rpc: warn and truncate grouplist if RPC/AUTH can not hold everything

The GlusterFS protocol currently uses AUTH_GLUSTERFS_V2 in the RPC/AUTH
header. This header contains the uid, gid and auxiliary groups of the
user/process that accesses the Gluster Volume.

The AUTH_GLUSTERFS_V2 structure allows up to 65535 auxiliary groups to
be passed on. Unfortunately, the RPC/AUTH header is limited to 400 bytes
by the RPC specification: http://tools.ietf.org/html/rfc5531#section-8.2

In order to not cause complete failures on the client-side when trying
to encode a AUTH_GLUSTERFS_V2 that would result in more than 400 bytes,
we can calculate the expected size of the other elements:

    1 | pid
    1 | uid
    1 | gid
    1 | groups_len
   XX | groups_val (GF_MAX_AUX_GROUPS=65535)
    1 | lk_owner_len
   YY | lk_owner_val (GF_MAX_LOCK_OWNER_LEN=1024)
  ----+-------------------------------------------
    5 | total xdr-units

  one XDR-unit is defined as BYTES_PER_XDR_UNIT = 4 bytes
  MAX_AUTH_BYTES = 400 is the maximum, this is 100 xdr-units.
  XX + YY can be 95 to fill the 100 xdr-units.

  Note that the on-wire protocol has tighter requirements than the
  internal structures. It is possible for xlators to use more groups and
  a bigger lk_owner than that can be sent by a GlusterFS-client.

This change prevents overflows when allocating the RPC/AUTH header. Two
new macros are introduced to calculate the number of groups that fit in
the RPC/AUTH header, when taking the size of the lk_owner in account. In
case the list of groups exceeds the maximum possible, only the first
groups are passed over the RPC/GlusterFS protocol to the bricks.
A warning is added to the logs, so that most system administrators will
get informed.

The reducing of the number of groups is not a new inventions. The
RPC/AUTH header (AUTH_SYS or AUTH_UNIX) that NFS uses has a limit of 16
groups. Most, if not all, NFS-clients will reduce any bigger number of
groups to 16. (nfs.server-aux-gids can be used to workaround the limit
of 16 groups, but the Gluster NFS-server will be limited to a maximum of
93 groups, or fewer in case the lk_owner structure contains more items.)

Change-Id: I8410e59d0fd246d601b54b961d3ae9cb5a858c10
BUG: 1053579
Signed-off-by: Niels de Vos <ndevos@redhat.com>
Reviewed-on: http://review.gluster.org/7202
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Harshavardhana <harsha@harshavardhana.net>
Reviewed-by: Santosh Pradhan <spradhan@redhat.com>
Reviewed-by: Vijay Bellur <vbellur@redhat.com>

3 files changed, 87 insertions, 8 deletions

diff --git a/rpc/rpc-lib/src/auth-glusterfs.c b/rpc/rpc-lib/src/auth-glusterfs.c
index 7bafa82fb82..c3fc166b731 100644
--- a/rpc/rpc-lib/src/auth-glusterfs.c
+++ b/rpc/rpc-lib/src/auth-glusterfs.c
@@ -171,8 +171,10 @@ auth_glusterfs_v2_request_init (rpcsvc_request_t *req, void *priv)
 int auth_glusterfs_v2_authenticate (rpcsvc_request_t *req, void *priv)
 {
         struct auth_glusterfs_parms_v2  au = {0,};
-        int ret = RPCSVC_AUTH_REJECT;
-        int i   = 0;
+        int ret                            = RPCSVC_AUTH_REJECT;
+        int i                              = 0;
+        int max_groups                     = 0;
+        int max_lk_owner_len               = 0;
 
         if (!req)
                 return ret;
@@ -191,17 +193,23 @@ int auth_glusterfs_v2_authenticate (rpcsvc_request_t *req, void *priv)
         req->lk_owner.len = au.lk_owner.lk_owner_len;
         req->auxgidcount = au.groups.groups_len;
 
-        if (req->auxgidcount > GF_MAX_AUX_GROUPS) {
+        /* the number of groups and size of lk_owner depend on each other */
+        max_groups = GF_AUTH_GLUSTERFS_MAX_GROUPS (req->lk_owner.len);
+        max_lk_owner_len = GF_AUTH_GLUSTERFS_MAX_LKOWNER (req->auxgidcount);
+
+        if (req->auxgidcount > max_groups) {
                 gf_log ("", GF_LOG_WARNING,
                         "more than max aux gids found (%d) , truncating it "
                         "to %d and continuing", au.groups.groups_len,
-                        GF_MAX_AUX_GROUPS);
-                req->auxgidcount = GF_MAX_AUX_GROUPS;
+                        max_groups);
+                req->auxgidcount = max_groups;
         }
 
-        if (req->lk_owner.len > GF_MAX_LOCK_OWNER_LEN) {
+        if (req->lk_owner.len > max_lk_owner_len) {
                 gf_log ("", GF_LOG_WARNING,
-                        "lkowner field > 1k, failing authentication");
+                        "lkowner field to big (%d), depends on the number of "
+                        "groups (%d), failing authentication",
+                        req->lk_owner.len, req->auxgidcount);
                 ret = RPCSVC_AUTH_REJECT;
                 goto err;
         }
diff --git a/rpc/rpc-lib/src/rpc-clnt.c b/rpc/rpc-lib/src/rpc-clnt.c
index 22513b789d6..e095c55b3d9 100644
--- a/rpc/rpc-lib/src/rpc-clnt.c
+++ b/rpc/rpc-lib/src/rpc-clnt.c
@@ -1133,17 +1133,34 @@ rpc_clnt_register_notify (struct rpc_clnt *rpc, rpc_clnt_notify_t fn,
         return 0;
 }
 
+/* used for GF_LOG_OCCASIONALLY() */
+static int gf_auth_max_groups_log = 0;
+
 ssize_t
 xdr_serialize_glusterfs_auth (char *dest, struct auth_glusterfs_parms_v2 *au)
 {
         ssize_t ret = -1;
         XDR     xdr;
+        uint64_t ngroups = 0;
+        int     max_groups = 0;
 
         if ((!dest) || (!au))
                 return -1;
 
+        max_groups = GF_AUTH_GLUSTERFS_MAX_GROUPS (au->lk_owner.lk_owner_len);
+
         xdrmem_create (&xdr, dest, GF_MAX_AUTH_BYTES, XDR_ENCODE);
 
+        if (au->groups.groups_len > max_groups) {
+                ngroups = au->groups.groups_len;
+                au->groups.groups_len = max_groups;
+
+                GF_LOG_OCCASIONALLY (gf_auth_max_groups_log,
+                                     THIS->name, GF_LOG_WARNING,
+                                     "too many groups, reducing %ld -> %d",
+                                     ngroups, max_groups);
+        }
+
         if (!xdr_auth_glusterfs_parms_v2 (&xdr, au)) {
                 gf_log (THIS->name, GF_LOG_WARNING,
                         "failed to encode auth glusterfs elements");
@@ -1154,6 +1171,9 @@ xdr_serialize_glusterfs_auth (char *dest, struct auth_glusterfs_parms_v2 *au)
         ret = (((size_t)(&xdr)->x_private) - ((size_t)(&xdr)->x_base));
 
 ret:
+        if (ngroups)
+                au->groups.groups_len = ngroups;
+
         return ret;
 }
 
@@ -1319,6 +1339,8 @@ rpc_clnt_record (struct rpc_clnt *clnt, call_frame_t *call_frame,
         struct auth_glusterfs_parms_v2  au          = {0, };
         struct iobuf                   *request_iob = NULL;
         char                            owner[4] = {0,};
+        int                             max_groups = 0;
+        int                             max_lkowner_len = 0;
 
         if (!prog || !rpchdr || !call_frame) {
                 goto out;
@@ -1345,6 +1367,27 @@ rpc_clnt_record (struct rpc_clnt *clnt, call_frame_t *call_frame,
                 au.lk_owner.lk_owner_len = 4;
         }
 
+        /* The number of groups and the size of lk_owner depend on oneother.
+         * We can truncate the groups, but should not touch the lk_owner. */
+        max_groups = GF_AUTH_GLUSTERFS_MAX_GROUPS (au.lk_owner.lk_owner_len);
+        if (au.groups.groups_len > max_groups) {
+                GF_LOG_OCCASIONALLY (gf_auth_max_groups_log, clnt->conn.name,
+                                     GF_LOG_WARNING, "truncating grouplist "
+                                     "from %d to %d", au.groups.groups_len,
+                                     max_groups);
+
+                au.groups.groups_len = max_groups;
+        }
+
+        max_lkowner_len = GF_AUTH_GLUSTERFS_MAX_LKOWNER (au.groups.groups_len);
+        if (au.lk_owner.lk_owner_len > max_lkowner_len) {
+                gf_log (clnt->conn.name, GF_LOG_ERROR, "lkowner field is too "
+                        "big (%d), it does not fit in the rpc-header",
+                        au.lk_owner.lk_owner_len);
+                errno = E2BIG;
+                goto out;
+        }
+
         gf_log (clnt->conn.name, GF_LOG_TRACE, "Auth Info: pid: %u, uid: %d"
                 ", gid: %d, owner: %s", au.pid, au.uid, au.gid,
                 lkowner_utoa (&call_frame->root->lk_owner));
diff --git a/rpc/rpc-lib/src/xdr-common.h b/rpc/rpc-lib/src/xdr-common.h
index 34dc9c6a228..f221192adbe 100644
--- a/rpc/rpc-lib/src/xdr-common.h
+++ b/rpc/rpc-lib/src/xdr-common.h
@@ -18,6 +18,7 @@
 
 #include <rpc/types.h>
 #include <sys/types.h>
+#include <rpc/auth.h>
 #include <rpc/xdr.h>
 #include <sys/uio.h>
 
@@ -34,7 +35,34 @@ enum gf_dump_procnum {
 #define GLUSTER_DUMP_PROGRAM 123451501 /* Completely random */
 #define GLUSTER_DUMP_VERSION 1
 
-#define GF_MAX_AUTH_BYTES   2048
+/* MAX_AUTH_BYTES is restricted to 400 bytes, see
+ * http://tools.ietf.org/html/rfc5531#section-8.2 */
+#define GF_MAX_AUTH_BYTES   MAX_AUTH_BYTES
+
+/* The size of an AUTH_GLUSTERFS_V2 structure:
+ *
+ *   1 | pid
+ *   1 | uid
+ *   1 | gid
+ *   1 | groups_len
+ *  XX | groups_val (GF_MAX_AUX_GROUPS=65535)
+ *   1 | lk_owner_len
+ *  YY | lk_owner_val (GF_MAX_LOCK_OWNER_LEN=1024)
+ * ----+-------------------------------------------
+ *   5 | total xdr-units
+ *
+ * one XDR-unit is defined as BYTES_PER_XDR_UNIT = 4 bytes
+ * MAX_AUTH_BYTES = 400 is the maximum, this is 100 xdr-units.
+ * XX + YY can be 95 to fill the 100 xdr-units.
+ *
+ * Note that the on-wire protocol has tighter requirements than the internal
+ * structures. It is possible for xlators to use more groups and a bigger
+ * lk_owner than that can be sent by a GlusterFS-client.
+ */
+#define GF_AUTH_GLUSTERFS_MAX_GROUPS(lk_owner_len) \
+           (95 - lk_owner_len)
+#define GF_AUTH_GLUSTERFS_MAX_LKOWNER(groups_len)  \
+           (95 - groups_len)
 
 #if GF_DARWIN_HOST_OS
 #define xdr_u_quad_t xdr_u_int64_t