sech1: can you take a look at monero-project/monero #9041 ?

jh hash? That's from the old Cryptonight

I had several of these issues in XMRig too

It's compiler bug

Anyway, the test case he found should be probably added to the unit tests to avoid buggy compilers

100% a compiler bug, 9.4.0 works: paste.debian.net/hidden/8d516b15

Well, at least it was entertaining. I commented there.

Someone should probably file a bug to GCC, they still support GCC 11

Actually, I found the fix for it: xmrig/xmrig 75c57f7

I'll make a PR

Good news is we already have plenty of tests in tests/hash/tests-extra-jh.txt

sech1: ty, can you add Closes #9041 to the PR description?

Added

jeffro256, thanks. I opened the files and saw that these are parameters for each test case. However, I could not really grasp their semantic meaning. For example, is N the number of recipient public keys? is T the number of inputs? what is W? I ask it for test_sig_clsag and also for test_bulletproof_plus and test_construct_tx. I'll be grateful if you can provide me with such info, cause I could not figure out