John R. Jackson writes:
>
> The first thing I'd do is build with -g (which it probably already is)
> and without -O. Then I'd try it again to see if the traceback points
> anywhere more useful.
>
> Then I'd try setting a breakpoint in tape_rdlabel and single stepping
> until the error to see if that showed anything odd going on.
Okay, I did this. With -g and without -Ox, here are the last few steps
before the SEGV:
(gdb) step
tape_open (filename=0x806d440 "file:/usr/local/amanda/test_amanda", mode=0,
mask=0) at tapeio.c:525
525 if(tape_info[fd].vtape_index < 0) {
(gdb) step
526 tape_info[fd].vtape_index = vslot;
(gdb) step
529 return fd;
(gdb) step
530 }
(gdb) step
tapefd_rdlabel (fd=3, datestamp=0xbfbfdc88, label=0xbfbfdc98) at tapeio.c:779
779 char *r = NULL;
(gdb) step
Program received signal SIGSEGV, Segmentation fault.
tapefd_rdlabel (fd=3, datestamp=0xbfbfdc88, label=0xbfbfdc98) at tapeio.c:779
779 char *r = NULL;
Well, I'm learning about gdb, anyway :/
Dan
--
Dan Debertin
[EMAIL PROTECTED]
www.nodewarrior.org
