ceph: improve readahead for file holes

When readahead encounters file holes, osd reply returns error -ENOENT,
finish_read() skips adding pages to the the page cache. So readahead
does not work for file holes. The fix is adding zero pages to the
page cache when -ENOENT is returned.

Signed-off-by: Yan, Zheng <zyan@redhat.com>

diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index 98933350331c47d362248a28fd539ad6bec507aa..6471e28e0586b443513fb5261d951d521464359b 100644 (file)
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -276,7 +276,7 @@ static void finish_read(struct ceph_osd_request *req, struct ceph_msg *msg)
        for (i = 0; i < num_pages; i++) {
                struct page *page = osd_data->pages[i];
 
-               if (rc < 0)
+               if (rc < 0 && rc != ENOENT)
                        goto unlock;
                if (bytes < (int)PAGE_CACHE_SIZE) {
                        /* zero (remainder of) page */