From: Jeff Layton <jlayton@redhat.com>
Subject: [RHEL5.1 PATCH] make static counters in new_inode and iunique be 32 bits (BZ 215356)
Date: Mon, 12 Feb 2007 09:50:16 -0500
Bugzilla: 215356
Message-Id: <200702121450.l1CEoGMX025675@dantu.rdu.redhat.com>
Changelog: [fs] make static counters in new_inode and iunique be 32 bits


When a 32-bit program that was not compiled with large file offsets does a
stat and gets a st_ino value back that won't fit in the 32 bit field, glibc
(correctly) generates an EOVERFLOW error. We can't do anything about fs's
with larger permanent inode numbers, but when we generate them on the fly,
we ought to try and have them fit within a 32 bit field.

This patch does this by making the static counters in these two functions be
32 bits. It's pretty much identical to the patches recently committed to RHEL3
and 4. This fixes the testcase in bz215356.

I had originally intended to make this part of a larger patchset. It would have
made sure that any callers of new_inode ensured that their i_ino values were
unique. Unfortunately, it's tough to do that without impacting performance,
and upstream has more or less decided not to worry about that unless and until
it becomes a problem.

-- Jeff

diff --git a/fs/inode.c b/fs/inode.c
index bf21dc6..23fc1fd 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -524,7 +524,8 @@ repeat:
  */
 struct inode *new_inode(struct super_block *sb)
 {
-	static unsigned long last_ino;
+	/* 32 bits for compatability mode stat calls */
+	static unsigned int last_ino;
 	struct inode * inode;
 
 	spin_lock_prefetch(&inode_lock);
@@ -683,7 +684,8 @@ static unsigned long hash(struct super_block *sb, unsigned long hashval)
  */
 ino_t iunique(struct super_block *sb, ino_t max_reserved)
 {
-	static ino_t counter;
+	/* 32 bits for compatability mode stat calls */
+	static unsigned int counter;
 	struct inode *inode;
 	struct hlist_head * head;
 	ino_t res;