This directory contains crash dumps from clang on an RPI3 running 12.0-ALPHA9 
FreeBSD 12.0-ALPHA9 r339305 GENERIC  arm64
attempting a -j4 buildworld. 

This particular failure was from a -DNO_CLEAN restart of 
buildworld after a previous failure of
the same kind, but clean restarts don't seem to work any better.

It's been possible to update the kernel slightly using make 
kernel-toolchain, but buildworld seems to reliably fail, generating 
files of the same size as shown here. Helpful options were
kldload filemon
setenv WITH_META_MODE
setenv NO_SILENT
courtesy of Mark Millard. An urelated tweak was to set
security.bsd.stack_guard_page=0, mostly on a hint from
https://github.com/JuliaLang/julia/issues/23328
which was probably grasping at straws, but it didn't hurt. 

In the end, just restarting buildworld with -DNO_CLEAN eventually 
allowed buildworld to complete. Near as I could tell the tendency to
segfault didn't depend on -j value or swap usage. I did eventually
disconnect all usb swap, relying only on a ~4GB partition on microSD.

2/1/2019
The signal 11's in clang persisted up to 342987, at which point I 
reverted the system to 339376.  The initial make kernel-toolchain
stopped once. Restarted, it ran to completion and produced a bootable
kernel.