if (!domain->domain_id) {
+		kfree(domain);
+		return -ENOMEM;
+	}
+
+	err = erofs_fscache_register_volume(sb);
+	if (err)
+		goto out;
+
+	domain->volume = sbi->volume;
+	refcount_set(&domain->ref, 1);
+	list_add(&domain->list, &erofs_domain_list);
+	sbi->domain = domain;
+	return 0;
+out:
+	kfree(domain->domain_id);
+	kfree(domain);
+	return err;
+}
+
+static int erofs_fscache_register_domain(struct super_block *sb)
+{
+	int err;
+	struct erofs_domain *domain;
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+
+	mutex_lock(&erofs_domain_list_lock);
+	list_for_each_entry(domain, &erofs_domain_list, list) {
+		if (!strcmp(domain->domain_id, sbi->opt.domain_id)) {
+			sbi->domain = domain;
+			sbi->volume = domain->volume;
+			refcount_inc(&domain->ref);
+			mutex_unlock(&erofs_domain_list_lock);
+			return 0;
+		}
+	}
+	err = erofs_fscache_init_domain(sb);
+	mutex_unlock(&erofs_domain_list_lock);
+	return err;
+}
+
 struct erofs_fscache *erofs_fscache_register_cookie(struct super_block *sb,
 						     char *name, bool need_inode)
 {
@@ -481,27 +578,18 @@ void erofs_fscache_unregister_cookie(struct erofs_fscache *ctx)
 int erofs_fscache_register_fs(struct super_block *sb)
 {
 	struct erofs_sb_info *sbi = EROFS_SB(sb);
-	struct fscache_volume *volume;
 	struct erofs_fscache *fscache;
-	char *name;
-	int ret = 0;
-
-	name = kasprintf(GFP_KERNEL, "erofs,%s", sbi->opt.fsid);
-	if (!name)
-		return -ENOMEM;
-
-	volume = fscache_acquire_volume(name, NULL, NULL, 0);
-	if (IS_ERR_OR_NULL(volume)) {
-		erofs_err(sb, "failed to register volume for %s", name);
-		kfree(name);
-		return volume ? PTR_ERR(volume) : -EOPNOTSUPP;
-	}
+	int ret;
 
-	sbi->volume = volume;
-	kfree(name);
+	if (sbi->opt.domain_id)
+		ret = erofs_fscache_register_domain(sb);
+	else
+		ret = erofs_fscache_register_volume(sb);
+	if (ret)
+		return ret;
 
+	/* acquired domain/volume will be relinquished in kill_sb() if error occurs */
 	fscache = erofs_fscache_register_cookie(sb, sbi->opt.fsid, true);
-	/* acquired volume will be relinquished in kill_sb() */
 	if (IS_ERR(fscache))
 		return PTR_ERR(fscache);
 
@@ -514,7 +602,13 @@ void erofs_fscache_unregister_fs(struct super_block *sb)
 	struct erofs_sb_info *sbi = EROFS_SB(sb);
 
 	erofs_fscache_unregister_cookie(sbi->s_fscache);
-	fscache_relinquish_volume(sbi->volume, NULL, false);
+
+	if (sbi->domain)
+		erofs_fscache_domain_put(sbi->domain);
+	else
+		fscache_relinquish_volume(sbi->volume, NULL, false);
+
 	sbi->s_fscache = NULL;
 	sbi->volume = NULL;
+	sbi->domain = NULL;
 }
diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
index aa71eb65e965..7f0939f4005b 100644
--- a/fs/erofs/internal.h
+++ b/fs/erofs/internal.h
@@ -76,6 +76,7 @@ struct erofs_mount_opts {
 #endif
 	unsigned int mount_opt;
 	char *fsid;
+	char *domain_id;
 };
 
 struct erofs_dev_context {
@@ -98,6 +99,13 @@ struct erofs_sb_lz4_info {
 	u16 max_pclusterblks;
 };
 
+struct erofs_domain {
+	refcount_t ref;
+	struct list_head list;
+	struct fscache_volume *volume;
+	char *domain_id;
+};
+
 struct erofs_fscache {
 	struct fscache_cookie *cookie;
 	struct inode *inode;
@@ -157,6 +165,7 @@ struct erofs_sb_info {
 	/* fscache support */
 	struct fscache_volume *volume;
 	struct erofs_fscache *s_fscache;
+	struct erofs_domain *domain;
 };
 
 #define EROFS_SB(sb) ((struct erofs_sb_info *)(sb)->s_fs_info)
-- 
2.20.1

On 9/15/22 8:42 PM, Jia Zhu wrote:
>  
> +static void erofs_fscache_domain_put(struct erofs_domain *domain)
> +{
> +	if (!domain)
> +		return;
> +	mutex_lock(&erofs_domain_list_lock);
> +	if (refcount_dec_and_test(&domain->ref)) {
> +		list_del(&domain->list);
> +		fscache_relinquish_volume(domain->volume, NULL, false);
> +		mutex_unlock(&erofs_domain_list_lock);

I think we can mutex_unlock as long as the domain is removed from the list.


Otherwise LGTM.
Reviewed-by: Jingbo Xu <jefflexu@linux.alibaba.com>

-- 
Thanks,
Jingbo

On 9/15/22 8:42 PM, Jia Zhu wrote:

>  struct erofs_fscache *erofs_fscache_register_cookie(struct super_block *sb,
>  						     char *name, bool need_inode)
>  {
> @@ -481,27 +578,18 @@ void erofs_fscache_unregister_cookie(struct erofs_fscache *ctx)
>  int erofs_fscache_register_fs(struct super_block *sb)
>  {
>  	struct erofs_sb_info *sbi = EROFS_SB(sb);
> -	struct fscache_volume *volume;
>  	struct erofs_fscache *fscache;
> -	char *name;
> -	int ret = 0;
> -
> -	name = kasprintf(GFP_KERNEL, "erofs,%s", sbi->opt.fsid);
> -	if (!name)
> -		return -ENOMEM;
> -
> -	volume = fscache_acquire_volume(name, NULL, NULL, 0);
> -	if (IS_ERR_OR_NULL(volume)) {
> -		erofs_err(sb, "failed to register volume for %s", name);
> -		kfree(name);
> -		return volume ? PTR_ERR(volume) : -EOPNOTSUPP;
> -	}
> +	int ret;
>  
> -	sbi->volume = volume;
> -	kfree(name);
> +	if (sbi->opt.domain_id)
> +		ret = erofs_fscache_register_domain(sb);
> +	else
> +		ret = erofs_fscache_register_volume(sb);
> +	if (ret)
> +		return ret;
>  
> +	/* acquired domain/volume will be relinquished in kill_sb() if error occurs */

"... in kill_sb() on error" is better to make this line within 80 chars
length.


>  	fscache = erofs_fscache_register_cookie(sb, sbi->opt.fsid, true);
> -	/* acquired volume will be relinquished in kill_sb() */
>  	if (IS_ERR(fscache))
>  		return PTR_ERR(fscache);
>  


-- 
Thanks,
Jingbo