Simon Jakobi pushed to branch wip/sjakobi/cfg-perf-test at Glasgow Haskell Compiler / GHC Commits: 7055fc01 by Simon Jakobi at 2026-06-03T17:57:11+02:00 testsuite: fix CmmControlFlowPerf to actually compile the .cmm The test passed a bare module name as top_mod to multimod_compile, so GHC ran `--make CmmControlFlowPerf` and failed to locate the module (.cmm sources aren't resolved from a bare module name). Even with the extension, `--make` links a non-existent main, and `-no-hs-main` does not suppress that. Use `compile` with the `cmm_src` setup instead: this resolves CmmControlFlowPerf.cmm and compiles with `-c` (no link), which is all a compiler perf test needs. Drop the now-unnecessary -no-hs-main. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> - - - - - 1 changed file: - testsuite/tests/perf/compiler/all.T Changes: ===================================== testsuite/tests/perf/compiler/all.T ===================================== @@ -830,10 +830,11 @@ test ('T26425', test('CmmControlFlowPerf', [ only_ways(['optasm']), unless(have_ncg(), skip), + cmm_src, collect_compiler_stats('bytes allocated', 5), collect_compiler_residency(15), pre_cmd('./genCmmControlFlowPerf'), extra_files(['genCmmControlFlowPerf']), ], - multimod_compile, - ['CmmControlFlowPerf', '-O -v0 -no-hs-main']) + compile, + ['-O -v0']) View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/7055fc01793352fa68cecd3c647adaea... -- View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/7055fc01793352fa68cecd3c647adaea... You're receiving this email because of your account on gitlab.haskell.org.