So I'm still observing irreproducible segfaults from time to time in rustc -- at all stages (stage0, stage1, and stage2). I'm keeping a log of the backtraces as they occur here for now.
They seem to mostly (but not exclusively) occur in LLVM-related code.
Do you have LLVM assertions enabled… (actually might not be that helpful if it is not reliably reproducible)
hmm, I do not, but I can do so
but yeah it is not reproducible
usually I just run
x.py again and things work fine
Could run memtest overnight to rule out memory-related shenanigans