[RFC PATCH 12/25] reducing rcu_read_lock() scopes in dput and friends, step 1

[Al Viro <viro@zeniv.linux.org.uk>]

rcu_read_lock() scopes in dentry eviction machinery are too wide;
worse, quite a few of the function involved are not neutral wrt
that, making them harder to reason about.

rcu_read_lock() scope is not the only thing establishing an
RCU read-side critical area - spin_lock scope does the same and
they can be mixed - the sequence
	rcu_read_lock()
	...
	spin_lock()
	...
	rcu_read_unlock()
	...
	rcu_read_lock()
	...
	spun_unlock()
	...
	rcu_read_unlock()
is an unbroken RCU read-side critical area.

Use of that observation allows to shrink rcu_read_lock() scopes;
it's too delicate to do that in a single step, so we'll split that
into several commits.

This one breaks rcu_read_lock() scopes in front of shrink_kill() call
in shrink_dentry_list() and finish_dput() call in dput(); in both
places we are withing ->d_lock scope, so doing that won't change
the RCU read-side critical areas.

With that done, we have all calls of shrink_kill() and finish_dput()
immediately preceded by rcu_read_lock(); next step will use that.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/dcache.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/fs/dcache.c b/fs/dcache.c
index 7cf965f9fb29..d690c18f0c33 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -978,6 +978,8 @@ void dput(struct dentry *dentry)
 		rcu_read_unlock();
 		return;
 	}
+	rcu_read_unlock();
+	rcu_read_lock();
 	finish_dput(dentry);
 }
 EXPORT_SYMBOL(dput);
@@ -1682,6 +1684,8 @@ static void shrink_dcache_tree(struct dentry *parent, bool for_umount)
 				wait_for_completion(&wait.completion);
 				continue;
 			}
+			rcu_read_unlock();
+			rcu_read_lock();
 			shrink_kill(v);
 		}
 		if (!list_empty(&data.dispose))
-- 
2.47.3