commit 3139f03a74ede3b3bd7cfc6ff219e6c9bc556632
author Gian Marco Iodice <gianmarco.iodice@arm.com> Mon Nov 05 14:26:32 2018 +0000
committer Gian Marco Iodice <gianmarco.iodice@arm.com> Thu Nov 08 13:31:33 2018 +0000
tree cd67b8c9ec1688fcf739c576553b10e435e6938e
parent 421405b6a21b124288a750e2da26dc01eb7391cb

COMPMID-1736: Fixed out-of-bound write in CLIm2Col

The issue was related to CLIm2Col when the number of input channels was less than
the number of elements processed by each thread.
The bug has been fixed in the validate_and_configure_window() function setting the correct number of elements accessed
in the output tensor.

Also fixed an issue GEMM3D when we have a single output channel

Change-Id: I094292d0c7662599c4a4c3916ec5f5821df5faef

src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp

diff --git a/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp b/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
index 715edae..5e02dda 100644
--- a/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
+++ b/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
@@ -110,7 +110,7 @@
     unsigned int &num_elems_processed_per_iteration_x = num_elements_processed[0];
     unsigned int &num_elems_processed_per_iteration_y = num_elements_processed[1];
     bool           reinterpret_input_as_3d             = reshape_info.reinterpret_input_as_3d();
-    bool           reinterpret_output_as_3d            = (reshape_info.depth_output_gemm3d() != 1);
+    bool           reinterpret_output_as_3d            = (reshape_info.depth_output_gemm3d() != 0);
 
     // In case both input and output have to be reinterpreted as 3D tensors,
     // force reinterpret_input_as_3d and reinterpret_output_as_3d to be false.
@@ -227,7 +227,7 @@
     _input1                   = input1;
     _output                   = output;
     _reinterpret_input_as_3d  = reshape_info.reinterpret_input_as_3d();
-    _reinterpret_output_as_3d = (reshape_info.depth_output_gemm3d() != 1);
+    _reinterpret_output_as_3d = (reshape_info.depth_output_gemm3d() != 0);
 
     // In case both input and output have to be reinterpreted as 3D tensors,
     // force reinterpret_input_as_3d and reinterpret_output_as_3d to be false.