ci: Run ctime_tests under Valgrind on arm64 #1975

pull fametrano wants to merge 1 commits into bitcoin-core:master from fametrano:ci-arm64-ctime changing 1 files +2 −2

fametrano commented at 8:43 AM on October 22, 2026: contributor

No CI job runs ctime_tests on ARM64. The Cirrus "ARM64: Linux (Debian stable), Valgrind" task did. #1689 moved it into valgrind_debian, whose job-wide CTIMETESTS: 'no' turned it off. This sets CTIMETESTS: 'yes' on the two arm64 entries.

I ran ci/ci.sh for both entries on aarch64 (Debian trixie, Valgrind 3.24, clang 19). Both pass. Both fail under Valgrind after I replace secp256k1_scalar_cond_negate in secp256k1_ecdsa_sig_sign with if (high) negate. Logs on request.

ci: Run ctime_tests under Valgrind on arm64

The Cirrus "ARM64: Linux (Debian stable), Valgrind" task ran
ctime_tests. 004f57fcd86d moved it into the `valgrind_debian` job,
whose `CTIMETESTS: 'no'` turned them off. No other arm64 job runs
ctime_tests.

26299c7672

real-or-random added the label assurance on Oct 22, 2026
real-or-random added the label ci on Oct 22, 2026
real-or-random added the label side-channel on Oct 22, 2026
real-or-random approved
real-or-random commented at 11:24 PM on October 22, 2026: contributor

ACK 26299c7672d4b61d2e17feecd97532d9abdb32c8

no idea why we missed this...
theStack approved
theStack commented at 3:44 AM on October 23, 2026: contributor

ACK 26299c7672d4b61d2e17feecd97532d9abdb32c8

Thanks for catching this. I think we should also bring back (either here or in a different PR) the ctime-tests on arm64 for gcc, where they historically have been the most relevant. Dropping the WITH_VALGRIND/CTIMETESTS overrides in the arm64-debian job seems to be one way to achieve that.

Contributors

fametrano

real-or-random

theStack

Labels

assurance ci side-channel

Linked (view graph)

#1689 ci: Convert `arm64` Cirrus tasks to GHA jobs