Re: Build failures in linux-4.9.y-stable-queue

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Fri, Dec 15, 2017 at 08:53:38AM +0100, Greg Kroah-Hartman wrote:
> On Fri, Dec 15, 2017 at 08:46:13AM +0100, Greg Kroah-Hartman wrote:
> > On Fri, Dec 15, 2017 at 08:43:32AM +0100, Greg Kroah-Hartman wrote:
> > > On Fri, Dec 15, 2017 at 08:42:25AM +0100, Greg Kroah-Hartman wrote:
> > > > On Thu, Dec 14, 2017 at 02:27:41PM -0800, Guenter Roeck wrote:
> > > > > Hi,
> > > > > 
> > > > > I thought the problem was fixed, but I still see:
> > > > > 
> > > > > arch/powerpc/include/asm/checksum.h:103:2: error:
> > > > > 	implicit declaration of function 'from64to32'
> > > > > 
> > > > > when building powerpc images (eg powerpc:defconfig).
> > > > > 
> > > > > This is with v4.9.69-20-g78542f2. v4.9.69 fails as well.
> > > > 
> > > > Yeah, I thought I took care of that already too.  I'm getting
> > > > conflicting reports from 0-day about this as well, it failed one build
> > > > cycle, then passed the next, and now failed again.  Ugh, let me try to
> > > > figure this out...
> > > 
> > > Ah, it passes powerpc builds on 0-day with the 'allnoconfig' build
> > > option, not with the defconfig, that explains that.
> > 
> > And the patch I tried to apply to fix this up didn't apply in the right
> > place, which is the problem here, let me go fix it now...
> > 
> > thanks for letting me know about this,
> 
> Ugh, this was bad.  Turns out that for the last release of 4.9 and 4.14,
> the last 5 or so patches never got applied when I created the git tree.
> They both stopped at the same place in the patch series, which is odd,
> but at least it means that they failed in the same way.

And now I know why.  They failed on the patch that modified the main
kernel Makefile, which makes sense as I hand-edit it to bump the version
number, so it was "dirty" and git would complain and stop applying
patches there.

I've "semi-scripted" the release process in the past few weeks, so an
error like this would not be caught, previously I would have noticed the
error as I applied the patches "by hand".  I'll go fix up my scripts to
not do this, and I'll go verify that I haven't dropped any other patches
like this recently either.

thanks,

greg k-h