fs: fs_struct use seqlock

Use a seqlock in the fs_struct to enable us to take an atomic copy of the
complete cwd and root paths. Use this in the RCU lookup path to avoid a
thread-shared spinlock in RCU lookup operations.

Multi-threaded apps may now perform path lookups with scalability matching
multi-process apps. Operations such as stat(2) become very scalable for
multi-threaded workload.

Signed-off-by: Nick Piggin <npiggin@kernel.dk>

1 files changed, 10 insertions, 0 deletions

diff --git a/fs/fs_struct.c b/fs/fs_struct.c
index ed45a9cf5f3d..60b8531f41c5 100644
--- a/fs/fs_struct.c
+++ b/fs/fs_struct.c
@@ -14,9 +14,11 @@ void set_fs_root(struct fs_struct *fs, struct path *path)
         struct path old_root;
 
         spin_lock(&fs->lock);
+        write_seqcount_begin(&fs->seq);
         old_root = fs->root;
         fs->root = *path;
         path_get(path);
+        write_seqcount_end(&fs->seq);
         spin_unlock(&fs->lock);
         if (old_root.dentry)
                 path_put(&old_root);
@@ -31,9 +33,11 @@ void set_fs_pwd(struct fs_struct *fs, struct path *path)
         struct path old_pwd;
 
         spin_lock(&fs->lock);
+        write_seqcount_begin(&fs->seq);
         old_pwd = fs->pwd;
         fs->pwd = *path;
         path_get(path);
+        write_seqcount_end(&fs->seq);
         spin_unlock(&fs->lock);
 
         if (old_pwd.dentry)
@@ -52,6 +56,7 @@ void chroot_fs_refs(struct path *old_root, struct path *new_root)
                 fs = p->fs;
                 if (fs) {
                         spin_lock(&fs->lock);
+                        write_seqcount_begin(&fs->seq);
                         if (fs->root.dentry == old_root->dentry
                             && fs->root.mnt == old_root->mnt) {
                                 path_get(new_root);
@@ -64,6 +69,7 @@ void chroot_fs_refs(struct path *old_root, struct path *new_root)
                                 fs->pwd = *new_root;
                                 count++;
                         }
+                        write_seqcount_end(&fs->seq);
                         spin_unlock(&fs->lock);
                 }
                 task_unlock(p);
@@ -88,8 +94,10 @@ void exit_fs(struct task_struct *tsk)
                 int kill;
                 task_lock(tsk);
                 spin_lock(&fs->lock);
+                write_seqcount_begin(&fs->seq);
                 tsk->fs = NULL;
                 kill = !--fs->users;
+                write_seqcount_end(&fs->seq);
                 spin_unlock(&fs->lock);
                 task_unlock(tsk);
                 if (kill)
@@ -105,6 +113,7 @@ struct fs_struct *copy_fs_struct(struct fs_struct *old)
                 fs->users = 1;
                 fs->in_exec = 0;
                 spin_lock_init(&fs->lock);
+                seqcount_init(&fs->seq);
                 fs->umask = old->umask;
                 get_fs_root_and_pwd(old, &fs->root, &fs->pwd);
         }
@@ -144,6 +153,7 @@ EXPORT_SYMBOL(current_umask);
 struct fs_struct init_fs = {
         .users          = 1,
         .lock           = __SPIN_LOCK_UNLOCKED(init_fs.lock),
+        .seq            = SEQCNT_ZERO,
         .umask          = 0022,
 };