unfuck proc_sysctl ->d_compare()

a) struct inode is not going to be freed under ->d_compare();
however, the thing PROC_I(inode)->sysctl points to just might.
Fortunately, it's enough to make freeing that sucker delayed,
provided that we don't step on its ->unregistering, clear
the pointer to it in PROC_I(inode) before dropping the reference
and check if it's NULL in ->d_compare().

b) I'm not sure that we *can* walk into NULL inode here (we recheck
dentry->seq between verifying that it's still hashed / fetching
dentry->d_inode and passing it to ->d_compare() and there's no
negative hashed dentries in /proc/sys/*), but if we can walk into
that, we really should not have ->d_compare() return 0 on it!
Said that, I really suspect that this check can be simply killed.
Nick?

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

4 files changed, 31 insertions, 13 deletions

diff --git a/fs/proc/inode.c b/fs/proc/inode.c
index 176ce4cda68a..d6a7ca1fdac5 100644
--- a/fs/proc/inode.c
+++ b/fs/proc/inode.c
@@ -27,6 +27,7 @@
 static void proc_evict_inode(struct inode *inode)
 {
         struct proc_dir_entry *de;
+        struct ctl_table_header *head;
 
         truncate_inode_pages(&inode->i_data, 0);
         end_writeback(inode);
@@ -38,8 +39,11 @@ static void proc_evict_inode(struct inode *inode)
         de = PROC_I(inode)->pde;
         if (de)
                 pde_put(de);
-        if (PROC_I(inode)->sysctl)
-                sysctl_head_put(PROC_I(inode)->sysctl);
+        head = PROC_I(inode)->sysctl;
+        if (head) {
+                rcu_assign_pointer(PROC_I(inode)->sysctl, NULL);
+                sysctl_head_put(head);
+        }
 }
 
 struct vfsmount *proc_mnt;
diff --git a/fs/proc/proc_sysctl.c b/fs/proc/proc_sysctl.c
index 09a1f92a34ef..8eb2522111c5 100644
--- a/fs/proc/proc_sysctl.c
+++ b/fs/proc/proc_sysctl.c
@@ -408,15 +408,18 @@ static int proc_sys_compare(const struct dentry *parent,
                 const struct dentry *dentry, const struct inode *inode,
                 unsigned int len, const char *str, const struct qstr *name)
 {
+        struct ctl_table_header *head;
         /* Although proc doesn't have negative dentries, rcu-walk means
          * that inode here can be NULL */
+        /* AV: can it, indeed? */
         if (!inode)
-                return 0;
+                return 1;
         if (name->len != len)
                 return 1;
         if (memcmp(name->name, str, len))
                 return 1;
-        return !sysctl_is_seen(PROC_I(inode)->sysctl);
+        head = rcu_dereference(PROC_I(inode)->sysctl);
+        return !head || !sysctl_is_seen(head);
 }
 
 static const struct dentry_operations proc_sys_dentry_operations = {
diff --git a/include/linux/sysctl.h b/include/linux/sysctl.h
index 7bb5cb64f3b8..bb7c2b086fa4 100644
--- a/include/linux/sysctl.h
+++ b/include/linux/sysctl.h
@@ -25,6 +25,7 @@
 #include <linux/kernel.h>
 #include <linux/types.h>
 #include <linux/compiler.h>
+#include <linux/rcupdate.h>
 
 struct completion;
 
@@ -1037,10 +1038,15 @@ struct ctl_table_root {
    struct ctl_table trees. */
 struct ctl_table_header
 {
-        struct ctl_table *ctl_table;
-        struct list_head ctl_entry;
-        int used;
-        int count;
+        union {
+                struct {
+                        struct ctl_table *ctl_table;
+                        struct list_head ctl_entry;
+                        int used;
+                        int count;
+                };
+                struct rcu_head rcu;
+        };
         struct completion *unregistering;
         struct ctl_table *ctl_table_arg;
         struct ctl_table_root *root;
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 0f1bd83db985..4eed0af5d144 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -194,9 +194,9 @@ static int sysrq_sysctl_handler(ctl_table *table, int write,
 static struct ctl_table root_table[];
 static struct ctl_table_root sysctl_table_root;
 static struct ctl_table_header root_table_header = {
-        .count = 1,
+        {{.count = 1,
         .ctl_table = root_table,
-        .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
+        .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
         .root = &sysctl_table_root,
         .set = &sysctl_table_root.default_set,
 };
@@ -1567,11 +1567,16 @@ void sysctl_head_get(struct ctl_table_header *head)
         spin_unlock(&sysctl_lock);
 }
 
+static void free_head(struct rcu_head *rcu)
+{
+        kfree(container_of(rcu, struct ctl_table_header, rcu));
+}
+
 void sysctl_head_put(struct ctl_table_header *head)
 {
         spin_lock(&sysctl_lock);
         if (!--head->count)
-                kfree(head);
+                call_rcu(&head->rcu, free_head);
         spin_unlock(&sysctl_lock);
 }
 
@@ -1948,10 +1953,10 @@ void unregister_sysctl_table(struct ctl_table_header * header)
         start_unregistering(header);
         if (!--header->parent->count) {
                 WARN_ON(1);
-                kfree(header->parent);
+                call_rcu(&header->parent->rcu, free_head);
         }
         if (!--header->count)
-                kfree(header);
+                call_rcu(&header->rcu, free_head);
         spin_unlock(&sysctl_lock);
 }