>What if R13 has been destroyed somehow? You can do forward chaining by >starting at TCBFSA to the first save area, then using the pointer in >word 3 of the save areas to >chain forward until it is zero. That is >what the SYSUDUMP formatter does. But I do agree that the forward chain >is not always reliable. It is for every compiled language that I'm >aware of. But it depends on the HLASM programmer following the >convention if there is any assembler along the line.
The only thing thiat is predictable these days is chaining backwards from the current save area. You cannot predictably chain forward unless you happen to know that all uses of your savearea were for 31-bit 72-byte cases (and not for 64-bit 144-byte cases, for example) and, of couse, that they follow the savearea chaining protocol.. >Back in the old days ISTR it set x'FF' to indicate the end of the savearea chain. Good memory, When the "T" parameter of RETURN was specified, prior to MVS/XA X'FF' was placed in the high byte of the reg 14 slot. As of MVS/XA (actually the point-release just after MVS/XA, perhaps no one had noticed for the original release of MVS/XA), that was changed to instead place x'01' in the low byte of the reg 14 slot. (for relatively obvious AMODE 31 reasons). Does anyone actually use this? (Just curious, not thinking of removing it) Someone mentioned clearing the chain when returning. Some of our best debuggers do some of their best work taking advantage of residual information, as that can give a hint of what had happened "just before". The data might not be predictable but sometimes is useful. Peter Relson z/OS Core Technology Design ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
