rpcgss: krb5: sanity check sealalg value in the downcall

The sealalg is checked in several places, giving the impression it could be
either SEAL_ALG_NONE or SEAL_ALG_DES.  But in fact SEAL_ALG_NONE seems to
be sufficient only for making mic's, and all the contexts we get must be
capable of wrapping as well.  So the sealalg must be SEAL_ALG_DES.  As
with signalg, just check for the right value on the downcall and ignore it
otherwise.  Similarly, tighten expectations for the sealalg on incoming
tokens, in case we do support other values eventually.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

1 files changed, 3 insertions, 1 deletions

diff --git a/net/sunrpc/auth_gss/gss_krb5_mech.c b/net/sunrpc/auth_gss/gss_krb5_mech.c
index 17587163fca..bf5435db878 100644
--- a/net/sunrpc/auth_gss/gss_krb5_mech.c
+++ b/net/sunrpc/auth_gss/gss_krb5_mech.c
@@ -148,9 +148,11 @@ gss_import_sec_context_kerberos(const void *p,
                 goto out_err_free_ctx;
         if (tmp != SGN_ALG_DES_MAC_MD5)
                 goto out_err_free_ctx;
-        p = simple_get_bytes(p, end, &ctx->sealalg, sizeof(ctx->sealalg));
+        p = simple_get_bytes(p, end, &tmp, sizeof(tmp));
         if (IS_ERR(p))
                 goto out_err_free_ctx;
+        if (tmp != SEAL_ALG_DES)
+                goto out_err_free_ctx;
         p = simple_get_bytes(p, end, &ctx->endtime, sizeof(ctx->endtime));
         if (IS_ERR(p))
                 goto out_err_free_ctx;