| 摘要 |
1,081,816. Automatic exchange systems. WESTERN ELECTRIC CO. Inc. Oct. 13, 1964 [Dec. 31, 1963], No. 41687/64. Heading H4K. [Also in Division G4] A data processing system comprises a program store containing three types of sequences of program order words, viz.: (1) base level (normal data processing) sequences, (2) timed interrupt sequences and (3) maintenance interrupt sequences, wherein the second sequences are executed at regular intervals for performing routine actions on the system and the third sequences are executed at any time as a result of a fault occurring in the system. The system is for use with a program-controlled ferreed switching system. The general method of operation is similar to that described in Specification 904,164. The base level program comprises five sublevels LA to LE (Fig. 74) of which the sequences associated with higher sub-levels i.e. LA, are performed more frequently than those of lower levels, i.e. LE. During the time allotted to sub-levels LA to LD, the various " jobs " associated with these sub-levels are performed. If at the end of the allotted time more jobs remain to be done then these jobs are tackled first when the sub-level containing them is next performed. In the case of sub-level E, however, a priority system of jobs is maintained. For example, if a fault, the location of which may be deferred, is detected, then this fault location takes priority every time sub-level LE is performed (Fig. 83, not shown) while other jobs executed in sub-level LE such as routine testing of the system are graded in order of priority and many machine cycles may pass before they are even considered for execution. The base-level program is used for all jobs which do not require a rigid timetable, e.g. the processing of information obtained from the line scanners which scan the lines at the rate of once every 100 m.secs. However, the actual gathering of information from the lines, e.g. request for service, impulsing, hang-up &c., must be performed to a rigid timetable and for this purpose the timed interrupt sequences are required. The timed interrupt sequences are initiated at 5 m.sec. intervals and for convenience they are divided into odd and even (H and J levels, Fig. 74) sequences. The various network points such as lines, trunks &c. which must be scanned and other points such as switches which require commands are addressed during these sequences. In addition, various malfunctions of the system may be detected during these sequences and these faults are " examined " in order to determine their nature and thus to decide how they should be dealt with. In general the particular program order word being executed is completed before the fault is dealt with. Some faults require almost immediate attention and these are dealt with by the maintenance interrupt sequences whilst others are not so urgent and may be deferred for remedial action to be taken during the course of the base level program as previously mentioned. The maintenance sequences comprise levels A to G (Fig. 74) in that order of priority. Any maintenance level currently being performed may be interrupted on a request occurring for a higher level. After the new level is executed successfully, the interrupted level is returned to and completion of that level causes a reversion to the base level. In most cases the A and B levels are of equal priority. The A level is set-up for manual testing of the system and is used very infrequently. The B level can interrupt the A level and of course all lower levels and occurs in the case of emergency action. The latter arises as a result of uncorrectable faults (i.e. lower level maintenance cannot cope) and leads to a rearrangement of the program and call stores, the central controls and the intercommunicating buses as is more fully described in Specification 1,081,812. The order word sequences for maintenance are in general stored in a program store but in the event that a fault occurs as a result of addressing such a store the maintenance sequence appropriate to this is stored in the call store. The latter contains special locations for storing information contained in a central control prior to maintenance. This information and the address of the order currently being executed are transferred to these locations before the data concerned with the maintenance function is brought into the central control. At the end of maintenace the original information is of course returned to central control. In general maintenance is delayed until the current order is executed, for example certain order words contain special bit positions indicating that no maintenance below a certain level is to be performed until the order is completed. If an interrupt is attempted during the processing of such words a flip-flop in central control is operated so that the relevant maintenance can be performed later. Numerous examples of deferable maintenance and of the methods used for protecting data in the stores are given in the Specification. |
