From: Ran Xiaokai <ran.xiaokai@zte.com.cn>

All tmpfs and ramfs pages have PG_dirty bit set
once they are allocated and added to the pagecache。So pages
can not be freed from tmpfs or ramfs. So skip tmpfs and ramfs
when drop caches.

Signed-off-by: Ran Xiaokai <ran.xiaokai@zte.com.cn>
---
 fs/drop_caches.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/fs/drop_caches.c b/fs/drop_caches.c
index e619c31b6bd9..600c043c1eb7 100644
--- a/fs/drop_caches.c
+++ b/fs/drop_caches.c
@@ -10,6 +10,7 @@
 #include <linux/writeback.h>
 #include <linux/sysctl.h>
 #include <linux/gfp.h>
+#include <linux/magic.h>
 #include "internal.h"
 
 /* A global variable is a bit ugly, but it keeps the code simple */
@@ -19,6 +20,9 @@ static void drop_pagecache_sb(struct super_block *sb, void *unused)
 {
 	struct inode *inode, *toput_inode = NULL;
 
+	if (sb->s_magic == TMPFS_MAGIC || sb->s_magic == RAMFS_MAGIC)
+		return;
+
 	spin_lock(&sb->s_inode_list_lock);
 	list_for_each_entry(inode, &sb->s_inodes, i_sb_list) {
 		spin_lock(&inode->i_lock);
-- 
2.15.2

On Fri, Jun 24, 2022 at 06:21:21PM +0000, cgel.zte@gmail.com wrote:
> From: Ran Xiaokai <ran.xiaokai@zte.com.cn>
> 
> All tmpfs and ramfs pages have PG_dirty bit set
> once they are allocated and added to the pagecache。So pages

minor,
                                                    ^
                                   Chinese full stop here..

Thanks,
Gao Xiang

On Fri, Jun 24, 2022 at 06:21:21PM +0000, cgel.zte@gmail.com wrote:
> From: Ran Xiaokai <ran.xiaokai@zte.com.cn>
> 
> All tmpfs and ramfs pages have PG_dirty bit set
> once they are allocated and added to the pagecache。So pages
> can not be freed from tmpfs or ramfs. So skip tmpfs and ramfs
> when drop caches.

Hard coding magic constants in VFS code is a bad idea.  If you have
numbers to justify this change (which should go into this commit log),
it can be properly solved by a flag on the super block,
file_system_type or even better bdi.