NonStop S-Series Server Description Manual (G06.27+)

TNS Execution Modes
HP NonStop S-Series Server Description Manual—520331-004
6-78
Gateway Tables
Gateway Tables
When a process that is operating in TNS mode calls a CALLABLE system procedure,
there is necessarily a transition to privileged mode.
In the case of programs that use TNS mode, such a transition to privileged mode takes
place in the millicode for the calling instruction (XCAL, PCAL, or DPCL). That is the
normal case for the TNS architecture, and it is based on the layout of the procedure
entry point (PEP) table for the segment containing the called procedure. However, in
the case of accelerated programs, use of the PEP table is generally skipped. Thus
there needs to be a separate mechanism for providing secure transitions to privileged
mode, and that is the function of the gateway tables.
Gateway tables are built by the Accelerator and, as shown in Figure 6-42, immediately
follow the accelerated code in the object code file. Gateway tables exist only for those
code areas that actually contain callable privileged procedures. User code (UC) and
user library (UL) seldom have such CALLABLE procedures and so rarely have
gateway tables. In fact, most CALLABLE procedures are in the system library (SL)
area. The example shown in Figure 6-42 illustrates the gateway table for the system
library, which resides in user space. (Calls to callable procedures in the SC area
require an additional mechanism to make the “far jump” to SC; for efficiency, the far-
jump mechanism is combined with the gateway table and so calls to SC are described
separately in the next topic.)
Note the major elements of the example shown, all highlighted in white. These are the
calling instructions in the accelerated user code (on the left) and the gateway table and
called procedure code in the accelerated code part of the system library code (on the
right). The example assumes the procedure being called is the READ procedure.
As indicated, the last instruction of the calling instructions in the user code is a jump to
the appropriate entry (READ) in the gateway table. Each entry of the gateway table
consists of two instructions: a special load-byte instruction and a jump instruction. The
load-byte instruction attempts to load a certain byte of the scratchpad page (SPAD),
which is located in privileged memory (Kseg2). Such a reference requires privileged
mode. For the privileged caller, the load occurs and the jump immediately follows. But
for the nonprivileged caller, an address error exception occurs. This invokes the
exception handler, which checks to assure that the exception occurred within the
gateway table, and allows the process to proceed in privileged state.
The exception handler returns to the first instruction of the gateway; this time the load-
byte instruction completes without incident, and the jump instruction sends control to
the entry point of the called procedure (READ in this example). The called procedure
now executes in privileged mode and performs the indicated operation. The EXIT
millicode, at procedure’s end, sets the RISC status register back to the state of the
caller (nonprivileged in this case) and exits back to the user code.