From: Josef Bacik <jbacik@redhat.com>
Date: Tue, 10 Feb 2009 13:49:34 -0500
Subject: [fs] fix symlink allocation context
Message-id: 1234291777-15344-22-git-send-email-jbacik@redhat.com
O-Subject: [PATCH 21/24] [RHEL 5.4] fs: fix symlink allocation context
Bugzilla: 445433
RH-Acked-by: Steven Whitehouse <swhiteho@redhat.com>
RH-Acked-by: Jeff Layton <jlayton@redhat.com>
RH-Acked-by: Jeff Layton <jlayton@redhat.com>

This is a backport of upstream commit

54566b2c1594c2326a645a3551f9d989f7ba3c5e

and is in reference to bz 445433.  This fixes a problem with the new aops where
page_symlink couldn't pass a GFP_NOFS type mask where the allocations occured,
which could cause a deadlock if we need to reclaim memory.

Signed-off-by: Josef Bacik <jbacik@redhat.com>

diff --git a/fs/buffer.c b/fs/buffer.c
index b3f5428..0ba5cce 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -2128,7 +2128,7 @@ int block_write_begin(struct file *file, struct address_space *mapping,
 	page = *pagep;
 	if (page == NULL) {
 		ownpage = 1;
-		page = __grab_cache_page(mapping, index);
+		page = grab_cache_page_write_begin(mapping, index, flags);
 		if (!page) {
 			status = -ENOMEM;
 			goto out;
diff --git a/fs/gfs2/ops_address.c b/fs/gfs2/ops_address.c
index 4fbfe1f..cb9156b 100644
--- a/fs/gfs2/ops_address.c
+++ b/fs/gfs2/ops_address.c
@@ -732,7 +732,7 @@ int gfs2_write_begin(struct file *file, struct address_space *mapping,
 		goto out_trans_fail;
 
 	error = -ENOMEM;
-	page = __grab_cache_page(mapping, index);
+	page = grab_cache_page_write_begin(mapping, index, flags);
 	*pagep = page;
 	if (unlikely(!page))
 		goto out_endtrans;
diff --git a/fs/libfs.c b/fs/libfs.c
index 87ef1dc..7c12fa4 100644
--- a/fs/libfs.c
+++ b/fs/libfs.c
@@ -359,7 +359,7 @@ int simple_write_begin(struct file *file, struct address_space *mapping,
 	index = pos >> PAGE_CACHE_SHIFT;
 	from = pos & (PAGE_CACHE_SIZE - 1);
 
-	page = __grab_cache_page(mapping, index);
+	page = grab_cache_page_write_begin(mapping, index, flags);
 	if (!page)
 		return -ENOMEM;
 
diff --git a/fs/namei.c b/fs/namei.c
index 33ffdd6..f9d5904 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -2686,18 +2686,23 @@ void page_put_link(struct dentry *dentry, struct nameidata *nd, void *cookie)
 	}
 }
 
-int __page_symlink(struct inode *inode, const char *symname, int len,
-		gfp_t gfp_mask)
+/*
+ * The nofs argument instructs pagecache_write_begin to pass AOP_FLAG_NOFS
+ */
+int __page_symlink(struct inode *inode, const char *symname, int len, int nofs)
 {
 	struct address_space *mapping = inode->i_mapping;
 	struct page *page;
 	void *fsdata;
 	int err;
 	char *kaddr;
+	unsigned int flags = AOP_FLAG_UNINTERRUPTIBLE;
+	if (nofs)
+		flags |= AOP_FLAG_NOFS;
 
 retry:
 	err = pagecache_write_begin(NULL, mapping, 0, len-1,
-				AOP_FLAG_UNINTERRUPTIBLE, &page, &fsdata);
+				flags, &page, &fsdata);
 	if (err)
 		goto fail;
 
@@ -2721,7 +2726,7 @@ fail:
 int page_symlink(struct inode *inode, const char *symname, int len)
 {
 	return __page_symlink(inode, symname, len,
-			mapping_gfp_mask(inode->i_mapping));
+			!(mapping_gfp_mask(inode->i_mapping) & __GFP_FS));
 }
 
 struct inode_operations page_symlink_inode_operations = {
diff --git a/include/linux/fs.h b/include/linux/fs.h
index e244844..5f5f524 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -399,6 +399,9 @@ enum positive_aop_returns {
 
 #define AOP_FLAG_UNINTERRUPTIBLE	0x0001 /* will not do a short write */
 #define AOP_FLAG_CONT_EXPAND		0x0002 /* called from cont_expand */
+#define AOP_FLAG_NOFS			0x0004 /* used by filesystem to direct
+						* helper code (eg buffer layer)
+						* to clear GFP_FS from alloc */
 
 /*
  * oh the beauties of C type declarations.
@@ -1956,7 +1959,7 @@ extern int page_readlink(struct dentry *, char __user *, int);
 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
 extern void page_put_link(struct dentry *, struct nameidata *, void *);
 extern int __page_symlink(struct inode *inode, const char *symname, int len,
-		gfp_t gfp_mask);
+		int nofs);
 extern int page_symlink(struct inode *inode, const char *symname, int len);
 extern struct inode_operations page_symlink_inode_operations;
 extern int generic_readlink(struct dentry *, char __user *, int);
diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
index f4a0aa5..78f2264 100644
--- a/include/linux/pagemap.h
+++ b/include/linux/pagemap.h
@@ -89,7 +89,8 @@ unsigned find_get_pages_contig(struct address_space *mapping, pgoff_t start,
 unsigned find_get_pages_tag(struct address_space *mapping, pgoff_t *index,
 			int tag, unsigned int nr_pages, struct page **pages);
 
-struct page *__grab_cache_page(struct address_space *mapping, pgoff_t index);
+struct page *grab_cache_page_write_begin(struct address_space *mapping,
+			pgoff_t index, unsigned flags);
 
 /*
  * Returns locked page at given index in given cache, creating it if needed.
diff --git a/mm/filemap.c b/mm/filemap.c
index c42c4da..15282de 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2141,7 +2141,7 @@ int pagecache_write_begin(struct file *file, struct address_space *mapping,
 		struct inode *inode = mapping->host;
 		struct page *page;
 again:
-		page = __grab_cache_page(mapping, index);
+		page = grab_cache_page_write_begin(mapping, index, flags);
 		*pagep = page;
 		if (!page)
 			return -ENOMEM;
@@ -2257,19 +2257,24 @@ EXPORT_SYMBOL(generic_file_direct_write);
  * Find or create a page at the given pagecache position. Return the locked
  * page. This function is specifically for buffered writes.
  */
-struct page *__grab_cache_page(struct address_space *mapping, pgoff_t index)
+struct page *grab_cache_page_write_begin(struct address_space *mapping,
+					pgoff_t index, unsigned flags)
 {
 	int status;
 	struct page *page;
+	gfp_t gfp_notmask = 0;
+	if (flags & AOP_FLAG_NOFS)
+		gfp_notmask = __GFP_FS;
 repeat:
 	page = find_lock_page(mapping, index);
 	if (likely(page))
 		return page;
 
-	page = page_cache_alloc(mapping);
+	page = __page_cache_alloc(mapping_gfp_mask(mapping) & ~gfp_notmask);
 	if (!page)
 		return NULL;
-	status = add_to_page_cache_lru(page, mapping, index, GFP_KERNEL);
+	status = add_to_page_cache_lru(page, mapping, index,
+						GFP_KERNEL & ~gfp_notmask);
 	if (unlikely(status)) {
 		page_cache_release(page);
 		if (status == -EEXIST)
@@ -2278,7 +2283,7 @@ repeat:
 	}
 	return page;
 }
-EXPORT_SYMBOL(__grab_cache_page);
+EXPORT_SYMBOL(grab_cache_page_write_begin);
 
 static ssize_t generic_perform_write_2copy(struct file *file,
 				struct iov_iter *i, loff_t pos)
@@ -2323,7 +2328,7 @@ static ssize_t generic_perform_write_2copy(struct file *file,
 			break;
 		}
 
-		page = __grab_cache_page(mapping, index);
+		page = grab_cache_page_write_begin(mapping, index, 0);
 		if (!page) {
 			status = -ENOMEM;
 			break;
@@ -2605,7 +2610,7 @@ generic_file_buffered_write_one_kernel_page(struct address_space *mapping,
 		       mapping->tree_lock.magic, RWLOCK_MAGIC);
 #endif
 
-	page = __grab_cache_page(mapping, index);
+	page = grab_cache_page_write_begin(mapping, index, 0);
 	if (!page)
 		return -ENOMEM;