Simple, kernel-only, 64bit OS, early stage. https://git.2137697.xyz/rs-dos/atom?h=qemu 2026-08-21T01:32:07Z 2026-08-21T01:32:07Z salmonik salmonik@tuta.io 2026-08-21T01:32:07Z urn:sha1:956093be4f3b7a0fe86a523e5f9d9e74b9a77e59 focus() only composites the title bar — the window content stays stale until the next write. Call term.flush() on Focus(true) so the full window is re-composited immediately when switching to it. 2026-08-20T11:47:10Z salmonik salmonik@tuta.io 2026-08-20T11:47:10Z urn:sha1:cb17c57452ccd2ccca77e99098402d4b9f61c0a5 Architecture: - vt: ACTIVE terminal redirect — when set, all gop_print!/put_char/write_str go there instead of HOME. Terminal::from_buffer() renders into a WM-owned pixel buffer. - tty: read_line_from_queue() — canonical-mode line input from the task's own message queue (TaskMsg::Key). Returns None on Close. - shell: run_windowed() — same shell as the console but reads from the message queue. Returns on ^D or window close. - wm: F1 calls SPAWN_TERMINAL_CB (set by rsdos). No shell logic in the WM. - rsdos: terminal_task callback + terminal_entry task — creates a 640x400 chrome window, wires up Terminal::from_buffer, sets ACTIVE, runs shell::run_windowed(), cleans up on exit. - shell: 'term' debug command that invokes the callback directly and logs each step to serial. 2026-08-19T16:46:35Z salmonik salmonik@tuta.io 2026-08-19T16:46:35Z urn:sha1:09a44f15678f27dd3d07f90e9ed0e71cf0e771ff - wm rewrite: chrome flag per window, create_home_window for the console (window 0, chrome-less, bottom-pinned, never closed), composite_rect fills desktop background then blits windows bottom-to-top, draw_window_intersect handles chrome-less windows, hit_test returns Content for chrome-less, destroy/move_window no-op for id 0. - vt: add window_params() so main.rs can wire the Terminal to wm window 0. - main.rs: create home terminal window 0 and attach Terminal after wm::init(). - Focus fix: focus(0) now stores CONSOLE sentinel (not raw 0) so focused() returns None when the home terminal has focus. Without this, closing a GUI window left FOCUSED=0, which tty::poll_key() interpreted as 'a GUI app has focus', spinning forever and swallowing all keyboard input. 2026-08-17T18:28:48Z salmonik salmonik@tuta.io 2026-08-17T18:28:48Z urn:sha1:7bd3459f4f8bae69509202e5aa0dbd04a42c656f - Refactor vt.rs into a per-instance Terminal struct (owned cell/rendered grids, cursor, ANSI parser, scroll state), rendering glyphs into a caller-owned pixel buffer with dirty-region blitting - Add gop buffer primitives (buffer_glyph_px, buffer_shift_up) and keep the legacy vt:: API as thin wrappers around the home terminal - Remove the now-obsolete scroll hook and shadow scroll path - Home terminal renders into its own buffer; window_id stays None until the WM integration lands 2026-08-17T17:08:47Z salmonik salmonik@tuta.io 2026-08-17T17:08:47Z urn:sha1:f4fab5a2b06c37df08d2d315bcb44d15af1786a4 The console's batched scroll (gop::shift_up_n) shifts the whole shared shadow, which dragged any composited window pixels up with it. Add a scroll hook so the window manager re-composites every visible window after a console scroll, restoring them at their fixed positions. 2026-08-17T16:56:04Z salmonik salmonik@tuta.io 2026-08-17T16:56:04Z urn:sha1:5558e3509d020a0e021a77f0617044497b2f2aee - Add the wm window-manager facility per ADR 0004: overlapping pixel-surface windows, damage-rectangle compositing, WM-drawn title bar, focus/minimize/close, and F1-F4 keybindings - Add gop shadow compositing primitives and vt::repaint_region so the console stays layer 0 beneath windows - Route mouse (absolute coords, press/release edges) and keyboard (focus-aware input pump) to the focused window - Add the rspaint drawing-program crate (pen, 8 colors, clear) and shell draw/windows/restore commands - Extend TaskMsg with AppMouse/Focus/Close and document the design in CONTEXT.md + ADR 0004 2026-08-17T12:53:52Z salmonik salmonik@tuta.io 2026-08-17T12:53:52Z urn:sha1:be868a43185826b02bf6d091019bb779678525bb - Add polled EHCI (USB2) and UHCI (USB1.1) host-controller drivers with HID boot mouse support, wired into cursor polling and the 'usb' shell cmd - Add USB Legacy ownership handoff (claim_usb_legacy) plus config_write_byte - Harden boot: verify the PIT/timer ticks before shell entry, EOI stray interrupts in the LAPIC to stop interrupt storms, bound PS/2 controller waits, and INTx-disable AHCI (polled driver) - Add boot stage markers, klog screen echo, and QR lock; scheduler mark_shell_running; NO_USB stopgap to keep BIOS USB-Legacy input alive - alloc_pages_above for DMA buffers that must avoid low conventional memory - Expand DEBUG_NOTES with EHCI/QEMU bring-up findings 2026-08-03T23:44:05Z salmonik salmonik@tuta.io 2026-08-03T23:44:05Z urn:sha1:3e2261d83176752f3ea21e0c97f330b4c1757437 Root cause: on real hardware SetMode can invalidate the pre-captured ExitBootServices map key, so EBS fails and the retry refreshed the map via capture_memory_map -> init_heap, which REBUILT the free list from the raw map and wiped the allocation state of every pre-EBS live object (the display shadow and the vt grids). Their spans became free while still live, later allocations carved inside the shadow, and the display writes clobbered them — zeroed back-pointers (family A ptr=0x404443e0 block=0x0) and gray glyph pixels (family B 0xaaaaaa00aaaaaa = 0x00AAAAAA = PALETTE[7] fg-7 text colour, not a firmware scrub). Fix: refresh_and_exit now re-captures the map and returns a fresh EBS key without rebuilding the free list (refresh_map_key); the free list is the authoritative live-allocation state. Verified in QEMU under a forced EBS failure (SIMULATE_EBS_FAILURE test hook): verify ok, probe NOT free, flhead outside the shadow — identical to a clean boot, and the full shell sequence is crash-free. Also: removed the crash-family debug markers from the screen (boot-time heapdbg lines, vt verify marker, shadow marker) — the vt grid self-check stays serial-only. Version 0.5.8 -> 0.5.9. 2026-08-03T00:23:35Z salmonik salmonik@tuta.io 2026-08-03T00:23:35Z urn:sha1:9fde94d068f87b34e0bc6bfa48388d89bb54f1e4 Real hardware has no serial console, so a kernel crash only leaves a frozen screen. Every dump_lines() report now also renders the crash text as a scannable QR below the white-on-red band (photograph the screen instead of relaying text). - rsnix/src/qr.rs: from-scratch no_std/no_alloc byte-mode QR encoder (EC-L, v1..v40, .bss statics only — safe on the fault path): RS over GF(256), codeword interleave, zigzag placement, 8 masks with N1-N4 penalty auto-selection, format/version info. - gop.rs: dump_qr() joins dump lines into a static buffer, truncates with '...' at the v40-L capacity, encodes, and renders centered black-on-white with a 4-module quiet zone. - tools/qrcheck: host harness (embedds qr.rs) + qrcheck.py cross-check against segno's placement primitives. Verified bit-exact: 12 versions x 8 masks forced, plus auto-mask choice matches segno's penalty evaluation. Caught 3 encoder bugs (multi-block interleave offsets, transposed version info, mask eval ordering). - QEMU e2e: page-fault dump screendump decodes via zbarimg, content matches the serial log line-for-line. - test-msg.sh and test-fault.sh both PASS; debug+release 0 warnings. 2026-08-02T23:34:31Z salmonik salmonik@tuta.io 2026-08-02T23:34:31Z urn:sha1:d5f41d6e057300c82b83653b4afadeffb45a9f19 The pre-verify 0x70020 'mystery' was a debug artifact, not corruption: BLANK = {ch:32, fg:7, bg:0} = raw 0x70020, and vec![BLANK; n] fills both grids with it at init. The old check compared cells against 0 and its probe loop break-ed on the first non-zero cell, so probes 1-3 stayed at their zero-initialized slots — c=[0x70020,0,0,0] was bogus on every healthy boot. A full-grid dump proves QEMU grids are 100% BLANK at shell entry (nz=3700 blank=3700). Fix: probes compare against blank_raw (0x70020), record all four values (no early break), probe both cells and rendered, and add a full-grid nz/blank summary line. The OptiPlex firmware 0x00AAAAAA scrub remains cleanly distinguishable (scrub != blank). Markers: the shell banner occupies grid rows 0-1 and the prompt row 2 (proven by screendump), and a stale crash.log replays at rows 2+, so the shadow and vt markers moved to the bottom of the screen via the new gop::text_rows()/bottom_row(off) helpers instead of rows 1-2. QEMU verified: verify: ok every boot; test-msg + test-fault PASS; debug and release builds 0 warnings.