[PATCH v2 0/3] add support for drop_caches for individual filesystem

[PATCH v2 0/3] add support for drop_caches for individual filesystem

[Ye Bin]

From: Ye Bin <yebin10@huawei.com>

In order to better analyze the issue of file system uninstallation caused
by kernel module opening files, it is necessary to perform dentry recycling
on a single file system. But now, apart from global dentry recycling, it is
not supported to do dentry recycling on a single file system separately.
This feature has usage scenarios in problem localization scenarios.At the
same time, it also provides users with a slightly fine-grained
pagecache/entry recycling mechanism.
This patchset supports the recycling of pagecache/entry for individual file
systems.

Diff v2 vs v1:
1. Fix possible live lock for shrink_icache_sb().
2. Introduce reclaim_dcache_sb() for reclaim dentry.
3. Fix potential deadlocks as follows:
https://lore.kernel.org/linux-fsdevel/00000000000098f75506153551a1@google.com/
After some consideration, it was decided that this feature would primarily
be used for debugging purposes. Instead of adding a new IOCTL command, the
task_work mechanism was employed to address potential deadlock issues.

Ye Bin (3):
  vfs: introduce reclaim_icache_sb() and reclaim_dcache_sb() helper
  sysctl: add support for drop_caches for individual filesystem
  Documentation: add instructions for using 'drop_fs_caches sysctl'
    sysctl

 Documentation/admin-guide/sysctl/vm.rst |  34 +++++++
 fs/dcache.c                             |  22 ++++
 fs/drop_caches.c                        | 127 ++++++++++++++++++++++++
 fs/inode.c                              |  21 ++++
 fs/internal.h                           |   1 +
 include/linux/dcache.h                  |   1 +
 6 files changed, 206 insertions(+)

-- 
2.34.1

[PATCH v2 1/3] vfs: introduce reclaim_icache_sb() and reclaim_dcache_sb() helper

[Ye Bin]

From: Ye Bin <yebin10@huawei.com>

This patch is prepare for support drop_caches for specify file system.
reclaim_icache_sb()/reclaim_dcache_sb() helper walk the superblock
inode/dentry LRU for freeable inodes/dentrys and attempt to free them.

Signed-off-by: Ye Bin <yebin10@huawei.com>
---
 fs/dcache.c            | 22 ++++++++++++++++++++++
 fs/inode.c             | 21 +++++++++++++++++++++
 fs/internal.h          |  1 +
 include/linux/dcache.h |  1 +
 4 files changed, 45 insertions(+)

diff --git a/fs/dcache.c b/fs/dcache.c
index de3e4e9777ea..d1b29b0f9062 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -1263,6 +1263,28 @@ void shrink_dcache_sb(struct super_block *sb)
 }
 EXPORT_SYMBOL(shrink_dcache_sb);
 
+/**
+ * reclaim_dcache_sb - reclaim dcache for a superblock
+ * @sb: superblock
+ *
+ * Reclaim the dcache for the specified super block. This is used to free
+ * the dcache via sysctl 'drop_fs_caches'.
+ */
+void reclaim_dcache_sb(struct super_block *sb)
+{
+	unsigned long count = list_lru_count(&sb->s_dentry_lru);
+
+	while (count > 0) {
+		LIST_HEAD(dispose);
+		unsigned long nr_to_walk = count >= 1024 ? 1024 : count;
+
+		count -= nr_to_walk;
+		list_lru_walk(&sb->s_dentry_lru, dentry_lru_isolate, &dispose,
+			      nr_to_walk);
+		shrink_dentry_list(&dispose);
+	}
+}
+
 /**
  * enum d_walk_ret - action to talke during tree walk
  * @D_WALK_CONTINUE:	contrinue walk
diff --git a/fs/inode.c b/fs/inode.c
index 7901c2896d78..325de5a51955 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -1073,6 +1073,27 @@ long prune_icache_sb(struct super_block *sb, struct shrink_control *sc)
 	return freed;
 }
 
+/*
+ * Walk the superblock inode LRU for freeable inodes and attempt to free them.
+ * Inodes to be freed are moved to a temporary list and then are freed outside
+ * inode_lock by dispose_list(). This is used to free the icache via sysctl
+ * 'drop_fs_caches'.
+ */
+void reclaim_icache_sb(struct super_block *sb)
+{
+	unsigned long count = list_lru_count(&sb->s_inode_lru);
+
+	while (count > 0) {
+		LIST_HEAD(dispose);
+		unsigned long nr_to_walk = count >= 1024 ? 1024 : count;
+
+		count -= nr_to_walk;
+		list_lru_walk(&sb->s_inode_lru, inode_lru_isolate, &dispose,
+			      nr_to_walk);
+		dispose_list(&dispose);
+	}
+}
+
 static void __wait_on_freeing_inode(struct inode *inode, bool is_inode_hash_locked);
 /*
  * Called with the inode lock held.
diff --git a/fs/internal.h b/fs/internal.h
index 9b2b4d116880..8d3101232fb4 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -204,6 +204,7 @@ extern int vfs_open(const struct path *, struct file *);
  * inode.c
  */
 extern long prune_icache_sb(struct super_block *sb, struct shrink_control *sc);
+extern void reclaim_icache_sb(struct super_block *sb);
 int dentry_needs_remove_privs(struct mnt_idmap *, struct dentry *dentry);
 bool in_group_or_capable(struct mnt_idmap *idmap,
 			 const struct inode *inode, vfsgid_t vfsgid);
diff --git a/include/linux/dcache.h b/include/linux/dcache.h
index c83e02b94389..fed46f694f54 100644
--- a/include/linux/dcache.h
+++ b/include/linux/dcache.h
@@ -257,6 +257,7 @@ extern struct dentry *d_find_any_alias(struct inode *inode);
 extern struct dentry * d_obtain_alias(struct inode *);
 extern struct dentry * d_obtain_root(struct inode *);
 extern void shrink_dcache_sb(struct super_block *);
+extern void reclaim_dcache_sb(struct super_block *sb);
 extern void shrink_dcache_parent(struct dentry *);
 extern void d_invalidate(struct dentry *);
 
-- 
2.34.1

[PATCH v2 2/3] sysctl: add support for drop_caches for individual filesystem

[Ye Bin]

From: Ye Bin <yebin10@huawei.com>

In order to better analyze the issue of file system uninstallation caused
by kernel module opening files, it is necessary to perform dentry recycling
on a single file system. But now, apart from global dentry recycling, it is
not supported to do dentry recycling on a single file system separately.
This feature has usage scenarios in problem localization scenarios.At the
same time, it also provides users with a slightly fine-grained
pagecache/entry recycling mechanism.
This patch supports the recycling of pagecache/entry for individual file
systems.

Signed-off-by: Ye Bin <yebin10@huawei.com>
---
 fs/drop_caches.c | 127 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 127 insertions(+)

diff --git a/fs/drop_caches.c b/fs/drop_caches.c
index 49f56a598ecb..3c7e624129ec 100644
--- a/fs/drop_caches.c
+++ b/fs/drop_caches.c
@@ -11,6 +11,10 @@
 #include <linux/sysctl.h>
 #include <linux/gfp.h>
 #include <linux/swap.h>
+#include <linux/ptrace.h>
+#include <asm/syscall.h>
+#include <linux/task_work.h>
+#include <linux/namei.h>
 #include "internal.h"
 
 /* A global variable is a bit ugly, but it keeps the code simple */
@@ -78,6 +82,124 @@ static int drop_caches_sysctl_handler(const struct ctl_table *table, int write,
 	return 0;
 }
 
+struct drop_fs_caches_work {
+	struct callback_head task_work;
+	dev_t dev;
+	char *path;
+	unsigned int ctl;
+};
+
+static void drop_fs_caches(struct callback_head *twork)
+{
+	int ret;
+	struct super_block *sb;
+	static bool suppress;
+	struct drop_fs_caches_work *work = container_of(twork,
+			struct drop_fs_caches_work, task_work);
+	unsigned int ctl = work->ctl;
+	dev_t dev = work->dev;
+
+	if (work->path) {
+		struct path path;
+
+		ret = kern_path(work->path, LOOKUP_FOLLOW, &path);
+		if (ret) {
+			syscall_set_return_value(current,
+						 current_pt_regs(),
+						 0, ret);
+			goto out;
+		}
+		dev = path.dentry->d_sb->s_dev;
+		/* Make this file's dentry and inode recyclable */
+		path_put(&path);
+	}
+
+	sb = user_get_super(dev, false);
+	if (!sb) {
+		syscall_set_return_value(current, current_pt_regs(), 0,
+					 -EINVAL);
+		goto out;
+	}
+
+	if (ctl & BIT(0)) {
+		lru_add_drain_all();
+		drop_pagecache_sb(sb, NULL);
+		count_vm_event(DROP_PAGECACHE);
+	}
+
+	if (ctl & BIT(1)) {
+		reclaim_dcache_sb(sb);
+		reclaim_icache_sb(sb);
+		count_vm_event(DROP_SLAB);
+	}
+
+	if (!READ_ONCE(suppress)) {
+		pr_info("%s (%d): %s: %d %u:%u\n", current->comm,
+			task_pid_nr(current), __func__, ctl,
+			MAJOR(sb->s_dev), MINOR(sb->s_dev));
+
+		if (ctl & BIT(2))
+			WRITE_ONCE(suppress, true);
+	}
+
+	drop_super(sb);
+out:
+	kfree(work->path);
+	kfree(work);
+}
+
+static int drop_fs_caches_sysctl_handler(const struct ctl_table *table,
+					 int write, void *buffer,
+					 size_t *length, loff_t *ppos)
+{
+	struct drop_fs_caches_work *work = NULL;
+	unsigned int major, minor;
+	unsigned int ctl;
+	int ret;
+	char *path = NULL;
+
+	if (!write)
+		return 0;
+
+	if (sscanf(buffer, "%u %u:%u", &ctl, &major, &minor) != 3) {
+		path = kstrdup(buffer, GFP_NOFS);
+		if (!path) {
+			ret = -ENOMEM;
+			goto out;
+		}
+
+		if (sscanf(buffer, "%u %s", &ctl, path) != 2) {
+			ret = -EINVAL;
+			goto out;
+		}
+	}
+
+	if (ctl < 1 || ctl > 7) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	work = kzalloc(sizeof(*work), GFP_KERNEL);
+	if (!work) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	init_task_work(&work->task_work, drop_fs_caches);
+	if (!path)
+		work->dev = MKDEV(major, minor);
+	work->path = path;
+	work->ctl = ctl;
+	ret = task_work_add(current, &work->task_work, TWA_RESUME);
+out:
+	if (ret) {
+		kfree(path);
+		kfree(work);
+	}
+
+	return ret;
+}
+
 static const struct ctl_table drop_caches_table[] = {
 	{
 		.procname	= "drop_caches",
@@ -88,6 +210,11 @@ static const struct ctl_table drop_caches_table[] = {
 		.extra1		= SYSCTL_ONE,
 		.extra2		= SYSCTL_FOUR,
 	},
+	{
+		.procname	= "drop_fs_caches",
+		.mode		= 0200,
+		.proc_handler	= drop_fs_caches_sysctl_handler,
+	},
 };
 
 static int __init init_vm_drop_caches_sysctls(void)
-- 
2.34.1

[PATCH v2 3/3] Documentation: add instructions for using 'drop_fs_caches sysctl' sysctl

[Ye Bin]

From: Ye Bin <yebin10@huawei.com>

Add instructions for 'drop_fs_caches sysctl' sysctl in 'vm.rst'.

Signed-off-by: Ye Bin <yebin10@huawei.com>
---
 Documentation/admin-guide/sysctl/vm.rst | 34 +++++++++++++++++++++++++
 1 file changed, 34 insertions(+)

diff --git a/Documentation/admin-guide/sysctl/vm.rst b/Documentation/admin-guide/sysctl/vm.rst
index ace73480eb9d..c6c29c8cf92e 100644
--- a/Documentation/admin-guide/sysctl/vm.rst
+++ b/Documentation/admin-guide/sysctl/vm.rst
@@ -37,6 +37,7 @@ Currently, these files are in /proc/sys/vm:
 - dirtytime_expire_seconds
 - dirty_writeback_centisecs
 - drop_caches
+- drop_fs_caches
 - enable_soft_offline
 - extfrag_threshold
 - highmem_is_dirtyable
@@ -284,6 +285,39 @@ used::
 These are informational only.  They do not mean that anything is wrong
 with your system.  To disable them, echo 4 (bit 2) into drop_caches.
 
+drop_fs_caches
+==============
+
+Writing to this will cause the kernel to drop clean for a specific file system
+caches, as well as reclaimable slab objects like dentries and inodes. Once
+dropped, their memory becomes free. Except for specifying the device number
+or file path for a specific file system, everything else is consistent with
+drop_caches. The device number can be viewed through "cat /proc/self/montinfo"
+or 'lsblk'.
+
+To free pagecache::
+
+	echo "1 MAJOR:MINOR" > /proc/sys/vm/drop_fs_caches
+Or
+	echo "1 /mnt/XX" > /proc/sys/vm/drop_fs_caches
+
+To free reclaimable slab objects (includes dentries and inodes)::
+
+	echo "2 MAJOR:MINOR" > /proc/sys/vm/drop_fs_caches
+Or
+	echo "2 /mnt/XX" > /proc/sys/vm/drop_fs_caches
+
+To free slab objects and pagecache::
+
+	echo "3 MAJOR:MINOR" > /proc/sys/vm/drop_fs_caches
+Or
+	echo "3 /mnt/XX" > /proc/sys/vm/drop_fs_caches
+
+You may see informational messages in your kernel log when this file is
+used::
+
+	echo (1234): drop_fs_caches: 3 MAJOR:MINOR
+
 enable_soft_offline
 ===================
 Correctable memory errors are very common on servers. Soft-offline is kernel's
-- 
2.34.1

Re: [PATCH v2 2/3] sysctl: add support for drop_caches for individual filesystem

[kernel test robot]

Hi Ye,

kernel test robot noticed the following build errors:

[auto build test ERROR on viro-vfs/for-next]
[also build test ERROR on linus/master brauner-vfs/vfs.all v6.18-rc6 next-20251117]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Ye-Bin/vfs-introduce-reclaim_icache_sb-and-reclaim_dcache_sb-helper/20251117-193502
base:   https://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs.git for-next
patch link:    https://lore.kernel.org/r/20251117112735.4170831-3-yebin%40huaweicloud.com
patch subject: [PATCH v2 2/3] sysctl: add support for drop_caches for individual filesystem
config: alpha-allnoconfig (https://download.01.org/0day-ci/archive/20251117/202511172139.326qNHOk-lkp@intel.com/config)
compiler: alpha-linux-gcc (GCC) 15.1.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251117/202511172139.326qNHOk-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202511172139.326qNHOk-lkp@intel.com/

All errors (new ones prefixed by >>):

   fs/drop_caches.c: In function 'drop_fs_caches':
>> fs/drop_caches.c:107:25: error: implicit declaration of function 'syscall_set_return_value'; did you mean 'syscall_get_return_value'? [-Wimplicit-function-declaration]
     107 |                         syscall_set_return_value(current,
         |                         ^~~~~~~~~~~~~~~~~~~~~~~~
         |                         syscall_get_return_value


vim +107 fs/drop_caches.c

    91	
    92	static void drop_fs_caches(struct callback_head *twork)
    93	{
    94		int ret;
    95		struct super_block *sb;
    96		static bool suppress;
    97		struct drop_fs_caches_work *work = container_of(twork,
    98				struct drop_fs_caches_work, task_work);
    99		unsigned int ctl = work->ctl;
   100		dev_t dev = work->dev;
   101	
   102		if (work->path) {
   103			struct path path;
   104	
   105			ret = kern_path(work->path, LOOKUP_FOLLOW, &path);
   106			if (ret) {
 > 107				syscall_set_return_value(current,
   108							 current_pt_regs(),
   109							 0, ret);
   110				goto out;
   111			}
   112			dev = path.dentry->d_sb->s_dev;
   113			/* Make this file's dentry and inode recyclable */
   114			path_put(&path);
   115		}
   116	
   117		sb = user_get_super(dev, false);
   118		if (!sb) {
   119			syscall_set_return_value(current, current_pt_regs(), 0,
   120						 -EINVAL);
   121			goto out;
   122		}
   123	
   124		if (ctl & BIT(0)) {
   125			lru_add_drain_all();
   126			drop_pagecache_sb(sb, NULL);
   127			count_vm_event(DROP_PAGECACHE);
   128		}
   129	
   130		if (ctl & BIT(1)) {
   131			reclaim_dcache_sb(sb);
   132			reclaim_icache_sb(sb);
   133			count_vm_event(DROP_SLAB);
   134		}
   135	
   136		if (!READ_ONCE(suppress)) {
   137			pr_info("%s (%d): %s: %d %u:%u\n", current->comm,
   138				task_pid_nr(current), __func__, ctl,
   139				MAJOR(sb->s_dev), MINOR(sb->s_dev));
   140	
   141			if (ctl & BIT(2))
   142				WRITE_ONCE(suppress, true);
   143		}
   144	
   145		drop_super(sb);
   146	out:
   147		kfree(work->path);
   148		kfree(work);
   149	}
   150	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH v2 2/3] sysctl: add support for drop_caches for individual filesystem

[kernel test robot]

Hi Ye,

kernel test robot noticed the following build errors:

[auto build test ERROR on viro-vfs/for-next]
[also build test ERROR on linus/master brauner-vfs/vfs.all v6.18-rc6 next-20251117]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Ye-Bin/vfs-introduce-reclaim_icache_sb-and-reclaim_dcache_sb-helper/20251117-193502
base:   https://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs.git for-next
patch link:    https://lore.kernel.org/r/20251117112735.4170831-3-yebin%40huaweicloud.com
patch subject: [PATCH v2 2/3] sysctl: add support for drop_caches for individual filesystem
config: x86_64-allnoconfig (https://download.01.org/0day-ci/archive/20251117/202511172149.q8geOAvk-lkp@intel.com/config)
compiler: clang version 20.1.8 (https://github.com/llvm/llvm-project 87f0227cb60147a26a1eeb4fb06e3b505e9c7261)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251117/202511172149.q8geOAvk-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202511172149.q8geOAvk-lkp@intel.com/

All errors (new ones prefixed by >>):

>> fs/drop_caches.c:108:8: error: call to undeclared function 'task_stack_page'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
     108 |                                                  current_pt_regs(),
         |                                                  ^
   include/linux/ptrace.h:389:27: note: expanded from macro 'current_pt_regs'
     389 | #define current_pt_regs() task_pt_regs(current)
         |                           ^
   arch/x86/include/asm/processor.h:650:39: note: expanded from macro 'task_pt_regs'
     650 |         unsigned long __ptr = (unsigned long)task_stack_page(task);     \
         |                                              ^
   fs/drop_caches.c:119:37: error: call to undeclared function 'task_stack_page'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
     119 |                 syscall_set_return_value(current, current_pt_regs(), 0,
         |                                                   ^
   include/linux/ptrace.h:389:27: note: expanded from macro 'current_pt_regs'
     389 | #define current_pt_regs() task_pt_regs(current)
         |                           ^
   arch/x86/include/asm/processor.h:650:39: note: expanded from macro 'task_pt_regs'
     650 |         unsigned long __ptr = (unsigned long)task_stack_page(task);     \
         |                                              ^
   2 errors generated.


vim +/task_stack_page +108 fs/drop_caches.c

    91	
    92	static void drop_fs_caches(struct callback_head *twork)
    93	{
    94		int ret;
    95		struct super_block *sb;
    96		static bool suppress;
    97		struct drop_fs_caches_work *work = container_of(twork,
    98				struct drop_fs_caches_work, task_work);
    99		unsigned int ctl = work->ctl;
   100		dev_t dev = work->dev;
   101	
   102		if (work->path) {
   103			struct path path;
   104	
   105			ret = kern_path(work->path, LOOKUP_FOLLOW, &path);
   106			if (ret) {
   107				syscall_set_return_value(current,
 > 108							 current_pt_regs(),
   109							 0, ret);
   110				goto out;
   111			}
   112			dev = path.dentry->d_sb->s_dev;
   113			/* Make this file's dentry and inode recyclable */
   114			path_put(&path);
   115		}
   116	
   117		sb = user_get_super(dev, false);
   118		if (!sb) {
   119			syscall_set_return_value(current, current_pt_regs(), 0,
   120						 -EINVAL);
   121			goto out;
   122		}
   123	
   124		if (ctl & BIT(0)) {
   125			lru_add_drain_all();
   126			drop_pagecache_sb(sb, NULL);
   127			count_vm_event(DROP_PAGECACHE);
   128		}
   129	
   130		if (ctl & BIT(1)) {
   131			reclaim_dcache_sb(sb);
   132			reclaim_icache_sb(sb);
   133			count_vm_event(DROP_SLAB);
   134		}
   135	
   136		if (!READ_ONCE(suppress)) {
   137			pr_info("%s (%d): %s: %d %u:%u\n", current->comm,
   138				task_pid_nr(current), __func__, ctl,
   139				MAJOR(sb->s_dev), MINOR(sb->s_dev));
   140	
   141			if (ctl & BIT(2))
   142				WRITE_ONCE(suppress, true);
   143		}
   144	
   145		drop_super(sb);
   146	out:
   147		kfree(work->path);
   148		kfree(work);
   149	}
   150	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki