Fix buffer overrun in SingularValueDecomposition::GetV

[CodeReclaimers]

GetV (line 236) loops over mUMatrix.size() (M*M) instead of mVMatrix.size() (N*N). When M > N, this reads past the end of mVMatrix.

Fix: change mUMatrix.size() to mVMatrix.size() on line 236.

Test (CMakeLists.txt)

cmake_minimum_required(VERSION 3.14)
project(test_issue_7_1 LANGUAGES CXX)
set(CMAKE_CXX_STANDARD 17)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
set(GTE_ROOT "${CMAKE_CURRENT_SOURCE_DIR}/../../GTE")
add_executable(test_issue_7_1 test_issue_7_1.cpp)
target_include_directories(test_issue_7_1 PRIVATE "${GTE_ROOT}")

Test (test_issue_7_1.cpp)

#include <cmath>
#include <cstdlib>
#include <iostream>
#include <vector>
#include <Mathematics/SingularValueDecomposition.h>

static bool test_getv_size()
{
    using Real = double;
    size_t const M = 5, N = 3;
    gte::SingularValueDecomposition<Real> svd(M, N, 1000);

    std::vector<Real> A(M * N);
    for (size_t r = 0; r < M; ++r)
        for (size_t c = 0; c < N; ++c)
            A[c + N * r] = static_cast<Real>((r+1)*(c+1));
    svd.Solve(A.data());

    // Surround buffer with sentinels to detect overrun.
    Real const sentinel = -999.0;
    std::vector<Real> vBuf(N*N + 4, sentinel);
    Real* vMatrix = vBuf.data() + 2;
    svd.GetV(vMatrix);

    return vBuf[0] == sentinel && vBuf[1] == sentinel
        && vBuf[N*N+2] == sentinel && vBuf[N*N+3] == sentinel;
}

int main()
{
    return test_getv_size() ? EXIT_SUCCESS : EXIT_FAILURE;
}