1. Patchew
  2. QEMU
  3. View series

[PATCH v16 00/14] macOS PV Graphics and new vmapple machine type

Phil Dennis-Jordan posted 14 patches 1 year, 1 month ago
Failed in applying to current master (apply log)
MAINTAINERS                         |  15 +
contrib/vmapple/uuid.sh             |   9 +
docs/system/arm/vmapple.rst         |  63 ++
docs/system/target-arm.rst          |   1 +
hw/Kconfig                          |   1 +
hw/arm/sbsa-ref.c                   |   2 +-
hw/arm/virt.c                       |   2 +-
hw/block/virtio-blk.c               |  58 +-
hw/core/qdev-properties-system.c    |   8 +
hw/display/Kconfig                  |  13 +
hw/display/apple-gfx-mmio.m         | 288 +++++++++
hw/display/apple-gfx-pci.m          | 156 +++++
hw/display/apple-gfx.h              |  77 +++
hw/display/apple-gfx.m              | 880 ++++++++++++++++++++++++++++
hw/display/meson.build              |   7 +
hw/display/trace-events             |  30 +
hw/i386/microvm.c                   |   2 +-
hw/loongarch/virt.c                 |  12 +-
hw/meson.build                      |   1 +
hw/mips/loongson3_virt.c            |   2 +-
hw/misc/Kconfig                     |   4 +
hw/misc/meson.build                 |   1 +
hw/misc/pvpanic-mmio.c              |  60 ++
hw/openrisc/virt.c                  |  12 +-
hw/pci-host/gpex.c                  |  43 +-
hw/riscv/virt.c                     |  12 +-
hw/vmapple/Kconfig                  |  32 +
hw/vmapple/aes.c                    | 581 ++++++++++++++++++
hw/vmapple/bdif.c                   | 274 +++++++++
hw/vmapple/cfg.c                    | 195 ++++++
hw/vmapple/meson.build              |   5 +
hw/vmapple/trace-events             |  21 +
hw/vmapple/trace.h                  |   1 +
hw/vmapple/virtio-blk.c             | 204 +++++++
hw/vmapple/vmapple.c                | 612 +++++++++++++++++++
hw/xen/xen-pvh-common.c             |   2 +-
hw/xtensa/virt.c                    |   2 +-
include/hw/misc/pvpanic.h           |   1 +
include/hw/pci-host/gpex.h          |   7 +-
include/hw/pci/pci_ids.h            |   1 +
include/hw/qdev-properties-system.h |   5 +
include/hw/virtio/virtio-blk.h      |  11 +-
include/hw/vmapple/vmapple.h        |  23 +
include/qemu-main.h                 |  14 +-
include/qemu/cutils.h               |  15 +
meson.build                         |   5 +
qapi/virtio.json                    |  14 +
system/main.c                       |  37 +-
ui/cocoa.m                          |  54 +-
ui/gtk.c                            |   4 +
ui/sdl2.c                           |   4 +
util/hexdump.c                      |  18 +
52 files changed, 3791 insertions(+), 110 deletions(-)
create mode 100755 contrib/vmapple/uuid.sh
create mode 100644 docs/system/arm/vmapple.rst
create mode 100644 hw/display/apple-gfx-mmio.m
create mode 100644 hw/display/apple-gfx-pci.m
create mode 100644 hw/display/apple-gfx.h
create mode 100644 hw/display/apple-gfx.m
create mode 100644 hw/misc/pvpanic-mmio.c
create mode 100644 hw/vmapple/Kconfig
create mode 100644 hw/vmapple/aes.c
create mode 100644 hw/vmapple/bdif.c
create mode 100644 hw/vmapple/cfg.c
create mode 100644 hw/vmapple/meson.build
create mode 100644 hw/vmapple/trace-events
create mode 100644 hw/vmapple/trace.h
create mode 100644 hw/vmapple/virtio-blk.c
create mode 100644 hw/vmapple/vmapple.c
create mode 100644 include/hw/vmapple/vmapple.h
[PATCH v16 00/14] macOS PV Graphics and new vmapple machine type
Posted by Phil Dennis-Jordan 1 year, 1 month ago 
This patch set introduces a new ARM and macOS HVF specific machine type
called "vmapple", as well as a family of display devices based on the
ParavirtualizedGraphics.framework in macOS. One of the display adapter
variants, apple-gfx-mmio, is required for the new machine type, while
apple-gfx-pci can be used to enable 3D graphics acceleration with x86-64
macOS guest OSes.

Previous versions of this patch set were submitted semi-separately:
the original vmapple patch set by Alexander Graf included a monolithic
implementation of apple-gfx-mmio. I subsequently reviewed and reworked
the latter to support the PCI variant of the device as well and submitted
the result in isolation. As requested in subsequent review, I have now
recombined this with the original vmapple patch set, which I have updated
and improved in a few ways as well.

The vmapple machine type approximates the configuration in macOS's own
Virtualization.framework when running arm64 macOS guests. In addition to
generic components such as a GICv3 and an XHCI USB controller, it
includes nonstandard extensions to the virtio block device, a special
"hardware" aes engine, a configuration device, a pvpanic variant, a
"backdoor" interface, and of course the apple-gfx paravirtualised display
adapter.

There are currently a few limitations to this which aren't intrinsic,
just imperfect emulation of the VZF, but it's good enough to be just
about usable for some purposes:

 * macOS 12 guests only. Versions 13+ currently fail during early boot.
 * macOS 11+ arm64 hosts only, with hvf accel. (Perhaps some differences
   between Apple M series CPUs and TCG's aarch64 implementation? macOS
   hosts only because ParavirtualizedGraphics.framework is a black box
   implementing most of the logic behind the apple-gfx device.)
 * PCI devices use legacy IRQs, not MSI/MSI-X. As far as I can tell,
   we'd need to include the GICv3 ITS, but it's unclear to me what
   exactly needs wiring up.
 * Due to a quirk (bug?) in the macOS XHCI driver when MSI-X is not
   available, correct functioning of the USB controller (and thus
   keyboard/tablet) requires a small workaround in the XHCI controller
   device. This is part of another patch series:
   https://patchew.org/QEMU/20241208191646.64857-1-phil@philjordan.eu/
 * The guest OS must first be provisioned using Virtualization.framework;
   the disk images can subsequently be used in Qemu. (See docs.)

The apple-gfx device can be used independently from the vmapple machine
type, at least in the PCI variant. It mainly targets x86-64 macOS guests
from version 11 on, but also includes a UEFI bootrom for basic
framebuffer mode. macOS 11 is also required on the host side, as well
as a GPU that supports the Metal API. On the guest side, this provides
3D acceleration/GPGPU support with a baseline Metal feature set,
irrespective of the host GPU's feature set. A few limitations in the
current integration:

 * Although it works fine with TCG, it does not work correctly
   cross-architecture: x86-64 guests on arm64 hosts appear to make
   some boot progress, but rendering is corrupted. I suspect
   incompatible texture memory layouts; I have no idea if this is
   fixable.
 * ParavirtualizedGraphics.framework and the guest driver support
   multi-headed configurations. The current Qemu integration always
   connects precisely 1 display.
 * State serialisation and deserialisation is currently not
   implemented, though supported in principle by the framework.
   Both apple-gfx variants thus set up a migration blocker.
 * Rendering efficiency could be better. The GPU-rendered guest
   framebuffer is copied to system memory and uses Qemu's usual
   CPU-based drawing. For maximum efficiency, the Metal texture
   containing the guest framebuffer could be drawn directly to
   a Metal view in the host window, staying on the GPU. (Similar
   to the OpenGL/virgl render path on other platforms.)

Some of my part of this work has been sponsored by Sauce Labs Inc.

---

v2 -> v3:

 * Merged the apple-gfx and vmapple patchsets.
 * Squashed a bunch of later apple-gfx patches into the main one.
   (dGPU support, queried MMIO area size, host GPU picking logic.)
 * Rebased on latest upstream, fixing any breakages due to internal
   Qemu API changes.
 * apple-gfx: Switched to re-entrant MMIO. This is supported by the
   underlying framework and simplifies the MMIO forwarding code which
   was previously different on x86-64 vs aarch64.
 * vmapple: Fixes for minor bugs and comments from the last round of
   review.
 * vmapple aes, conf, apple-gfx: Switched reset methods to implement
   the ResettableClass base's interface.
 * vmapple: switched from virtio-hid to an XHCI USB controller and
   USB mouse and tablet devices. macOS does not provide drivers for
   virtio HID devices, at least not in version 12's vmapple kernel.
   So input now sort of works (interrupt issues) rather than not
   at all. Use network-based remote access to the guest OS as a
   work-around.

v3 -> v4:

 * Complete rework of the mechanism for handling runloop/libdispatch
   events on the main thread. PV graphics now work with the SDL UI.
 * Renamed 'apple-gfx-vmapple' device to 'apple-gfx-mmio'
 * hw/display/apple-gfx: threading model overhaul to be more consistent,
   safer, and more QEMU-idiomatic.
 * display-modes property on the apple-gfx devices now uses the
   native array property mechanism and works on both device variants.
 * hw/vmapple/aes: Improvements to logging and error handling.
 * hw/vmapple/cfg: Bug fixes around device property default values.
 * hw/vmapple/{aes,cfg,virtio-blk/vmapple}: Most header code moved into
   .c files, only a single vmapple.h now contains the #defines for the
   vmapple machine model-specific device type names.
 * hw/block/virtio-blk: New patch for replacing virtio_blk_free_request
   with g_free. (Optional)
 * Various smaller changes following comments in v3 code review in
   apple-gfx, aes, cfg, bdif, virtio-blk-vmapple, and the vmapple
   machine type itself. See patch-specific v4 change notes for details.

v4 -> v5:

 * Simplified the main thread runloop mechanism. Back to setting
	 qemu_main directly, but narrowing the scope of what it needs to do,
	 and it can now be NULL. (Meaning run the QEMU main event loop on
	 the main thread as is traditional.)
 * hw/display/apple-gfx: Further improvements to the BH based job code bridging
   the libdispatch & QEMU thread synchronisation impedance mismatch.
 * hw/display/apple-gfx: Thread safety and object lifetime improvements.
 * hw/display/apple-gfx-*: Better buffer and error handling in display mode
   property setters and getters.
 * hw/vmapple/aes: More consistent and safer logging/tracing
 * hw/vmapple/cfg: Better error reporting on overlong property strings.
 * hw/vmapple/virtio-blk: Fixed theoretically-unaligned write to config buffer.
 * vmapple machine type: Moved ecam region into machine state, improved device
   property setting error handling, improved ECID/UUID extraction script and
   docs.
 * Various smaller fixes in apple-gfx/-mmio, apple-gfx-pci, vmapple/aes,
   vmapple/cfg, vmapple/virtio-blk, and vmapple machine type.
 * Added SPDX license identifiers where they were missing.

v5 -> v6:

 * 01/15 (main/Cocoa/runloop): Combined functions, fixed whitespace
 * 02/15 (apple-gfx): Further refinement of PVG threading: reduced some callback
   tasks from BHs to merely acquiring RCU read lock; replaced some libdispatch
   tasks with BHs; last remaining synchronous BH now uses emphemeral
   QemuSemaphore.
 * 02/15 (apple-gfx): Readability improvements and other smaller tweaks
   (see patch change notes for details)
 * 04/15 (display modes): Replaced use of alloca() with NSMutableArray.

v6 -> v7:

 * 02/15 (apple-gfx): Use g_ptr_array_find() helper function, coding style tweak
 * 03/15 (apple-gfx-pci): Removed an unused function parameter
 * 04/15 (apple-gfx display mode property): Simplified error handling in
   property parsing.
 * 10/15 (vmapple/aes): Coding style tweaks.
 * 12/15 (vmapple/cfg): Changed error messages for overrun of properties with
   fixed-length strings to be more useful to users than developers.
 * 15/15 (vmapple machine type): Tiny error handling fix, un-inlined function

v7 -> v8:

 * 02/15 (apple-gfx): Naming and type use improvements, fixes for a bug and a
   leak.
 * 04/15 (apple-gfx display mode property): Type use improvement
 * 10/15 (vmapple/aes): Guest error logging tweaks.
 * 11/15 (vmapple/bdif): Replaced uses of cpu_physical_memory_read with
   dma_memory_read, and a g_free call with g_autofree.
 * 12/15 (vmapple/cfg): Macro hygiene fix: consistently enclosing arguments in
   parens.
 * 15/15 (vmapple machine type): Use less verbose pattern for defining uuid
   property.

v8 -> v9:

 * 01/16 (ui & main loop): Set qemu_main to NULL for GTK UI as well.
 * 02/16 (apple-gfx): Pass device pointer to graphic_console_init(), various
	 non-functional changes.
 * 03/16 (apple-gfx-pci): Fixup of changed common call, whitespace and comment
   formatting tweaks.
 * 04/16 (apple-gfx display modes): Re-ordered type definitions so we can drop
   a 'struct' keyword.
 * 10/16 (vmapple/aes): Replaced a use of cpu_physical_memory_write with
   dma_memory_write, minor style tweak.
 * 11/16 (vmapple/bdif): Replaced uses of cpu_physical_memory_write with
   dma_memory_write.
 * 13/16 (vmapple/virtio-blk): Correctly specify class_size for
   VMAppleVirtIOBlkClass.
 * 15/16 (vmapple machine type): Documentation improvements, fixed variable
   name and struct field used during pvpanic device creation.
 * 16/16 (NEW/RFC vmapple/virtio-blk): Proposed change to replace type hierarchy
   with a variant property. This seems cleaner and less confusing than the
   original approach to me, but I'm not sure if it warrants creation of a new
   QAPI enum and property type definition.

v9 -> v10:

 * 01/15 (ui & main loop): Added comments to qemu_main declaration and GTK.
 * 02/15 (apple-gfx): Reworked the way frame rendering code is threaded to use
   BHs for sections requiring BQL.
 * 02/15 (apple-gfx): Fixed ./configure error on non-macOS platforms.
 * 10/15 (vmapple/aes): Code style and comment improvements.
 * 12/15 (vmapple/cfg): Slightly tidier error reporting for overlong property
   values.
 * 13/15 (vmapple/virtio-blk): Folded v9 patch 16/16 into this one, changing
   the device type design to provide a single device type with a variant
	 property instead of 2 different subtypes for aux and root volumes.
 * 15/15 (vmapple machine type): Documentation fixup for changed virtio-blk
   device type; small improvements to shell commands in documentation;
   improved propagation of errors during cfg device instantiation.

v10 -> v11:

 * 01/15 (ui & main loop): Simplified main.c, better comments & commit message
 * 02/15 (apple-gfx): Give each PV display instance a unique serial number.
 * 02 & 03/15 (apple-gfx, -pci): Formatting/style tweaks
 * 15/15 (vmapple machine type): Improvements to shell code in docs

v11 -> v12:

 * 01/15 (ui & main loop): More precise wording of code comments.
 * 02/15 (apple-gfx): Fixed memory management regressions introduced in v10;
   improved error handling; various more conmetic code adjustments
 * 09/15 (GPEX): Fixed uses of deleted GPEX_NUM_IRQS constant that have been
   added to QEMU since this patch was originally written.

v12 -> v13:

 * 15/15 (vmapple machine type): Bumped the machine type version from 9.2
   to 10.0.
 * All patches in the series now have been positively reviewed and received
   corresponding reviewed-by tags.

v13 -> v14:

 * 6/15 (hw/vmapple directory): Changed myself from reviewer
   to maintainer, as that seemed appropriate at this point.
 * 15/15 (vmapple machine type): Gate creation of XHCI and
   USB HID devices behind if (defaults_enabled()).

v14 -> v15

 * Constified property tables to match Richard Henderson's recent project-
   wide convention change. (patches 4/15, 7/15, 11/15, 12/15, & 13/15)

v15 -> v16

 * 14 patches now, as patch 8 has already been pulled. (Thanks Philippe!)
 * Fixed a bunch of conflicts with upstream code motion:
    - DEFINE_PROP_END_OF_LIST removal (4/14 - apple-gfx mode list, 7/14 -
      pvpanic-mmio, 10/14 - bdif, 11/14 - cfg device, and
      12/14 - vmapple-virtio-blk)
    - sysemu->system move/rename: (1/14 - ui/qemu-main, 2/14 - apple-gfx,
      9/14 - aes, 10/14 - bdif, 14/14 - vmapple machine type)
 * 14/14 (vmapple machine type):
    - Moved compatibility setting for removing legacy mode from virtio-pci
      to proper global property table rather than (ab)using sugar property.
    - Removed a few superfluous #includes during sysemu rename cleanup.
    - Removed machine type versioning as it's not necessary (yet?)
    - Made memory map array const

Alexander Graf (8):
  hw: Add vmapple subdir
  hw/misc/pvpanic: Add MMIO interface
  gpex: Allow more than 4 legacy IRQs
  hw/vmapple/aes: Introduce aes engine
  hw/vmapple/bdif: Introduce vmapple backdoor interface
  hw/vmapple/cfg: Introduce vmapple cfg region
  hw/vmapple/virtio-blk: Add support for apple virtio-blk
  hw/vmapple/vmapple: Add vmapple machine type

Phil Dennis-Jordan (6):
  ui & main loop: Redesign of system-specific main thread event handling
  hw/display/apple-gfx: Introduce ParavirtualizedGraphics.Framework
    support
  hw/display/apple-gfx: Adds PCI implementation
  hw/display/apple-gfx: Adds configurable mode list
  MAINTAINERS: Add myself as maintainer for apple-gfx, reviewer for HVF
  hw/block/virtio-blk: Replaces request free function with g_free

 MAINTAINERS                         |  15 +
 contrib/vmapple/uuid.sh             |   9 +
 docs/system/arm/vmapple.rst         |  63 ++
 docs/system/target-arm.rst          |   1 +
 hw/Kconfig                          |   1 +
 hw/arm/sbsa-ref.c                   |   2 +-
 hw/arm/virt.c                       |   2 +-
 hw/block/virtio-blk.c               |  58 +-
 hw/core/qdev-properties-system.c    |   8 +
 hw/display/Kconfig                  |  13 +
 hw/display/apple-gfx-mmio.m         | 288 +++++++++
 hw/display/apple-gfx-pci.m          | 156 +++++
 hw/display/apple-gfx.h              |  77 +++
 hw/display/apple-gfx.m              | 880 ++++++++++++++++++++++++++++
 hw/display/meson.build              |   7 +
 hw/display/trace-events             |  30 +
 hw/i386/microvm.c                   |   2 +-
 hw/loongarch/virt.c                 |  12 +-
 hw/meson.build                      |   1 +
 hw/mips/loongson3_virt.c            |   2 +-
 hw/misc/Kconfig                     |   4 +
 hw/misc/meson.build                 |   1 +
 hw/misc/pvpanic-mmio.c              |  60 ++
 hw/openrisc/virt.c                  |  12 +-
 hw/pci-host/gpex.c                  |  43 +-
 hw/riscv/virt.c                     |  12 +-
 hw/vmapple/Kconfig                  |  32 +
 hw/vmapple/aes.c                    | 581 ++++++++++++++++++
 hw/vmapple/bdif.c                   | 274 +++++++++
 hw/vmapple/cfg.c                    | 195 ++++++
 hw/vmapple/meson.build              |   5 +
 hw/vmapple/trace-events             |  21 +
 hw/vmapple/trace.h                  |   1 +
 hw/vmapple/virtio-blk.c             | 204 +++++++
 hw/vmapple/vmapple.c                | 612 +++++++++++++++++++
 hw/xen/xen-pvh-common.c             |   2 +-
 hw/xtensa/virt.c                    |   2 +-
 include/hw/misc/pvpanic.h           |   1 +
 include/hw/pci-host/gpex.h          |   7 +-
 include/hw/pci/pci_ids.h            |   1 +
 include/hw/qdev-properties-system.h |   5 +
 include/hw/virtio/virtio-blk.h      |  11 +-
 include/hw/vmapple/vmapple.h        |  23 +
 include/qemu-main.h                 |  14 +-
 include/qemu/cutils.h               |  15 +
 meson.build                         |   5 +
 qapi/virtio.json                    |  14 +
 system/main.c                       |  37 +-
 ui/cocoa.m                          |  54 +-
 ui/gtk.c                            |   4 +
 ui/sdl2.c                           |   4 +
 util/hexdump.c                      |  18 +
 52 files changed, 3791 insertions(+), 110 deletions(-)
 create mode 100755 contrib/vmapple/uuid.sh
 create mode 100644 docs/system/arm/vmapple.rst
 create mode 100644 hw/display/apple-gfx-mmio.m
 create mode 100644 hw/display/apple-gfx-pci.m
 create mode 100644 hw/display/apple-gfx.h
 create mode 100644 hw/display/apple-gfx.m
 create mode 100644 hw/misc/pvpanic-mmio.c
 create mode 100644 hw/vmapple/Kconfig
 create mode 100644 hw/vmapple/aes.c
 create mode 100644 hw/vmapple/bdif.c
 create mode 100644 hw/vmapple/cfg.c
 create mode 100644 hw/vmapple/meson.build
 create mode 100644 hw/vmapple/trace-events
 create mode 100644 hw/vmapple/trace.h
 create mode 100644 hw/vmapple/virtio-blk.c
 create mode 100644 hw/vmapple/vmapple.c
 create mode 100644 include/hw/vmapple/vmapple.h

-- 
2.39.5 (Apple Git-154)
Re: [PATCH v16 00/14] macOS PV Graphics and new vmapple machine type
Posted by Philippe Mathieu-Daudé 1 year, 1 month ago 
On 23/12/24 23:16, Phil Dennis-Jordan wrote:
> This patch set introduces a new ARM and macOS HVF specific machine type
> called "vmapple", as well as a family of display devices based on the
> ParavirtualizedGraphics.framework in macOS. One of the display adapter
> variants, apple-gfx-mmio, is required for the new machine type, while
> apple-gfx-pci can be used to enable 3D graphics acceleration with x86-64
> macOS guest OSes.
> 
> Previous versions of this patch set were submitted semi-separately:
> the original vmapple patch set by Alexander Graf included a monolithic
> implementation of apple-gfx-mmio. I subsequently reviewed and reworked
> the latter to support the PCI variant of the device as well and submitted
> the result in isolation. As requested in subsequent review, I have now
> recombined this with the original vmapple patch set, which I have updated
> and improved in a few ways as well.
> 
> The vmapple machine type approximates the configuration in macOS's own
> Virtualization.framework when running arm64 macOS guests. In addition to
> generic components such as a GICv3 and an XHCI USB controller, it
> includes nonstandard extensions to the virtio block device, a special
> "hardware" aes engine, a configuration device, a pvpanic variant, a
> "backdoor" interface, and of course the apple-gfx paravirtualised display
> adapter.
> 
> There are currently a few limitations to this which aren't intrinsic,
> just imperfect emulation of the VZF, but it's good enough to be just
> about usable for some purposes:
> 
>   * macOS 12 guests only. Versions 13+ currently fail during early boot.
>   * macOS 11+ arm64 hosts only, with hvf accel. (Perhaps some differences
>     between Apple M series CPUs and TCG's aarch64 implementation? macOS
>     hosts only because ParavirtualizedGraphics.framework is a black box
>     implementing most of the logic behind the apple-gfx device.)
>   * PCI devices use legacy IRQs, not MSI/MSI-X. As far as I can tell,
>     we'd need to include the GICv3 ITS, but it's unclear to me what
>     exactly needs wiring up.
>   * Due to a quirk (bug?) in the macOS XHCI driver when MSI-X is not
>     available, correct functioning of the USB controller (and thus
>     keyboard/tablet) requires a small workaround in the XHCI controller
>     device. This is part of another patch series:
>     https://patchew.org/QEMU/20241208191646.64857-1-phil@philjordan.eu/
>   * The guest OS must first be provisioned using Virtualization.framework;
>     the disk images can subsequently be used in Qemu. (See docs.)
> 
> The apple-gfx device can be used independently from the vmapple machine
> type, at least in the PCI variant. It mainly targets x86-64 macOS guests
> from version 11 on, but also includes a UEFI bootrom for basic
> framebuffer mode. macOS 11 is also required on the host side, as well
> as a GPU that supports the Metal API. On the guest side, this provides
> 3D acceleration/GPGPU support with a baseline Metal feature set,
> irrespective of the host GPU's feature set. A few limitations in the
> current integration:
> 
>   * Although it works fine with TCG, it does not work correctly
>     cross-architecture: x86-64 guests on arm64 hosts appear to make
>     some boot progress, but rendering is corrupted. I suspect
>     incompatible texture memory layouts; I have no idea if this is
>     fixable.
>   * ParavirtualizedGraphics.framework and the guest driver support
>     multi-headed configurations. The current Qemu integration always
>     connects precisely 1 display.
>   * State serialisation and deserialisation is currently not
>     implemented, though supported in principle by the framework.
>     Both apple-gfx variants thus set up a migration blocker.
>   * Rendering efficiency could be better. The GPU-rendered guest
>     framebuffer is copied to system memory and uses Qemu's usual
>     CPU-based drawing. For maximum efficiency, the Metal texture
>     containing the guest framebuffer could be drawn directly to
>     a Metal view in the host window, staying on the GPU. (Similar
>     to the OpenGL/virgl render path on other platforms.)


> v15 -> v16
> 
>   * 14 patches now, as patch 8 has already been pulled. (Thanks Philippe!)
>   * Fixed a bunch of conflicts with upstream code motion:
>      - DEFINE_PROP_END_OF_LIST removal (4/14 - apple-gfx mode list, 7/14 -
>        pvpanic-mmio, 10/14 - bdif, 11/14 - cfg device, and
>        12/14 - vmapple-virtio-blk)
>      - sysemu->system move/rename: (1/14 - ui/qemu-main, 2/14 - apple-gfx,
>        9/14 - aes, 10/14 - bdif, 14/14 - vmapple machine type)
>   * 14/14 (vmapple machine type):
>      - Moved compatibility setting for removing legacy mode from virtio-pci
>        to proper global property table rather than (ab)using sugar property.
>      - Removed a few superfluous #includes during sysemu rename cleanup.
>      - Removed machine type versioning as it's not necessary (yet?)
>      - Made memory map array const

Great.

> Alexander Graf (8):
>    hw: Add vmapple subdir
>    hw/misc/pvpanic: Add MMIO interface
>    gpex: Allow more than 4 legacy IRQs
>    hw/vmapple/aes: Introduce aes engine
>    hw/vmapple/bdif: Introduce vmapple backdoor interface
>    hw/vmapple/cfg: Introduce vmapple cfg region
>    hw/vmapple/virtio-blk: Add support for apple virtio-blk
>    hw/vmapple/vmapple: Add vmapple machine type
> 
> Phil Dennis-Jordan (6):
>    ui & main loop: Redesign of system-specific main thread event handling
>    hw/display/apple-gfx: Introduce ParavirtualizedGraphics.Framework
>      support
>    hw/display/apple-gfx: Adds PCI implementation
>    hw/display/apple-gfx: Adds configurable mode list
>    MAINTAINERS: Add myself as maintainer for apple-gfx, reviewer for HVF
>    hw/block/virtio-blk: Replaces request free function with g_free

If there are no objection or further comments, I'm taking this
series and will post a PR after xmas & testing.

Thanks all!

Phil.
Re: [PATCH v16 00/14] macOS PV Graphics and new vmapple machine type
Posted by Phil Dennis-Jordan 1 year, 1 month ago 
On Mon, 23 Dec 2024 at 23:58, Philippe Mathieu-Daudé <philmd@linaro.org>
wrote:

> On 23/12/24 23:16, Phil Dennis-Jordan wrote:
> > This patch set introduces a new ARM and macOS HVF specific machine type
> > called "vmapple", as well as a family of display devices based on the
> > ParavirtualizedGraphics.framework in macOS. One of the display adapter
> > variants, apple-gfx-mmio, is required for the new machine type, while
> > apple-gfx-pci can be used to enable 3D graphics acceleration with x86-64
> > macOS guest OSes.
> >
> > Previous versions of this patch set were submitted semi-separately:
> > the original vmapple patch set by Alexander Graf included a monolithic
> > implementation of apple-gfx-mmio. I subsequently reviewed and reworked
> > the latter to support the PCI variant of the device as well and submitted
> > the result in isolation. As requested in subsequent review, I have now
> > recombined this with the original vmapple patch set, which I have updated
> > and improved in a few ways as well.
> >
> > The vmapple machine type approximates the configuration in macOS's own
> > Virtualization.framework when running arm64 macOS guests. In addition to
> > generic components such as a GICv3 and an XHCI USB controller, it
> > includes nonstandard extensions to the virtio block device, a special
> > "hardware" aes engine, a configuration device, a pvpanic variant, a
> > "backdoor" interface, and of course the apple-gfx paravirtualised display
> > adapter.
> >
> > There are currently a few limitations to this which aren't intrinsic,
> > just imperfect emulation of the VZF, but it's good enough to be just
> > about usable for some purposes:
> >
> >   * macOS 12 guests only. Versions 13+ currently fail during early boot.
> >   * macOS 11+ arm64 hosts only, with hvf accel. (Perhaps some differences
> >     between Apple M series CPUs and TCG's aarch64 implementation? macOS
> >     hosts only because ParavirtualizedGraphics.framework is a black box
> >     implementing most of the logic behind the apple-gfx device.)
> >   * PCI devices use legacy IRQs, not MSI/MSI-X. As far as I can tell,
> >     we'd need to include the GICv3 ITS, but it's unclear to me what
> >     exactly needs wiring up.
> >   * Due to a quirk (bug?) in the macOS XHCI driver when MSI-X is not
> >     available, correct functioning of the USB controller (and thus
> >     keyboard/tablet) requires a small workaround in the XHCI controller
> >     device. This is part of another patch series:
> >     https://patchew.org/QEMU/20241208191646.64857-1-phil@philjordan.eu/
> >   * The guest OS must first be provisioned using
> Virtualization.framework;
> >     the disk images can subsequently be used in Qemu. (See docs.)
> >
> > The apple-gfx device can be used independently from the vmapple machine
> > type, at least in the PCI variant. It mainly targets x86-64 macOS guests
> > from version 11 on, but also includes a UEFI bootrom for basic
> > framebuffer mode. macOS 11 is also required on the host side, as well
> > as a GPU that supports the Metal API. On the guest side, this provides
> > 3D acceleration/GPGPU support with a baseline Metal feature set,
> > irrespective of the host GPU's feature set. A few limitations in the
> > current integration:
> >
> >   * Although it works fine with TCG, it does not work correctly
> >     cross-architecture: x86-64 guests on arm64 hosts appear to make
> >     some boot progress, but rendering is corrupted. I suspect
> >     incompatible texture memory layouts; I have no idea if this is
> >     fixable.
> >   * ParavirtualizedGraphics.framework and the guest driver support
> >     multi-headed configurations. The current Qemu integration always
> >     connects precisely 1 display.
> >   * State serialisation and deserialisation is currently not
> >     implemented, though supported in principle by the framework.
> >     Both apple-gfx variants thus set up a migration blocker.
> >   * Rendering efficiency could be better. The GPU-rendered guest
> >     framebuffer is copied to system memory and uses Qemu's usual
> >     CPU-based drawing. For maximum efficiency, the Metal texture
> >     containing the guest framebuffer could be drawn directly to
> >     a Metal view in the host window, staying on the GPU. (Similar
> >     to the OpenGL/virgl render path on other platforms.)
>
>
> > v15 -> v16
> >
> >   * 14 patches now, as patch 8 has already been pulled. (Thanks
> Philippe!)
> >   * Fixed a bunch of conflicts with upstream code motion:
> >      - DEFINE_PROP_END_OF_LIST removal (4/14 - apple-gfx mode list, 7/14
> -
> >        pvpanic-mmio, 10/14 - bdif, 11/14 - cfg device, and
> >        12/14 - vmapple-virtio-blk)
> >      - sysemu->system move/rename: (1/14 - ui/qemu-main, 2/14 -
> apple-gfx,
> >        9/14 - aes, 10/14 - bdif, 14/14 - vmapple machine type)
> >   * 14/14 (vmapple machine type):
> >      - Moved compatibility setting for removing legacy mode from
> virtio-pci
> >        to proper global property table rather than (ab)using sugar
> property.
> >      - Removed a few superfluous #includes during sysemu rename cleanup.
> >      - Removed machine type versioning as it's not necessary (yet?)
> >      - Made memory map array const
>
> Great.
>
> > Alexander Graf (8):
> >    hw: Add vmapple subdir
> >    hw/misc/pvpanic: Add MMIO interface
> >    gpex: Allow more than 4 legacy IRQs
> >    hw/vmapple/aes: Introduce aes engine
> >    hw/vmapple/bdif: Introduce vmapple backdoor interface
> >    hw/vmapple/cfg: Introduce vmapple cfg region
> >    hw/vmapple/virtio-blk: Add support for apple virtio-blk
> >    hw/vmapple/vmapple: Add vmapple machine type
> >
> > Phil Dennis-Jordan (6):
> >    ui & main loop: Redesign of system-specific main thread event handling
> >    hw/display/apple-gfx: Introduce ParavirtualizedGraphics.Framework
> >      support
> >    hw/display/apple-gfx: Adds PCI implementation
> >    hw/display/apple-gfx: Adds configurable mode list
> >    MAINTAINERS: Add myself as maintainer for apple-gfx, reviewer for HVF
> >    hw/block/virtio-blk: Replaces request free function with g_free
>
> If there are no objection or further comments, I'm taking this
> series and will post a PR after xmas & testing.


Thanks Phil, much appreciated! Enjoy your time off.

I've just posted an updated -v3 of my xhci patch series as v2 suffered from
the same breakage as this patch set. I recommend applying that one on top
for testing vmapple. It helps when you can use keyboard and mouse properly
and don't need to mess around with VNC.

All the best,
Phil
Re: [PATCH v16 00/14] macOS PV Graphics and new vmapple machine type
Posted by Philippe Mathieu-Daudé 1 year, 1 month ago 
On 27/12/24 13:19, Phil Dennis-Jordan wrote:
> 
> 
> On Mon, 23 Dec 2024 at 23:58, Philippe Mathieu-Daudé <philmd@linaro.org 
> <mailto:philmd@linaro.org>> wrote:


>      > Alexander Graf (8):
>      >    hw: Add vmapple subdir
>      >    hw/misc/pvpanic: Add MMIO interface
>      >    gpex: Allow more than 4 legacy IRQs
>      >    hw/vmapple/aes: Introduce aes engine
>      >    hw/vmapple/bdif: Introduce vmapple backdoor interface
>      >    hw/vmapple/cfg: Introduce vmapple cfg region
>      >    hw/vmapple/virtio-blk: Add support for apple virtio-blk
>      >    hw/vmapple/vmapple: Add vmapple machine type
>      >
>      > Phil Dennis-Jordan (6):
>      >    ui & main loop: Redesign of system-specific main thread event
>     handling
>      >    hw/display/apple-gfx: Introduce ParavirtualizedGraphics.Framework
>      >      support
>      >    hw/display/apple-gfx: Adds PCI implementation
>      >    hw/display/apple-gfx: Adds configurable mode list
>      >    MAINTAINERS: Add myself as maintainer for apple-gfx, reviewer
>     for HVF
>      >    hw/block/virtio-blk: Replaces request free function with g_free
> 
>     If there are no objection or further comments, I'm taking this
>     series and will post a PR after xmas & testing.
> 
> 
> Thanks Phil, much appreciated! Enjoy your time off.
> 
> I've just posted an updated -v3 of my xhci patch series as v2 suffered 
> from the same breakage as this patch set. I recommend applying that one 
> on top for testing vmapple. It helps when you can use keyboard and mouse 
> properly and don't need to mess around with VNC.

I'm first taking patches 1-5 & 8 [*] squashing mostly style fixups,
TypeInfo rebase and header reduction ("qemu/osdep.h" is only for .c):

-- >8 --
diff --git a/hw/display/apple-gfx.h b/hw/display/apple-gfx.h
index a1160bf6619..3900cdbabbb 100644
--- a/hw/display/apple-gfx.h
+++ b/hw/display/apple-gfx.h
@@ -8,17 +8,14 @@
  #ifndef QEMU_APPLE_GFX_H
  #define QEMU_APPLE_GFX_H

-#define TYPE_APPLE_GFX_MMIO         "apple-gfx-mmio"
-#define TYPE_APPLE_GFX_PCI          "apple-gfx-pci"
-
-#include "qemu/osdep.h"
-#include <dispatch/dispatch.h>
-#import <ParavirtualizedGraphics/ParavirtualizedGraphics.h>
-#include "qemu/typedefs.h"
+#include "qemu/queue.h"
  #include "exec/memory.h"
  #include "hw/qdev-properties.h"
  #include "ui/surface.h"

+#define TYPE_APPLE_GFX_MMIO         "apple-gfx-mmio"
+#define TYPE_APPLE_GFX_PCI          "apple-gfx-pci"
+
  @class PGDeviceDescriptor;
  @protocol PGDevice;
  @protocol PGDisplay;
@@ -71,7 +68,7 @@ void *apple_gfx_host_ptr_for_gpa_range(uint64_t 
guest_physical,
                                         uint64_t length, bool read_only,
                                         MemoryRegion **mapping_in_region);

-extern const PropertyInfo qdev_prop_display_mode;
+extern const PropertyInfo qdev_prop_apple_gfx_display_mode;

  #endif

diff --git a/hw/display/apple-gfx-mmio.m b/hw/display/apple-gfx-mmio.m
index 60580a373d8..b2e0e7a30fa 100644
--- a/hw/display/apple-gfx-mmio.m
+++ b/hw/display/apple-gfx-mmio.m
@@ -3,9 +3,6 @@
   *
   * Copyright © 2023 Amazon.com, Inc. or its affiliates. All Rights 
Reserved.
   *
- * This work is licensed under the terms of the GNU GPL, version 2 or 
later.
- * See the COPYING file in the top-level directory.
- *
   * SPDX-License-Identifier: GPL-2.0-or-later
   *
   * ParavirtualizedGraphics.framework is a set of libraries that macOS 
provides
@@ -16,13 +13,14 @@
   */

  #include "qemu/osdep.h"
-#import <ParavirtualizedGraphics/ParavirtualizedGraphics.h>
-#include "apple-gfx.h"
-#include "monitor/monitor.h"
+#include "qemu/log.h"
+#include "block/aio-wait.h"
  #include "hw/sysbus.h"
  #include "hw/irq.h"
+#include "apple-gfx.h"
  #include "trace.h"
-#include "qemu/log.h"
+
+#import <ParavirtualizedGraphics/ParavirtualizedGraphics.h>

  OBJECT_DECLARE_SIMPLE_TYPE(AppleGFXMMIOState, APPLE_GFX_MMIO)

@@ -32,10 +30,9 @@
   * their definitions here so that we can also work with the ARM version.
   */
  typedef bool(^IOSFCRaiseInterrupt)(uint32_t vector);
-typedef bool(^IOSFCUnmapMemory)(
-    void *, void *, void *, void *, void *, void *);
-typedef bool(^IOSFCMapMemory)(
-    uint64_t phys, uint64_t len, bool ro, void **va, void *, void *);
+typedef bool(^IOSFCUnmapMemory)(void *, void *, void *, void *, void *, 
void *);
+typedef bool(^IOSFCMapMemory)(uint64_t phys, uint64_t len, bool ro, 
void **va,
+                              void *, void *);

  @interface PGDeviceDescriptor (IOSurfaceMapper)
  @property (readwrite, nonatomic) bool usingIOSurfaceMapper;
@@ -168,8 +165,8 @@ static bool apple_gfx_mmio_unmap_surface_memory(void 
*ptr)
      RCU_READ_LOCK_GUARD();
      region = memory_region_from_host(ptr, &offset);
      if (!region) {
-        qemu_log_mask(LOG_GUEST_ERROR, "%s: memory at %p to be unmapped 
not "
-                      "found.\n",
+        qemu_log_mask(LOG_GUEST_ERROR,
+                      "%s: memory at %p to be unmapped not found.\n",
                        __func__, ptr);
          return false;
      }
@@ -261,7 +258,7 @@ static void apple_gfx_mmio_reset(Object *obj, 
ResetType type)
  static const Property apple_gfx_mmio_properties[] = {
      DEFINE_PROP_ARRAY("display-modes", AppleGFXMMIOState,
                        common.num_display_modes, common.display_modes,
-                      qdev_prop_display_mode, AppleGFXDisplayMode),
+                      qdev_prop_apple_gfx_display_mode, 
AppleGFXDisplayMode),
  };

  static void apple_gfx_mmio_class_init(ObjectClass *klass, void *data)
@@ -276,7 +273,7 @@ static void apple_gfx_mmio_class_init(ObjectClass 
*klass, void *data)
      device_class_set_props(dc, apple_gfx_mmio_properties);
  }

-static TypeInfo apple_gfx_mmio_types[] = {
+static const TypeInfo apple_gfx_mmio_types[] = {
      {
          .name          = TYPE_APPLE_GFX_MMIO,
          .parent        = TYPE_SYS_BUS_DEVICE,
diff --git a/hw/display/apple-gfx-pci.m b/hw/display/apple-gfx-pci.m
index 4cfc01fc725..b939bb9b233 100644
--- a/hw/display/apple-gfx-pci.m
+++ b/hw/display/apple-gfx-pci.m
@@ -12,11 +12,12 @@
   * aimed primarily at x86-64 macOS VMs.
   */

-#include "apple-gfx.h"
+#include "qemu/osdep.h"
  #include "hw/pci/pci_device.h"
  #include "hw/pci/msi.h"
-#include "qapi/error.h"
+#include "apple-gfx.h"
  #include "trace.h"
+
  #import <ParavirtualizedGraphics/ParavirtualizedGraphics.h>

  OBJECT_DECLARE_SIMPLE_TYPE(AppleGFXPCIState, APPLE_GFX_PCI)
@@ -27,7 +28,7 @@
      AppleGFXState common;
  };

-static const char* apple_gfx_pci_option_rom_path = NULL;
+static const char *apple_gfx_pci_option_rom_path = NULL;

  static void apple_gfx_init_option_rom_path(void)
  {
@@ -117,7 +118,7 @@ static void apple_gfx_pci_reset(Object *obj, 
ResetType type)
  static const Property apple_gfx_pci_properties[] = {
      DEFINE_PROP_ARRAY("display-modes", AppleGFXPCIState,
                        common.num_display_modes, common.display_modes,
-                      qdev_prop_display_mode, AppleGFXDisplayMode),
+                      qdev_prop_apple_gfx_display_mode, 
AppleGFXDisplayMode),
  };

  static void apple_gfx_pci_class_init(ObjectClass *klass, void *data)
@@ -139,7 +140,7 @@ static void apple_gfx_pci_class_init(ObjectClass 
*klass, void *data)
      device_class_set_props(dc, apple_gfx_pci_properties);
  }

-static TypeInfo apple_gfx_pci_types[] = {
+static const TypeInfo apple_gfx_pci_types[] = {
      {
          .name          = TYPE_APPLE_GFX_PCI,
          .parent        = TYPE_PCI_DEVICE,
diff --git a/hw/display/apple-gfx.m b/hw/display/apple-gfx.m
index b00c72fc4dd..aa1455b6295 100644
--- a/hw/display/apple-gfx.m
+++ b/hw/display/apple-gfx.m
@@ -3,9 +3,6 @@
   *
   * Copyright © 2023 Amazon.com, Inc. or its affiliates. All Rights 
Reserved.
   *
- * This work is licensed under the terms of the GNU GPL, version 2 or 
later.
- * See the COPYING file in the top-level directory.
- *
   * SPDX-License-Identifier: GPL-2.0-or-later
   *
   * ParavirtualizedGraphics.framework is a set of libraries that macOS 
provides
@@ -15,21 +12,24 @@
   */

  #include "qemu/osdep.h"
-#import <ParavirtualizedGraphics/ParavirtualizedGraphics.h>
-#include <mach/mach_vm.h>
-#include "apple-gfx.h"
-#include "trace.h"
-#include "qemu-main.h"
-#include "exec/address-spaces.h"
-#include "migration/blocker.h"
-#include "monitor/monitor.h"
-#include "qemu/main-loop.h"
+#include "qemu/lockable.h"
  #include "qemu/cutils.h"
  #include "qemu/log.h"
  #include "qapi/visitor.h"
  #include "qapi/error.h"
+#include "block/aio-wait.h"
+#include "exec/address-spaces.h"
  #include "system/dma.h"
+#include "migration/blocker.h"
  #include "ui/console.h"
+#include "apple-gfx.h"
+#include "trace.h"
+
+#include <mach/mach.h>
+#include <mach/mach_vm.h>
+#include <dispatch/dispatch.h>
+
+#import <ParavirtualizedGraphics/ParavirtualizedGraphics.h>

  static const AppleGFXDisplayMode apple_gfx_default_modes[] = {
      { 1920, 1080, 60 },
@@ -419,7 +419,7 @@ static void set_cursor_glyph(void *opaque)
          s->cursor = NULL;
      }

-    if (bpp == 32) { /* Shouldn't be anything else, but just to be 
safe...*/
+    if (bpp == 32) { /* Shouldn't be anything else, but just to be 
safe... */
          s->cursor = cursor_alloc(width, height);
          s->cursor->hot_x = job->hotspot.x;
          s->cursor->hot_y = job->hotspot.y;
@@ -467,7 +467,7 @@ static void apple_gfx_do_read_memory(void *opaque)

      r = dma_memory_read(&address_space_memory, job->physical_address,
                          job->dst, job->length, MEMTXATTRS_UNSPECIFIED);
-    job->success = r == MEMTX_OK;
+    job->success = (r == MEMTX_OK);

      qemu_sem_post(&job->sem);
  }
@@ -694,12 +694,11 @@ static void new_frame_handler_bh(void *opaque)
  static NSArray<PGDisplayMode *> *apple_gfx_create_display_mode_array(
      const AppleGFXDisplayMode display_modes[], uint32_t 
display_mode_count)
  {
-    uint32_t i;
      PGDisplayMode *mode_obj;
      NSMutableArray<PGDisplayMode *> *mode_array =
          [[NSMutableArray alloc] initWithCapacity:display_mode_count];

-    for (i = 0; i < display_mode_count; i++) {
+    for (unsigned i = 0; i < display_mode_count; i++) {
          const AppleGFXDisplayMode *mode = &display_modes[i];
          trace_apple_gfx_display_mode(i, mode->width_px, mode->height_px);
          PGDisplayCoord_t mode_size = { mode->width_px, mode->height_px };
@@ -836,8 +835,8 @@ static void apple_gfx_set_display_mode(Object *obj, 
Visitor *v,

      ret = qemu_strtoi(endptr, &endptr, 10, &val);
      if (ret || val > UINT16_MAX || val <= 0) {
-        error_setg(errp, "width in '%s' must be a decimal integer number "
-                   "of pixels in the range 1..65535", name);
+        error_setg(errp, "width in '%s' must be a decimal integer number"
+                         " of pixels in the range 1..65535", name);
          return;
      }
      mode->width_px = val;
@@ -847,8 +846,8 @@ static void apple_gfx_set_display_mode(Object *obj, 
Visitor *v,

      ret = qemu_strtoi(endptr + 1, &endptr, 10, &val);
      if (ret || val > UINT16_MAX || val <= 0) {
-        error_setg(errp, "height in '%s' must be a decimal integer number "
-                   "of pixels in the range 1..65535", name);
+        error_setg(errp, "height in '%s' must be a decimal integer number"
+                         " of pixels in the range 1..65535", name);
          return;
      }
      mode->height_px = val;
@@ -859,18 +858,18 @@ static void apple_gfx_set_display_mode(Object 
*obj, Visitor *v,
      ret = qemu_strtoi(endptr + 1, &endptr, 10, &val);
      if (ret || val > UINT16_MAX || val <= 0) {
          error_setg(errp, "refresh rate in '%s'"
-                   " must be a positive decimal integer (Hertz)", name);
+                         " must be a positive decimal integer (Hertz)", 
name);
          return;
      }
      mode->refresh_rate_hz = val;
      return;

  separator_error:
-    error_setg(errp, "Each display mode takes the format "
-               "'<width>x<height>@<rate>'");
+    error_setg(errp,
+               "Each display mode takes the format 
'<width>x<height>@<rate>'");
  }

-const PropertyInfo qdev_prop_display_mode = {
+const PropertyInfo qdev_prop_apple_gfx_display_mode = {
      .name  = "display_mode",
      .description =
          "Display mode in pixels and Hertz, as 
<width>x<height>@<refresh-rate> "
---

Regards,

Phil.

[*] I'll share the rebased branch later.
Re: [PATCH v16 00/14] macOS PV Graphics and new vmapple machine type
Posted by Philippe Mathieu-Daudé 1 year, 1 month ago 
Cc'ing Joelle (FYI https://github.com/utmapp/UTM/issues/3491)

On 23/12/24 23:16, Phil Dennis-Jordan wrote:
> This patch set introduces a new ARM and macOS HVF specific machine type
> called "vmapple", as well as a family of display devices based on the
> ParavirtualizedGraphics.framework in macOS. One of the display adapter
> variants, apple-gfx-mmio, is required for the new machine type, while
> apple-gfx-pci can be used to enable 3D graphics acceleration with x86-64
> macOS guest OSes.
> 
> Previous versions of this patch set were submitted semi-separately:
> the original vmapple patch set by Alexander Graf included a monolithic
> implementation of apple-gfx-mmio. I subsequently reviewed and reworked
> the latter to support the PCI variant of the device as well and submitted
> the result in isolation. As requested in subsequent review, I have now
> recombined this with the original vmapple patch set, which I have updated
> and improved in a few ways as well.
> 
> The vmapple machine type approximates the configuration in macOS's own
> Virtualization.framework when running arm64 macOS guests. In addition to
> generic components such as a GICv3 and an XHCI USB controller, it
> includes nonstandard extensions to the virtio block device, a special
> "hardware" aes engine, a configuration device, a pvpanic variant, a
> "backdoor" interface, and of course the apple-gfx paravirtualised display
> adapter.
> 
> There are currently a few limitations to this which aren't intrinsic,
> just imperfect emulation of the VZF, but it's good enough to be just
> about usable for some purposes:
> 
>   * macOS 12 guests only. Versions 13+ currently fail during early boot.
>   * macOS 11+ arm64 hosts only, with hvf accel. (Perhaps some differences
>     between Apple M series CPUs and TCG's aarch64 implementation? macOS
>     hosts only because ParavirtualizedGraphics.framework is a black box
>     implementing most of the logic behind the apple-gfx device.)
>   * PCI devices use legacy IRQs, not MSI/MSI-X. As far as I can tell,
>     we'd need to include the GICv3 ITS, but it's unclear to me what
>     exactly needs wiring up.
>   * Due to a quirk (bug?) in the macOS XHCI driver when MSI-X is not
>     available, correct functioning of the USB controller (and thus
>     keyboard/tablet) requires a small workaround in the XHCI controller
>     device. This is part of another patch series:
>     https://patchew.org/QEMU/20241208191646.64857-1-phil@philjordan.eu/
>   * The guest OS must first be provisioned using Virtualization.framework;
>     the disk images can subsequently be used in Qemu. (See docs.)
> 
> The apple-gfx device can be used independently from the vmapple machine
> type, at least in the PCI variant. It mainly targets x86-64 macOS guests
> from version 11 on, but also includes a UEFI bootrom for basic
> framebuffer mode. macOS 11 is also required on the host side, as well
> as a GPU that supports the Metal API. On the guest side, this provides
> 3D acceleration/GPGPU support with a baseline Metal feature set,
> irrespective of the host GPU's feature set. A few limitations in the
> current integration:
> 
>   * Although it works fine with TCG, it does not work correctly
>     cross-architecture: x86-64 guests on arm64 hosts appear to make
>     some boot progress, but rendering is corrupted. I suspect
>     incompatible texture memory layouts; I have no idea if this is
>     fixable.

Zoltan, does that ring a bell?

Phil, should we display a warning in this configuration case? Or only
allow it with some developper option, like:

     -device '{"driver":"apple-gfx-pci", \
               "display-modes":["3840x2160@60"], \
               "x-force-cross-rendering":"true"}'

>   * ParavirtualizedGraphics.framework and the guest driver support
>     multi-headed configurations. The current Qemu integration always
>     connects precisely 1 display.
>   * State serialisation and deserialisation is currently not
>     implemented, though supported in principle by the framework.
>     Both apple-gfx variants thus set up a migration blocker.
>   * Rendering efficiency could be better. The GPU-rendered guest
>     framebuffer is copied to system memory and uses Qemu's usual
>     CPU-based drawing. For maximum efficiency, the Metal texture
>     containing the guest framebuffer could be drawn directly to
>     a Metal view in the host window, staying on the GPU. (Similar
>     to the OpenGL/virgl render path on other platforms.)
> 
> Some of my part of this work has been sponsored by Sauce Labs Inc.
> 
> ---


> Alexander Graf (8):
>    hw: Add vmapple subdir
>    hw/misc/pvpanic: Add MMIO interface
>    gpex: Allow more than 4 legacy IRQs
>    hw/vmapple/aes: Introduce aes engine
>    hw/vmapple/bdif: Introduce vmapple backdoor interface
>    hw/vmapple/cfg: Introduce vmapple cfg region
>    hw/vmapple/virtio-blk: Add support for apple virtio-blk
>    hw/vmapple/vmapple: Add vmapple machine type
> 
> Phil Dennis-Jordan (6):
>    ui & main loop: Redesign of system-specific main thread event handling
>    hw/display/apple-gfx: Introduce ParavirtualizedGraphics.Framework
>      support
>    hw/display/apple-gfx: Adds PCI implementation
>    hw/display/apple-gfx: Adds configurable mode list
>    MAINTAINERS: Add myself as maintainer for apple-gfx, reviewer for HVF
>    hw/block/virtio-blk: Replaces request free function with g_free
> 
>   MAINTAINERS                         |  15 +
>   contrib/vmapple/uuid.sh             |   9 +
>   docs/system/arm/vmapple.rst         |  63 ++
>   docs/system/target-arm.rst          |   1 +
>   hw/Kconfig                          |   1 +
>   hw/arm/sbsa-ref.c                   |   2 +-
>   hw/arm/virt.c                       |   2 +-
>   hw/block/virtio-blk.c               |  58 +-
>   hw/core/qdev-properties-system.c    |   8 +
>   hw/display/Kconfig                  |  13 +
>   hw/display/apple-gfx-mmio.m         | 288 +++++++++
>   hw/display/apple-gfx-pci.m          | 156 +++++
>   hw/display/apple-gfx.h              |  77 +++
>   hw/display/apple-gfx.m              | 880 ++++++++++++++++++++++++++++
>   hw/display/meson.build              |   7 +
>   hw/display/trace-events             |  30 +
>   hw/i386/microvm.c                   |   2 +-
>   hw/loongarch/virt.c                 |  12 +-
>   hw/meson.build                      |   1 +
>   hw/mips/loongson3_virt.c            |   2 +-
>   hw/misc/Kconfig                     |   4 +
>   hw/misc/meson.build                 |   1 +
>   hw/misc/pvpanic-mmio.c              |  60 ++
>   hw/openrisc/virt.c                  |  12 +-
>   hw/pci-host/gpex.c                  |  43 +-
>   hw/riscv/virt.c                     |  12 +-
>   hw/vmapple/Kconfig                  |  32 +
>   hw/vmapple/aes.c                    | 581 ++++++++++++++++++
>   hw/vmapple/bdif.c                   | 274 +++++++++
>   hw/vmapple/cfg.c                    | 195 ++++++
>   hw/vmapple/meson.build              |   5 +
>   hw/vmapple/trace-events             |  21 +
>   hw/vmapple/trace.h                  |   1 +
>   hw/vmapple/virtio-blk.c             | 204 +++++++
>   hw/vmapple/vmapple.c                | 612 +++++++++++++++++++
>   hw/xen/xen-pvh-common.c             |   2 +-
>   hw/xtensa/virt.c                    |   2 +-
>   include/hw/misc/pvpanic.h           |   1 +
>   include/hw/pci-host/gpex.h          |   7 +-
>   include/hw/pci/pci_ids.h            |   1 +
>   include/hw/qdev-properties-system.h |   5 +
>   include/hw/virtio/virtio-blk.h      |  11 +-
>   include/hw/vmapple/vmapple.h        |  23 +
>   include/qemu-main.h                 |  14 +-
>   include/qemu/cutils.h               |  15 +
>   meson.build                         |   5 +
>   qapi/virtio.json                    |  14 +
>   system/main.c                       |  37 +-
>   ui/cocoa.m                          |  54 +-
>   ui/gtk.c                            |   4 +
>   ui/sdl2.c                           |   4 +
>   util/hexdump.c                      |  18 +
>   52 files changed, 3791 insertions(+), 110 deletions(-)
>   create mode 100755 contrib/vmapple/uuid.sh
>   create mode 100644 docs/system/arm/vmapple.rst
>   create mode 100644 hw/display/apple-gfx-mmio.m
>   create mode 100644 hw/display/apple-gfx-pci.m
>   create mode 100644 hw/display/apple-gfx.h
>   create mode 100644 hw/display/apple-gfx.m
>   create mode 100644 hw/misc/pvpanic-mmio.c
>   create mode 100644 hw/vmapple/Kconfig
>   create mode 100644 hw/vmapple/aes.c
>   create mode 100644 hw/vmapple/bdif.c
>   create mode 100644 hw/vmapple/cfg.c
>   create mode 100644 hw/vmapple/meson.build
>   create mode 100644 hw/vmapple/trace-events
>   create mode 100644 hw/vmapple/trace.h
>   create mode 100644 hw/vmapple/virtio-blk.c
>   create mode 100644 hw/vmapple/vmapple.c
>   create mode 100644 include/hw/vmapple/vmapple.h
>
Re: [PATCH v16 00/14] macOS PV Graphics and new vmapple machine type
Posted by Phil Dennis-Jordan 1 year, 1 month ago 
On Mon, 30 Dec 2024 at 19:55, Philippe Mathieu-Daudé <philmd@linaro.org>
wrote:

> Cc'ing Joelle (FYI https://github.com/utmapp/UTM/issues/3491)
>
> On 23/12/24 23:16, Phil Dennis-Jordan wrote:
> > This patch set introduces a new ARM and macOS HVF specific machine type
> > called "vmapple", as well as a family of display devices based on the
> > ParavirtualizedGraphics.framework in macOS. One of the display adapter
> > variants, apple-gfx-mmio, is required for the new machine type, while
> > apple-gfx-pci can be used to enable 3D graphics acceleration with x86-64
> > macOS guest OSes.
> >
> > Previous versions of this patch set were submitted semi-separately:
> > the original vmapple patch set by Alexander Graf included a monolithic
> > implementation of apple-gfx-mmio. I subsequently reviewed and reworked
> > the latter to support the PCI variant of the device as well and submitted
> > the result in isolation. As requested in subsequent review, I have now
> > recombined this with the original vmapple patch set, which I have updated
> > and improved in a few ways as well.
> >
> > The vmapple machine type approximates the configuration in macOS's own
> > Virtualization.framework when running arm64 macOS guests. In addition to
> > generic components such as a GICv3 and an XHCI USB controller, it
> > includes nonstandard extensions to the virtio block device, a special
> > "hardware" aes engine, a configuration device, a pvpanic variant, a
> > "backdoor" interface, and of course the apple-gfx paravirtualised display
> > adapter.
> >
> > There are currently a few limitations to this which aren't intrinsic,
> > just imperfect emulation of the VZF, but it's good enough to be just
> > about usable for some purposes:
> >
> >   * macOS 12 guests only. Versions 13+ currently fail during early boot.
> >   * macOS 11+ arm64 hosts only, with hvf accel. (Perhaps some differences
> >     between Apple M series CPUs and TCG's aarch64 implementation? macOS
> >     hosts only because ParavirtualizedGraphics.framework is a black box
> >     implementing most of the logic behind the apple-gfx device.)
> >   * PCI devices use legacy IRQs, not MSI/MSI-X. As far as I can tell,
> >     we'd need to include the GICv3 ITS, but it's unclear to me what
> >     exactly needs wiring up.
> >   * Due to a quirk (bug?) in the macOS XHCI driver when MSI-X is not
> >     available, correct functioning of the USB controller (and thus
> >     keyboard/tablet) requires a small workaround in the XHCI controller
> >     device. This is part of another patch series:
> >     https://patchew.org/QEMU/20241208191646.64857-1-phil@philjordan.eu/
> >   * The guest OS must first be provisioned using
> Virtualization.framework;
> >     the disk images can subsequently be used in Qemu. (See docs.)
> >
> > The apple-gfx device can be used independently from the vmapple machine
> > type, at least in the PCI variant. It mainly targets x86-64 macOS guests
> > from version 11 on, but also includes a UEFI bootrom for basic
> > framebuffer mode. macOS 11 is also required on the host side, as well
> > as a GPU that supports the Metal API. On the guest side, this provides
> > 3D acceleration/GPGPU support with a baseline Metal feature set,
> > irrespective of the host GPU's feature set. A few limitations in the
> > current integration:
> >
> >   * Although it works fine with TCG, it does not work correctly
> >     cross-architecture: x86-64 guests on arm64 hosts appear to make
> >     some boot progress, but rendering is corrupted. I suspect
> >     incompatible texture memory layouts; I have no idea if this is
> >     fixable.
>
> Zoltan, does that ring a bell?
>
> Phil, should we display a warning in this configuration case? Or only
> allow it with some developper option, like:
>
>      -device '{"driver":"apple-gfx-pci", \
>                "display-modes":["3840x2160@60"], \
>                "x-force-cross-rendering":"true"}'
>
>
This is a good idea. I think the override option is probably better, as
warnings are easily missed and the resulting behaviour seems to be
generally unusable. Do you want me to put this together? If so, what's the
most acceptable way to check the target and host architectures respectively
from (I guess?) the device instance init function?


> >   * ParavirtualizedGraphics.framework and the guest driver support
> >     multi-headed configurations. The current Qemu integration always
> >     connects precisely 1 display.
> >   * State serialisation and deserialisation is currently not
> >     implemented, though supported in principle by the framework.
> >     Both apple-gfx variants thus set up a migration blocker.
> >   * Rendering efficiency could be better. The GPU-rendered guest
> >     framebuffer is copied to system memory and uses Qemu's usual
> >     CPU-based drawing. For maximum efficiency, the Metal texture
> >     containing the guest framebuffer could be drawn directly to
> >     a Metal view in the host window, staying on the GPU. (Similar
> >     to the OpenGL/virgl render path on other platforms.)
> >
> > Some of my part of this work has been sponsored by Sauce Labs Inc.
> >
> > ---
>
>
> > Alexander Graf (8):
> >    hw: Add vmapple subdir
> >    hw/misc/pvpanic: Add MMIO interface
> >    gpex: Allow more than 4 legacy IRQs
> >    hw/vmapple/aes: Introduce aes engine
> >    hw/vmapple/bdif: Introduce vmapple backdoor interface
> >    hw/vmapple/cfg: Introduce vmapple cfg region
> >    hw/vmapple/virtio-blk: Add support for apple virtio-blk
> >    hw/vmapple/vmapple: Add vmapple machine type
> >
> > Phil Dennis-Jordan (6):
> >    ui & main loop: Redesign of system-specific main thread event handling
> >    hw/display/apple-gfx: Introduce ParavirtualizedGraphics.Framework
> >      support
> >    hw/display/apple-gfx: Adds PCI implementation
> >    hw/display/apple-gfx: Adds configurable mode list
> >    MAINTAINERS: Add myself as maintainer for apple-gfx, reviewer for HVF
> >    hw/block/virtio-blk: Replaces request free function with g_free
> >
> >   MAINTAINERS                         |  15 +
> >   contrib/vmapple/uuid.sh             |   9 +
> >   docs/system/arm/vmapple.rst         |  63 ++
> >   docs/system/target-arm.rst          |   1 +
> >   hw/Kconfig                          |   1 +
> >   hw/arm/sbsa-ref.c                   |   2 +-
> >   hw/arm/virt.c                       |   2 +-
> >   hw/block/virtio-blk.c               |  58 +-
> >   hw/core/qdev-properties-system.c    |   8 +
> >   hw/display/Kconfig                  |  13 +
> >   hw/display/apple-gfx-mmio.m         | 288 +++++++++
> >   hw/display/apple-gfx-pci.m          | 156 +++++
> >   hw/display/apple-gfx.h              |  77 +++
> >   hw/display/apple-gfx.m              | 880 ++++++++++++++++++++++++++++
> >   hw/display/meson.build              |   7 +
> >   hw/display/trace-events             |  30 +
> >   hw/i386/microvm.c                   |   2 +-
> >   hw/loongarch/virt.c                 |  12 +-
> >   hw/meson.build                      |   1 +
> >   hw/mips/loongson3_virt.c            |   2 +-
> >   hw/misc/Kconfig                     |   4 +
> >   hw/misc/meson.build                 |   1 +
> >   hw/misc/pvpanic-mmio.c              |  60 ++
> >   hw/openrisc/virt.c                  |  12 +-
> >   hw/pci-host/gpex.c                  |  43 +-
> >   hw/riscv/virt.c                     |  12 +-
> >   hw/vmapple/Kconfig                  |  32 +
> >   hw/vmapple/aes.c                    | 581 ++++++++++++++++++
> >   hw/vmapple/bdif.c                   | 274 +++++++++
> >   hw/vmapple/cfg.c                    | 195 ++++++
> >   hw/vmapple/meson.build              |   5 +
> >   hw/vmapple/trace-events             |  21 +
> >   hw/vmapple/trace.h                  |   1 +
> >   hw/vmapple/virtio-blk.c             | 204 +++++++
> >   hw/vmapple/vmapple.c                | 612 +++++++++++++++++++
> >   hw/xen/xen-pvh-common.c             |   2 +-
> >   hw/xtensa/virt.c                    |   2 +-
> >   include/hw/misc/pvpanic.h           |   1 +
> >   include/hw/pci-host/gpex.h          |   7 +-
> >   include/hw/pci/pci_ids.h            |   1 +
> >   include/hw/qdev-properties-system.h |   5 +
> >   include/hw/virtio/virtio-blk.h      |  11 +-
> >   include/hw/vmapple/vmapple.h        |  23 +
> >   include/qemu-main.h                 |  14 +-
> >   include/qemu/cutils.h               |  15 +
> >   meson.build                         |   5 +
> >   qapi/virtio.json                    |  14 +
> >   system/main.c                       |  37 +-
> >   ui/cocoa.m                          |  54 +-
> >   ui/gtk.c                            |   4 +
> >   ui/sdl2.c                           |   4 +
> >   util/hexdump.c                      |  18 +
> >   52 files changed, 3791 insertions(+), 110 deletions(-)
> >   create mode 100755 contrib/vmapple/uuid.sh
> >   create mode 100644 docs/system/arm/vmapple.rst
> >   create mode 100644 hw/display/apple-gfx-mmio.m
> >   create mode 100644 hw/display/apple-gfx-pci.m
> >   create mode 100644 hw/display/apple-gfx.h
> >   create mode 100644 hw/display/apple-gfx.m
> >   create mode 100644 hw/misc/pvpanic-mmio.c
> >   create mode 100644 hw/vmapple/Kconfig
> >   create mode 100644 hw/vmapple/aes.c
> >   create mode 100644 hw/vmapple/bdif.c
> >   create mode 100644 hw/vmapple/cfg.c
> >   create mode 100644 hw/vmapple/meson.build
> >   create mode 100644 hw/vmapple/trace-events
> >   create mode 100644 hw/vmapple/trace.h
> >   create mode 100644 hw/vmapple/virtio-blk.c
> >   create mode 100644 hw/vmapple/vmapple.c
> >   create mode 100644 include/hw/vmapple/vmapple.h
> >
>
>
Re: [PATCH v16 00/14] macOS PV Graphics and new vmapple machine type
Posted by Michael S. Tsirkin 1 year, 1 month ago 
On Mon, Dec 23, 2024 at 11:16:31PM +0100, Phil Dennis-Jordan wrote:
> This patch set introduces a new ARM and macOS HVF specific machine type
> called "vmapple", as well as a family of display devices based on the
> ParavirtualizedGraphics.framework in macOS. One of the display adapter
> variants, apple-gfx-mmio, is required for the new machine type, while
> apple-gfx-pci can be used to enable 3D graphics acceleration with x86-64
> macOS guest OSes.
> 
> Previous versions of this patch set were submitted semi-separately:
> the original vmapple patch set by Alexander Graf included a monolithic
> implementation of apple-gfx-mmio. I subsequently reviewed and reworked
> the latter to support the PCI variant of the device as well and submitted
> the result in isolation. As requested in subsequent review, I have now
> recombined this with the original vmapple patch set, which I have updated
> and improved in a few ways as well.
> 
> The vmapple machine type approximates the configuration in macOS's own
> Virtualization.framework when running arm64 macOS guests. In addition to
> generic components such as a GICv3 and an XHCI USB controller, it
> includes nonstandard extensions to the virtio block device, a special
> "hardware" aes engine, a configuration device, a pvpanic variant, a
> "backdoor" interface, and of course the apple-gfx paravirtualised display
> adapter.
> 
> There are currently a few limitations to this which aren't intrinsic,
> just imperfect emulation of the VZF, but it's good enough to be just
> about usable for some purposes:
> 
>  * macOS 12 guests only. Versions 13+ currently fail during early boot.
>  * macOS 11+ arm64 hosts only, with hvf accel. (Perhaps some differences
>    between Apple M series CPUs and TCG's aarch64 implementation? macOS
>    hosts only because ParavirtualizedGraphics.framework is a black box
>    implementing most of the logic behind the apple-gfx device.)
>  * PCI devices use legacy IRQs, not MSI/MSI-X. As far as I can tell,
>    we'd need to include the GICv3 ITS, but it's unclear to me what
>    exactly needs wiring up.
>  * Due to a quirk (bug?) in the macOS XHCI driver when MSI-X is not
>    available, correct functioning of the USB controller (and thus
>    keyboard/tablet) requires a small workaround in the XHCI controller
>    device. This is part of another patch series:
>    https://patchew.org/QEMU/20241208191646.64857-1-phil@philjordan.eu/
>  * The guest OS must first be provisioned using Virtualization.framework;
>    the disk images can subsequently be used in Qemu. (See docs.)
> 
> The apple-gfx device can be used independently from the vmapple machine
> type, at least in the PCI variant. It mainly targets x86-64 macOS guests
> from version 11 on, but also includes a UEFI bootrom for basic
> framebuffer mode. macOS 11 is also required on the host side, as well
> as a GPU that supports the Metal API. On the guest side, this provides
> 3D acceleration/GPGPU support with a baseline Metal feature set,
> irrespective of the host GPU's feature set. A few limitations in the
> current integration:

virtio things:

Reviewed-by: Michael S. Tsirkin <mst@redhat.com>

feel free to merge, if if you want me to pick up
virtio bits, let me know.


>  * Although it works fine with TCG, it does not work correctly
>    cross-architecture: x86-64 guests on arm64 hosts appear to make
>    some boot progress, but rendering is corrupted. I suspect
>    incompatible texture memory layouts; I have no idea if this is
>    fixable.
>  * ParavirtualizedGraphics.framework and the guest driver support
>    multi-headed configurations. The current Qemu integration always
>    connects precisely 1 display.
>  * State serialisation and deserialisation is currently not
>    implemented, though supported in principle by the framework.
>    Both apple-gfx variants thus set up a migration blocker.
>  * Rendering efficiency could be better. The GPU-rendered guest
>    framebuffer is copied to system memory and uses Qemu's usual
>    CPU-based drawing. For maximum efficiency, the Metal texture
>    containing the guest framebuffer could be drawn directly to
>    a Metal view in the host window, staying on the GPU. (Similar
>    to the OpenGL/virgl render path on other platforms.)
> 
> Some of my part of this work has been sponsored by Sauce Labs Inc.
> 
> ---
> 
> v2 -> v3:
> 
>  * Merged the apple-gfx and vmapple patchsets.
>  * Squashed a bunch of later apple-gfx patches into the main one.
>    (dGPU support, queried MMIO area size, host GPU picking logic.)
>  * Rebased on latest upstream, fixing any breakages due to internal
>    Qemu API changes.
>  * apple-gfx: Switched to re-entrant MMIO. This is supported by the
>    underlying framework and simplifies the MMIO forwarding code which
>    was previously different on x86-64 vs aarch64.
>  * vmapple: Fixes for minor bugs and comments from the last round of
>    review.
>  * vmapple aes, conf, apple-gfx: Switched reset methods to implement
>    the ResettableClass base's interface.
>  * vmapple: switched from virtio-hid to an XHCI USB controller and
>    USB mouse and tablet devices. macOS does not provide drivers for
>    virtio HID devices, at least not in version 12's vmapple kernel.
>    So input now sort of works (interrupt issues) rather than not
>    at all. Use network-based remote access to the guest OS as a
>    work-around.
> 
> v3 -> v4:
> 
>  * Complete rework of the mechanism for handling runloop/libdispatch
>    events on the main thread. PV graphics now work with the SDL UI.
>  * Renamed 'apple-gfx-vmapple' device to 'apple-gfx-mmio'
>  * hw/display/apple-gfx: threading model overhaul to be more consistent,
>    safer, and more QEMU-idiomatic.
>  * display-modes property on the apple-gfx devices now uses the
>    native array property mechanism and works on both device variants.
>  * hw/vmapple/aes: Improvements to logging and error handling.
>  * hw/vmapple/cfg: Bug fixes around device property default values.
>  * hw/vmapple/{aes,cfg,virtio-blk/vmapple}: Most header code moved into
>    .c files, only a single vmapple.h now contains the #defines for the
>    vmapple machine model-specific device type names.
>  * hw/block/virtio-blk: New patch for replacing virtio_blk_free_request
>    with g_free. (Optional)
>  * Various smaller changes following comments in v3 code review in
>    apple-gfx, aes, cfg, bdif, virtio-blk-vmapple, and the vmapple
>    machine type itself. See patch-specific v4 change notes for details.
> 
> v4 -> v5:
> 
>  * Simplified the main thread runloop mechanism. Back to setting
> 	 qemu_main directly, but narrowing the scope of what it needs to do,
> 	 and it can now be NULL. (Meaning run the QEMU main event loop on
> 	 the main thread as is traditional.)
>  * hw/display/apple-gfx: Further improvements to the BH based job code bridging
>    the libdispatch & QEMU thread synchronisation impedance mismatch.
>  * hw/display/apple-gfx: Thread safety and object lifetime improvements.
>  * hw/display/apple-gfx-*: Better buffer and error handling in display mode
>    property setters and getters.
>  * hw/vmapple/aes: More consistent and safer logging/tracing
>  * hw/vmapple/cfg: Better error reporting on overlong property strings.
>  * hw/vmapple/virtio-blk: Fixed theoretically-unaligned write to config buffer.
>  * vmapple machine type: Moved ecam region into machine state, improved device
>    property setting error handling, improved ECID/UUID extraction script and
>    docs.
>  * Various smaller fixes in apple-gfx/-mmio, apple-gfx-pci, vmapple/aes,
>    vmapple/cfg, vmapple/virtio-blk, and vmapple machine type.
>  * Added SPDX license identifiers where they were missing.
> 
> v5 -> v6:
> 
>  * 01/15 (main/Cocoa/runloop): Combined functions, fixed whitespace
>  * 02/15 (apple-gfx): Further refinement of PVG threading: reduced some callback
>    tasks from BHs to merely acquiring RCU read lock; replaced some libdispatch
>    tasks with BHs; last remaining synchronous BH now uses emphemeral
>    QemuSemaphore.
>  * 02/15 (apple-gfx): Readability improvements and other smaller tweaks
>    (see patch change notes for details)
>  * 04/15 (display modes): Replaced use of alloca() with NSMutableArray.
> 
> v6 -> v7:
> 
>  * 02/15 (apple-gfx): Use g_ptr_array_find() helper function, coding style tweak
>  * 03/15 (apple-gfx-pci): Removed an unused function parameter
>  * 04/15 (apple-gfx display mode property): Simplified error handling in
>    property parsing.
>  * 10/15 (vmapple/aes): Coding style tweaks.
>  * 12/15 (vmapple/cfg): Changed error messages for overrun of properties with
>    fixed-length strings to be more useful to users than developers.
>  * 15/15 (vmapple machine type): Tiny error handling fix, un-inlined function
> 
> v7 -> v8:
> 
>  * 02/15 (apple-gfx): Naming and type use improvements, fixes for a bug and a
>    leak.
>  * 04/15 (apple-gfx display mode property): Type use improvement
>  * 10/15 (vmapple/aes): Guest error logging tweaks.
>  * 11/15 (vmapple/bdif): Replaced uses of cpu_physical_memory_read with
>    dma_memory_read, and a g_free call with g_autofree.
>  * 12/15 (vmapple/cfg): Macro hygiene fix: consistently enclosing arguments in
>    parens.
>  * 15/15 (vmapple machine type): Use less verbose pattern for defining uuid
>    property.
> 
> v8 -> v9:
> 
>  * 01/16 (ui & main loop): Set qemu_main to NULL for GTK UI as well.
>  * 02/16 (apple-gfx): Pass device pointer to graphic_console_init(), various
> 	 non-functional changes.
>  * 03/16 (apple-gfx-pci): Fixup of changed common call, whitespace and comment
>    formatting tweaks.
>  * 04/16 (apple-gfx display modes): Re-ordered type definitions so we can drop
>    a 'struct' keyword.
>  * 10/16 (vmapple/aes): Replaced a use of cpu_physical_memory_write with
>    dma_memory_write, minor style tweak.
>  * 11/16 (vmapple/bdif): Replaced uses of cpu_physical_memory_write with
>    dma_memory_write.
>  * 13/16 (vmapple/virtio-blk): Correctly specify class_size for
>    VMAppleVirtIOBlkClass.
>  * 15/16 (vmapple machine type): Documentation improvements, fixed variable
>    name and struct field used during pvpanic device creation.
>  * 16/16 (NEW/RFC vmapple/virtio-blk): Proposed change to replace type hierarchy
>    with a variant property. This seems cleaner and less confusing than the
>    original approach to me, but I'm not sure if it warrants creation of a new
>    QAPI enum and property type definition.
> 
> v9 -> v10:
> 
>  * 01/15 (ui & main loop): Added comments to qemu_main declaration and GTK.
>  * 02/15 (apple-gfx): Reworked the way frame rendering code is threaded to use
>    BHs for sections requiring BQL.
>  * 02/15 (apple-gfx): Fixed ./configure error on non-macOS platforms.
>  * 10/15 (vmapple/aes): Code style and comment improvements.
>  * 12/15 (vmapple/cfg): Slightly tidier error reporting for overlong property
>    values.
>  * 13/15 (vmapple/virtio-blk): Folded v9 patch 16/16 into this one, changing
>    the device type design to provide a single device type with a variant
> 	 property instead of 2 different subtypes for aux and root volumes.
>  * 15/15 (vmapple machine type): Documentation fixup for changed virtio-blk
>    device type; small improvements to shell commands in documentation;
>    improved propagation of errors during cfg device instantiation.
> 
> v10 -> v11:
> 
>  * 01/15 (ui & main loop): Simplified main.c, better comments & commit message
>  * 02/15 (apple-gfx): Give each PV display instance a unique serial number.
>  * 02 & 03/15 (apple-gfx, -pci): Formatting/style tweaks
>  * 15/15 (vmapple machine type): Improvements to shell code in docs
> 
> v11 -> v12:
> 
>  * 01/15 (ui & main loop): More precise wording of code comments.
>  * 02/15 (apple-gfx): Fixed memory management regressions introduced in v10;
>    improved error handling; various more conmetic code adjustments
>  * 09/15 (GPEX): Fixed uses of deleted GPEX_NUM_IRQS constant that have been
>    added to QEMU since this patch was originally written.
> 
> v12 -> v13:
> 
>  * 15/15 (vmapple machine type): Bumped the machine type version from 9.2
>    to 10.0.
>  * All patches in the series now have been positively reviewed and received
>    corresponding reviewed-by tags.
> 
> v13 -> v14:
> 
>  * 6/15 (hw/vmapple directory): Changed myself from reviewer
>    to maintainer, as that seemed appropriate at this point.
>  * 15/15 (vmapple machine type): Gate creation of XHCI and
>    USB HID devices behind if (defaults_enabled()).
> 
> v14 -> v15
> 
>  * Constified property tables to match Richard Henderson's recent project-
>    wide convention change. (patches 4/15, 7/15, 11/15, 12/15, & 13/15)
> 
> v15 -> v16
> 
>  * 14 patches now, as patch 8 has already been pulled. (Thanks Philippe!)
>  * Fixed a bunch of conflicts with upstream code motion:
>     - DEFINE_PROP_END_OF_LIST removal (4/14 - apple-gfx mode list, 7/14 -
>       pvpanic-mmio, 10/14 - bdif, 11/14 - cfg device, and
>       12/14 - vmapple-virtio-blk)
>     - sysemu->system move/rename: (1/14 - ui/qemu-main, 2/14 - apple-gfx,
>       9/14 - aes, 10/14 - bdif, 14/14 - vmapple machine type)
>  * 14/14 (vmapple machine type):
>     - Moved compatibility setting for removing legacy mode from virtio-pci
>       to proper global property table rather than (ab)using sugar property.
>     - Removed a few superfluous #includes during sysemu rename cleanup.
>     - Removed machine type versioning as it's not necessary (yet?)
>     - Made memory map array const
> 
> Alexander Graf (8):
>   hw: Add vmapple subdir
>   hw/misc/pvpanic: Add MMIO interface
>   gpex: Allow more than 4 legacy IRQs
>   hw/vmapple/aes: Introduce aes engine
>   hw/vmapple/bdif: Introduce vmapple backdoor interface
>   hw/vmapple/cfg: Introduce vmapple cfg region
>   hw/vmapple/virtio-blk: Add support for apple virtio-blk
>   hw/vmapple/vmapple: Add vmapple machine type
> 
> Phil Dennis-Jordan (6):
>   ui & main loop: Redesign of system-specific main thread event handling
>   hw/display/apple-gfx: Introduce ParavirtualizedGraphics.Framework
>     support
>   hw/display/apple-gfx: Adds PCI implementation
>   hw/display/apple-gfx: Adds configurable mode list
>   MAINTAINERS: Add myself as maintainer for apple-gfx, reviewer for HVF
>   hw/block/virtio-blk: Replaces request free function with g_free
> 
>  MAINTAINERS                         |  15 +
>  contrib/vmapple/uuid.sh             |   9 +
>  docs/system/arm/vmapple.rst         |  63 ++
>  docs/system/target-arm.rst          |   1 +
>  hw/Kconfig                          |   1 +
>  hw/arm/sbsa-ref.c                   |   2 +-
>  hw/arm/virt.c                       |   2 +-
>  hw/block/virtio-blk.c               |  58 +-
>  hw/core/qdev-properties-system.c    |   8 +
>  hw/display/Kconfig                  |  13 +
>  hw/display/apple-gfx-mmio.m         | 288 +++++++++
>  hw/display/apple-gfx-pci.m          | 156 +++++
>  hw/display/apple-gfx.h              |  77 +++
>  hw/display/apple-gfx.m              | 880 ++++++++++++++++++++++++++++
>  hw/display/meson.build              |   7 +
>  hw/display/trace-events             |  30 +
>  hw/i386/microvm.c                   |   2 +-
>  hw/loongarch/virt.c                 |  12 +-
>  hw/meson.build                      |   1 +
>  hw/mips/loongson3_virt.c            |   2 +-
>  hw/misc/Kconfig                     |   4 +
>  hw/misc/meson.build                 |   1 +
>  hw/misc/pvpanic-mmio.c              |  60 ++
>  hw/openrisc/virt.c                  |  12 +-
>  hw/pci-host/gpex.c                  |  43 +-
>  hw/riscv/virt.c                     |  12 +-
>  hw/vmapple/Kconfig                  |  32 +
>  hw/vmapple/aes.c                    | 581 ++++++++++++++++++
>  hw/vmapple/bdif.c                   | 274 +++++++++
>  hw/vmapple/cfg.c                    | 195 ++++++
>  hw/vmapple/meson.build              |   5 +
>  hw/vmapple/trace-events             |  21 +
>  hw/vmapple/trace.h                  |   1 +
>  hw/vmapple/virtio-blk.c             | 204 +++++++
>  hw/vmapple/vmapple.c                | 612 +++++++++++++++++++
>  hw/xen/xen-pvh-common.c             |   2 +-
>  hw/xtensa/virt.c                    |   2 +-
>  include/hw/misc/pvpanic.h           |   1 +
>  include/hw/pci-host/gpex.h          |   7 +-
>  include/hw/pci/pci_ids.h            |   1 +
>  include/hw/qdev-properties-system.h |   5 +
>  include/hw/virtio/virtio-blk.h      |  11 +-
>  include/hw/vmapple/vmapple.h        |  23 +
>  include/qemu-main.h                 |  14 +-
>  include/qemu/cutils.h               |  15 +
>  meson.build                         |   5 +
>  qapi/virtio.json                    |  14 +
>  system/main.c                       |  37 +-
>  ui/cocoa.m                          |  54 +-
>  ui/gtk.c                            |   4 +
>  ui/sdl2.c                           |   4 +
>  util/hexdump.c                      |  18 +
>  52 files changed, 3791 insertions(+), 110 deletions(-)
>  create mode 100755 contrib/vmapple/uuid.sh
>  create mode 100644 docs/system/arm/vmapple.rst
>  create mode 100644 hw/display/apple-gfx-mmio.m
>  create mode 100644 hw/display/apple-gfx-pci.m
>  create mode 100644 hw/display/apple-gfx.h
>  create mode 100644 hw/display/apple-gfx.m
>  create mode 100644 hw/misc/pvpanic-mmio.c
>  create mode 100644 hw/vmapple/Kconfig
>  create mode 100644 hw/vmapple/aes.c
>  create mode 100644 hw/vmapple/bdif.c
>  create mode 100644 hw/vmapple/cfg.c
>  create mode 100644 hw/vmapple/meson.build
>  create mode 100644 hw/vmapple/trace-events
>  create mode 100644 hw/vmapple/trace.h
>  create mode 100644 hw/vmapple/virtio-blk.c
>  create mode 100644 hw/vmapple/vmapple.c
>  create mode 100644 include/hw/vmapple/vmapple.h
> 
> -- 
> 2.39.5 (Apple Git-154)

Patchew 2.1-devel-b67e4c2

© 2016 - 2026 Red Hat, Inc.