Regression testing buildworld on RPI3

[bob prohaska]

On Fri, Feb 01, 2019 at 02:08:03PM -0800, bob prohaska wrote:
> After much trial by error, it's looking as if -j4 buildworld runs 
> without error on r339376 and stops with signal 11 in cc somewhere
> near or after r342204. 
> 

On retry, buildworld on r339376 stopped with signal 11 in buildworld.

Just for fun, the system was updated to r343731. This time, the
kernel toolchain and kernel were updated first, but after reboot
the old toolchain was deleted and both world and kernel rebuilt
from scratch. Former practice was to build world, then kernel,
then reboot, so the toolchain was made with the old kernel. This
sequence was meant to ensure the new kernel supervised building
the whole system.

Buildworld seemed to take much longer than normal and both buildkernel
and buildworld log files seemed 5-10% larger than usual, but there
were no errors of any kind:

bob at www:/usr/src % ls -l *.log
-rw-r--r--  1 root  wheel   5605091 Feb  6 10:24 buildkernel.log
-rw-r--r--  1 root  wheel  35441641 Feb  6 08:15 buildworld.log
-rw-r--r--  1 root  wheel    212624 Feb  6 11:46 installkernel.log
-rw-r--r--  1 root  wheel   5809452 Feb  6 11:41 installworld.log

I don't understand why it works, but am glad it does work.

Thanks for reading, and any thoughts.

bob prohaska