On Thu, Jul 17, 2003 at 11:30:43PM +0200, Ulrich Weigand wrote: > Jay Maynard wrote: > >Does Linux/390 (and/or z/Linux) use disabled wait codes in the same manner > >as MVS and VM? Is there any significance to the instruction address part of > >the disabled wait PSW? If not, anyone have a guess as to how hard it would > >be to add, at least for common cases? > I don't know how MVS and VM use disabled wait codes.
On VM and MVS, if a disabled wait state is loaded, the address part of the PSW is not an instruction address, but rather a code that points to a description of the problem in the OS's Messages and Codes book. The code itself is the low-order 12 bits of the PSW. For example, disabled wait 007 is used by MVS to indicate a failure to communicate with the master console. > On Linux, the (usually) only sitation where a disabled wait PSW is loaded > is after a kernel panic happened. In that case, the PSW address is set to > the address of the kernel code that performed the panic () call; if you > have a kernel disassembly (or at least the System.map), you can at least > find out which panic it was if for some reason the console output didn't > make it through anymore ... Ah. The disabled wait state PSW had always seemed random to me. > What 'common cases' are you thinking about? Things like "cannot mount root filesystem", where the console messages may not be going anywhere at all due to no device being connected and similar causes. The idea is that the code, by itself (plus a doc file that's available outside the system), is enough to get the admin on the right track to fixing the problem.
