Barbara, Just after I sent my question to the list, I searched the bookshelves using the keyword 'loop' and surprisingly to find some useful information in a diagnostic book.
I think what the manual says is the same as what you said. In a MP system, a disabled loop is observed as a spin loop by other processors. However, if you have only one processor, a disabled loop can bring a system outage. I must say it's very interesting. Suppose in a single-processor system, I write some codes which cause a disabled loop. How to handle it? Press 'restart' key like what we usually react in a windows system? :) On 10/24/07, Barbara Nitz <[EMAIL PROTECTED]> wrote: > > > > Disabling oneself for interrupts (if done correctly) means holding a spin > lock. So sooner or later (if you have at least two processors in that lpar) > excessive spin recovery will kick in and if set up correctly issue the > restart interrupt. That usually takes care of the 'easier' forms of bugs. > Won't work on a uni lpar, that one will just become unresponsive and you > have to *know* that it may be a spin loop and you have to use the right icon > on the HMC to interrupt that one. > > The spin loops that are caused by unexpected circular chains (SRM, RSM, > dispatcher itself) are resolved by that excessive spin setup, too, but the > spin will occur again right away on another processor because the code will > run into the same circular chain. (The dispatcher protects itself against > that by going into recovery and preventing every other processor from > executing dispatcher code. On the one processor it will attempt to sort > things out.) > > > -- Best Regards, Johnny Luo ---------------------------------------------------------------------- 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
