[Octave-bug-tracker] [bug #35229] imread (read_images) fails when QuantumDepth is 32

[anonymous]

URL:
  <http://savannah.gnu.org/bugs/?35229>

                 Summary: imread (read_images) fails when QuantumDepth is 32
                 Project: GNU Octave
            Submitted by: None
            Submitted on: Tue 03 Jan 2012 06:53:27 PM UTC
                Category: Interpreter
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Crash
                  Status: None
             Assigned to: None
         Originator Name: Bernard Kozioziemski
        Originator Email: address@hidden
             Open/Closed: Open
         Discussion Lock: Any
                 Release: 3.4.3
        Operating System: GNU/Linux

    _______________________________________________________

Details:

in __magick_read__.cc, I found the following bug.  I was getting crashes with
floating point exception (division by zero) when attempting to read a
grayscale image.  I traced the problem to line 215 in
src/DLD-FUNCTIONS/__magick_read__.cc


const int divisor = (((1 << QuantumDepth) - 1) / ((1 << depth) - 1));


When graphicsmagick has been compiled with QuantumDepth=32, divisor is always
zero as 1 is implicitly assumed to be a 32 bit integer.  I made the following
change:


const unsigned int divisor = (((((long long)1) << QuantumDepth) - 1) / ((1 <<
depth) - 1));


This casts '1' to a long long int, allowing the 32 bit shift.  




    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?35229>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/