[PATCH V7 02/10] nsfs: switch to dedicated inode pool

From: Richard Guy Briggs
Date: Tue May 12 2015 - 16:03:34 EST


Instead of borrowing proc's inode pool and starting at a custom offset, give
nsfs its own dedicated inode pool without an offset.

Signed-off-by: Richard Guy Briggs <rgb@xxxxxxxxxx>
---
fs/nsfs.c | 44 ++++++++++++++++++++++++++++++++++++++++++++
fs/proc/internal.h | 2 ++
include/linux/proc_ns.h | 20 ++++++++------------
init/version.c | 2 +-
ipc/msgutil.c | 2 +-
kernel/pid.c | 2 +-
kernel/user.c | 2 +-
7 files changed, 58 insertions(+), 16 deletions(-)

diff --git a/fs/nsfs.c b/fs/nsfs.c
index af1b24f..1fcd529 100644
--- a/fs/nsfs.c
+++ b/fs/nsfs.c
@@ -159,3 +159,47 @@ void __init nsfs_init(void)
panic("can't set nsfs up\n");
nsfs_mnt->mnt_sb->s_flags &= ~MS_NOUSER;
}
+
+static DEFINE_IDA(ns_inum_ida);
+static DEFINE_SPINLOCK(ns_inum_lock); /* protects the above */
+
+/*
+ * Return an inode number between PROC_DYNAMIC_FIRST and
+ * 0xffffffff, or zero on failure.
+ */
+int ns_alloc_inum(struct ns_common *ns)
+{
+ unsigned int i;
+ int error;
+
+ atomic_long_set(&ns->stashed, 0);
+retry:
+ if (!ida_pre_get(&ns_inum_ida, GFP_KERNEL))
+ return -ENOMEM;
+
+ spin_lock_irq(&ns_inum_lock);
+ error = ida_get_new(&ns_inum_ida, &i);
+ spin_unlock_irq(&ns_inum_lock);
+ if (error == -EAGAIN)
+ goto retry;
+ else if (error)
+ return error;
+
+ if (i > UINT_MAX) {
+ spin_lock_irq(&ns_inum_lock);
+ ida_remove(&ns_inum_ida, i);
+ spin_unlock_irq(&ns_inum_lock);
+ return -ENOSPC;
+ }
+ ns->inum = NS_DYNAMIC_FIRST + i;
+ return 0;
+}
+
+void ns_free_inum(struct ns_common *ns)
+{
+ unsigned long flags;
+ spin_lock_irqsave(&ns_inum_lock, flags);
+ ida_remove(&ns_inum_ida, ns->inum - NS_DYNAMIC_FIRST);
+ spin_unlock_irqrestore(&ns_inum_lock, flags);
+}
+
diff --git a/fs/proc/internal.h b/fs/proc/internal.h
index c835b94..8f89cbb 100644
--- a/fs/proc/internal.h
+++ b/fs/proc/internal.h
@@ -18,6 +18,8 @@
struct ctl_table_header;
struct mempolicy;

+#define PROC_ROOT_INO 1
+
/*
* This is not completely implemented yet. The idea is to
* create an in-memory tree (like the actual /proc filesystem
diff --git a/include/linux/proc_ns.h b/include/linux/proc_ns.h
index 2654550..721d0e2 100644
--- a/include/linux/proc_ns.h
+++ b/include/linux/proc_ns.h
@@ -30,11 +30,12 @@ extern const struct proc_ns_operations *ns_entries[];
* We always define these enumerators
*/
enum {
- PROC_ROOT_INO = 1,
- PROC_IPC_INIT_INO = 0xEFFFFFFFU,
- PROC_UTS_INIT_INO = 0xEFFFFFFEU,
- PROC_USER_INIT_INO = 0xEFFFFFFDU,
- PROC_PID_INIT_INO = 0xEFFFFFFCU,
+ NS_NULL_INIT_INO,
+ NS_IPC_INIT_INO,
+ NS_UTS_INIT_INO,
+ NS_USER_INIT_INO,
+ NS_PID_INIT_INO,
+ NS_DYNAMIC_FIRST,
};

#ifdef CONFIG_PROC_FS
@@ -58,13 +59,8 @@ static inline void proc_free_inum(unsigned int inum) {}

#endif /* CONFIG_PROC_FS */

-static inline int ns_alloc_inum(struct ns_common *ns)
-{
- atomic_long_set(&ns->stashed, 0);
- return proc_alloc_inum(&ns->inum);
-}
-
-#define ns_free_inum(ns) proc_free_inum((ns)->inum)
+extern int ns_alloc_inum(struct ns_common *ns);
+extern void ns_free_inum(struct ns_common *ns);

extern struct file *proc_ns_fget(int fd);
#define get_proc_ns(inode) ((struct ns_common *)(inode)->i_private)
diff --git a/init/version.c b/init/version.c
index fe41a63..050dda0 100644
--- a/init/version.c
+++ b/init/version.c
@@ -35,7 +35,7 @@ struct uts_namespace init_uts_ns = {
.domainname = UTS_DOMAINNAME,
},
.user_ns = &init_user_ns,
- .ns.inum = PROC_UTS_INIT_INO,
+ .ns.inum = NS_UTS_INIT_INO,
#ifdef CONFIG_UTS_NS
.ns.ops = &utsns_operations,
#endif
diff --git a/ipc/msgutil.c b/ipc/msgutil.c
index 2b49159..aca24bb 100644
--- a/ipc/msgutil.c
+++ b/ipc/msgutil.c
@@ -31,7 +31,7 @@ DEFINE_SPINLOCK(mq_lock);
struct ipc_namespace init_ipc_ns = {
.count = ATOMIC_INIT(1),
.user_ns = &init_user_ns,
- .ns.inum = PROC_IPC_INIT_INO,
+ .ns.inum = NS_IPC_INIT_INO,
#ifdef CONFIG_IPC_NS
.ns.ops = &ipcns_operations,
#endif
diff --git a/kernel/pid.c b/kernel/pid.c
index cd36a5e..a06abc2 100644
--- a/kernel/pid.c
+++ b/kernel/pid.c
@@ -79,7 +79,7 @@ struct pid_namespace init_pid_ns = {
.level = 0,
.child_reaper = &init_task,
.user_ns = &init_user_ns,
- .ns.inum = PROC_PID_INIT_INO,
+ .ns.inum = NS_PID_INIT_INO,
#ifdef CONFIG_PID_NS
.ns.ops = &pidns_operations,
#endif
diff --git a/kernel/user.c b/kernel/user.c
index b069ccb..deaf107 100644
--- a/kernel/user.c
+++ b/kernel/user.c
@@ -50,7 +50,7 @@ struct user_namespace init_user_ns = {
.count = ATOMIC_INIT(3),
.owner = GLOBAL_ROOT_UID,
.group = GLOBAL_ROOT_GID,
- .ns.inum = PROC_USER_INIT_INO,
+ .ns.inum = NS_USER_INIT_INO,
#ifdef CONFIG_USER_NS
.ns.ops = &userns_operations,
#endif
--
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/