Re: Part II: Revenge of the CoreDumps

SaTriGuy wrote: > > >Ran it this week (6:30AM); it ran fine for Saturday and Monday, errored > >out. > > > >OK, I adjusted the source code to the point where I was getting > >predictable coredumps and recompiled (keeping the .ec file). I ran the > >executable and got a core file in return. I attempted to start xdb with > >the executable name and core file, and it can't read the core file. > >Here's the error: > > > > <<<< XDB Version A.10.00 HP-UX >>>> > > Do you want to save a backup copy of the core file? y > > Core file saved as "core23303" > > Registers bad in core file (UE644) > > Error trying to read "core"; ignoring it (UE646) > > Procedures: 17 > > Files: > >1 > > > >No problem, I execute the program from within xdb with the appropriate > >arguments; that's where the source code map is pointing me. It loads > >the shared memory maps and executes the program. Here's what pops up > >next: > > > > >r -h 90 -i N > > Starting process 23646: "db_rpt.4ge -h 90 -i N" > > Wait...loading shared-library map tables. Done. > > > > segmentation violation (no ignore) at 0xc009ccf0 > > (file unknown): TMEM@libc.1 +0x00088cf0: (line unknown) > > > >What now? I've been around 4gl for a little while (4 1/2 years) but > > I think that this is telling you where you actually aborted, with a SEGV in > libc. libc is one of the system libraries. At this point, could you enter the > "bt" command? Did . . . here's the result: Starting process 18416: "db_rpt_cp.4ge -h 90 -i N" Wait...loading shared-library map tables. Done. segmentation violation (no ignore) at 0xc009ccf0 (file unknown): TMEM@libc.1 +0x00088cf0: (line unknown) >bt Can't set breakpoint (invalid address) (UE310) > I tried to use the core file as well, but xdb tells me that the core file is bad, so it ignores it. > > If you want, you could submitt the core file and executable to the case owner > and request that I take a look at them. It'd be easier with both the core and > the executable that produced the core. I can't do it for a couple of days as > I'm currently not in my office, but I'd be more than glad to see what I can do. > Madison Pruet I'll get the test case together and EMail it ASAP. John Carlson Informix DBA WHSmith USA