b9780

ggml-org/llama.cppb9780Jun 24, 2026by github-actions[bot]same day as b9777

What changed since b9777

low risk

Vulkan builds now fail fast on shader compile errors instead of silently producing a broken backend.

Improvements

  • Vulkan shader generation now propagates subprocess exit codes (WEXITSTATUS on POSIX, GetExitCodeProcess on Windows) so failures are detected reliably.
  • Build aborts with EXIT_FAILURE before output files are written, preventing a broken libggml-vulkan from being shipped.
  • Launch exceptions and non-zero exits with empty stderr are now treated as compile failures rather than successes.

Bug fixes

  • Fixed silent Vulkan shader compilation failures that previously surfaced only at runtime (issue #24393).

Upgrade notes

No API or config changes; this only affects the Vulkan build path. If your CI previously passed despite broken shaders, expect it to fail now and surface the underlying compiler error. Rebuild Vulkan binaries to pick up the stricter check.

Analyzed Oct 31, 2026

Upgrading?

Release notes summary

Improves build reliability for the Vulkan backend by ensuring the build fails immediately if a shader fails to compile, rather than silently succeeding and producing broken binaries at runtime.

Key Highlights

  • Vulkan shader compilation failure detection
  • Build system robustness improvement
  • Fixes issue #24393

New Features

  • Vulkan build failure detection

Release Notes (excerpt)

<details open> vulkan: fail the build when a shader fails to compile (#24450) * vulkan-shaders-gen: fail the build when a shader fails to compile vulkan-shaders-gen did not detect shader-compile subprocess failures, so a broken libggml-vulkan could be produced while the build reported success and the breakage only surfaced at run time. execute_command() discarded the child exit code (POSIX waitpid passed nullptr for status; the Windows branch never…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories