siegel@SCIVAX.STSCI.EDU ("HOWARD SIEGEL") (07/07/88)
Jerry Leichter (LEICHTER-JERRY@CS.YALE.EDU) writes >> [In response to a message asking about methods of identifying a CPU for >> software protection, the author writes:] >> > - How can we get the cpu identification (F$GETSYI("SID")??) from >> within >> >> With the system service SYS$GETSYI. >> However, the >> format of the SID register is different for different processors (in >> the case of the 11/78x, I know that in addition to the serial number >> the register also contains the ECO/FCO level, so you have to be >> carefull to look at only the serial number portion unless you want >> your software to stop working every time the machine has a field >> upgrade), also all microVAX IIs have the same serial number. >> >> FALSE. The ONLY VAX CPU ever to have a serial number in the SID was the 780 >> (and, of course, the 785). Even then, there were only 12 bits of serial >> number, and there were a lot more than 4096 780's produced! >> The 780 and 785 were NOT the only VAXen that had serial numbers in the SID. The following is from the 1986 VAX Hardware Handbooks, Volumes 1 and 2 (order codes EB-25949-46 and EB-28348-46). The format of the SID is as follows: 31 24 23 16 15 8 7 0 +------------+------------+------------+------------+ | | | | | +------------+------------+------------+------------+ \__________/ \____________________________________/ Type Code Type Specific Information Type Code - Binary number that identifies the model of VAX processor: Type Code Processor --------- ------------------------------------------- 0 Reserved for DEC 1 11/780 or 11/785 per bits 23:22 2 11/750, 11/751 3 11/725, 11/730 4 8600 5 - 127 Reserved for DEC 128 - 255 Reserved for Digital CSS group and customers Note that I only had the 1986 manuals available, so the list of processor type codes is incomplete. Type specific information for the older series VAXen: (Note the 86xx are transition machines. They are new technology versions of the 78x architecture, not true BI machines) 11/725, 11/730, 11/750, 11/751: Bits 23:16 - Not used Bits 15:8 - CPU microcode software rev level Bits 7:0 - CPU hardware rev level 11/780, 11/785: Bits 23:22 - CPU type, 1 = 780, 0 = 785 Bits 21:18 - CPU hardware rev level Bits 17:15 - letter assigned to system rev level Bits 14:12 - Manufacturing facility number Bits 11:0 - CPU serial number 11/782, 8600, 8650 Bits 23:16 - CPU hardware rev level Bits 15:12 - Manufacturing facility number Bits 11:0 - CPU serial number Type specific information for the newer series VAXen: 8800/8700, 8550/8500: Bit 23 - Left/Right processor flag Bits 22:16 - Processor rev level Bits 15:0 - system serial number 8300/8200: Bits 23:19 - Processor rev code Bits 18:9 - Microcode patch rev level Bit 8 - Secondary patch bit Bits 7:0 - Control store ROM/RAM rev level Thus, not only did the 780 and 785 contain system serial numbers, but also the 782, 86xx, 88xx, 87xx, and 85xx processors. These handbooks did not contain information on the microVAX series processor registers, so I can not point to anyhting in writing, but I have heard that the older processors (the microVAX-I and -II) each had the same serial number (eg. all uVAX-I had the same number...). The issue of CPU serial numbers has become important to me because the system that we inheritted and are maintaining has the feature that it uses the CPU serial number to determine, via a lookup into a hardcoded table, what software configuration to bring up. This has caused us grief because we would like to distribute the system across more CPU's than the software currently allows, and the newer systems either have the same SN for all processors, or there really is no SN to use. The system is also not designed to easily convert to using a different CPU unique identifier such as the DECnet node number, the DECnet node name, the SCS system id, etc. etc. +--------------------------------------+--------------------------+ ! Howard Siegel - TRW, Inc. ! "I am quite sure that ! ! ! you could elevate an ! ! ARPAnet: siegel@scivax.arpa ! earwig to the level of ! ! SPAN: SCIVAX::SIEGEL ! a nuclear scientist, ! ! USnail: One Space Park, O2/2751 ! but it would still be ! ! Redondo Beach, Ca. 90278 ! a very stupid thing to ! ! AT&T: 213/535-5083 ! do!" ! +--------------------------------------+--------------------------+ ------