Write boot milestones directly to VGA text buffer (0xB8000) at each
init stage so the user can see exactly where the kernel stops on
platforms where the VGA driver may not initialize.
Milestones: Magic OK, GDT, IDT, PIC, PMM, PAGING, HEAP, CPIO, VFS,
INITRD, TSS. Each appears as white-on-blue text.
If the multiboot2 magic check fails, display the received magic value
in red-on-white on screen and halt (instead of silently returning).
This helps diagnose if GRUB matched the multiboot1 header instead.
Display fixes for UTM/Mac black screen issue:
- Remove MULTIBOOT_HEADER_TAG_FRAMEBUFFER from multiboot2 header (was required,
requesting text mode depth=0 which confused GRUB on some platforms)
- Add COM1 serial port (0x3F8) output alongside debugcon for UTM serial capture
- Change VGA background from black to dark blue for diagnostics
- Add early canary write to 0xB8000 ('COS' in magenta) before subsystem init
- print_hex now outputs to both debugcon and COM1
New ls command and SYS_READDIR syscall:
- SYS_READDIR (10): reads directory entries via VFS
- VFS root listing: vfs_readdir handles '/' by iterating mount table
- apps/ls: lists CWD contents, appends '/' for directories
- apps/libc/syscalls.h: readdir() wrapper
When GRUB is configured with multiboot2, it may provide a graphical
(RGB) framebuffer instead of legacy VGA text mode. This happens on
UTM/QEMU on macOS and other configurations where the bootloader
switches to a pixel-based display.
- Parse multiboot2 framebuffer tag in kernel_main to detect display mode
- Identity-map the framebuffer address (often at 0xFD000000+) after
paging is enabled, with write-through caching
- Add framebuffer.h describing the boot-time display info
- Embed 8x16 VGA bitmap font (font8x16.h) for pixel-mode rendering
- Rewrite VGA driver to support both text mode and pixel mode:
- Text mode: unchanged behavior writing to 0xB8000
- Pixel mode: renders characters using the bitmap font to the
GRUB-provided framebuffer, with proper VGA color palette mapping
- Auto-detects mode from fb_info at init time
- Multiboot2 header now requests text mode via framebuffer tag, but
gracefully falls back to pixel rendering if GRUB provides RGB
- Reverted grub.cfg gfxpayload=text (caused display issues on UTM)
Tested: boots in both text mode and graphical framebuffer mode.
When using multiboot2, GRUB may set a graphical framebuffer instead
of text mode, causing VGA text writes to 0xB8000 to be invisible.
- Add MULTIBOOT_HEADER_TAG_FRAMEBUFFER to multiboot2 header requesting
80x25 text mode (depth=0)
- Add 'set gfxpayload=text' to grub.cfg as additional safeguard
This fixes blank display when running under UTM/QEMU with certain
VGA device configurations.
- Created env.c/h: key=value store with ENV_MAX_VARS=32 entries per
process. Supports get, set, unset, and copy operations.
- Added env_block_t and cwd[128] fields to process_t struct.
- process_create() initializes CWD=/ by default.
- Fork inherits environment via memcpy of the entire process_t.
- Added SYS_GETENV (8) and SYS_SETENV (9) system calls.
- Created env-test user app that verifies: CWD default, set/get,
and inheritance across fork.
- Updated kernel.c to launch 'sh' as init process (for next task).
- Updated README.md to check off environment variables task.
Tested: env-test reads CWD=/, sets TEST=hello, reads it back,
forks child which inherits TEST=hello. All verified via QEMU.
- Added paging_clone_directory_from(): deep-copies user-space pages so
parent and child have independent memory. Kernel pages are shared.
- Fixed process_fork() to accept registers_t* for accurate child state,
and to clone from the parent's page directory (not the kernel's).
- Refactored process_exit() to properly context-switch to next process
using new process_switch_to_user assembly stub (loads full registers_t
and performs iret), instead of halting unconditionally.
- Fixed sys_waitpid() to use proper blocking: marks process BLOCKED,
invokes scheduler, and resumes with exit code when child dies.
- Added SYSCALL_SWITCHED mechanism to prevent syscall_handler from
clobbering the next process's EAX after a context switch.
- Created fork-test user app that validates fork + waitpid.
- Added docs/fork.md with architecture documentation.
Tested: fork-test creates child, both print messages, parent waits for
child exit (code 7), parent reaps and exits (code 0). hello-world also
verified to still work correctly after the process_exit refactor.
- Created apps/hello-world/hello-world.S: AT&T assembly user program that
calls SYS_WRITE to print 'Hello, World' then SYS_EXIT(0)
- Created apps/user.ld: linker script for user apps at 0x08048000
- Created scripts/build_apps.sh: builds each app dir into flat binary
- Updated CMakeLists.txt: added apps build target in ISO pipeline
- Updated gen_initrd.sh: packs built binaries from build/apps_bin/
- Updated kernel.c: replaced inline machine code with VFS-based loading
of hello-world from /initrd/hello-world via cpio_find()
Verified: hello-world binary (49 bytes) loads from CPIO initrd,
prints 'Hello, World', and exits with code 0.
VFS subsystem (vfs.c/h):
- Mount table with up to 16 mount points, longest-prefix path matching.
- File descriptor table (256 entries, fds 0-2 reserved for std streams).
- Path resolution walks mount table then delegates to filesystem's
finddir() for each path component.
- Operations: open, close, read, write, seek, readdir, stat.
- Each filesystem driver provides a vfs_fs_ops_t with callbacks.
Initrd filesystem driver (initrd_fs.c/h):
- Read-only VFS driver backed by the CPIO ramdisk.
- Mounted at '/initrd' during boot.
- Zero-copy reads: file data points directly into the CPIO archive
memory, no allocation or copying needed.
- Supports readdir (flat iteration) and finddir (name lookup).
Bug fix: resolve_path was overwriting file-specific fs_data (set by
finddir, e.g. pointer to CPIO file data) with the mount's fs_data
(NULL). Fixed to preserve fs_data from finddir.
Verified in QEMU: kernel reads /initrd/README via VFS and prints its
contents. Ring 3 user process continues to work.
Build system changes:
- scripts/gen_initrd.sh packs all files from apps/ into a newc-format
CPIO archive at build/isodir/boot/initrd.cpio.
- CMakeLists.txt adds 'initrd' target as ISO dependency. GRUB loads the
archive as a Multiboot2 module via 'module2 /boot/initrd.cpio'.
- apps/README added as placeholder file for initial ramdisk content.
Kernel changes:
- kernel.c scans Multiboot2 tags for MULTIBOOT_TAG_TYPE_MODULE to find
the initrd's physical address range, then passes it to cpio_init().
- cpio.c/h implements a parser for the SVR4/newc CPIO format:
- cpio_init(): lists archive contents on startup
- cpio_find(): look up a file by name (handles ./ prefix)
- cpio_next(): iterate through all entries
- cpio_count(): count files in archive
- The initrd lives in identity-mapped physical memory, so no additional
mapping is needed to access it.
Verified in QEMU: GRUB loads the module at 0x0014A000, CPIO parser
finds the README file (38 bytes). All existing functionality (Ring 3
processes, syscalls) continues to work.
Add complete user-mode process support:
- TSS (tss.c/h): Task State Segment for Ring 3->0 transitions, installed
as GDT entry 5 (selector 0x28). ESP0 updated per-process for kernel
stack switching.
- Process management (process.c/h): Process table with up to 64 processes.
process_create() clones kernel page directory, maps user code at
0x08048000 and user stack at 0xBFFFF000, copies flat binary code.
Round-robin scheduler via schedule_tick() modifies the interrupt frame
in-place for zero-copy context switching.
- System calls (syscall.c/h): INT 0x80 dispatcher with 8 syscalls:
SYS_EXIT, SYS_WRITE (to debug port + VGA), SYS_READ, SYS_FORK,
SYS_GETPID, SYS_YIELD, SYS_WAITPID, SYS_EXEC. IDT gate at 0x80
uses DPL=3 (flags 0xEE) so user code can invoke it.
- Assembly stubs (interrupts.S): isr128 for INT 0x80, tss_flush for
loading the Task Register, enter_usermode for initial iret to Ring 3.
- Paging extensions (paging.c/h): paging_clone_directory() to create
per-process page directories, paging_map_page_in() for mapping into
non-active directories, paging_switch_directory() for CR3 switching.
- GDT expanded from 5 to 6 entries to accommodate TSS descriptor.
gdt_set_gate() exposed in header for TSS initialization.
- ISR handler routes timer IRQ (32) to scheduler and INT 0x80 to
syscall dispatcher. Exception handler now prints EIP/CS/ERR for
debugging.
- Kernel boots a test user program that writes 'Hello from Ring 3!'
via SYS_WRITE and exits with code 42 via SYS_EXIT. Verified working
in QEMU.
Context switching approach: Timer/syscall interrupts save all registers
via the ISR stub. schedule_tick() copies saved_regs between PCBs and
overwrites the interrupt frame, so the existing iret restores the next
process's state without separate switch assembly.
- Created VGA driver that writes to the 0xB8000 text-mode framebuffer.
Supports 80x25 display with 16 foreground/background colors, scrolling,
hardware cursor updates, and special characters (\n, \r, \t, \b).
- Provides vga_puts, vga_putchar, vga_put_hex, vga_put_dec, vga_set_color.
- Displays boot banner ("ClaudeOS v0.1 booting...") on screen clear.
- vga_show_mem_stats() prints total RAM, kernel start/end addresses, and
kernel size on the VGA display during boot.
- Registered as the first driver using REGISTER_DRIVER, proving the
driver framework works end-to-end (probe -> init lifecycle).
Tested: driver loads successfully, debug port confirms vga probe/init.
- Created driver framework with probe/init lifecycle. Drivers register via
REGISTER_DRIVER macro which places pointers in a .drivers linker section.
- During boot, init_drivers() iterates the section, probes each driver
(checking if hardware is present), and initializes those that respond OK.
- Added .drivers section to linker.ld with __drivers_start/__drivers_end
symbols for iteration.
- Also added .rodata.* pattern to the .rodata section for string literals
placed in sub-sections by the compiler.
- No drivers are registered yet; the VGA driver will be the first.
Tested: boots cleanly with driver scan completing (0 registered, 0 loaded).
- Added first-fit free-list allocator with block splitting and coalescing.
Provides kmalloc(), kfree(), and kcalloc() for kernel-space dynamic memory.
- Each block carries an inline header with a magic value (0xCAFEBABE) for
heap corruption detection, plus double-free checking.
- Memory is obtained from the paging subsystem in 4 KiB page increments.
All allocations are 8-byte aligned with a 16-byte minimum block size.
- Created freestanding string.c with memset, memcpy, memmove, memcmp,
strlen, strcmp, strncmp, strcpy, strncpy — replacing the unavailable
libc implementations.
- Added documentation in docs/kmalloc.md.
Tested: kmalloc(64) returns 0xD0001010 (in kernel heap) and kfree succeeds.
Works with both 4 MiB and 128 MiB RAM.
- Created two-level x86 paging (page directory + page tables) with 4 KiB pages.
- Identity maps all detected physical memory in two phases:
1) Static: first 16 MiB using 4 BSS-allocated page tables (avoids
chicken-and-egg with PMM bitmap in BSS).
2) Dynamic: memory above 16 MiB using PMM-allocated page tables,
created before paging is enabled so physical addresses still work.
- Provides kernel heap at 0xD0000000–0xF0000000 for virtual page allocation.
- API: paging_map_page, paging_unmap_page, paging_alloc_page, paging_free_page,
paging_get_physical.
- Added pmm_get_memory_size() to expose detected RAM for paging init.
- Kernel tests paging by allocating a virtual page, writing 0xDEADBEEF, and
reading it back, then freeing it.
- Added documentation in docs/paging.md.
Tested: boots and passes paging test with both 4 MiB and 128 MiB RAM in QEMU.
- Changed grub.cfg from 'multiboot' to 'multiboot2' command. The PMM parses
Multiboot2 tag structures, but GRUB was booting with Multiboot1 protocol,
causing the memory map parsing to silently fail (all memory stayed marked
as used, leading to OOM on every allocation).
- Fixed BITMAP_SIZE calculation to properly round up instead of truncating,
ensuring the last few pages of the address space are covered.
- Fixed sign comparison warning in bitmap init loop.
- Added debug output to PMM init (mem_upper, region count) for diagnostics.
- Removed stale Multiboot1 magic constant and (void)addr cast from kernel.c.
- Added documentation for the interrupt subsystem and PMM in docs/.
- Checked off 'Implement a PIC handler' and 'Create a physical memory
allocator' in the task list.
Tested: kernel boots in QEMU with both 4MB and 128MB RAM, PMM correctly
allocates from NORMAL zone (0x01000000) and DMA zone (0x00001000).
- Added bitmap-based physical memory manager (PMM) that parses the Multiboot2
memory map to discover available RAM regions.
- Supports two allocation zones: PMM_ZONE_DMA (below 16MB) and PMM_ZONE_NORMAL
(above 16MB), with automatic fallback from NORMAL to DMA when the preferred
zone is exhausted.
- Marks kernel memory region and multiboot info structure as reserved using
_kernel_start/_kernel_end linker symbols.
- Page 0 is always marked as used to prevent returning NULL as a valid address.
- Added linker script symbols (_kernel_start, _kernel_end) to track kernel
memory boundaries for the allocator.
- Kernel now initializes PMM after PIC and performs test allocations to verify
the subsystem works.
- Reworked IDT initialization to register all 32 CPU exception handlers (ISR 0-31)
and 16 hardware interrupt handlers (IRQ 0-15, mapped to IDT entries 32-47).
- Created assembly stubs in interrupts.S using macros for ISRs with and without
error codes, plus IRQ stubs. All route through a common stub that saves
registers, loads kernel data segment, and calls the C handler.
- Added isr.c with a unified interrupt dispatcher that handles both exceptions
(halts on fault) and hardware IRQs (sends EOI via PIC).
- Implemented PIC (8259) driver in pic.c with full initialization sequence that
remaps IRQ 0-7 to IDT 32-39 and IRQ 8-15 to IDT 40-47. Includes mask/unmask
and EOI support.
- Extracted port I/O primitives (inb, outb, io_wait) into port_io.h header for
reuse across drivers.
- Kernel now initializes PIC after IDT and enables interrupts with STI.
This commit adds GDT initialization with proper code/data segments and reloads CS.
It also adds the initial IDT structure and loads an empty IDT.
Build configuration updated to disable SSE/MMX to prevent compiler generation of unsupported instructions in early boot.
