Re: [PULL 00/13] qtests, kconfig and misc patches

[Peter Maydell]

On Mon, 3 Feb 2020 at 14:30, Thomas Huth <address@hidden> wrote:
>
> On 03/02/2020 15.04, Peter Maydell wrote:
> > On Mon, 3 Feb 2020 at 12:38, Thomas Huth <address@hidden> wrote:
> >>
> >>  Hi Peter,
> >>
> >> the following changes since commit 
> >> 28db64fce555a03b4ca256d5b6f4290abdfbd9e8:
> >>
> >>   Merge remote-tracking branch 
> >> 'remotes/stefanha/tags/tracing-pull-request' into staging (2020-01-31 
> >> 17:37:00 +0000)
> >>
> >> are available in the Git repository at:
> >>
> >>   https://gitlab.com/huth/qemu.git tags/pull-request-2020-02-03
> >>
> >> for you to fetch changes up to 585c138628bbf22ea8e740b2f4f1a3ed0274ebe8:
> >>
> >>   trivial: Remove xenfb_enabled from sysemu.h (2020-02-03 10:33:57 +0100)
> >>
> >> ----------------------------------------------------------------
> >> * Current qtests queue
> >> * Some Kconfig updates
> >> * Some trivial clean-ups here and there
> >> ----------------------------------------------------------------
> >
> > All the incremental rebuilds failed:
> >
> > Linux cam-vm-266 4.15.0-70-generic x86_64
> > From git://git-us.linaro.org/people/pmaydell/qemu-arm
> >    f31160c7d1..f9e931a1d9  staging    -> pmaydell/staging
> > make: Entering directory '/home/petmay01/qemu-for-merges/build/w64'
> > make[1]: Entering directory '/home/petmay01/qemu-for-merges/slirp'
> > make[1]: Nothing to be done for 'all'.
> > make[1]: Leaving directory '/home/petmay01/qemu-for-merges/slirp'
> >   CC      qga/main.o
> >   CC      stubs/machine-init-done.o
> >   CC      stubs/replay-user.o
> >   CC      stubs/semihost.o
> >   CC      qemu-img.o
> >   CC      qemu-io.o
> >   CC      chardev/char.o
> > make: *** No rule to make target
> > '/home/petmay01/qemu-for-merges/hw/bt/Kconfig', needed by
> > 'aarch64-softmmu/config-devices.mak'.  Stop.
> > make: *** Waiting for unfinished jobs....
> >   CC      chardev/char-mux.o
> > make: Leaving directory '/home/petmay01/qemu-for-merges/build/w64'
>
> Oh, they are still failing??? Why are there still references to
> hw/bt/Kconfig in these config-devices.mak files, I'd expect that they
> would have been regenerated at least once during the past week?

build/all/aarch64-softmmu/config-devices.mak.d was most recently
touched this morning, and it still includes hw/bt/Kconfig in its
dependency list. I think this is because minikconf will still put
a Kconfig file into the .d file it generates even if the Kconfig
file happens to be empty.

And make doesn't have any rules that tell it that config-devices.mak.d
need to be updated either:
$ make -C build/all -n aarch64-softmmu/config-devices.mak.d
make: Entering directory '/home/petmay01/linaro/qemu-for-merges/build/all'
make[1]: Entering directory '/home/petmay01/linaro/qemu-for-merges/slirp'
make[1]: Nothing to be done for 'all'.
make[1]: Leaving directory '/home/petmay01/linaro/qemu-for-merges/slirp'
make: Nothing to be done for 'aarch64-softmmu/config-devices.mak.d'.
make: Leaving directory '/home/petmay01/linaro/qemu-for-merges/build/all'

or that it needs to rerun minikconf, which would update the .mak.d.

An extremely cheesy workaround would be if the commit which
removes the hw/bt/Kconfig also touches configure; then Make
will know it needs to rerun configure, which will (among
other things) blow away all the config-devices.mak.d and
force rerunning of minikconf.

I don't know what the correct additional makefile magic
would be that would cause us to automatically get deletion
of a Kconfig file right; maybe Paolo does?

thanks
-- PMM