From: Jeff Layton <jlayton@redhat.com>
Date: Thu, 13 Sep 2007 11:49:12 -0400
Subject: [nfs] nfs_symlink: allocate page with GFP_HIGHUSER
Message-id: 200709131549.l8DFnCMA032079@dantu.rdu.redhat.com
O-Subject: [RHEL5 PATCH] nfs_symlink should use GFP_HIGHUSER to allocate page (BZ#245042)
Bugzilla: 245042

This patch is intended for 5.2...

The patch for bug 218718 makes it so that nfs_symlink() allocates a page
destined for pagecache and uses that to hold the name to which the symlink
points. It allocates it using GFP_KERNEL but most pagecache pages are
allocated using GFP_HIGHUSER.

There's no reason to use normal zone memory for this, and a machine doing a
storm of NFS symlinks could potentially end up under normal zone memory
pressure. A patch to change the code to allocate the page out of highmem
recently went upstream. The following patch does the same for RHEL5.

diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 10b0f66..3783bd5 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1503,7 +1503,7 @@ static int nfs_symlink(struct inode *dir, struct dentry *dentry, const char *sym
 
 	lock_kernel();
 
-	page = alloc_page(GFP_KERNEL);
+	page = alloc_page(GFP_HIGHUSER);
 	if (!page) {
 		unlock_kernel();
 		return -ENOMEM;