Re: Automated report: NetBSD-current/i386 build success

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Tue,  6 Mar 2018 17:08:40 +0000 (UTC)
    From:        NetBSD Test Fixture <bracket%NetBSD.org@localhost>
    Message-ID:  <152035611931.24762.16467812075534166319%babylon5.NetBSD.org@localhost>

  | The NetBSD-current/i386 build is working again.
  |
  | The following commits were made between the last failed build and the
  | successful build:
  |
  |     2018.03.06.13.37.43 christos src/crypto/external/bsd/openssl/lib/ ...    [2 of them]

Perhaps needless to say that Christos' openssl commits did not fix the INET6 nd
bug reported in the "build failure" automated report mail - that was fixed 
(just before that report actualy) by Martin.

However, after his fix, the build failed again in a way I have observed in the 
b5 builds a few times recently (Christos' openssl changes did not fix this one 
either).

It is a gcc related compile error (I mean, compiling gcc) ...

--- dependall-gpl3 ---
/tmp/bracket/build/2018.03.06.11.21.31-i386/src/external/gpl3/gcc/dist/gcc/config/i386/sse.md:24127:47: error: control reaches end of non-void function [-Werror=return-type]
cc1plus: all warnings being treated as errors
*** [insn-latencytab.o] Error code 1
nbmake[9]: stopped in /tmp/bracket/build/2018.03.06.11.21.31-i386/src/external/gpl3/gcc/usr.bin/backend

This one seems to be non-deterministic - it comes and goes, which to me 
suggests something related to the parallel builds - or perhaps to the build
aborting from the previous run (due to some other error, the nd6 one in this
instance) leaving something "half done" in the gcc build, causing a subsequest
build to fail (but probably not, as I believe all the b5 builds start with clean 
objdirs.)

Does anyone who understands the gcc build process have any idea what
might be happening there?

What causes this to fail (always the same way, as best I can tell) sometimes,
and with no apparent related changes, be OK in a later build attempt?

kre