Re: [Qemu-devel] [RFC PATCH v5 00/52] Support Kconfig in QEMU

[Philippe Mathieu-Daudé]

Hi,

On 1/25/19 11:06 AM, Paolo Bonzini wrote:
> (I'm only momentarily at the helm and will give control back to Yang after
> this iteration.)
> 
> This is still RFC mostly because of the lack of documentation, and because
> only x86 is fully converted, but it's converging.  Other targets still
> enable embedded devices in default-configs/ instead of using "select"
> directives.  For many targets, the conversion will be trivial because
> they only support one board.  The complex ones are ARM, MIPS and PPC
> of course.  s390 as usual is just different in some respects, but all
> of its issues are sorted out already in this series and so it's just
> yet another single-board target.
> 
> It supports defconfig (default-configs file chooses boards only)
> and allnoconfig (default-configs file chooses devices too) and builds
> all targets.  I haven't yet checked that the configuration is the same
> before and after the conversion, but at least device-introspection-test
> and other qtests all pass, which did catch some errors.
> 
> As mentioned in the previous versions, this is only a replacement
> for default-configs, in order to simplify configuration and remove
> the need to track dependencies between configuration symbols.  In
> fact, even with the current incomplete conversion the diffstat
> for default-configs is already
> 
>  31 files changed, 108 insertions(+), 241 deletions(-)
> 
> Devices can be disabled by adding for example
> 
>    CONFIG_HPET=n
> 
> to default-configs/i386-softmmu.mak.  If you prefer they can be
> listed manually and "make allnoconfig" can be executed before
> building.  This probably should become a configure option
> "--without-default-devices" instead.
> 
> For the previous discussions on the Kconfig design, see
> http://lists.nongnu.org/archive/html/qemu-devel/2018-09/msg02827.html
> 
> Patches 1-27 should probably be committed now, so please review!  Yang,
> once they are reviewed you can extract them and post them to the
> mailing list as non-RFC!
> 
> As to the rest, you're welcome to try them, post conversions for the
> simple targets, suggest usability improvements, and whatever.  Thanks to
> everyone for the work on the previous iterations.  It's great to see
> the work on this GSoC project come back to life after five years!

How to express "depends of (TARGET_LONG_BITS > 32)"?