Skip to content

fix: extend CCE 19.0.0 -Oipa0 workaround to m_mpi_common.fpp#1409

Merged
sbryngelson merged 11 commits intomasterfrom
fix/cce19-ipa-m_mpi_common
May 9, 2026
Merged

fix: extend CCE 19.0.0 -Oipa0 workaround to m_mpi_common.fpp#1409
sbryngelson merged 11 commits intomasterfrom
fix/cce19-ipa-m_mpi_common

Conversation

@sbryngelson
Copy link
Copy Markdown
Member

Summary

  • m_mpi_common.fpp triggers the same CCE 19.0.0 optcg IPA crash (bring_routine_resident SIGSEGV) as m_phase_change.fpp, consistently failing the Oak Ridge | Frontier (cpu) CI job (e.g. run 25553342717)
  • Adds -Oipa0 for pre_process/m_mpi_common.fpp.f90 (confirmed failing target) and simulation/m_mpi_common.fpp.f90 (precautionary — pre_process fails first, masking whether simulation also crashes)
  • Same workaround pattern as PR Work around CCE 19.0.0 compiler bugs for Cray+OpenACC builds #1286; all 6 precheck tests pass; build verified on Frontier batch node with CCE 19.0.0

Test plan

  • Oak Ridge | Frontier (cpu) CI job passes (this is the only currently failing job on master)
  • Oak Ridge | Frontier (gpu-acc) CI job still passes (existing workaround unchanged)
  • All other CI jobs unaffected

m_mpi_common.fpp triggers the same bring_routine_resident SIGSEGV in
optcg during ipa_processing as m_phase_change.fpp. Adds -Oipa0 for the
pre_process build (where CI consistently fails) and for simulation as a
precaution, since pre_process fails first and blocks visibility of whether
simulation also crashes.
@qodo-code-review
Copy link
Copy Markdown
Contributor

ⓘ You've reached your Qodo monthly free-tier limit. Reviews pause until next month — upgrade your plan to continue now, or link your paid account if you already have one.

@codecov
Copy link
Copy Markdown

codecov Bot commented May 8, 2026

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 64.90%. Comparing base (b3fcc2a) to head (2f83f0a).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files
@@           Coverage Diff           @@
##           master    #1409   +/-   ##
=======================================
  Coverage   64.90%   64.90%           
=======================================
  Files          72       72           
  Lines       18874    18874           
  Branches     1571     1571           
=======================================
  Hits        12250    12250           
  Misses       5649     5649           
  Partials      975      975           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

Spencer Bryngelson added 10 commits May 8, 2026 15:42
…ARY_PATH ordering

On Frontier, the compute-node build step previously inherited whatever
CMake state the login-node --deps-only step left behind. Wipe compiled
Fortran staging/install slugs before each build while preserving dep
dirs (silo, hdf5, lapack, fftw) that require internet to fetch.

Also remove trailing whitespace in famd LD_LIBRARY_PATH export.
Per-file set_source_files_properties is insufficient because the IPA pass
inlines across all files in the target — any file can become the crash site.
Apply -Oipa0 target-wide for pre_process and simulation (Cray, not OpenMP).
OpenMP excluded: thermochem requires IPA for !DIR$ INLINEALWAYS device calls.
@sbryngelson sbryngelson merged commit 141ff01 into master May 9, 2026
83 of 84 checks passed
@sbryngelson sbryngelson deleted the fix/cce19-ipa-m_mpi_common branch May 9, 2026 02:37
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Development

Successfully merging this pull request may close these issues.

1 participant