SMB3: fix readpage for large swap cache

readpage was calculating the offset of the page incorrectly
for the case of large swapcaches.

    loff_t offset = (loff_t)page->index << PAGE_SHIFT;

As pointed out by Matthew Wilcox, this needs to use
page_file_offset() to calculate the offset instead.
Pages coming from the swap cache have page->index set
to their index within the swapcache, not within the backing
file.  For a sufficiently large swapcache, we could have
overlapping values of page->index within the same backing file.

Suggested by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: <stable@vger.kernel.org> # v5.7+
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>

diff --git a/fs/cifs/file.c b/fs/cifs/file.c
index cd108607a070678c52aa8e0e860a92dda4ed97ee..0a72840a88f1a014fcfe2db71734fc7a4f691fdc 100644 (file)
--- a/fs/cifs/file.c
+++ b/fs/cifs/file.c
@@ -4619,7 +4619,7 @@ read_complete:
 
 static int cifs_readpage(struct file *file, struct page *page)
 {
-       loff_t offset = (loff_t)page->index << PAGE_SHIFT;
+       loff_t offset = page_file_offset(page);
        int rc = -EACCES;
        unsigned int xid;