summaryrefslogtreecommitdiff
path: root/trunk/TAO/orbsvcs/orbsvcs/SSLIOP/SSLIOP_Accept_Strategy.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'trunk/TAO/orbsvcs/orbsvcs/SSLIOP/SSLIOP_Accept_Strategy.cpp')
-rw-r--r--trunk/TAO/orbsvcs/orbsvcs/SSLIOP/SSLIOP_Accept_Strategy.cpp66
1 files changed, 66 insertions, 0 deletions
diff --git a/trunk/TAO/orbsvcs/orbsvcs/SSLIOP/SSLIOP_Accept_Strategy.cpp b/trunk/TAO/orbsvcs/orbsvcs/SSLIOP/SSLIOP_Accept_Strategy.cpp
new file mode 100644
index 00000000000..5de9b828787
--- /dev/null
+++ b/trunk/TAO/orbsvcs/orbsvcs/SSLIOP/SSLIOP_Accept_Strategy.cpp
@@ -0,0 +1,66 @@
+#include "orbsvcs/SSLIOP/SSLIOP_Accept_Strategy.h"
+
+
+ACE_RCSID (SSLIOP,
+ SSLIOP_Accept_Strategy,
+ "$Id$")
+
+TAO_BEGIN_VERSIONED_NAMESPACE_DECL
+
+TAO::SSLIOP::Accept_Strategy::Accept_Strategy (
+ TAO_ORB_Core * orb_core,
+ const ACE_Time_Value & timeout)
+ : TAO_Accept_Strategy<TAO::SSLIOP::Connection_Handler,
+ ACE_SSL_SOCK_ACCEPTOR> (orb_core),
+ timeout_ (timeout)
+{
+}
+
+int
+TAO::SSLIOP::Accept_Strategy::accept_svc_handler (handler_type * svc_handler)
+{
+ ACE_TRACE ("TAO::SSLIOP::Accept_Strategy::accept_svc_handler");
+
+ // The following code is basically the same code found in
+ // ACE_Accept_Strategy::accept_svc_handler(). The only difference
+ // is that a timeout value is passed to the peer acceptor's accept()
+ // method. A timeout is necessary to prevent malicious or
+ // misbehaved clients from only completing the TCP handshake and not
+ // the SSL handshake. Without the timeout, a denial-of-service
+ // vulnerability would exist where multiple incomplete SSL passive
+ // connections (i.e. where only the TCP handshake is completed)
+ // could result in the server process running out of file
+ // descriptors. That would be due to the SSL handshaking process
+ // blocking/waiting for the handshake to complete.
+
+ // The timeout value will be modified. Make a copy.
+ ACE_Time_Value timeout (this->timeout_);
+
+ // Try to find out if the implementation of the reactor that we are
+ // using requires us to reset the event association for the newly
+ // created handle. This is because the newly created handle will
+ // inherit the properties of the listen handle, including its event
+ // associations.
+ const int reset_new_handle = this->reactor_->uses_event_associations ();
+
+ if (this->peer_acceptor_.accept (svc_handler->peer (), // stream
+ 0, // remote address
+ &timeout, // timeout
+ 1, // restart
+ reset_new_handle // reset new handler
+ ) == -1)
+ {
+ // Ensure that errno is preserved in case the svc_handler
+ // close() method resets it.
+ ACE_Errno_Guard error (errno);
+
+ // Close down handler to avoid memory leaks.
+ svc_handler->close (0);
+
+ return -1;
+ }
+ else
+ return 0;
+}
+
+TAO_END_VERSIONED_NAMESPACE_DECL