Re: Build regressions/improvements in v2.6.39-rc3

From: Michael Ellerman
Date: Wed Apr 13 2011 - 21:26:25 EST


On Thu, 2011-04-14 at 10:05 +1000, Michael Ellerman wrote:
> On Wed, 2011-04-13 at 19:38 +0200, Geert Uytterhoeven wrote:
> > On Wed, Apr 13, 2011 at 19:32, David Miller <davem@xxxxxxxxxxxxx> wrote:
> > > From: Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>
> > > Date: Wed, 13 Apr 2011 18:35:45 +0200 (CEST)
> > >
> > >> 47 regressions:
> > >> + build/linus_sparc64-allmodconfig_sparc64/usr/include/linux/ethtool.h: error: expected specifier-qualifier-list before '__u32': 104, 90, 21, 282, 389, 460, 521, 444, 96, 297, 82, 57, 291, 258, 231, 482, 209, 510, 265, 113 => 104, 538, 90, 21, 523, 552, 266, 445, 292, 96, 82, 512, 563, 57, 259, 283, 298, 484, 390, 231, 209, 462, 574, 113
> > >
> > > I do allmodconfig builds on sparc64 systems every single day using
> > > various versions of gcc-4.4, gcc-4.5, and gcc-4.6
> > >
> > > And I never ever see this build error. Never.
> > >
> > > If you're using a cross build environment, maybe something isn't quite
> > > right with that?
> >
> > As it mentions `/usr/include/linux/ethtool.h', it seems to be an
> > include that's part of the
> > cross build environment.
> >
> > I'm just using the logs from the linux-next build service...
>
> Which is just using the cross compilers that Tony builds ...
>
> I'll have a look and see if we can work out what the problem is.

So allmodconfig enables CONFIG_BUILD_DOCSRC, which builds the code in
Documentation, which is userspace code.

In our case we know that won't work because our cross tool chain doesn't
have libc, so I've taught kisskb to disable CONFIG_BUILD_DOCSRC for
allmodconfig builds.

cheers

--
Michael Ellerman
OzLabs, LTC Prism Kernel Team

email: michaele@xxxxxxxxxx
stime: ellerman@xxxxxxxxxxx
notes: Michael Ellerman/Australia/IBM

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/