[PATCH v2 0/3] Fixups for l_pid

[PATCH v2 0/3] Fixups for l_pid

[Benjamin Coddington]

LTP fcntl tests (fcntl11 fcntl14 fcntl17 fcntl19 fcntl20 fcntl21) have been
failing for NFSv4 mounts due to an unexpected l_pid.  What follows are some
fixups:

on v2:
	- Rebase onto linux-next
	- Revert back to using the stack in locks_mandatory_area(), and fixup
	patch description for 1/3
	- Add a comment to 3/3 explaining the seemingly-redundant fl_pid and
	fl_nspid

These three patches can be pulled from the branch named "fixups_for_l_pid"
here: git://bcodding.com/~bcodding/linux

Benjamin Coddington (3):
  fs/locks: Use allocation rather than the stack in fcntl_getlk()
  fs/locks: Set fl_nspid at file_lock allocation
  fs/locks: Use fs-specific l_pid for remote locks

 fs/locks.c         | 100 ++++++++++++++++++++++++++++++++++-------------------
 include/linux/fs.h |   7 ++++
 2 files changed, 72 insertions(+), 35 deletions(-)

-- 
2.9.3

[PATCH v2 1/3] fs/locks: Use allocation rather than the stack in fcntl_getlk()

[Benjamin Coddington]

Struct file_lock is fairly large, so let's save some space on the stack by
using an allocation for struct file_lock in fcntl_getlk(), just as we do
for fcntl_setlk().

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
---
 fs/locks.c | 46 ++++++++++++++++++++++++++--------------------
 1 file changed, 26 insertions(+), 20 deletions(-)

diff --git a/fs/locks.c b/fs/locks.c
index afefeb4ad6de..d7daa6c8932f 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -2086,14 +2086,17 @@ static void posix_lock_to_flock64(struct flock64 *flock, struct file_lock *fl)
  */
 int fcntl_getlk(struct file *filp, unsigned int cmd, struct flock *flock)
 {
-	struct file_lock file_lock;
+	struct file_lock *fl;
 	int error;
 
+	fl = locks_alloc_lock();
+	if (fl == NULL)
+		return -ENOMEM;
 	error = -EINVAL;
 	if (flock->l_type != F_RDLCK && flock->l_type != F_WRLCK)
 		goto out;
 
-	error = flock_to_posix_lock(filp, &file_lock, flock);
+	error = flock_to_posix_lock(filp, fl, flock);
 	if (error)
 		goto out;
 
@@ -2103,23 +2106,22 @@ int fcntl_getlk(struct file *filp, unsigned int cmd, struct flock *flock)
 			goto out;
 
 		cmd = F_GETLK;
-		file_lock.fl_flags |= FL_OFDLCK;
-		file_lock.fl_owner = filp;
+		fl->fl_flags |= FL_OFDLCK;
+		fl->fl_owner = filp;
 	}
 
-	error = vfs_test_lock(filp, &file_lock);
+	error = vfs_test_lock(filp, fl);
 	if (error)
 		goto out;
  
-	flock->l_type = file_lock.fl_type;
-	if (file_lock.fl_type != F_UNLCK) {
-		error = posix_lock_to_flock(flock, &file_lock);
+	flock->l_type = fl->fl_type;
+	if (fl->fl_type != F_UNLCK) {
+		error = posix_lock_to_flock(flock, fl);
 		if (error)
-			goto rel_priv;
+			goto out;
 	}
-rel_priv:
-	locks_release_private(&file_lock);
 out:
+	locks_free_lock(fl);
 	return error;
 }
 
@@ -2298,14 +2300,18 @@ int fcntl_setlk(unsigned int fd, struct file *filp, unsigned int cmd,
  */
 int fcntl_getlk64(struct file *filp, unsigned int cmd, struct flock64 *flock)
 {
-	struct file_lock file_lock;
+	struct file_lock *fl;
 	int error;
 
+	fl = locks_alloc_lock();
+	if (fl == NULL)
+		return -ENOMEM;
+
 	error = -EINVAL;
 	if (flock->l_type != F_RDLCK && flock->l_type != F_WRLCK)
 		goto out;
 
-	error = flock64_to_posix_lock(filp, &file_lock, flock);
+	error = flock64_to_posix_lock(filp, fl, flock);
 	if (error)
 		goto out;
 
@@ -2315,20 +2321,20 @@ int fcntl_getlk64(struct file *filp, unsigned int cmd, struct flock64 *flock)
 			goto out;
 
 		cmd = F_GETLK64;
-		file_lock.fl_flags |= FL_OFDLCK;
-		file_lock.fl_owner = filp;
+		fl->fl_flags |= FL_OFDLCK;
+		fl->fl_owner = filp;
 	}
 
-	error = vfs_test_lock(filp, &file_lock);
+	error = vfs_test_lock(filp, fl);
 	if (error)
 		goto out;
 
-	flock->l_type = file_lock.fl_type;
-	if (file_lock.fl_type != F_UNLCK)
-		posix_lock_to_flock64(flock, &file_lock);
+	flock->l_type = fl->fl_type;
+	if (fl->fl_type != F_UNLCK)
+		posix_lock_to_flock64(flock, fl);
 
-	locks_release_private(&file_lock);
 out:
+	locks_free_lock(fl);
 	return error;
 }
 
-- 
2.9.3

[PATCH v2 2/3] fs/locks: Set fl_nspid at file_lock allocation

[Benjamin Coddington]

Since commit c69899a17ca4 "NFSv4: Update of VFS byte range lock must be
atomic with the stateid update", NFSv4 has been inserting locks in rpciod
worker context.  The result is that the file_lock's fl_nspid is the
kworker's pid instead of the original userspace pid.  We can fix that up by
setting fl_nspid in locks_allocate_lock, and tranfer it to the file_lock
that's eventually recorded.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
---
 fs/locks.c | 29 ++++++++++++++++++++---------
 1 file changed, 20 insertions(+), 9 deletions(-)

diff --git a/fs/locks.c b/fs/locks.c
index d7daa6c8932f..0f5a461b8da6 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -249,7 +249,9 @@ locks_dump_ctx_list(struct list_head *list, char *list_type)
 	struct file_lock *fl;
 
 	list_for_each_entry(fl, list, fl_list) {
-		pr_warn("%s: fl_owner=%p fl_flags=0x%x fl_type=0x%x fl_pid=%u\n", list_type, fl->fl_owner, fl->fl_flags, fl->fl_type, fl->fl_pid);
+		pr_warn("%s: fl_owner=%p fl_flags=0x%x fl_type=0x%x fl_pid=%u fl_nspid=%u\n",
+			list_type, fl->fl_owner, fl->fl_flags, fl->fl_type,
+			fl->fl_pid, pid_vnr(fl->fl_nspid));
 	}
 }
 
@@ -294,8 +296,10 @@ struct file_lock *locks_alloc_lock(void)
 {
 	struct file_lock *fl = kmem_cache_zalloc(filelock_cache, GFP_KERNEL);
 
-	if (fl)
+	if (fl) {
 		locks_init_lock_heads(fl);
+		fl->fl_nspid = get_pid(task_tgid(current));
+	}
 
 	return fl;
 }
@@ -328,6 +332,8 @@ void locks_free_lock(struct file_lock *fl)
 	BUG_ON(!hlist_unhashed(&fl->fl_link));
 
 	locks_release_private(fl);
+	if (fl->fl_nspid)
+		put_pid(fl->fl_nspid);
 	kmem_cache_free(filelock_cache, fl);
 }
 EXPORT_SYMBOL(locks_free_lock);
@@ -357,8 +363,15 @@ EXPORT_SYMBOL(locks_init_lock);
  */
 void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
 {
+	struct pid *replace_pid = new->fl_nspid;
+
 	new->fl_owner = fl->fl_owner;
 	new->fl_pid = fl->fl_pid;
+	if (fl->fl_nspid) {
+		new->fl_nspid = get_pid(fl->fl_nspid);
+		if (replace_pid)
+			put_pid(replace_pid);
+	}
 	new->fl_file = NULL;
 	new->fl_flags = fl->fl_flags;
 	new->fl_type = fl->fl_type;
@@ -733,7 +746,6 @@ static void locks_wake_up_blocks(struct file_lock *blocker)
 static void
 locks_insert_lock_ctx(struct file_lock *fl, struct list_head *before)
 {
-	fl->fl_nspid = get_pid(task_tgid(current));
 	list_add_tail(&fl->fl_list, before);
 	locks_insert_global_locks(fl);
 }
@@ -743,10 +755,6 @@ locks_unlink_lock_ctx(struct file_lock *fl)
 {
 	locks_delete_global_locks(fl);
 	list_del_init(&fl->fl_list);
-	if (fl->fl_nspid) {
-		put_pid(fl->fl_nspid);
-		fl->fl_nspid = NULL;
-	}
 	locks_wake_up_blocks(fl);
 }
 
@@ -823,8 +831,6 @@ posix_test_lock(struct file *filp, struct file_lock *fl)
 	list_for_each_entry(cfl, &ctx->flc_posix, fl_list) {
 		if (posix_locks_conflict(fl, cfl)) {
 			locks_copy_conflock(fl, cfl);
-			if (cfl->fl_nspid)
-				fl->fl_pid = pid_vnr(cfl->fl_nspid);
 			goto out;
 		}
 	}
@@ -2452,6 +2458,7 @@ void locks_remove_posix(struct file *filp, fl_owner_t owner)
 	lock.fl_end = OFFSET_MAX;
 	lock.fl_owner = owner;
 	lock.fl_pid = current->tgid;
+	lock.fl_nspid = get_pid(task_tgid(current));
 	lock.fl_file = filp;
 	lock.fl_ops = NULL;
 	lock.fl_lmops = NULL;
@@ -2460,6 +2467,7 @@ void locks_remove_posix(struct file *filp, fl_owner_t owner)
 
 	if (lock.fl_ops && lock.fl_ops->fl_release_private)
 		lock.fl_ops->fl_release_private(&lock);
+	put_pid(lock.fl_nspid);
 	trace_locks_remove_posix(inode, &lock, error);
 }
 
@@ -2482,6 +2490,8 @@ locks_remove_flock(struct file *filp, struct file_lock_context *flctx)
 	if (list_empty(&flctx->flc_flock))
 		return;
 
+	fl.fl_nspid = get_pid(task_tgid(current));
+
 	if (filp->f_op->flock && is_remote_lock(filp))
 		filp->f_op->flock(filp, F_SETLKW, &fl);
 	else
@@ -2489,6 +2499,7 @@ locks_remove_flock(struct file *filp, struct file_lock_context *flctx)
 
 	if (fl.fl_ops && fl.fl_ops->fl_release_private)
 		fl.fl_ops->fl_release_private(&fl);
+	put_pid(fl.fl_nspid);
 }
 
 /* The i_flctx must be valid when calling into here */
-- 
2.9.3

[PATCH v2 3/3] fs/locks: Use fs-specific l_pid for remote locks

[Benjamin Coddington]

Now that we're setting fl_nspid at file_lock allocation, we need to handle
the case where a remote filesystem directly sets fl_pid.  If the filesystem
implements the lock operation, set a flag to return the lock's fl_pid,
rather than fl_nspid.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
---
 fs/locks.c         | 25 +++++++++++++++++++------
 include/linux/fs.h |  7 +++++++
 2 files changed, 26 insertions(+), 6 deletions(-)

diff --git a/fs/locks.c b/fs/locks.c
index 0f5a461b8da6..c7a222a8eaa7 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -2047,16 +2047,28 @@ SYSCALL_DEFINE2(flock, unsigned int, fd, unsigned int, cmd)
  */
 int vfs_test_lock(struct file *filp, struct file_lock *fl)
 {
-	if (filp->f_op->lock && is_remote_lock(filp))
+	if (filp->f_op->lock && is_remote_lock(filp)) {
+		fl->fl_flags |= FL_PID_PRIV;
 		return filp->f_op->lock(filp, F_GETLK, fl);
+	}
 	posix_test_lock(filp, fl);
 	return 0;
 }
 EXPORT_SYMBOL_GPL(vfs_test_lock);
 
+static inline int flock_translate_pid(struct file_lock *fl)
+{
+	if (IS_OFDLCK(fl))
+		return -1;
+	if (fl->fl_flags & FL_PID_PRIV)
+		return fl->fl_pid;
+	return pid_vnr(fl->fl_nspid);
+}
+
+
 static int posix_lock_to_flock(struct flock *flock, struct file_lock *fl)
 {
-	flock->l_pid = IS_OFDLCK(fl) ? -1 : fl->fl_pid;
+	flock->l_pid = flock_translate_pid(fl);
 #if BITS_PER_LONG == 32
 	/*
 	 * Make sure we can represent the posix lock via
@@ -2078,7 +2090,7 @@ static int posix_lock_to_flock(struct flock *flock, struct file_lock *fl)
 #if BITS_PER_LONG == 32
 static void posix_lock_to_flock64(struct flock64 *flock, struct file_lock *fl)
 {
-	flock->l_pid = IS_OFDLCK(fl) ? -1 : fl->fl_pid;
+	flock->l_pid = flock_translate_pid(fl);
 	flock->l_start = fl->fl_start;
 	flock->l_len = fl->fl_end == OFFSET_MAX ? 0 :
 		fl->fl_end - fl->fl_start + 1;
@@ -2596,7 +2608,9 @@ static void lock_get_status(struct seq_file *f, struct file_lock *fl,
 	struct inode *inode = NULL;
 	unsigned int fl_pid;
 
-	if (fl->fl_nspid) {
+	if (fl->fl_flags & FL_PID_PRIV) {
+		fl_pid = fl->fl_pid;
+	} else {
 		struct pid_namespace *proc_pidns = file_inode(f->file)->i_sb->s_fs_info;
 
 		/* Don't let fl_pid change based on who is reading the file */
@@ -2609,8 +2623,7 @@ static void lock_get_status(struct seq_file *f, struct file_lock *fl,
 		 */
 		if (fl_pid == 0)
 			return;
-	} else
-		fl_pid = fl->fl_pid;
+	}
 
 	if (fl->fl_file != NULL)
 		inode = locks_inode(fl->fl_file);
diff --git a/include/linux/fs.h b/include/linux/fs.h
index aa4affb38c39..11794cbaaed0 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -908,6 +908,7 @@ static inline struct file *get_file(struct file *f)
 #define FL_UNLOCK_PENDING	512 /* Lease is being broken */
 #define FL_OFDLCK	1024	/* lock is "owned" by struct file */
 #define FL_LAYOUT	2048	/* outstanding pNFS layout */
+#define FL_PID_PRIV	4096	/* F_GETLK should report fl_pid */
 
 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
 
@@ -973,6 +974,12 @@ int opens_in_grace(struct net *);
  * 3) lock range end
  *
  * Obviously, the last two criteria only matter for POSIX locks.
+ *
+ * fl_pid and fl_nspid appear redundant, but fl_pid is used to handle cases
+ * where a local F_GETLK returns a lock that was granted to a remote owner,
+ * as might exist with a remotely available filesystem like NFS.  For that
+ * case, the lock manager should set fl_pid, and that value is returned to a
+ * local F_GETLK.
  */
 struct file_lock {
 	struct file_lock *fl_next;	/* singly linked list for this inode  */
-- 
2.9.3

Re: [PATCH v2 0/3] Fixups for l_pid

[Jeff Layton]

On Tue, 2017-05-30 at 12:31 -0400, Benjamin Coddington wrote:
> LTP fcntl tests (fcntl11 fcntl14 fcntl17 fcntl19 fcntl20 fcntl21) have been
> failing for NFSv4 mounts due to an unexpected l_pid.  What follows are some
> fixups:
> 
> on v2:
> 	- Rebase onto linux-next
> 	- Revert back to using the stack in locks_mandatory_area(), and fixup
> 	patch description for 1/3
> 	- Add a comment to 3/3 explaining the seemingly-redundant fl_pid and
> 	fl_nspid
> 
> These three patches can be pulled from the branch named "fixups_for_l_pid"
> here: git://bcodding.com/~bcodding/linux
> 
> Benjamin Coddington (3):
>   fs/locks: Use allocation rather than the stack in fcntl_getlk()
>   fs/locks: Set fl_nspid at file_lock allocation
>   fs/locks: Use fs-specific l_pid for remote locks
> 
>  fs/locks.c         | 100 ++++++++++++++++++++++++++++++++++-------------------
>  include/linux/fs.h |   7 ++++
>  2 files changed, 72 insertions(+), 35 deletions(-)
> 

This all looks good to me. Nice work!

I've gone ahead and merged it into my linux-next branch, and it should
make v4.13 (barring any problems).

Thanks!
-- 
Jeff Layton <jlayton@redhat.com>