Debian Patches
Status for hipblaslt/7.2.4+dfsg-1~exp1
| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| 0009-fix-amdclang.patch | Use versioned LLVM 22 toolchain names in Tensile defaults Replace amdclang/amdclang++ with versioned clang-22/clang++-22 and amdgpu-arch with amdgpu-arch-22 in the Tensile toolchain defaults. Using unversioned names (clang++, amdgpu-arch) relies on PATH ordering which is controlled during the build (d/rules prepends /usr/lib/llvm-22/bin) but is not guaranteed at autopkgtest time. Explicit versioned names are robust regardless of PATH. Also add the versioned names to the supported- component validator lists so Tensile does not reject them. |
Christian Bayle <bayle@debian.org> | not-needed | 2026-06-05 | ||
| 0012-Use-local-mathjax.patch | Use local mathjax | Christian Bayle <bayle@debian.org> | not-needed | 2025-11-01 | ||
| 0004-optional-system-nanobind.patch | optional-system-nanobind | Christian Bayle <bayle@debian.org> | not-needed | 2025-11-11 | ||
| 0005-origami-mut.patch | Use local origami component when building as Debian package When building inside the Debian/Ubuntu packaging environment (DEBIAN_MUT=ON), origami is provided as a separate orig component tarball placed at ${CMAKE_CURRENT_SOURCE_DIR}/origami rather than at the shared monorepo path ../../shared/origami used by the ROCm superbuild. Add a DEBIAN_MUT conditional so dpkg-buildpackage finds the local component, while the upstream ROCM_LIBS_SUPERBUILD guard (with its ORIGAMI_ENABLE_INSTALL workaround) is preserved for non-Debian builds. |
Christian Bayle <bayle@debian.org> | not-needed | 2026-06-04 | ||
| 0013-fix-arm64-immintrin-include.patch | Guard immintrin.h include for x86 architectures only The header <immintrin.h> provides x86 SIMD intrinsics (SSE, AVX, etc.) and is only available on x86/x86_64 platforms. Including it unconditionally causes build failures on arm64 with the error: "This header is only meant to be used on x86 and x64 architecture" . Since hipblaslt_math.hpp does not actually use any x86 intrinsics, the include appears to be unused. This patch guards it with architecture detection macros to allow building on arm64 while preserving the include on x86 systems where it might be needed. =================================================================== |
Igor Luppi <igor.luppi@canonical.com> | invalid | 2026-01-29 | ||
| 0014-fix-hipconfig-version-parsing.patch | Fix hipconfig --version parsing in Tensile Component.py hipconfig --version now outputs "HIP version: 7.2.4" rather than a bare version string. The regex r'(.+)' captured the full line, causing int() conversion to fail on "HIP version: 7". Use a numeric-only regex to extract just the x.y.z version number. =================================================================== |
Igor Luppi <igor.luppi@canonical.com> | yes | 2026-06-04 |
All known versions for source package 'hipblaslt'
- 7.2.4+dfsg-1~exp1 (experimental)
- 7.1.1+dfsg-4 (sid)
