On Tue, Jan 16, 2018 at 12:47:56PM +0100, Arnd Bergmann wrote:
> On Mon, Jan 15, 2018 at 2:30 PM, Olof's autobuilder <build@xxxxxxxxx> wrote:
> > Here are the build results from automated periodic testing.
> >
> > The tree being built was stable-rc, found at:
> >
> > URL: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git
> >
> > Branch: linux-3.18.y
> >
> > Topmost commits:
> > 4dc79b5 Linux 3.18.92-rc1
> > 00d0655 e1000e: Fix e1000_check_for_copper_link_ich8lan return value.
> > b944e64 uas: ignore UAS for Norelsys NS1068(X) chips
> >
> > Runtime: 28m 37s
> >
> > Passed: 131
> >
> > Warnings: 20680
> >
>
> > No errors
> >
> >
> > Warnings:
> >
> > arm64.allmodconfig:
> > /tmp/ccmFH2sM.s:77: Warning: ignoring incorrect section type for .init_array.00100
> > /tmp/ccmFH2sM.s:90: Warning: ignoring incorrect section type for .fini_array.00100
> > /tmp/ccCNAAf2.s:50: Warning: ignoring incorrect section type for .init_array.00100
> > /tmp/ccCNAAf2.s:63: Warning: ignoring incorrect section type for .fini_array.00100
> > /tmp/cckbGzsX.s:337: Warning: ignoring incorrect section type for .init_array.00100
> > /tmp/cckbGzsX.s:350: Warning: ignoring incorrect section type for .fini_array.00100
>
> This is the result of a bug in the assembler that has since been
> fixed. The warning
> itself is apparently harmless, but it's annoying to get 20000 lines of
> warnings for
> a simple allmodconfig build. I would suggest backporting commit
>
> cc622420798c ("gcov: disable for COMPILE_TEST")
>
> to all stable kernels 3.18, 4.1 and 4.4 that are affected by this.
> Olof's build bot
> only reported it for 3.18, but my interpretation is that he uses an
> older toolchain
> for that kernel, which triggers this warning, while newer assemblers are fixed.
>
> The warning showed up in the past few days after Olof's build scripts
> got adapted
> to also report assembler warnings, rather than just compiler warnings.
>
> The intention of the cc622420798c commit was to help with other issues of
> compile testing, fixing this particular warning was an unintended side-effect.
> Adding it to stable kernels will also help with the other issues it addressed at
> the time, in particular CPU usage during 'allmodconfig' build testing.
Good idea, I had to turn that off in my local build configs by hand,
this is much nicer :)
Now queued up, thanks.
greg k-h