Build failure when optimization is disabled

[robxnano]

The build fails in resize_impl_avx2.cpp when built with ./configure --enable-debug. It seems to be related to optimization, using make DEBUGCFLAGS="-O1 -g -ggdb" is enough to make it compile successfully.

Branch: master
Operating System: Fedora 41
GCC version: 14.2.1
Architecture: x86_64 (building with SIMD enabled)

make[1]: Entering directory '/home/builder/projects/zimg-master'
/bin/sh ./libtool  --tag=CXX   --mode=compile g++ -DPACKAGE_NAME=\"zimg\" -DPACKAGE_TARNAME=\"zimg\" -DPACKAGE_VERSION=\"3.0.5\" -DPACKAGE_STRING=\"zimg\ 3.0.5\" -DPACKAGE_BUGREPORT=\"https://github.com/sekrit-twc/zimg/pulls\" -DPACKAGE_URL=\"https://github.com/sekrit-twc/zimg\" -DHAVE_STDIO_H=1 -DHAVE_STDLIB_H=1 -DHAVE_STRING_H=1 -DHAVE_INTTYPES_H=1 -DHAVE_STDINT_H=1 -DHAVE_STRINGS_H=1 -DHAVE_SYS_STAT_H=1 -DHAVE_SYS_TYPES_H=1 -DHAVE_UNISTD_H=1 -DSTDC_HEADERS=1 -DHAVE_DLFCN_H=1 -DLT_OBJDIR=\".libs/\" -DHAVE_PTHREAD_PRIO_INHERIT=1 -DHAVE_PTHREAD=1 -DZIMG_X86=1 -I.   -I./src/zimg -I./graphengine/include -DGRAPHENGINE_IMPL_NAMESPACE=zimg  -std=c++17 -O2 -Wall -Wextra -Wno-unused-parameter -Wno-missing-field-initializers -O0 -g3 -ggdb -pthread -mavx2 -mf16c -mfma -mtune=haswell -fvisibility=hidden  -MT src/zimg/resize/x86/libavx2_la-resize_impl_avx2.lo -MD -MP -MF src/zimg/resize/x86/.deps/libavx2_la-resize_impl_avx2.Tpo -c -o src/zimg/resize/x86/libavx2_la-resize_impl_avx2.lo `test -f 'src/zimg/resize/x86/resize_impl_avx2.cpp' || echo './'`src/zimg/resize/x86/resize_impl_avx2.cpp
libtool: compile:  g++ -DPACKAGE_NAME=\"zimg\" -DPACKAGE_TARNAME=\"zimg\" -DPACKAGE_VERSION=\"3.0.5\" "-DPACKAGE_STRING=\"zimg 3.0.5\"" -DPACKAGE_BUGREPORT=\"https://github.com/sekrit-twc/zimg/pulls\" -DPACKAGE_URL=\"https://github.com/sekrit-twc/zimg\" -DHAVE_STDIO_H=1 -DHAVE_STDLIB_H=1 -DHAVE_STRING_H=1 -DHAVE_INTTYPES_H=1 -DHAVE_STDINT_H=1 -DHAVE_STRINGS_H=1 -DHAVE_SYS_STAT_H=1 -DHAVE_SYS_TYPES_H=1 -DHAVE_UNISTD_H=1 -DSTDC_HEADERS=1 -DHAVE_DLFCN_H=1 -DLT_OBJDIR=\".libs/\" -DHAVE_PTHREAD_PRIO_INHERIT=1 -DHAVE_PTHREAD=1 -DZIMG_X86=1 -I. -I./src/zimg -I./graphengine/include -DGRAPHENGINE_IMPL_NAMESPACE=zimg -std=c++17 -O2 -Wall -Wextra -Wno-unused-parameter -Wno-missing-field-initializers -O0 -g3 -ggdb -pthread -mavx2 -mf16c -mfma -mtune=haswell -fvisibility=hidden -MT src/zimg/resize/x86/libavx2_la-resize_impl_avx2.lo -MD -MP -MF src/zimg/resize/x86/.deps/libavx2_la-resize_impl_avx2.Tpo -c src/zimg/resize/x86/resize_impl_avx2.cpp  -fPIC -DPIC -o src/zimg/resize/x86/.libs/libavx2_la-resize_impl_avx2.o
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:517:44: error: the last argument must be an 8-bit immediate
  517 |                         __m256i x = k == 0 ? x0 : k == 4 ? x8 : _mm256_alignr_epi8(x8, x0, static_cast<unsigned>(k * 4));
      |                                            ^
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:517:44: error: the last argument must be an 8-bit immediate
  517 |                         __m256i x = k == 0 ? x0 : k == 4 ? x8 : _mm256_alignr_epi8(x8, x0, static_cast<unsigned>(k * 4));
      |                                            ^
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:517:44: error: the last argument must be an 8-bit immediate
  517 |                         __m256i x = k == 0 ? x0 : k == 4 ? x8 : _mm256_alignr_epi8(x8, x0, static_cast<unsigned>(k * 4));
      |                                            ^
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:517:44: error: the last argument must be an 8-bit immediate
  517 |                         __m256i x = k == 0 ? x0 : k == 4 ? x8 : _mm256_alignr_epi8(x8, x0, static_cast<unsigned>(k * 4));
      |                                            ^
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:517:44: error: the last argument must be an 8-bit immediate
  517 |                         __m256i x = k == 0 ? x0 : k == 4 ? x8 : _mm256_alignr_epi8(x8, x0, static_cast<unsigned>(k * 4));
      |                                            ^
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:517:44: error: the last argument must be an 8-bit immediate
  517 |                         __m256i x = k == 0 ? x0 : k == 4 ? x8 : _mm256_alignr_epi8(x8, x0, static_cast<unsigned>(k * 4));
      |                                            ^
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:517:44: error: the last argument must be an 8-bit immediate
  517 |                         __m256i x = k == 0 ? x0 : k == 4 ? x8 : _mm256_alignr_epi8(x8, x0, static_cast<unsigned>(k * 4));
      |                                            ^
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:517:44: error: the last argument must be an 8-bit immediate
  517 |                         __m256i x = k == 0 ? x0 : k == 4 ? x8 : _mm256_alignr_epi8(x8, x0, static_cast<unsigned>(k * 4));
      |                                            ^
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:517:44: error: the last argument must be an 8-bit immediate
  517 |                         __m256i x = k == 0 ? x0 : k == 4 ? x8 : _mm256_alignr_epi8(x8, x0, static_cast<unsigned>(k * 4));
      |                                            ^
src/zimg/resize/x86/resize_impl_avx2.cpp: In lambda function:
src/zimg/resize/x86/resize_impl_avx2.cpp:612:43: error: the last argument must be an 8-bit immediate
  612 |                                 x = k % 4 ? mm256_alignr_epi8_ps(x4, x0, (k % 4) * 4) : x0;
      |                                     ~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

...

make[1]: *** [Makefile:2599: src/zimg/resize/x86/libavx2_la-resize_impl_avx2.lo] Error 1
make[1]: Leaving directory '/home/builder/projects/zimg-master'
make: *** [Makefile:3693: all-recursive] Error 1

[robxnano]

Looks like e2cfb5b was the commit which introduced the issue.

[sekrit-twc]

This is a GCC bug. Report upstream.