Stack smashing in IPPICV norm implementation on systems without AVX2 #27380
Description
System Information
OS: Ubuntu 24.04 x64
CPU: Core i5 2500k (Sandy Bridge)
Both debug and release tests crashed with stack smashing error in the test Core_Norm/ElemWiseTest.accuracy/0.
Stack trace from GDB:
Note: Google Test filter = Core_Norm/ElemWiseTest.accuracy/0
[==========] Running 1 test from 1 test case.
[----------] Global test environment set-up.
[----------] 1 test from Core_Norm/ElemWiseTest
[ RUN ] Core_Norm/ElemWiseTest.accuracy/0, where GetParam() = 16-byte object <B0-E9 F9-56 55-55 00-00 30-D7 F9-56 55-55 00-00>
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff55f3bd6 in icv_y8_ownNorm_Inf_16u_C1MR_P8Y8 () from /home/arina/Projects/opencv-build/lib/libopencv_core.so.412
(gdb) bt
#0 0x00007ffff55f3bd6 in icv_y8_ownNorm_Inf_16u_C1MR_P8Y8 () at /home/arina/Projects/opencv-build/lib/libopencv_core.so.412
#1 0x000000000000000e in ??? ()
#2 0x000000000000000e in ??? ()
#3 0x000000000000001e in ??? ()
#4 0x0000000000000002 in ??? ()
#5 0x00007fffffffd850 in ??? ()
#6 0x00007ffff55e3bc7 in icv_y8_ippiNorm_Inf_16u_C1MR () at /home/arina/Projects/opencv-build/lib/libopencv_core.so.412
#7 0x00000000fffffff0 in ??? ()
#8 0x00007ffff55407f5 in ipp_hal_norm (src=0x5555572c9cdc "", src_step=28, mask=0x55555719608e "", mask_step=14, width=14, height=30, type=2, norm_type=1, result=0x7fffffffc6d0) at /home/arina/Projects/opencv/hal/ipp/src/norm_ipp.cpp:35
#9 0x00007ffff4cf22f7 in cv::norm (_src=..., normType=0, _mask=...) at /home/arina/Projects/opencv/modules/core/src/norm.dispatch.cpp:310
The issue is not visible on newer platform with the same OS and compiler.
Detailed description
Steps to reproduce
./bin/opencv_test_core --gtest_filter=Core_Norm/ElemWiseTest.accuracy/0
Issue submission checklist
- I report the issue, it's not a question
- I checked the problem with documentation, FAQ, open issues, forum.opencv.org, Stack Overflow, etc and have not found any solution
- I updated to the latest OpenCV version and the issue is still there
- There is reproducer code and related data files (videos, images, onnx, etc)