svn commit: r255187 - in head/sys: conf crypto/aesni modules/aesni

[David O'Brien]

On Mon, Sep 09, 2013 at 11:41:49PM -0700, John-Mark Gurney wrote:
> David O'Brien wrote this message on Mon, Sep 09, 2013 at 23:01 -0700:
> Please back that out until you have an understand of what the real
> problem is...

Folks use FreeBSD for real work -- I have a window of opportunity to
install a new world + kernel.  When I tried updating this laptop in
early August the kernel build was broken.  It appeared broken again
this time.

The issue is you added a new feature to the GCC compiler
(r255185 2013-09-03), and then immediately consumed it in the kernel
build (r255187 2013-09-03 11:31:23).  You did not put anything in
UPDATING warning folks of this.


Robert Watson used to often remind folks that when we add features to
the toolchain we give a reasonable amount of time for them to work
their way into folks userland before using them in the kernel build.


My userland sources are from Sunday September 1st when I started my
'make buildworld' + 'mergemaster -p' + 'make installworld' +
'mergemaster -i' sequence.  Sorry it takes so long to build world with
clang on a T60.  I updated my kernel sources and tried to build a new
kernel before rebooting.  I had no reason to not expect a 2 day old
compiler could not build a kernel during a code freeze.

I've backed out r255440 as r255452.

Please add a note to UPDATING.

-- 
-- David  (obrien at FreeBSD.org)