more boost unit test framework bugs. Was: Corrupt hardware on warp CI runners? #36026

I keep seeing a bunch of intermittent memory issues:

Like:

<!--85328a0da195eb286784d51f73fa0af9-->

🚧 At least one of the CI tasks failed. _{Task Alpine (musl): https://github.com/bitcoin/bitcoin/actions/runs/32290632833/job/96190361218} _{LLM reason (✨ experimental): CI failed because the test_kernel CTest job aborted during setup with an out-of-memory (exp: Out of memory) system_error.}

Originally posted by @DrahtBot in #36025 (comment)

Or:

🚧 At least one of the CI tasks failed. _{Task previous releases: https://github.com/bitcoin/bitcoin/actions/runs/32101345101/job/95602400388} _{LLM reason (✨ experimental): CI failed due to a GCC 12 internal compiler error (segmentation fault) while compiling src/rpc/blockchain.cpp.}

Edit: Looks like the ICE was due to a corrupt VM/hardware

Originally posted by @DrahtBot in #36005 (comment)

I have experienced this often on my fork, can you please check if it's related to https://github.com/l0rinc/bitcoin/pull/252? I can push it upstream if it helps.

Heh, I am not too excited to look into boost test, with the list of issues there growing. Can't we just remove it?

Edit: With removal I meant https://github.com/bitcoin/bitcoin/issues/34666

The reason I didn't push this upstream was exactly because it felt like magic that I wasn't comfortable with.

Can't we just remove it?

My understanding is that it affected our Alpine infra, maybe we can update those as a workaround...

Ok, took a look here. Seems fine to disable it, because I don't think we need it for anything. The semgrep analysis seems right, but I don't think we need any of the details. The altstack seems pretty useless or at least has questionable benefit?

  // Avoid test failures from undersized alternate signal stacks on some platforms.
  #define BOOST_TEST_DISABLE_ALT_STACK

Thanks for checking, pushed as #36045, added you as coauthor.