releases/4.19.31/cifs-fix-read-after-write-for-files-with-read-caching.patch

   1 From 6dfbd84684700cb58b34e8602c01c12f3d2595c8 Mon Sep 17 00:00:00 2001
   2 From: Pavel Shilovsky <piastryyy@gmail.com>
   3 Date: Mon, 4 Mar 2019 17:48:01 -0800
   4 Subject: CIFS: Fix read after write for files with read caching
   5
   6 From: Pavel Shilovsky <piastryyy@gmail.com>
   7
   8 commit 6dfbd84684700cb58b34e8602c01c12f3d2595c8 upstream.
   9
  10 When we have a READ lease for a file and have just issued a write
  11 operation to the server we need to purge the cache and set oplock/lease
  12 level to NONE to avoid reading stale data. Currently we do that
  13 only if a write operation succedeed thus not covering cases when
  14 a request was sent to the server but a negative error code was
  15 returned later for some other reasons (e.g. -EIOCBQUEUED or -EINTR).
  16 Fix this by turning off caching regardless of the error code being
  17 returned.
  18
  19 The patches fixes generic tests 075 and 112 from the xfs-tests.
  20
  21 Cc: <stable@vger.kernel.org>
  22 Signed-off-by: Pavel Shilovsky <pshilov@microsoft.com>
  23 Signed-off-by: Steve French <stfrench@microsoft.com>
  24 Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
  25 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
  26
  27 ---
  28  fs/cifs/file.c |   12 +++++++-----
  29  1 file changed, 7 insertions(+), 5 deletions(-)
  30
  31 --- a/fs/cifs/file.c
  32 +++ b/fs/cifs/file.c
  33 @@ -2871,14 +2871,16 @@ cifs_strict_writev(struct kiocb *iocb, s
  34          * these pages but not on the region from pos to ppos+len-1.
  35          */
  36         written = cifs_user_writev(iocb, from);
  37 -       if (written > 0 && CIFS_CACHE_READ(cinode)) {
  38 +       if (CIFS_CACHE_READ(cinode)) {
  39                 /*
  40 -                * Windows 7 server can delay breaking level2 oplock if a write
  41 -                * request comes - break it on the client to prevent reading
  42 -                * an old data.
  43 +                * We have read level caching and we have just sent a write
  44 +                * request to the server thus making data in the cache stale.
  45 +                * Zap the cache and set oplock/lease level to NONE to avoid
  46 +                * reading stale data from the cache. All subsequent read
  47 +                * operations will read new data from the server.
  48                  */
  49                 cifs_zap_mapping(inode);
  50 -               cifs_dbg(FYI, "Set no oplock for inode=%p after a write operation\n",
  51 +               cifs_dbg(FYI, "Set Oplock/Lease to NONE for inode=%p after write\n",
  52                          inode);
  53                 cinode->oplock = 0;
  54         }