feature: add architecture-agnostic kernel panic (BSOD) engine

Introduce core.panic, a new architecture-agnostic kernel panic engine
with a pre-allocated LVGL BSOD screen. The screen and all widgets are
created at init time so that zero LVGL allocation occurs at panic time.

Layout features:
- Top banner with decoded teapot TGA image (128x128) and title/subtitle
- Two-column content area with dark semi-transparent cards:
  - Left: Fault Details, CPU Registers, Process Info, System Info
  - Right: Call Stack (full height)
- Pastel red background (#7A2828), monospace text for data sections

Key changes:
- Add core.panic.pas with pre-built BSOD screen, register/trace/system
  info formatting, hex conversion helpers, and syslog fallback
- Add arch.x86.panic.pas as the x86 bridge: packs interrupt registers
  into a TRegisterSnapshot and calls core.panic.panic()
- Simplify all 19 x86 fault handlers to one-liner panic calls
- Remove legacy BSOD code from arch.x86.util.pas
- Add teapot.tga (256x256) embedded via splash_tga.asm
- Add LVGL bindings: lv_refr_now, lv_image_create, lv_image_set_src,
  lv_image_set_scale, lv_image_set_inner_align, and style helpers
- Wire up panic init and test command in asuro.pas

.drone.yml

@@ -22,8 +22,8 @@ steps:
     commands:
       - git fetch --tags
       - find . -type f -print0 | xargs -0 dos2unix
-      - chmod +x /drone/src/*.sh
+      - chmod +x /drone/src/toolchain/*.sh
-      - /drone/src/compile.sh
+      - /drone/src/toolchain/compile.sh
       
   - name: upload-iso-artifact
     image: alpine/git

.gitignore

@@ -14,5 +14,8 @@ localenv.json
 /lvgl/
 dockerout.txt
 AGENTS.md
+lessons_learnt.md
 *.log
 /doc/*.md
+wasuro
+src/core/core.version.pas

Dockerfile

@@ -14,10 +14,9 @@ RUN curl -sL https://sourceforge.net/projects/freepascal/files/Linux/$FPC_VERSIO
 	pushd fpc-$FPC_VERSION.i386-linux && ./install.sh && popd && \
 	rm -rf fpc-$FPC_VERSION.i386-linux
 
-COPY compile.sh /compile.sh
 ADD https://raw.githubusercontent.com/fsaintjacques/semver-tool/master/src/semver /usr/bin/semver
 RUN chmod +x /usr/bin/semver
 WORKDIR /code
 RUN find . -type f -print0 | xargs -0 dos2unix
 ENTRYPOINT ["/bin/bash", "-c"]
-CMD ["/compile.sh"]
+CMD ["find toolchain -name '*.sh' -exec dos2unix {} + 2>/dev/null; find toolchain -name '*.sh' -exec chmod +x {} +; bash toolchain/compile.sh"]

compat/lmemorymanager.pas

@@ -0,0 +1,75 @@
+{ Compatibility shim: wasuro references the old unit name 'lmemorymanager'.
+  Since wasuro/ cannot be modified, this unit re-exports the public API of
+  memory.heap so that 'uses lmemorymanager' continues to compile. }
+unit lmemorymanager;
+
+interface
+
+uses
+    memory.heap;
+
+const
+    ALLOC_UNIT        = memory.heap.ALLOC_UNIT;
+    DATA_OFFSET       = memory.heap.DATA_OFFSET;
+    PAGE_SIZE_LMM     = memory.heap.PAGE_SIZE_LMM;
+    TOTAL_UNITS       = memory.heap.TOTAL_UNITS;
+    BITMAP_DWORDS     = memory.heap.BITMAP_DWORDS;
+    SIZE_PREFIX       = memory.heap.SIZE_PREFIX;
+    LARGE_ALLOC_MAGIC = memory.heap.LARGE_ALLOC_MAGIC;
+
+type
+    PHeapPageHeader = memory.heap.PHeapPageHeader;
+    THeapPageHeader = memory.heap.THeapPageHeader;
+
+procedure init;
+function kalloc(size : uint32) : void;
+function klalloc(size : uint32) : void;
+procedure klfree(address : uint32);
+function kpalloc(address : uint32) : void;
+procedure kfree(area : void);
+function lmm_total_free : uint32;
+function lmm_page_count : uint32;
+
+implementation
+
+procedure init;
+begin
+    memory.heap.init;
+end;
+
+function kalloc(size : uint32) : void; inline;
+begin
+    kalloc := memory.heap.kalloc(size);
+end;
+
+function klalloc(size : uint32) : void; inline;
+begin
+    klalloc := memory.heap.klalloc(size);
+end;
+
+procedure klfree(address : uint32); inline;
+begin
+    memory.heap.klfree(address);
+end;
+
+function kpalloc(address : uint32) : void; inline;
+begin
+    kpalloc := memory.heap.kpalloc(address);
+end;
+
+procedure kfree(area : void); inline;
+begin
+    memory.heap.kfree(area);
+end;
+
+function lmm_total_free : uint32; inline;
+begin
+    lmm_total_free := memory.heap.lmm_total_free;
+end;
+
+function lmm_page_count : uint32; inline;
+begin
+    lmm_page_count := memory.heap.lmm_page_count;
+end;
+
+end.

compat/types.pas

@@ -0,0 +1,7 @@
+unit types;
+
+interface
+
+implementation
+
+end.

compile_sources.sh

@@ -1,13 +0,0 @@
-#!/usr/bin/env bash
-echo " "
-echo "======================="
-echo " "
-echo "Compiling FPC Sources..."
-echo " "
-# Build -Fu flags for all directories under src/ and wasuro/
-FU_PATHS="-Fusrc -Fuwasuro"
-for dir in $(find src wasuro -type d 2>/dev/null); do
-    FU_PATHS="$FU_PATHS -Fu$dir"
-done
-
-fpc -Aelf -gw -g -gl -n -v0ew -O3 -OpPENTIUM3 -Si -Sc -Sg -Xd -CX -XXs -CfSSE -CfSSE2 -Rintel -Pi386 -Tlinux -FElib/ $FU_PATHS src/kernel.pas

compile_stub.sh

@@ -1,7 +0,0 @@
-#!/usr/bin/env bash
-echo " "
-echo "======================="
-echo " "
-echo "Compiling Stub..."
-echo " "
-nasm -f elf src/stub/stub.asm -o lib/stub.o

lessons_learnt.md

@@ -1,40 +0,0 @@
-# Lessons Learnt
-
-## 1. File Pickers Must Be Proper Reusable OS Components
-Inline pickers hacked into app code are broken and not reusable. Build file pickers as a standalone unit (`filepicker.pas`) that any program can call via a callback.
-
-## 2. All Errors Need Visible In-App Feedback
-Never use `syslog.logln` as the only error output. Always show the user a dialog, msgbox, or status bar message. Syslog is supplemental only.
-
-## 4. Shell Command Output Goes to stdout_buf, Not syslog
-Use `stdio.bufWriteStr(stdout_buf, ...)` for all shell command output. `syslog` is for kernel diagnostics only.
-
-## 5. Unit Test Convention
-New subsystem units get a `procedure UnitTest` (called at boot from kernel.pas) and a `procedure init` that registers a CLI command so tests can be re-run at runtime. Tests use a nested `Assert(condition, name)` that increments `passed`/`failed` and logs failures via `syslog.logln`. At-boot tests must not do disk I/O — only test in-memory/virtual structures.
-
-## 6. ISR-Visible Spin-Wait Flags Must Use Pointer Dereferencing
-FPC has no `volatile` keyword. Any `while flag = 0 do asm hlt end` loop where `flag` is set by an ISR will hang forever because FPC hoists the load into a register before the loop and never re-reads memory. Always access the flag through a pointer: `while puint32(@flag)^ = 0 do proc_yield()`. This applies everywhere in the kernel that spins waiting for ISR-written state (e.g. `submit_io_wait`, `proc_await`-style loops, DMA done flags).
-
-## 7. Never Spawn a Process Per I/O Operation
-Spawning a new kernel process for each FAT32 read/write/dir operation is pure overhead: process creation (kalloc 8 KB stack, fake IRET frame, DList insertion), an extra context switch, and a double-wait (VFS spins on the worker, worker spins on AHCI). Instead, submit the I/O directly and park the calling process via `psAwaiting` until the AHCI ISR fires the completion callback — zero extra processes, zero wasted scheduler ticks.
-
-## 8. Use psAwaiting to Park a Process Pending Async I/O (not proc_yield spin)
-The correct way to block a process on an async I/O result is: (1) set up a completion callback that sets `State := psReady` then `Done := 1`, (2) atomically check if the operation is already done (under CLI) and set `CurrentProcess^.State := psAwaiting` if not, (3) spin on `Done` via `puint32(@wait.Done)^` per Lesson #6 with `asm hlt end` inside the loop. The process is removed from the runnable set while hardware works. **Do NOT call `proc_await` from `submit_io_wait`** — `proc_await` spins on `CurrentProcess^.State = psAwaiting` without pointer indirection and FPC caches the value in a register, causing it to return immediately before the I/O completes. The calling code then reads garbage from the DMA buffer, causing invalid pointer dereferences and page faults.
-
-## 9. VFS GetDirectoryListing Must Return a Snapshot, Not a Live Tree Reference
-`GetDirectoryListing` for `otVDIRECTORY` nodes previously returned `PHashMap(Obj^.Reference)` — a direct pointer into the live VFS tree. When callers called `FreeDirectoryListing` on the returned map, it freed the live VFSObjects and their name strings, destroying the VFS tree. Any subsequent `GetObjectFromPath` call would then dereference the trashed memory and page fault. **Fix**: always return a heap-allocated snapshot copy of the directory (new `PHashMap`, new `PVFSObject` entries with `stringCopy`'d keys). `FreeDirectoryListing` can then safely own and free the snapshot without touching the live tree. The rule: any function that returns a `PHashMap` to a caller who will call `FreeDirectoryListing` must allocate a fresh map — never return a raw interior pointer to a live data structure.
-
-## 10. Never Call Synchronous Disk I/O From an LVGL Timer or Event Callback
-LVGL timer callbacks (`lv_timer_create`) and event callbacks run synchronously inside the LVGL render loop. Calling any function that parks the current process waiting for disk I/O (e.g. `vfs.FileSize`, `vfs.OpenFile`, `vfs.ReadFile`) from inside a timer or event callback will hang the entire UI — the process blocks mid-callback and the LVGL loop never resumes. All disk I/O that must occur in a UI context must either: (a) use the async VFS API (`OpenFileAsync`, `WriteFileAsync`) with a completion callback that triggers a deferred LVGL update, or (b) be deferred to a dedicated worker process that signals back via a flag the UI can poll. `GetDirectoryListingFrom` is the one bulk exception — it was already present in the original picker and works because it loads into a snapshot synchronously as a single operation; per-file calls like `FileSize` in a loop are the problem.
-
-## 11. Never Allocate Large Arrays as Local Variables in Deep Kernel Call Chains
-Local variables are stack-allocated in Pascal. Kernel processes have a fixed, limited stack (≈8 KB). In deep LVGL callback chains (kernel → LVGL render loop → timer callback → app code), there may only be a few KB of usable stack remaining by the time app code executes. Declaring local arrays larger than ~200–400 bytes (e.g. `array[0..255] of pchar` = 1024 bytes) inside any procedure that runs in this context will silently overflow the stack, corruptung adjacent memory and causing a page fault with a garbled call trace. **Rule**: any local variable array larger than ~128 bytes in a UI callback or procedure called from one must be moved to the heap with `kalloc`/`kfree`. The corruption pattern is a page fault with a call stack full of completely unrelated function names.
-
-## 12. Pascal Silently Ignores Everything After `end.` — Dead Code Masquerades as Live Code
-FPC compiles everything up to and including the first `end.` and silently discards the rest of the file. If a file is partially rewritten by prepending a new implementation but the old implementation is not deleted, the old code survives after `end.`. Forward declarations resolve to the **last** implementation before `end.` — so if the new code was added before the old `end.` but the old procedures were in the "dead" zone they have no effect. However the reverse is dangerous: if the old `end.` was replaced/removed and the new code placed before it, the old procedures in the dead zone still compile to nothing but pollute the source. **Rule**: after any large-scale replacement of a Pascal unit, explicitly check where `end.` falls with a text search and PowerShell-truncate the file at the first `end.`. Never assume a multi-step replacement tool correctly removed all old code — always verify with `Select-String -Pattern "^end\.$"`.
-
-## 13. `for i := 0 to uint32_count - 1` Wraps to 4 Billion When Count is Zero
-In Free Pascal, `uint32` arithmetic wraps: if `dir_count` is `uint32` and equals `0`, then `dir_count - 1 = 0xFFFFFFFF`. A `for i := 0 to dir_count - 1 do` loop with zero count does NOT skip — it runs 4 billion iterations, immediately accessing invalid memory and page faulting. FPC does not emit a warning for this. **Rule**: always guard `uint32`-bounded for loops: `if count > 0 then for i := 0 to count - 1 do`. Alternatively, use `sint32` for loop counters that might legitimately be zero.
-
-## 14. FAT32 Deleted Entries ($E5) Must Be Filtered in getDirEntries
-When a FAT32 directory entry is deleted, `fileName[0]` is overwritten with `$E5` but the rest of the entry (including the cluster pointer) remains intact on disk. `getDirEntries` must explicitly skip entries where `fileName[0] = char($E5)` — stopping only at `char(0)` (end-of-directory) is not enough. If `$E5` entries are returned to callers, they appear in directory listings with a garbage first character, and navigating into them follows a stale/freed cluster chain, which causes reads into arbitrary memory or FAT data, leading to a triple fault. Fix location: the `while i < maxEntries` loop in `fat32.getDirEntries`.

src/app/app.base64.pas

@@ -17,12 +17,12 @@
 	
 	@author(Angus C <[email protected]>)
 }
-unit base64_prog;
+unit app.base64;
 
 interface
 
 uses
-    stdio, util, strings, tracer, base64, lmemorymanager;
+    io.stdio, core.util, arch.x86.util, core.strings, debug.tracer, core.enc.base64, memory.heap;
 
 procedure init();
 
@@ -38,7 +38,7 @@ var
     PSize     : uInt32;
 
 begin
-    tracer.push_trace('base64_prog.run');
+    debug.tracer.push_trace('base64_prog.run');
     if paramCount(Params) > 1 then begin
         encdec := getParam(0, Params);
         if stringEquals(encdec, 'encode') then begin
@@ -58,23 +58,23 @@ begin
             dec(pinput);
             pinput^ := #0;
             result := b64_encode_str(input);
-            stdio.bufWriteStrLn(stdout_buf, result);
+            io.stdio.bufWriteStrLn(stdout_buf, result);
             kfree(void(result));
         end else if stringEquals(encdec, 'decode') then begin
             input := getParam(1, Params);
             result := b64_decode_str(input);
-            stdio.bufWriteStrLn(stdout_buf, result);
+            io.stdio.bufWriteStrLn(stdout_buf, result);
             kfree(void(result));
-        end else stdio.bufWriteStrLn(stderr_buf, 'Usage: base64 <encode/decode> <text>');
+        end else io.stdio.bufWriteStrLn(stderr_buf, 'Usage: core.enc.base64 <encode/decode> <text>');
     end else begin
-        stdio.bufWriteStrLn(stderr_buf, 'Usage: base64 <encode/decode> <text>');
+        io.stdio.bufWriteStrLn(stderr_buf, 'Usage: core.enc.base64 <encode/decode> <text>');
     end;
 end;
 
 procedure init();
 begin
-    tracer.push_trace('base64_prog.init');
+    debug.tracer.push_trace('base64_prog.init');
-    stdio.registerCommand('BASE64', @Run, 'Perform Base64 Encode/Decode.');
+    io.stdio.registerCommand('BASE64', @Run, 'Perform Base64 Encode/Decode.');
 end;
 
 end.

src/app/app.dhclient.pas

@@ -17,12 +17,12 @@
 	
 	@author(Kieron Morris <[email protected]>)
 }
-unit dhclient;
+unit app.dhclient;
 
 interface
 
 uses
-    stdio, util, strings, tracer, dhcp;
+    io.stdio, core.util, arch.x86.util, core.strings, debug.tracer, driver.net.proto.dhcp;
 
 procedure init();
 
@@ -30,14 +30,14 @@ implementation
 
 procedure run(Params : PParamList; stdin_buf, stdout_buf, stderr_buf : POutBuf);
 begin
-    tracer.push_trace('dhclient.run');
+    debug.tracer.push_trace('dhclient.run');
     DHCPDiscover();
 end;
 
 procedure init();
 begin
-    tracer.push_trace('dhclient.init');
+    debug.tracer.push_trace('dhclient.init');
-    stdio.registerCommand('DHClient', @Run, 'Run the DHCP configuration utility.');
+    io.stdio.registerCommand('DHClient', @Run, 'Run the DHCP configuration utility.');
 end;
 
 end.