Channelpadding tests rely on actual time; can fail
We're seeing an intermittent failure on Jenkins:
14:46:53 channelpadding/channelpadding_consensus: [forking]
14:46:53 FAIL ../tor/src/test/test_channelpadding.c:445: assert(decision OP_EQ CHANNELPADDING_PADDING_SCHEDULED): 4 vs 2
14:46:53 [channelpadding_consensus FAILED]
Looking at the code, it seems that the underlying channelpadding code depends on the actual time (from monotime_coarse_abosolute_*()
) to make its decisions. But we aren't doing anything to mock those functions from inside the test case, which may be making the outcome of this test dependent on the code running fast enough.