On 11/12/26 6:30 AM, Daniel P. Berrangé wrote:
When building a multi-target binary, QOM types from multiple targets
will be linked together. QOM types are required to have unique names
when registered, so linking multiple targets in turn requires QOM type
names to be unique across the QEMU codebase.

Global QOM type name uniqueness is a desirable property for QEMU
regardless of multi-target work, so that QEMU presents a non-ambiguous
object model to applications regardless of context.

In a multi-target binary, it is desirable to applications are able to
introspect all types linked into QEMU, when launching with '-machine
none'. Dynamic filtering of registered classes at runtime, to resolving
QOM naming conflicts is undesirable, as it requires introspection
to be repeated for each available target in order to discover the full
set of available types in a binary.

Introspection is a fairly heavy weight process due to QEMU's vast size,
so enabling introspection to be done once only for all targets will be a
very significant benefit for management applications with a multi-target
binary.

The glaring problem is that our historical design has led to multiple
targets all providing a 'virt' machine type, which leads to multiple QOM
types call 'virt-machine' which are all different.

Having a common short name "virt" is a useful characteristic for users,
but this does not imply we need to use that as the basis for primary QOM
typename. It is sufficient to have an alias called 'virt'.

In terms of implementation there are two core pieces

  * An 'Aliased' interface which can be implemented by QOM types which
    want to support type aliases.

  * A 'TargetSpecific' interface which can be implemented by QOM types
    which want to declare  they can only be used with a subset of targets.

With those foundational pieces

  * Machine implements Aliased and TargetSpecific

  * query-machines reports multiple aliases and declared supported
    targets per machine for introspection

  * All the 'virt' machines are renamed to $target-virt, and aliases
    added for the shorter 'virt' names to retain compatibility and
    convenience when a target is selected

  * All the machines which have 'is_available' set now report their
    targets via the TargetSpecific  interface

  * Creating machines validates declared targets from TargetSpecific

An apparent design mistake in TYPE_ACCEL_CPU is resolved which allows
removal of the is_available hooks from target specific CPU impls, which
were working around the bug.

The original proposed patches to unconditionally register some object
properties is pulled in.

Overall this brings us closer to having a single unified object model
across all targets, without having to make runtime variations of what
is exposed/reported by a given QOM class name.

Daniel P. Berrangé (43):
   iotests: better heuristics to find I/O tests build/source dirs
   qom: define a TargetSpecific interface
   qom: define an Aliased interface
   hw: tweak approach to setting alias in machine macros
   hw/core: implement TYPE_ALIASED interface in machines
   qapi: report multiple aliases in MachineInfo
   system: report multiple machine aliases in help output
   system: increase space for machine name in '-machine help'
   system: resolve multiple aliases when creating machines
   qapi: deprecate 'alias' field in MachineInfo
   hw/arm: convert machines to use AliasedClass interface
   hw/avr: convert machines to use AliasedClass interface
   hw/hexagon: convert machines to use AliasedClass interface
   hw/i386: convert machines to use AliasedClass interface
   hw/loongarch: convert machines to use AliasedClass interface
   hw/m68k: convert machines to use AliasedClass interface
   hw/ppc: convert machines to use AliasedClass interface
   hw/s390x: convert machines to use AliasedClass interface
   include/hw: convert machine docs to use AliasedClass interface
   hw: drop 'alias' field from MachineClass
   hw/arm: rename the 'virt' machines to 'arm-virt'
   hw/riscv: rename the 'virt' machines to 'riscv-virt'
   hw/hexagon: rename the 'virt' machine to 'hexagon-virt'
   hw/loongarch: rename the 'virt' machine to 'loongarch-virt'
   hw/m68k: rename the 'virt' machine to 'm68k-virt'
   hw/or1k: rename the 'virt' machine to 'or1k-virt'
   hw/xtensa: rename the 'virt' machine to 'xtensa-virt'
   system: stop sorting machine names based on 'family'
   system: put machine aliases in alphbetical order
   system: filter machine type list if a target is selected
   hw/core: filter machine type list if a target is selected
   system: filter machines based on target during creation
   hw/core: implement TYPE_TARGET_SPECIFIC for machine
   qapi: report target compatibility for machine types
   hw/arm: declare arm machine targets
   hw/hexagon: declare hexagon machine target
   hw/m68k: declare m68k machine target
   hw/or1k: declare or1k machine target
   hw/riscv: declare riscv machine target
   accel: make TYPE_ACCEL_CPU name target independent
   Revert "multi-target: restrict whpx-accel-*-cpu to matching target"
   Revert "target/riscv/kvm: restrict kvm-accel-riscv-cpu to riscv
     targets"
   qom: drop the 'is_available' callback from TypeInfo

Pierrick Bouvier (1):
   Revert "target/riscv/tcg: restrict tcg-accel-riscv-cpu to riscv
     targets"

Yonggang Luo (4):
   target-info: add target_mask_selected()
   hw/pci: unconditionally expose the 'loadparm' property
   hw/scsi: unconditionally expose the 'loadparm' property
   hw/virtio/virtio-mem: unconditionally expose 'unplugged-inaccessible'

  accel/accel-common.c             |   3 +-
  accel/whpx/whpx-common.c         |   8 --
  docs/about/deprecated.rst        |   7 ++
  hw/arm/aspeed_ast27x0-fc.c       |   5 +-
  hw/arm/aspeed_ast27x0_evb.c      |  14 ++-
  hw/arm/raspi.c                   |   9 +-
  hw/arm/raspi4b.c                 |   6 +-
  hw/arm/virt.c                    |  43 +++++++--
  hw/arm/xlnx-versal-virt.c        |   7 +-
  hw/avr/arduino.c                 |  30 +++++-
  hw/core/machine-qmp-cmds.c       |  52 ++++++++++-
  hw/core/machine.c                |   5 +
  hw/hexagon/hexagon_dsp.c         |   7 +-
  hw/hexagon/virt.c                |  11 ++-
  hw/i386/pc_piix.c                |  25 ++++-
  hw/i386/pc_q35.c                 |  20 +++-
  hw/loongarch/virt.c              |  39 ++++++--
  hw/m68k/virt.c                   |  46 +++++++---
  hw/or1k/virt.c                   |  14 ++-
  hw/pci/pci.c                     |  18 ++--
  hw/ppc/pnv.c                     |   7 +-
  hw/ppc/spapr.c                   |  31 ++++---
  hw/riscv/virt.c                  |  14 ++-
  hw/s390x/s390-virtio-ccw.c       |  24 +++--
  hw/scsi/scsi-disk.c              |  18 ++--
  hw/virtio/virtio-mem.c           |  12 +--
  hw/xtensa/virt.c                 |   9 +-
  include/accel/accel-cpu-target.h |   3 +-
  include/accel/accel-cpu.h        |   2 +
  include/hw/arm/virt.h            |   2 +-
  include/hw/core/boards.h         |  29 ++++--
  include/hw/hexagon/virt.h        |   2 +-
  include/hw/i386/pc.h             |   5 +-
  include/hw/loongarch/virt.h      |   2 +-
  include/hw/riscv/virt.h          |   2 +-
  include/qemu/target-info-init.h  |  15 ++-
  include/qemu/target-info.h       |   9 ++
  include/qom/object.h             |   3 -
  include/qom/object_interfaces.h  | 152 +++++++++++++++++++++++++++++++
  meson.build                      |   1 +
  qapi/machine.json                |  19 +++-
  qom/object.c                     |   4 -
  qom/object_interfaces.c          |  88 ++++++++++++++++++
  rust/qom/src/qom.rs              |   1 -
  stubs/meson.build                |   1 +
  stubs/target-mask-selected.c     |  13 +++
  system/vl.c                      | 115 ++++++++++++-----------
  target-info-def.c                |   3 +
  target-info-qom.c                |  10 ++
  target/riscv/kvm/kvm-cpu.c       |   2 -
  target/riscv/tcg/tcg-cpu.c       |   1 -
  tests/qemu-iotests/check         |  52 ++++++++---
  52 files changed, 804 insertions(+), 216 deletions(-)
  create mode 100644 stubs/target-mask-selected.c


Hi Daniel,

thanks for taking a look at the whole problem, and sending this great series. I was busy for KVM Forum this week, and I'll be happy to take a look at it this week. Overall, it looks clean and the right thing to do.

By the way, I noticed one configuration is not building: oss-fuzz.

../tests/qtest/fuzz/qos_fuzz.c:49:62: error: too few arguments to function call, expected 5, have 3
   49 |     mach_info = qmp_query_machines(false, false, &error_abort);
      |                 ~~~~~~~~~~~~~~~~~~                           ^
./qapi/qapi-commands-machine.h:22:18: note: 'qmp_query_machines' declared here 22 | MachineInfoList *qmp_query_machines(bool has_compat_props, bool compat_props, bool has_all_targets, bool all_targets, Error **errp);
      |

It's a very small nit issue, and not worth respinning for that right now, just mentioning it.

Regards,
Pierrick

Reply via email to