[PATCH] eCryptfs: zero out last page for llseek/write

[Michael Halcrow <mhalcrow@us.ibm.com>]

When one llseek's past the end of the file and then writes, every page
past the previous end of the file should be cleared. Trevor found that
the code, as is, does not assure that the very last page is always
cleared. This patch takes care of that.

This patch, including the two that I sent on June 13th, are bugfixes
to try to clean up the recent llseek() hack job; hopefully they can be
merged before the release of 2.6.22 (I've noticed that we're already
at -rc6).

Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com>

---
 fs/ecryptfs/mmap.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/fs/ecryptfs/mmap.c b/fs/ecryptfs/mmap.c
index 6df410c..7d5a43c 100644
--- a/fs/ecryptfs/mmap.c
+++ b/fs/ecryptfs/mmap.c
@@ -422,6 +422,8 @@ static int ecryptfs_prepare_write(struct file *file, struct page *page,
 				goto out;
 			}
 		}
+		if (end_of_prev_pg_pos + 1 > i_size_read(page->mapping->host))
+			zero_user_page(page, 0, PAGE_CACHE_SIZE, KM_USER0);
 	}
 out:
 	return rc;
-- 
1.4.4.4