} else
                        SetPageError(page);
 
-               unlock_page(page);
-
                if (rdata->result == 0 ||
                    (rdata->result == -EAGAIN && got_bytes))
                        cifs_readpage_to_fscache(rdata->mapping->host, page);
 
+               unlock_page(page);
+
                got_bytes -= min_t(unsigned int, PAGE_SIZE, got_bytes);
 
                put_page(page);
        struct cifs_sb_info *cifs_sb = CIFS_FILE_SB(ractl->file);
        struct TCP_Server_Info *server;
        pid_t pid;
-       unsigned int xid, last_batch_size = 0;
+       unsigned int xid, nr_pages, last_batch_size = 0, cache_nr_pages = 0;
+       pgoff_t next_cached = ULONG_MAX;
+       bool caching = fscache_cookie_enabled(cifs_inode_cookie(ractl->mapping->host)) &&
+               cifs_inode_cookie(ractl->mapping->host)->cache_priv;
+       bool check_cache = caching;
 
        xid = get_xid();
 
        /*
         * Chop the readahead request up into rsize-sized read requests.
         */
-       while (readahead_count(ractl) - last_batch_size) {
-               unsigned int i, nr_pages, got, rsize;
+       while ((nr_pages = readahead_count(ractl) - last_batch_size)) {
+               unsigned int i, got, rsize;
                struct page *page;
                struct cifs_readdata *rdata;
                struct cifs_credits credits_on_stack;
                struct cifs_credits *credits = &credits_on_stack;
+               pgoff_t index = readahead_index(ractl) + last_batch_size;
+
+               /*
+                * Find out if we have anything cached in the range of
+                * interest, and if so, where the next chunk of cached data is.
+                */
+               if (caching) {
+                       if (check_cache) {
+                               rc = cifs_fscache_query_occupancy(
+                                       ractl->mapping->host, index, nr_pages,
+                                       &next_cached, &cache_nr_pages);
+                               if (rc < 0)
+                                       caching = false;
+                               check_cache = false;
+                       }
+
+                       if (index == next_cached) {
+                               /*
+                                * TODO: Send a whole batch of pages to be read
+                                * by the cache.
+                                */
+                               page = readahead_page(ractl);
+
+                               if (cifs_readpage_from_fscache(ractl->mapping->host,
+                                                              page) < 0) {
+                                       /*
+                                        * TODO: Deal with cache read failure
+                                        * here, but for the moment, delegate
+                                        * that to readpage.
+                                        */
+                                       caching = false;
+                               }
+                               unlock_page(page);
+                               next_cached++;
+                               cache_nr_pages--;
+                               if (cache_nr_pages == 0)
+                                       check_cache = true;
+                               continue;
+                       }
+               }
 
                if (open_file->invalidHandle) {
                        rc = cifs_reopen_file(open_file, true);
                if (rc)
                        break;
                nr_pages = min_t(size_t, rsize / PAGE_SIZE, readahead_count(ractl));
+               nr_pages = min_t(size_t, nr_pages, next_cached - index);
 
                /*
                 * Give up immediately if rsize is too small to read an entire
 
        }
 }
 
+static inline void fscache_end_operation(struct netfs_cache_resources *cres)
+{
+       const struct netfs_cache_ops *ops = fscache_operation_valid(cres);
+
+       if (ops)
+               ops->end_operation(cres);
+}
+
 /*
- * Retrieve a page from FS-Cache
+ * Fallback page reading interface.
  */
-int __cifs_readpage_from_fscache(struct inode *inode, struct page *page)
+static int fscache_fallback_read_page(struct inode *inode, struct page *page)
 {
-       cifs_dbg(FYI, "%s: (fsc:%p, p:%p, i:0x%p\n",
-                __func__, CIFS_I(inode)->fscache, page, inode);
-       return -ENOBUFS; // Needs conversion to using netfslib
+       struct netfs_cache_resources cres;
+       struct fscache_cookie *cookie = cifs_inode_cookie(inode);
+       struct iov_iter iter;
+       struct bio_vec bvec[1];
+       int ret;
+
+       memset(&cres, 0, sizeof(cres));
+       bvec[0].bv_page         = page;
+       bvec[0].bv_offset       = 0;
+       bvec[0].bv_len          = PAGE_SIZE;
+       iov_iter_bvec(&iter, READ, bvec, ARRAY_SIZE(bvec), PAGE_SIZE);
+
+       ret = fscache_begin_read_operation(&cres, cookie);
+       if (ret < 0)
+               return ret;
+
+       ret = fscache_read(&cres, page_offset(page), &iter, NETFS_READ_HOLE_FAIL,
+                          NULL, NULL);
+       fscache_end_operation(&cres);
+       return ret;
 }
 
 /*
- * Retrieve a set of pages from FS-Cache
+ * Fallback page writing interface.
  */
-int __cifs_readpages_from_fscache(struct inode *inode,
-                               struct address_space *mapping,
-                               struct list_head *pages,
-                               unsigned *nr_pages)
+static int fscache_fallback_write_page(struct inode *inode, struct page *page,
+                                      bool no_space_allocated_yet)
 {
-       cifs_dbg(FYI, "%s: (0x%p/%u/0x%p)\n",
-                __func__, CIFS_I(inode)->fscache, *nr_pages, inode);
-       return -ENOBUFS; // Needs conversion to using netfslib
+       struct netfs_cache_resources cres;
+       struct fscache_cookie *cookie = cifs_inode_cookie(inode);
+       struct iov_iter iter;
+       struct bio_vec bvec[1];
+       loff_t start = page_offset(page);
+       size_t len = PAGE_SIZE;
+       int ret;
+
+       memset(&cres, 0, sizeof(cres));
+       bvec[0].bv_page         = page;
+       bvec[0].bv_offset       = 0;
+       bvec[0].bv_len          = PAGE_SIZE;
+       iov_iter_bvec(&iter, WRITE, bvec, ARRAY_SIZE(bvec), PAGE_SIZE);
+
+       ret = fscache_begin_write_operation(&cres, cookie);
+       if (ret < 0)
+               return ret;
+
+       ret = cres.ops->prepare_write(&cres, &start, &len, i_size_read(inode),
+                                     no_space_allocated_yet);
+       if (ret == 0)
+               ret = fscache_write(&cres, page_offset(page), &iter, NULL, NULL);
+       fscache_end_operation(&cres);
+       return ret;
 }
 
-void __cifs_readpage_to_fscache(struct inode *inode, struct page *page)
+/*
+ * Retrieve a page from FS-Cache
+ */
+int __cifs_readpage_from_fscache(struct inode *inode, struct page *page)
 {
-       struct cifsInodeInfo *cifsi = CIFS_I(inode);
+       int ret;
 
-       WARN_ON(!cifsi->fscache);
+       cifs_dbg(FYI, "%s: (fsc:%p, p:%p, i:0x%p\n",
+                __func__, cifs_inode_cookie(inode), page, inode);
 
+       ret = fscache_fallback_read_page(inode, page);
+       if (ret < 0)
+               return ret;
+
+       /* Read completed synchronously */
+       SetPageUptodate(page);
+       return 0;
+}
+
+void __cifs_readpage_to_fscache(struct inode *inode, struct page *page)
+{
        cifs_dbg(FYI, "%s: (fsc: %p, p: %p, i: %p)\n",
-                __func__, cifsi->fscache, page, inode);
+                __func__, cifs_inode_cookie(inode), page, inode);
+
+       fscache_fallback_write_page(inode, page, true);
+}
+
+/*
+ * Query the cache occupancy.
+ */
+int __cifs_fscache_query_occupancy(struct inode *inode,
+                                  pgoff_t first, unsigned int nr_pages,
+                                  pgoff_t *_data_first,
+                                  unsigned int *_data_nr_pages)
+{
+       struct netfs_cache_resources cres;
+       struct fscache_cookie *cookie = cifs_inode_cookie(inode);
+       loff_t start, data_start;
+       size_t len, data_len;
+       int ret;
 
-       // Needs conversion to using netfslib
+       ret = fscache_begin_read_operation(&cres, cookie);
+       if (ret < 0)
+               return ret;
+
+       start = first * PAGE_SIZE;
+       len = nr_pages * PAGE_SIZE;
+       ret = cres.ops->query_occupancy(&cres, start, len, PAGE_SIZE,
+                                       &data_start, &data_len);
+       if (ret == 0) {
+               *_data_first = data_start / PAGE_SIZE;
+               *_data_nr_pages = len / PAGE_SIZE;
+       }
+
+       fscache_end_operation(&cres);
+       return ret;
 }
 
 #ifndef _CIFS_FSCACHE_H
 #define _CIFS_FSCACHE_H
 
+#include <linux/swap.h>
 #include <linux/fscache.h>
 
 #include "cifsglob.h"
 }
 
 
-extern int cifs_fscache_release_page(struct page *page, gfp_t gfp);
-extern int __cifs_readpage_from_fscache(struct inode *, struct page *);
-extern int __cifs_readpages_from_fscache(struct inode *,
-                                        struct address_space *,
-                                        struct list_head *,
-                                        unsigned *);
-extern void __cifs_readpage_to_fscache(struct inode *, struct page *);
-
 static inline struct fscache_cookie *cifs_inode_cookie(struct inode *inode)
 {
        return CIFS_I(inode)->fscache;
                           i_size_read(inode), flags);
 }
 
-static inline int cifs_readpage_from_fscache(struct inode *inode,
-                                            struct page *page)
-{
-       if (CIFS_I(inode)->fscache)
-               return __cifs_readpage_from_fscache(inode, page);
+extern int __cifs_fscache_query_occupancy(struct inode *inode,
+                                         pgoff_t first, unsigned int nr_pages,
+                                         pgoff_t *_data_first,
+                                         unsigned int *_data_nr_pages);
 
-       return -ENOBUFS;
+static inline int cifs_fscache_query_occupancy(struct inode *inode,
+                                              pgoff_t first, unsigned int nr_pages,
+                                              pgoff_t *_data_first,
+                                              unsigned int *_data_nr_pages)
+{
+       if (!cifs_inode_cookie(inode))
+               return -ENOBUFS;
+       return __cifs_fscache_query_occupancy(inode, first, nr_pages,
+                                             _data_first, _data_nr_pages);
 }
 
-static inline int cifs_readpages_from_fscache(struct inode *inode,
-                                             struct address_space *mapping,
-                                             struct list_head *pages,
-                                             unsigned *nr_pages)
+extern int __cifs_readpage_from_fscache(struct inode *pinode, struct page *ppage);
+extern void __cifs_readpage_to_fscache(struct inode *pinode, struct page *ppage);
+
+
+static inline int cifs_readpage_from_fscache(struct inode *inode,
+                                            struct page *page)
 {
-       if (CIFS_I(inode)->fscache)
-               return __cifs_readpages_from_fscache(inode, mapping, pages,
-                                                    nr_pages);
+       if (cifs_inode_cookie(inode))
+               return __cifs_readpage_from_fscache(inode, page);
        return -ENOBUFS;
 }
 
 static inline void cifs_readpage_to_fscache(struct inode *inode,
                                            struct page *page)
 {
-       if (PageFsCache(page))
+       if (cifs_inode_cookie(inode))
                __cifs_readpage_to_fscache(inode, page);
 }
 
+static inline int cifs_fscache_release_page(struct page *page, gfp_t gfp)
+{
+       if (PageFsCache(page)) {
+               if (current_is_kswapd() || !(gfp & __GFP_FS))
+                       return false;
+               wait_on_page_fscache(page);
+               fscache_note_page_release(cifs_inode_cookie(page->mapping->host));
+       }
+       return true;
+}
+
 #else /* CONFIG_CIFS_FSCACHE */
 static inline
 void cifs_fscache_fill_coherency(struct inode *inode,
 static inline struct fscache_cookie *cifs_inode_cookie(struct inode *inode) { return NULL; }
 static inline void cifs_invalidate_cache(struct inode *inode, unsigned int flags) {}
 
-static inline int
-cifs_readpage_from_fscache(struct inode *inode, struct page *page)
+static inline int cifs_fscache_query_occupancy(struct inode *inode,
+                                              pgoff_t first, unsigned int nr_pages,
+                                              pgoff_t *_data_first,
+                                              unsigned int *_data_nr_pages)
 {
+       *_data_first = ULONG_MAX;
+       *_data_nr_pages = 0;
        return -ENOBUFS;
 }
 
-static inline int cifs_readpages_from_fscache(struct inode *inode,
-                                             struct address_space *mapping,
-                                             struct list_head *pages,
-                                             unsigned *nr_pages)
+static inline int
+cifs_readpage_from_fscache(struct inode *inode, struct page *page)
 {
        return -ENOBUFS;
 }
 
-static inline void cifs_readpage_to_fscache(struct inode *inode,
-                       struct page *page) {}
+static inline
+void cifs_readpage_to_fscache(struct inode *inode, struct page *page) {}
+
+static inline int nfs_fscache_release_page(struct page *page, gfp_t gfp)
+{
+       return true; /* May release page */
+}
 
 #endif /* CONFIG_CIFS_FSCACHE */