Sysctls are being moved out of kernel/sysctl.c and out to
their own respective subsystems / users to help with easier
maintance and avoid merge conflicts. But when we move just
one entry and to its own new file the last entry for this
new file must be empty, so we are essentialy bloating the
kernel one extra empty entry per each newly moved sysctl.
To help with this, this adds support for registering just
one ctl_table, therefore not bloating the kernel when we
move a single ctl_table to its own file.
Since the process of registering just one single table is the
same as that of registering an array table, so the code is
similar to registering an array table. The difference between
registering just one table and registering an array table is
that we no longer traversal through pointers when registering
a single table. These lead to that we have to add a complete
implementation process for register just one ctl_table, so we
have to add so much code.
Suggested-by: Matthew Wilcox <[email protected]>
Signed-off-by: Meng Tang <[email protected]>
---
fs/proc/proc_sysctl.c | 313 +++++++++++++++++++++++++++++++++++++++++
include/linux/sysctl.h | 6 +
2 files changed, 319 insertions(+)
diff --git a/fs/proc/proc_sysctl.c b/fs/proc/proc_sysctl.c
index 6c87c99f0856..5eb6ddf9dfd7 100644
--- a/fs/proc/proc_sysctl.c
+++ b/fs/proc/proc_sysctl.c
@@ -101,7 +101,9 @@ static void drop_sysctl_table(struct ctl_table_header *header);
static int sysctl_follow_link(struct ctl_table_header **phead,
struct ctl_table **pentry);
static int insert_links(struct ctl_table_header *head);
+static int insert_links_single(struct ctl_table_header *head);
static void put_links(struct ctl_table_header *header);
+static void put_links_single(struct ctl_table_header *header);
static void sysctl_print_dir(struct ctl_dir *dir)
{
@@ -198,6 +200,25 @@ static void erase_entry(struct ctl_table_header *head, struct ctl_table *entry)
rb_erase(node, &head->parent->root);
}
+static void init_header_single(struct ctl_table_header *head,
+ struct ctl_table_root *root, struct ctl_table_set *set,
+ struct ctl_node *node, struct ctl_table *table)
+{
+ head->ctl_table = table;
+ head->ctl_table_arg = table;
+ head->used = 0;
+ head->count = 1;
+ head->nreg = 1;
+ head->unregistering = NULL;
+ head->root = root;
+ head->set = set;
+ head->parent = NULL;
+ head->node = node;
+ INIT_HLIST_HEAD(&head->inodes);
+ if (node)
+ node->header = head;
+}
+
static void init_header(struct ctl_table_header *head,
struct ctl_table_root *root, struct ctl_table_set *set,
struct ctl_node *node, struct ctl_table *table)
@@ -227,6 +248,42 @@ static void erase_header(struct ctl_table_header *head)
erase_entry(head, entry);
}
+static int insert_header_single(struct ctl_dir *dir, struct ctl_table_header *header)
+{
+ int err;
+
+ /* Is this a permanently empty directory? */
+ if (is_empty_dir(&dir->header))
+ return -EROFS;
+
+ /* Am I creating a permanently empty directory? */
+ if (header->ctl_table == sysctl_mount_point) {
+ if (!RB_EMPTY_ROOT(&dir->root))
+ return -EINVAL;
+ set_empty_dir(dir);
+ }
+
+ dir->header.nreg++;
+ header->parent = dir;
+ err = insert_links_single(header);
+ if (err)
+ goto fail_links;
+
+ err = insert_entry(header, header->ctl_table);
+ if (err)
+ goto fail;
+
+ return 0;
+fail:
+ erase_entry(header, header->ctl_table);
+ put_links_single(header);
+fail_links:
+ if (header->ctl_table == sysctl_mount_point)
+ clear_empty_dir(dir);
+ header->parent = NULL;
+ drop_sysctl_table(&dir->header);
+ return err;
+}
static int insert_header(struct ctl_dir *dir, struct ctl_table_header *header)
{
struct ctl_table *entry;
@@ -1128,6 +1185,40 @@ static int sysctl_check_table_array(const char *path, struct ctl_table *table)
return err;
}
+static int sysctl_check_table_single(const char *path, struct ctl_table *table)
+{
+ int err = 0;
+
+ if (table->child)
+ err |= sysctl_err(path, table, "Not a file");
+
+ if ((table->proc_handler == proc_dostring) ||
+ (table->proc_handler == proc_dointvec) ||
+ (table->proc_handler == proc_douintvec) ||
+ (table->proc_handler == proc_douintvec_minmax) ||
+ (table->proc_handler == proc_dointvec_minmax) ||
+ (table->proc_handler == proc_dou8vec_minmax) ||
+ (table->proc_handler == proc_dointvec_jiffies) ||
+ (table->proc_handler == proc_dointvec_userhz_jiffies) ||
+ (table->proc_handler == proc_dointvec_ms_jiffies) ||
+ (table->proc_handler == proc_doulongvec_minmax) ||
+ (table->proc_handler == proc_doulongvec_ms_jiffies_minmax)) {
+ if (!table->data)
+ err |= sysctl_err(path, table, "No data");
+ if (!table->maxlen)
+ err |= sysctl_err(path, table, "No maxlen");
+ else
+ err |= sysctl_check_table_array(path, table);
+ }
+ if (!table->proc_handler)
+ err |= sysctl_err(path, table, "No proc_handler");
+
+ if ((table->mode & (S_IRUGO|S_IWUGO)) != table->mode)
+ err |= sysctl_err(path, table, "bogus .mode 0%o",
+ table->mode);
+ return err;
+}
+
static int sysctl_check_table(const char *path, struct ctl_table *table)
{
int err = 0;
@@ -1163,6 +1254,41 @@ static int sysctl_check_table(const char *path, struct ctl_table *table)
return err;
}
+static struct ctl_table_header *new_links_single(struct ctl_dir *dir, struct ctl_table *table,
+ struct ctl_table_root *link_root)
+{
+ struct ctl_table *link_table;
+ struct ctl_table_header *links;
+ struct ctl_node *node;
+ char *link_name;
+ int name_bytes = 0;
+
+ name_bytes += strlen(table->procname) + 1;
+
+ links = kzalloc(sizeof(struct ctl_table_header) +
+ sizeof(struct ctl_node) +
+ sizeof(struct ctl_table)*2 +
+ name_bytes,
+ GFP_KERNEL);
+
+ if (!links)
+ return NULL;
+
+ node = (struct ctl_node *)(links + 1);
+ link_table = (struct ctl_table *)(node + 1);
+ link_name = (char *)&link_table[2];
+
+ memcpy(link_name, table->procname, strlen(table->procname) + 1);
+ link_table->procname = link_name;
+ link_table->mode = S_IFLNK|S_IRWXUGO;
+ link_table->data = link_root;
+ link_name += strlen(table->procname) + 1;
+
+ init_header_single(links, dir->header.root, dir->header.set, node, link_table);
+ links->nreg = 1;
+
+ return links;
+}
static struct ctl_table_header *new_links(struct ctl_dir *dir, struct ctl_table *table,
struct ctl_table_root *link_root)
{
@@ -1206,6 +1332,27 @@ static struct ctl_table_header *new_links(struct ctl_dir *dir, struct ctl_table
return links;
}
+static bool get_links_single(struct ctl_dir *dir,
+ struct ctl_table *table, struct ctl_table_root *link_root)
+{
+ struct ctl_table_header *head;
+ struct ctl_table *link;
+
+ /* Is there link available for table? */
+ const char *procname = table->procname;
+
+ link = find_entry(&head, dir, procname, strlen(procname));
+ if (!link)
+ return false;
+ if ((S_ISDIR(link->mode) && S_ISDIR(table->mode)) ||
+ (S_ISLNK(link->mode) && (link->data == link_root))) {
+ head->nreg++;
+ return true;
+ }
+
+ return false;
+}
+
static bool get_links(struct ctl_dir *dir,
struct ctl_table *table, struct ctl_table_root *link_root)
{
@@ -1234,6 +1381,47 @@ static bool get_links(struct ctl_dir *dir,
return true;
}
+static int insert_links_single(struct ctl_table_header *head)
+{
+ struct ctl_table_set *root_set = &sysctl_table_root.default_set;
+ struct ctl_dir *core_parent = NULL;
+ struct ctl_table_header *links;
+ int err;
+
+ if (head->set == root_set)
+ return 0;
+
+ core_parent = xlate_dir(root_set, head->parent);
+ if (IS_ERR(core_parent))
+ return 0;
+
+ if (get_links_single(core_parent, head->ctl_table, head->root))
+ return 0;
+
+ core_parent->header.nreg++;
+ spin_unlock(&sysctl_lock);
+
+ links = new_links_single(core_parent, head->ctl_table, head->root);
+
+ spin_lock(&sysctl_lock);
+ err = -ENOMEM;
+ if (!links)
+ goto out;
+
+ err = 0;
+ if (get_links_single(core_parent, head->ctl_table, head->root)) {
+ kfree(links);
+ goto out;
+ }
+
+ err = insert_header_single(core_parent, links);
+ if (err)
+ kfree(links);
+out:
+ drop_sysctl_table(&core_parent->header);
+ return err;
+}
+
static int insert_links(struct ctl_table_header *head)
{
struct ctl_table_set *root_set = &sysctl_table_root.default_set;
@@ -1401,6 +1589,102 @@ struct ctl_table_header *register_sysctl(const char *path, struct ctl_table *tab
}
EXPORT_SYMBOL(register_sysctl);
+/**
+ * __register_sysctl_table_single - register a leaf sysctl table
+ * @set: Sysctl tree to register on
+ * @path: The path to the directory the sysctl table is in.
+ * @table: the top-level table structure
+ *
+ * Register extraly one sysctl table. @table should be a filled in extraly one
+ * ctl_table. If ctl_table which need to register have child or is not single,
+ * we should use register_sysctl_init or register_sysctl instead of
+ * register_sysctl_single.
+ *
+ * The members of the &struct ctl_table structure are used refer to
+ * __register_sysctl_table.
+ */
+struct ctl_table_header *__register_sysctl_table_single(
+ struct ctl_table_set *set,
+ const char *path, struct ctl_table *table)
+{
+ struct ctl_table_root *root = set->dir.header.root;
+ struct ctl_table_header *header;
+ const char *name, *nextname;
+ struct ctl_dir *dir;
+ struct ctl_node *node;
+
+ header = kzalloc(sizeof(struct ctl_table_header) +
+ sizeof(struct ctl_node), GFP_KERNEL);
+ if (!header)
+ return NULL;
+
+ node = (struct ctl_node *)(header + 1);
+ init_header_single(header, root, set, node, table);
+ if (sysctl_check_table_single(path, table))
+ goto fail;
+
+ spin_lock(&sysctl_lock);
+ dir = &set->dir;
+ /* Reference moved down the diretory tree get_subdir */
+ dir->header.nreg++;
+ spin_unlock(&sysctl_lock);
+
+ /* Find the directory for the ctl_table */
+ for (name = path; name; name = nextname) {
+ int namelen;
+
+ nextname = strchr(name, '/');
+ if (nextname) {
+ namelen = nextname - name;
+ nextname++;
+ } else {
+ namelen = strlen(name);
+ }
+ if (namelen == 0)
+ continue;
+
+ dir = get_subdir(dir, name, namelen);
+ if (IS_ERR(dir))
+ goto fail;
+ }
+
+ spin_lock(&sysctl_lock);
+ if (insert_header_single(dir, header))
+ goto fail_put_dir_locked;
+
+ drop_sysctl_table(&dir->header);
+ spin_unlock(&sysctl_lock);
+
+ return header;
+
+fail_put_dir_locked:
+ drop_sysctl_table(&dir->header);
+ spin_unlock(&sysctl_lock);
+fail:
+ kfree(header);
+ dump_stack();
+ return NULL;
+}
+
+/**
+ * __register_sysctl_single - register extraly one sysctl table
+ * @path: The path to the directory the sysctl table is in.
+ * @table: the table structure
+ *
+ * Register extraly one sysctl table. @table should be a filled in extraly one
+ * ctl_table. If ctl_table which need to register have child or is not single,
+ * we should use register_sysctl_init or register_sysctl instead of
+ * register_sysctl_single.
+ *
+ * See __register_sysctl_table_single for more details.
+ */
+struct ctl_table_header *__register_sysctl_single(const char *path, struct ctl_table *table)
+{
+ return __register_sysctl_table_single(&sysctl_table_root.default_set,
+ path, table);
+}
+EXPORT_SYMBOL(__register_sysctl_single);
+
/**
* __register_sysctl_init() - register sysctl table to path
* @path: path name for sysctl base
@@ -1655,6 +1939,35 @@ int __register_sysctl_base(struct ctl_table *base_table)
return 0;
}
+static void put_links_single(struct ctl_table_header *header)
+{
+ struct ctl_table_set *root_set = &sysctl_table_root.default_set;
+ struct ctl_table_root *root = header->root;
+ struct ctl_dir *parent = header->parent;
+ struct ctl_dir *core_parent;
+ struct ctl_table_header *link_head;
+ struct ctl_table *link;
+
+ if (header->set == root_set)
+ return;
+
+ core_parent = xlate_dir(root_set, parent);
+ if (IS_ERR(core_parent))
+ return;
+
+ link = find_entry(&link_head, core_parent, header->ctl_table->procname,
+ strlen(header->ctl_table->procname));
+ if (link &&
+ ((S_ISDIR(link->mode) && S_ISDIR(header->ctl_table->mode)) ||
+ (S_ISLNK(link->mode) && (link->data == root)))) {
+ drop_sysctl_table(link_head);
+ } else {
+ pr_err("sysctl link missing during unregister: ");
+ sysctl_print_dir(parent);
+ pr_cont("%s\n", header->ctl_table->procname);
+ }
+}
+
static void put_links(struct ctl_table_header *header)
{
struct ctl_table_set *root_set = &sysctl_table_root.default_set;
diff --git a/include/linux/sysctl.h b/include/linux/sysctl.h
index 644fd53ad5f1..2f1754aad68c 100644
--- a/include/linux/sysctl.h
+++ b/include/linux/sysctl.h
@@ -218,6 +218,9 @@ extern void setup_sysctl_set(struct ctl_table_set *p,
int (*is_seen)(struct ctl_table_set *));
extern void retire_sysctl_set(struct ctl_table_set *set);
+struct ctl_table_header *__register_sysctl_table_single(
+ struct ctl_table_set *set,
+ const char *path, struct ctl_table *table);
struct ctl_table_header *__register_sysctl_table(
struct ctl_table_set *set,
const char *path, struct ctl_table *table);
@@ -235,6 +238,9 @@ extern int sysctl_init_bases(void);
extern void __register_sysctl_init(const char *path, struct ctl_table *table,
const char *table_name);
#define register_sysctl_init(path, table) __register_sysctl_init(path, table, #table)
+extern struct ctl_table_header *__register_sysctl_single(const char *path,
+ struct ctl_table *table);
+#define register_sysctl_single(path, table) __register_sysctl_single(path, table)
extern struct ctl_table_header *register_sysctl_mount_point(const char *path);
void do_sysctl_args(void);
--
2.20.1
Sysctls are being moved out of kernel/sysctl.c and out to
their own respective subsystems / users to help with easier
maintance and avoid merge conflicts. But when we move just
one entry and to its own new file the last entry for this
new file must be empty, so we are essentialy bloating the
kernel one extra empty entry per each newly moved sysctl.
To help with this, I have added support for registering just
one ctl_table, therefore not bloating the kernel when we
move a single ctl_table to its own file.
The optimization has been implemented in the previous patch,
here use register_sysctl_single() to register single one
ctl_table.
In this modification, I counted the size changes of each
object file during the compilation process.
When there is no strip, size changes are as follows:
before now save space
fs/dcache.o 904936 904760 176bytes
fs/exec.o 883584 883440 144bytes
fs/namespace.o 1614776 1614616 160bytes
fs/notify/dnotify/dnotify.o 255992 255872 120bytes
init/do_mounts_initrd.o 296552 296392 160bytes
kernel/acct.o 459184 459032 152bytes
kernel/delayacct.o 208680 208536 144bytes
kernel/kprobes.o 794968 794936 32bytes
kernel/panic.o 367696 367560 136bytes
When there is exec with 'strip -d', size changes are as follows:
before now save space
fs/dcache.o 79040 78952 88bytes
fs/exec.o 57960 57864 96bytes
fs/namespace.o 111904 111824 80bytes
fs/notify/dnotify/dnotify.o 8816 8736 80bytes
init/do_mounts_initrd.o 4872 4760 112bytes
kernel/acct.o 18104 18000 104bytes
kernel/delayacct.o 8768 8664 104bytes
kernel/kprobes.o 63192 63104 88bytes
kernel/panic.o 26760 26672 88bytes
Suggested-by: Matthew Wilcox <[email protected]>
Signed-off-by: Meng Tang <[email protected]>
---
fs/dcache.c | 5 ++---
fs/exec.c | 5 ++---
fs/namespace.c | 5 ++---
fs/notify/dnotify/dnotify.c | 5 ++---
init/do_mounts_initrd.c | 5 ++---
kernel/acct.c | 5 ++---
kernel/delayacct.c | 5 ++---
kernel/kprobes.c | 5 ++---
kernel/panic.c | 5 ++---
9 files changed, 18 insertions(+), 27 deletions(-)
diff --git a/fs/dcache.c b/fs/dcache.c
index c84269c6e8bf..29fed2df79d1 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -190,13 +190,12 @@ static struct ctl_table fs_dcache_sysctls[] = {
.maxlen = 6*sizeof(long),
.mode = 0444,
.proc_handler = proc_nr_dentry,
- },
- { }
+ }
};
static int __init init_fs_dcache_sysctls(void)
{
- register_sysctl_init("fs", fs_dcache_sysctls);
+ register_sysctl_single("fs", fs_dcache_sysctls);
return 0;
}
fs_initcall(init_fs_dcache_sysctls);
diff --git a/fs/exec.c b/fs/exec.c
index c2586b791b87..58e9e50b9d98 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -2140,13 +2140,12 @@ static struct ctl_table fs_exec_sysctls[] = {
.proc_handler = proc_dointvec_minmax_coredump,
.extra1 = SYSCTL_ZERO,
.extra2 = SYSCTL_TWO,
- },
- { }
+ }
};
static int __init init_fs_exec_sysctls(void)
{
- register_sysctl_init("fs", fs_exec_sysctls);
+ register_sysctl_single("fs", fs_exec_sysctls);
return 0;
}
diff --git a/fs/namespace.c b/fs/namespace.c
index df172818e1f8..1384fa7f8c79 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -4673,13 +4673,12 @@ static struct ctl_table fs_namespace_sysctls[] = {
.mode = 0644,
.proc_handler = proc_dointvec_minmax,
.extra1 = SYSCTL_ONE,
- },
- { }
+ }
};
static int __init init_fs_namespace_sysctls(void)
{
- register_sysctl_init("fs", fs_namespace_sysctls);
+ register_sysctl_single("fs", fs_namespace_sysctls);
return 0;
}
fs_initcall(init_fs_namespace_sysctls);
diff --git a/fs/notify/dnotify/dnotify.c b/fs/notify/dnotify/dnotify.c
index 829dd4a61b66..813a22825be5 100644
--- a/fs/notify/dnotify/dnotify.c
+++ b/fs/notify/dnotify/dnotify.c
@@ -28,12 +28,11 @@ static struct ctl_table dnotify_sysctls[] = {
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
- },
- {}
+ }
};
static void __init dnotify_sysctl_init(void)
{
- register_sysctl_init("fs", dnotify_sysctls);
+ register_sysctl_single("fs", dnotify_sysctls);
}
#else
#define dnotify_sysctl_init() do { } while (0)
diff --git a/init/do_mounts_initrd.c b/init/do_mounts_initrd.c
index 327962ea354c..d37f24959aa3 100644
--- a/init/do_mounts_initrd.c
+++ b/init/do_mounts_initrd.c
@@ -28,13 +28,12 @@ static struct ctl_table kern_do_mounts_initrd_table[] = {
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
- },
- { }
+ }
};
static __init int kernel_do_mounts_initrd_sysctls_init(void)
{
- register_sysctl_init("kernel", kern_do_mounts_initrd_table);
+ register_sysctl_single("kernel", kern_do_mounts_initrd_table);
return 0;
}
late_initcall(kernel_do_mounts_initrd_sysctls_init);
diff --git a/kernel/acct.c b/kernel/acct.c
index 62200d799b9b..c628808c9213 100644
--- a/kernel/acct.c
+++ b/kernel/acct.c
@@ -83,13 +83,12 @@ static struct ctl_table kern_acct_table[] = {
.maxlen = 3*sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec,
- },
- { }
+ }
};
static __init int kernel_acct_sysctls_init(void)
{
- register_sysctl_init("kernel", kern_acct_table);
+ register_sysctl_single("kernel", kern_acct_table);
return 0;
}
late_initcall(kernel_acct_sysctls_init);
diff --git a/kernel/delayacct.c b/kernel/delayacct.c
index 2c1e18f7c5cf..6b776cbcb559 100644
--- a/kernel/delayacct.c
+++ b/kernel/delayacct.c
@@ -73,13 +73,12 @@ static struct ctl_table kern_delayacct_table[] = {
.proc_handler = sysctl_delayacct,
.extra1 = SYSCTL_ZERO,
.extra2 = SYSCTL_ONE,
- },
- { }
+ }
};
static __init int kernel_delayacct_sysctls_init(void)
{
- register_sysctl_init("kernel", kern_delayacct_table);
+ register_sysctl_single("kernel", kern_delayacct_table);
return 0;
}
late_initcall(kernel_delayacct_sysctls_init);
diff --git a/kernel/kprobes.c b/kernel/kprobes.c
index 94cab8c9ce56..1cf54662e2ed 100644
--- a/kernel/kprobes.c
+++ b/kernel/kprobes.c
@@ -970,13 +970,12 @@ static struct ctl_table kprobe_sysctls[] = {
.proc_handler = proc_kprobes_optimization_handler,
.extra1 = SYSCTL_ZERO,
.extra2 = SYSCTL_ONE,
- },
- {}
+ }
};
static void __init kprobe_sysctls_init(void)
{
- register_sysctl_init("debug", kprobe_sysctls);
+ register_sysctl_single("debug", kprobe_sysctls);
}
#endif /* CONFIG_SYSCTL */
diff --git a/kernel/panic.c b/kernel/panic.c
index ae5c0ca86016..90f1a0f25139 100644
--- a/kernel/panic.c
+++ b/kernel/panic.c
@@ -85,13 +85,12 @@ static struct ctl_table kern_panic_table[] = {
.proc_handler = proc_dointvec_minmax,
.extra1 = SYSCTL_ZERO,
.extra2 = SYSCTL_ONE,
- },
- { }
+ }
};
static __init int kernel_panic_sysctls_init(void)
{
- register_sysctl_init("kernel", kern_panic_table);
+ register_sysctl_single("kernel", kern_panic_table);
return 0;
}
late_initcall(kernel_panic_sysctls_init);
--
2.20.1
On Wed, Mar 02, 2022 at 10:55:10AM +0800, Meng Tang wrote:
> Sysctls are being moved out of kernel/sysctl.c and out to
> their own respective subsystems / users to help with easier
> maintance and avoid merge conflicts. But when we move just
> one entry and to its own new file the last entry for this
> new file must be empty, so we are essentialy bloating the
> kernel one extra empty entry per each newly moved sysctl.
>
> To help with this, this adds support for registering just
> one ctl_table, therefore not bloating the kernel when we
> move a single ctl_table to its own file.
>
> Since the process of registering just one single table is the
> same as that of registering an array table, so the code is
> similar to registering an array table. The difference between
> registering just one table and registering an array table is
> that we no longer traversal through pointers when registering
> a single table. These lead to that we have to add a complete
> implementation process for register just one ctl_table, so we
> have to add so much code.
>
> Suggested-by: Matthew Wilcox <[email protected]>
> Signed-off-by: Meng Tang <[email protected]>
Is there really no helpers which you can add to share code?
Now that you did this work, look carefully.
Luis