gdb/19: gdb v3 demangler fails on hairyfunc5 hairyfunc6 hairyfunc7

[chastain at redhat dot com]

>Number:         19
>Category:       gdb
>Synopsis:       gdb v3 demangler fails on hairyfunc5 hairyfunc6 hairyfunc7
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Feb 04 18:18:00 PST 2001
>Closed-Date:
>Last-Modified:
>Originator:     Michael Chastain
>Release:        cvs 2001-02-02
>Organization:
>Environment:
Red Hat Linux 7 native (stabs debugging format)
>Description:
gdb has two demanglers for g++ names.
The new demangler produces incorrect results on the symbols "hairyfunc5" "hairyfunc6" "hairyfunc7" in testsuite/gdb.c++/cplusfuncs.cc.
The old demangler works fine.

The stabs in the generated assembly code are:

  .stabs "_Z10hairyfunc5PFPFilEPcE:F(0,1)",36,0,183,_Z10hairyfunc5PFPFilEPcE
  .stabs "_Z10hairyfunc6PFPFilEPiE:F(0,1)",36,0,184,_Z10hairyfunc6PFPFilEPiE
  .stabs "_Z10hairyfunc7PFPFilEPFiPcEE:F(0,1)",36,0,185,_Z10hairyfunc7PFPFilEPFiPcEE

These stabs are correct.  gdb's demangler interprets them wrong.
  
>How-To-Repeat:
Build testsuite/gdb.c++/cplusfuncs with a recent g++ compiler (such as cvs g++ 2001-02-02).

Invoke gdb and enter the following commands:

  (gdb) print &'hairyfunc5'
  $1 = (int (*)(PFPc_PFl_i)) 0x8049120 <hairyfunc5(int (*)(long) (*)(char*))>
  (gdb) print &'hairyfunc6'
  $2 = (int (*)(PFPi_PFl_i)) 0x8049140 <hairyfunc6(int (*)(long) (*)(int*))
  (gdb) print &'hairyfunc7'
  $3 = (int (*)(PFPFPc_i_PFL_i)) 0x8049160 <hairyfunc7(int (*)(long (*)(int (*)(char*)))>



>Fix:
Take a PATH train to New Jersey and geek-slap DMR for designing declarators with both prefix and postfix operators.
>Release-Note:
>Audit-Trail:
>Unformatted: