segmentation fault when compling c++ code - c++

Question

I am now days struggling at this question from my code,the code could be complied successfully but when I ran the binary file, the segmentation fault would occured and here below was the problem:
Program received signal SIGSEGV, Segmentation fault. _int_malloc (av=av#entry=0x7ffff6adfb20 <main_arena>, bytes=bytes#entry=15859713) at malloc.c:3802 malloc.c: No such file or directory.
Env:ubuntu 16.04 VM workstation
Com: g++, version:5.4.0
c++:c++11
Lib: imebra 5.0.1
Here is my code:
#include <imebra/imebra.h>
#include <iostream>
#include <fstream>
#include <stdlib.h>
#define img_height 2816
#define img_width 2816
#define img_bit 2
#define img_size img_height*img_width*img_bit //15.1MB
using namespace std;
//MONOCHROME1: indicates that the greyscale ranges from bright to dark with ascending pixel values
//MONOCHROME2: indicates that the greyscale ranges from dark to bright with ascending pixel values
/*
create an Image object
fill the image object with raw data
create a DICOM dataset
add the image to the DICOM dataset
fill all the necessary DICOM tags (e.g. sop class, instance, patient name, etc)
save the DICOM dataset
*/
int main()
{
//ifstream mydata("/home/lixingyu/GH1.raw",ios::binary);
//uint32_t *pImgData = (uint32_t *)malloc(img_size*sizeof(uint32_t));
//mydata.read(pImgData,img_size);
FILE *fp = NULL;
fp = fopen("/home/lixingyu/123.raw","rb");
uint32_t *pImgData = new (std::nothrow) uint32_t (img_size);
fread(pImgData,sizeof(uint32_t),img_size,fp);
cout<<"success"<<endl;
/*---------program stop here -------*/
// Creat an image 500 pixels wide , 400 pixels height
// each sample is a 16 bit unsigned value, the colorspace
// is monochrome_2, the higher bit used is 15
// imebra ::MutableImage image(500,400,imebra::bitDepth_t::depthU16,"MONOCHROME_2",15);
imebra ::MutableImage image(img_height,img_width,imebra::bitDepth_t::depthU16,"MONOCHROME2",15);
// 1. Fill the image with data
// We use a writing data handler to write into the image.
// The data is committed into the image only when the writing
// data handler goes out of scope.
imebra::WritingDataHandlerNumeric writeIntoImage(image.getWritingDataHandler());
for (size_t y=0;y!=img_width;++y)
{
for (size_t x=0; x!= img_height; ++x)
{
writeIntoImage.setUnsignedLong(y*img_height+x,pImgData[y*img_height+x]);
}
}
// specify the tansfer syntax and the charset
imebra::charsetsList_t charsets;
charsets.push_back("ISO 2022 IR 6");
//Explicit VR little endian
imebra::MutableDataSet dataSet("1.2.840.10008.1.2.1",charsets);
// add the image to the dataSet
dataSet.setImage(0,image,imebra::imageQuality_t::veryHigh);
// set the patient name
dataSet.setUnicodePatientName(imebra::TagId(imebra::tagId_t::PatientName_0010_0010),imebra::UnicodePatientName(L"fjx",L"",L""));
// save to a file
imebra::CodecFactory::save(dataSet,"GH1.dcm",imebra::codecType_t::dicom);
free(pImgData);
}
When I use gdb to debug my code, the question occured and I have changed my stack size to 100MB,but then the segementation fault would occure.
Maybe something wrong with Dynamic memory application？？
Could anyone help me out?
FYI,The func of imebra::XXX are all from imebra lib.

Answer 1

You are not allowed to call free on memory allocated by new. That causes undefined behavior. You must call delete instead.
You are also allocating only one uint32_t (and initialing it with the value img_size), not an array of img_size many. For that you would need new (std::nothrow) uint32_t[img_size]; instead (and later delete[] instead of delete). So you are going to write out-of-bounds with fread.
You also need to check that the return value of new(std::nothrow) is not a null pointer, which would happen on allocation failure. If you use the throwing version, then you won't need that check.
Please don't use new like this though and use std::vector instead. malloc in C++ is even worse than new.
Similarly, don't use the C IO library in C++. Use std::ifstream instead.