Re: 22- polgalois()

[Igor Schein <igor@txc.com>]

On Wed, Jun 12, 2002 at 08:39:20PM +0200, Karim BELABAS wrote:
> On Wed, 12 Jun 2002, Igor Schein wrote:
> > ? \d
> [...]
> >    new_galois_format = 0 (off)
> >   ***   bug in GP (Segmentation Fault), please report
> >
> > I see that with debug 64bit binaries on both Solaris and HPUX.
> 
> Igor, you're amazing. Can you try and debug this on Solaris ? Stack trace ?
> I can't reproduce it on alpha + linux ( I had assumed it was careless memory
> handling, which Linux is generally quite unforgiving about, but apparently
> it's not )

with Force cc + dbx, I get this:

signal SEGV (no mapping at the fault address) in sd_ulong at line 488 in file "gp.c"
  488           pariputsf("   %s = %lu %s\n", s, n, *msg);

(/opt/SUNWspro/WS6/bin/sparcv9/dbx) dump
ptn = 0x100503630
msg = 0x77fffef20
v = 0x1005472c0 ""
Min = 0
s = 0x100547140 "output"
flag = 0
n = 4294967296U
Max = 3U
__func__ = "sd_ulong"

Here's indirect way of showing the stack - I haven't have an equivalent
of gdb's backtrace command in dbx:

(/opt/SUNWspro/WS6/bin/sparcv9/dbx) up
Current function is sd_output
  783     return sd_ulong(v,flag,"output",(ulong*) &(GP_DATA->fmt->prettyp), 0,3,msg);
(/opt/SUNWspro/WS6/bin/sparcv9/dbx) up
Current function is help_default
 1021       ((void (*)(char*,int)) dft->fun)("", d_ACKNOWLEDGE);
(/opt/SUNWspro/WS6/bin/sparcv9/dbx) up
Current function is escape0
 1782       case 'd': help_default(); break;
(/opt/SUNWspro/WS6/bin/sparcv9/dbx) up
Current function is escape
 1851     escape0(tch);
(/opt/SUNWspro/WS6/bin/sparcv9/dbx) up
Current function is check_meta
 2336       case '\\': escape(buf); break;
(/opt/SUNWspro/WS6/bin/sparcv9/dbx) up
Current function is gp_main_loop
 2402         else if (!check_meta(b->buf)) break;
(/opt/SUNWspro/WS6/bin/sparcv9/dbx) up
Current function is main
 2696     (void)gp_main_loop(1);

Hope that's sufficient

Igor