Tested on ppc64el, arm64, and amd64 in PPA (~steelf/ubuntu/ppa): - ppc64el: https://launchpad.net/~steelf/+archive/ubuntu/ppa/+build/33836194 (Successfully built) - arm64: https://launchpad.net/~steelf/+archive/ubuntu/ppa/+build/33836193 (Successfully built) - amd64: https://launchpad.net/~steelf/+archive/ubuntu/ppa/+build/33836192 (Successfully built)
Forwarded upstream: https://gitlab.gnome.org/GNOME/gegl/-/issues/480 Submitted to Debian BTS via email (bugs.debian.org is experiencing 503 backend delays, will update with bug ID once assigned). Debdiff attached for stonking. ** Description changed: + gegl 1:0.4.72-2 fails to build from source on ppc64el due to test + failure in format-sensing. + + Cause: + GCC 15 on ppc64el miscompiles the union-based bitwise sign-bit clearing helper (int_fabsf) in operations/common/opacity.c, gegl/gegl-math.h, and gegl/buffer/gegl-sampler-cubic.c. The mask load is corrupted, causing int_fabsf(value - 1.0f) <= EPSILON to evaluate to true for value = 0.5f, which bypasses the format conversion in gegl:opacity. + + Fix: + Replace union type-punning hacks with __builtin_fabsf / __builtin_fabs on GCC/Clang with standard fabsf/fabs fallback for MSVC. + + Upstream issue: + https://gitlab.gnome.org/GNOME/gegl/-/issues/480 + + Debian BTS: + Submitted via email (bugs.debian.org backend is delayed with 503, will link bug ID once assigned). + + --- [ Original Report ] gegl 1:0.4.72-2 fails to build from source in stonking. Regressing architectures (built before, fail now): ppc64el. (1:0.4.70-4 built successfully on ppc64el; 1:0.4.72-2 built on all other arches.) Build log: - https://launchpad.net/ubuntu/+source/gegl/1:0.4.72-2/+build/33591983/+files/buildlog_ubuntu-stonking-ppc64el.gegl_1%3A0.4.72-2_BUILDING.txt.gz - Diagnosis (deterministic matcher): testsuite-failure — dh_auto_test: error: + Diagnosis (deterministic matcher): testsuite-failure - dh_auto_test: error: cd obj-powerpc64le-linux-gnu && DEB_PYTHON_INSTALL_LAYOUT=deb LC_ALL=C.UTF-8 MESON_TESTTHREADS=4 meson test --verbose --timeout-multiplier 3 returned exit code 1 Root cause analysis: meson test 14/241 "gegl:format_sensing" (tests/simple/test-format-sensing.c) fails with exit status 255: 4 of 7 gegl:opacity cases (test_opacity_linear_001/002/004, test_opacity_gamma_001/002/004) negotiate the wrong output format (e.g. "Got RGB float expected RGBA float", "Got RGBA u8 expected RGBA float"). All other tests pass. The failing ppc64el build (a retry, 2026-09-24) used gcc-15 15.3.0-4ubuntu1 from stonking-proposed; the successful builds on 2026-09-12 used gcc-15 15.3.0-3ubuntu1. babl is identical (1:0.1.128-1). The same gegl version builds fine on ppc64el in Debian unstable. This suggests a possible toolchain (gcc-15) miscompilation or ppc64el-specific regression rather than a gegl source bug; needs reproduction (e.g. rebuild with gcc-15 15.3.0-3 vs -4 on ppc64el). No existing report found in Launchpad, Debian BTS, or upstream GNOME GitLab. ** Changed in: gegl (Ubuntu) Status: New => Confirmed ** Tags added: patch ** Bug watch added: gitlab.gnome.org/GNOME/gegl/-/issues #480 https://gitlab.gnome.org/GNOME/gegl/-/issues/480 ** Patch added: "gegl_0.4.72-2ubuntu1_stonking.debdiff" https://bugs.launchpad.net/ubuntu/+source/gegl/+bug/2168951/+attachment/6008482/+files/gegl_0.4.72-2ubuntu1_stonking.debdiff ** Description changed: gegl 1:0.4.72-2 fails to build from source on ppc64el due to test failure in format-sensing. Cause: GCC 15 on ppc64el miscompiles the union-based bitwise sign-bit clearing helper (int_fabsf) in operations/common/opacity.c, gegl/gegl-math.h, and gegl/buffer/gegl-sampler-cubic.c. The mask load is corrupted, causing int_fabsf(value - 1.0f) <= EPSILON to evaluate to true for value = 0.5f, which bypasses the format conversion in gegl:opacity. Fix: Replace union type-punning hacks with __builtin_fabsf / __builtin_fabs on GCC/Clang with standard fabsf/fabs fallback for MSVC. Upstream issue: https://gitlab.gnome.org/GNOME/gegl/-/issues/480 Debian BTS: Submitted via email (bugs.debian.org backend is delayed with 503, will link bug ID once assigned). + + --- [ Original Report ] gegl 1:0.4.72-2 fails to build from source in stonking. Regressing architectures (built before, fail now): ppc64el. (1:0.4.70-4 built successfully on ppc64el; 1:0.4.72-2 built on all other arches.) Build log: - - https://launchpad.net/ubuntu/+source/gegl/1:0.4.72-2/+build/33591983/+files/buildlog_ubuntu-stonking-ppc64el.gegl_1%3A0.4.72-2_BUILDING.txt.gz + - https://launchpad.net/ubuntu/+source/gegl/1:0.4.72-2/+build/33591983/+files/buildlog_ubuntu-stonking-ppc64el.gegl_1%3A0.4.72-2_BUILDING.txt.gz Diagnosis (deterministic matcher): testsuite-failure - dh_auto_test: error: cd obj-powerpc64le-linux-gnu && DEB_PYTHON_INSTALL_LAYOUT=deb LC_ALL=C.UTF-8 MESON_TESTTHREADS=4 meson test --verbose --timeout-multiplier 3 returned exit code 1 Root cause analysis: meson test 14/241 "gegl:format_sensing" (tests/simple/test-format-sensing.c) fails with exit status 255: 4 of 7 gegl:opacity cases (test_opacity_linear_001/002/004, test_opacity_gamma_001/002/004) negotiate the wrong output format (e.g. "Got RGB float expected RGBA float", "Got RGBA u8 expected RGBA float"). All other tests pass. The failing ppc64el build (a retry, 2026-09-24) used gcc-15 15.3.0-4ubuntu1 from stonking-proposed; the successful builds on 2026-09-12 used gcc-15 15.3.0-3ubuntu1. babl is identical (1:0.1.128-1). The same gegl version builds fine on ppc64el in Debian unstable. This suggests a possible toolchain (gcc-15) miscompilation or ppc64el-specific regression rather than a gegl source bug; needs reproduction (e.g. rebuild with gcc-15 15.3.0-3 vs -4 on ppc64el). No existing report found in Launchpad, Debian BTS, or upstream GNOME GitLab. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2168951 Title: gegl 1:0.4.72-2 FTBFS on ppc64el (format_sensing test failure) To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/gegl/+bug/2168951/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
