Re: libstdc++/34: Segmentation fault when resizing a valarry.

[Gabriel Dos Reis <Gabriel dot Dos-Reis at cmla dot ens-cachan dot fr>]

jsk29@cornell.edu writes:

| >Number:         34
| >Category:       libstdc++
| >Synopsis:       Segmentation fault when resizing a valarry.
| >Confidential:   no
| >Severity:       serious
| >Priority:       medium
| >Responsible:    unassigned
| >State:          open
| >Class:          sw-bug
| >Submitter-Id:   net
| >Arrival-Date:   Sat Feb 05 13:07:00 PST 2000
| >Closed-Date:
| >Last-Modified:
| >Originator:     John Karcz
| >Release:        gcc version 2.95.2 19991024 (release)
| >Organization:
| >Environment:
| Redhat 6.1 on a Pentium 200, gcc 2.95.2 in /usr/local.
| >Description:
| First, I hope this is the correct place to report an STL
| in gcc 2.95.2.  If not, please let me know who to contact.
| 
| Resizing a valarray of valarrays can cause a segmentation
| fault under certain circumstances.
| 
| I am very new to using the STL, and to C++ in general.  This
| is the first large program I have written using the STL, so
| this could well be a feature and not a bug. :)  However, it
| looks like a bug to me, and I can't figure out a reason why
| it would occur othersize, so I'm reporting it to the list.

Thank you very much for your bug report.

| Let me know if you need more information, or if I should
| redirect this report to someone else.
| >How-To-Repeat:
| A test program is included, which should reproduce the
| crash.  I've also included tha Makefile, so you can see the
| (simple) command line switches I used, and a backtrace, in
| case that is helpful.
| 
| To reproduce the core dump:
| 
| 1.  Define a valarray<valarray<double>>.  (I believe it
|     also works with types other than double, but I haven't
|     thoroughly tested any other types.)

valarray<valarray<double> > is not expected to work in general,
because:

26.1/1
  The complex and valarray components are parameterized by the type of
  information they contain and manipulate.  A C++ program shall
  instantiate these components only with a type T that satisfies the
  following requirements
  [...]
  -- If T is a class, its assignment operator, copy and default
  constructors, and destructor shall correspond to each other in the
  following sense: Initialization of raw storage using the default
  constructor, followed by assignment, is semantically equivalent to
  initialization of raw storage using the copy constructor. [...]

valarrays do not meet this requirement.

If you want to build a two dimensional array out of valarray, please
consider using gslice and gslice_array in conjunction with a plain
valarray. 

[...]

| #4  0x804cd96 in valarray<valarray<double> >::resize (this=0xbffffc10, __n=10, 
|     __c={_M_size = 3221224456, _M_data = 0x0})
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
I'm wondering how this can be possible, given:

template<typename _Tp>
inline valarray<_Tp>::valarray () : _M_size (0), _M_data (0) {}


Thanks a lot.

-- Gaby