Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 3 additions & 2 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@ ZAR-Zig-Agent-Runtime is the Zig runtime port of OpenClaw, with parity-first del
- Original OpenClaw baseline (`v2026.3.13-1`): `100/100` covered
- Original OpenClaw beta baseline (`v2026.3.13-beta.1`): `100/100` covered
- Union baseline: `141/141` covered (`MISSING_IN_ZIG=0`)
- Latest local validation: `zig build test --summary all` -> `398/398` passed
- Latest local validation: `zig build test --summary all` -> `399/399` passed
- Current edge release target tag: `v0.2.0-zig-edge.31`
- License posture: repo-wide `GPL-2.0-only` with Linux-style SPDX headers on repo-owned source and script files
- Toolchain policy: Codeberg `master` is canonical; `adybag14-cyber/zig` publishes rolling `latest-master` and immutable `upstream-<sha>` Windows releases for refresh and reproducibility.
Expand Down Expand Up @@ -44,8 +44,9 @@ ZAR-Zig-Agent-Runtime is the Zig runtime port of OpenClaw, with parity-first del
- `src/pal/framebuffer.zig` exposes the framebuffer path through the bare-metal PAL and the bare-metal ABI now supports bounded mode switching plus supported-mode table export through `oc_framebuffer_set_mode`, `oc_framebuffer_supported_mode_count`, `oc_framebuffer_supported_mode_width`, and `oc_framebuffer_supported_mode_height`
- `src/baremetal/edid.zig`, `src/baremetal/display_output.zig`, and `src/baremetal/virtio_gpu.zig` now provide EDID-backed display capability export plus bounded render/present/flush proof for the first real controller-specific path, `virtio-gpu-pci`, including digital-input, preferred-timing, CEA, DisplayID, HDMI-vendor-data, and basic-audio capability flags when present in EDID data
- `src/baremetal/display_output.zig` now derives the exported connector type from EDID capability flags instead of hard-coding the virtio-gpu path as always `virtual`
- the same display-output surface now exports bounded per-output entries through `oc_display_output_entry_count` and `oc_display_output_entry`, and the broad host/tool surface now exposes `display-outputs` and `display-output <index>` on top of that table
- `scripts/baremetal-qemu-framebuffer-console-probe-check.ps1` now proves live MMIO banner pixels plus exported adapter metadata against the freestanding PVH artifact at `640x400`, `1024x768`, and `1280x720`
- `scripts/baremetal-qemu-virtio-gpu-display-probe-check.ps1` now proves live `virtio-gpu-pci` EDID/controller capability export plus resource-create/attach/set-scanout/flush behavior with non-zero scanout pixel readback over QEMU with `edid=on`
- `scripts/baremetal-qemu-virtio-gpu-display-probe-check.ps1` now proves live `virtio-gpu-pci` EDID/controller capability export plus output-entry count/entry metadata and resource-create/attach/set-scanout/flush behavior with non-zero scanout pixel readback over QEMU with `edid=on`
- real HDMI/DisplayPort connector-specific scanout paths are still future depth and are not claimed by this branch
- keyboard/mouse is now strict-closed in [`docs/zig-port/FS5_5_HARDWARE_DRIVERS_SYSTEMS.md`](docs/zig-port/FS5_5_HARDWARE_DRIVERS_SYSTEMS.md)
- `src/baremetal/ps2_input.zig` now contains a real x86 port-I/O backed PS/2 controller path
Expand Down
8 changes: 4 additions & 4 deletions docs/zig-port/FS5_5_HARDWARE_DRIVERS_SYSTEMS.md
Original file line number Diff line number Diff line change
Expand Up @@ -145,10 +145,10 @@ Current local source-of-truth evidence:
- a real EDID-backed display capability path now exists beyond the rendered BGA console:
- `src/baremetal/edid.zig` provides bounded EDID header/checksum/timing/name parsing
- EDID parsing now also exports capability flags for digital input, preferred timing, CEA extension presence, DisplayID extension presence, HDMI vendor data, and basic audio when those descriptors are present
- `src/baremetal/display_output.zig` provides the exported display-output ABI surface plus EDID byte export
- `src/baremetal/virtio_gpu.zig` probes the first real controller-specific path, `virtio-gpu-pci`, through modern virtio PCI capabilities plus `GET_DISPLAY_INFO`, `GET_EDID`, bounded 2D resource creation, guest-backing attach, scanout selection, transfer-to-host, and flush
- `src/baremetal/display_output.zig` provides the exported display-output ABI surface plus EDID byte export and a bounded per-output entry table
- `src/baremetal/virtio_gpu.zig` probes the first real controller-specific path, `virtio-gpu-pci`, through modern virtio PCI capabilities plus `GET_DISPLAY_INFO`, `GET_EDID`, bounded multi-scanout enumeration, connector-aware scanout selection, bounded 2D resource creation, guest-backing attach, transfer-to-host, and flush
- `src/pal/framebuffer.zig` now also exposes the display-output state and EDID byte surface through the PAL seam
- host regressions now prove the framebuffer export surface updates host-backed framebuffer state, glyph pixels, supported-mode enumeration, high-resolution mode switching, and preservation of the last valid mode on unsupported requests
- host regressions now prove the framebuffer export surface updates host-backed framebuffer state, glyph pixels, supported-mode enumeration, high-resolution mode switching, per-output entry export, and preservation of the last valid mode on unsupported requests
- a live bare-metal PVH/QEMU proof now passes:
- `scripts/baremetal-qemu-framebuffer-console-probe-check.ps1`
- exported framebuffer state has `magic=framebuffer_magic`, `api_version=2`, and now proves `640x400` (`cols=80`, `rows=25`), `1024x768` (`cols=128`, `rows=48`), and `1280x720` (`cols=160`, `rows=45`) surfaces over the same BGA path
Expand All @@ -159,7 +159,7 @@ Current local source-of-truth evidence:
- a second live bare-metal PVH/QEMU proof now passes:
- `scripts/baremetal-qemu-virtio-gpu-display-probe-check.ps1`
- exported display-output state has `magic=display_output_magic`, `api_version=2`, `backend=virtio_gpu`, `controller=virtio_gpu`, an EDID-derived connector type, and a real EDID header over `virtio-gpu-pci,edid=on`
- runtime now also reports the selected virtio-gpu PCI vendor/device, PCI location, active scanout, current mode, preferred mode, physical dimensions, manufacturer/product IDs, exported EDID byte surface, and the exported capability flags derived from the EDID payload
- runtime now also reports the selected virtio-gpu PCI vendor/device, PCI location, active scanout, current mode, preferred mode, physical dimensions, manufacturer/product IDs, exported EDID byte surface, the exported capability flags derived from the EDID payload, and the bounded per-output entry export for the selected scanout
- the same proof now also validates non-zero present statistics plus non-zero scanout pixels from the guest-backed render pattern after resource-create/attach/set-scanout/flush
- current real source-of-truth rendered display support now covers bounded Bochs/QEMU BGA mode-setting plus virtio-gpu present/flush over the virtual scanout path
- real HDMI/DisplayPort connector-specific scanout paths are not yet implemented and are not claimed by this branch
Expand Down
Loading
Loading