bug#24358: 25.1.50; re-search-forward errors with "Variable binding depth exceeds max-specpdl-size"

[Sam Halliday <sam.halliday@gmail.com>]

The failure occurs at 20% if I delete the line (which seems to have
moved to line 412).

Some more debugging, human readable first two frames:

(gdb) f 1
#1  0x00000000006351d9 in xsignal (error_symbol=..., data=...) at eval.c:1582
1582  Fsignal (error_symbol, data);
(gdb) p error_symbol
$8 = {
  i = 9600
}
(gdb) xpr
Lisp_Symbol
$9 = (struct Lisp_Symbol *) 0xdb0610 <lispsym+9600>
"args-out-of-range"
(gdb) p data
$10 = {
  i = 116316307
}
(gdb) xpr
Lisp_Cons
$11 = (struct Lisp_Cons *) 0x6eed890
{
  car = {
    i = 0x3cff3c4
  },
  u = {
    cdr = {
      i = 0x6eed8a3
    },
    chain = 0x6eed8a3
  }
}

(gdb) f 2
#2  0x00000000006352ae in xsignal3 (error_symbol=..., arg1=...,
arg2=..., arg3=...) at eval.c:1609
1609  xsignal (error_symbol, list3 (arg1, arg2, arg3));
(gdb) p error_symbol
$12 = {
  i = 9600
}
(gdb) xpr
Lisp_Symbol
$13 = (struct Lisp_Symbol *) 0xdb0610 <lispsym+9600>
"args-out-of-range"
(gdb) p arg1
$14 = {
  i = 63960004
}
(gdb) xpr
Lisp_String
$15 = (struct Lisp_String *) 0x3cff3c0
"七宝焼/"
(gdb) p arg2
$16 = {
  i = -18
}
(gdb) xpr
Lisp_Int1
$17 = -5
(gdb) p arg3
$18 = {
  i = 0
}
(gdb) xpr
Lisp_Symbol
$19 = (struct Lisp_Symbol *) 0xdae090 <lispsym>
"nil"




On 23 October 2016 at 18:19, Eli Zaretskii <eliz@gnu.org> wrote:
>> From: Sam Halliday <sam.halliday@gmail.com>
>> Date: Sun, 23 Oct 2016 16:58:07 +0100
>> Cc: npostavs@users.sourceforge.net, 24358@debbugs.gnu.org
>>
>> Processing OKURI-NASI entries ...
>> processed 10% ...
>> processed 20% ...
>> processed 30% ...
>> processed 40% ...
>>
>> Thread 1 "emacs" hit Breakpoint 2, Fsignal (error_symbol=...,
>> data=...) at eval.c:1476
>> 1476    = (NILP (error_symbol) ? Fcar (data) : error_symbol);
>> (gdb) xbacktrace
>> "substring" (0xffff6f00)
>> "string=" (0xffff70a0)
>> "if" (0xffff7230)
>> "while" (0xffff7400)
>> "catch" (0xffff7600)
>> "let" (0xffff7860)
>> "and" (0xffff79f0)
>> "let" (0xffff7c50)
>> "and" (0xffff7de0)
>> "or" (0xffff7f70)
>> "let" (0xffff81d0)
>> "skkdic-breakup-string" (0xffff8380)
>
> Thanks.
>
> Does it help to delete this line:
>
>   re_match_object = Qnil;       /* Stop protecting string from GC.  */
>
> in src/search.c:string_match_1 (should be line 404)?