feat(exec): enable loading ELF binaries from ext2 filesystem #90
+1,313
−7
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add load_elf_from_ext2() helper function that loads ELF binaries from
the ext2 filesystem with proper permission checks:
- Resolves path to inode using ext2 driver
- Validates file is not a directory (ENOTDIR)
- Checks execute permission bit S_IXUSR (EACCES)
- Returns file content as Vec<u8>
Modify sys_execv_with_frame() to try ext2 filesystem first:
- For paths containing '/': load directly from ext2
- For bare names: try /bin/<name> on ext2, fall back to test disk
This enables exec("/bin/ls", argv) to load programs from the ext2
filesystem, preparing for full Unix-like program execution. The
test disk fallback ensures backward compatibility during transition.
Co-Authored-By: Claude Opus 4.5 <[email protected]>
|
You have run out of free Bugbot PR reviews for this billing cycle. This will reset on February 12. To receive reviews on all of your PRs, visit the Cursor dashboard to activate Pro and start your 14-day free trial. |
Fixed critical bugs in ext2 block allocation that caused filesystem
corruption when writing files:
1. allocate_block: Was calculating global_block as (bg_index * blocks_per_group + local_block),
but ext2 bitmap bit 0 represents block s_first_data_block (usually 1 for 1KB blocks),
not block 0. Fixed to add s_first_data_block to the calculation.
2. free_block: Same off-by-one error in reverse - was calculating local_block as
(block_num % blocks_per_group) instead of ((block_num - s_first_data_block) % blocks_per_group).
3. truncate_file: Was clearing inode block pointers without freeing the blocks in the
bitmap, causing block leaks. When O_TRUNC reopened a file, the old block stayed
marked "in use" so allocate_block would return a different (already-in-use) block,
corrupting unrelated files.
The combination of these bugs caused /bin directory corruption: when fs_write_test
truncated and rewrote hello.txt, it would allocate /bin's data block (296) instead
of hello.txt's original block (290), overwriting the directory entries.
Also adds exec_from_ext2_test which verifies:
- exec("/bin/hello_world") loads and runs binary from ext2
- exec("/nonexistent") returns ENOENT
- exec("/hello.txt") returns EACCES (not executable)
- exec("/test") returns ENOTDIR/EACCES for directories
All 216 boot stages now pass.
Co-Authored-By: Claude Opus 4.5 <[email protected]>
Add kernel unit tests and userspace regression tests to validate the ext2 block allocation fixes from commit d190da5. Kernel unit tests (kernel/src/fs/ext2/block_group.rs): - 11 tests with MockBlockDevice for allocate_block/free_block arithmetic - test_allocate_block_1kb_first_data_block_offset: bit 0 → block 1 - test_allocate_block_bit_295_returns_block_296: exact bug scenario - test_free_block_1kb_offset: block 296 clears bit 295 - test_truncate_frees_blocks_to_bitmap: multi-block free + realloc Userspace regression test (userspace/tests/fs_block_alloc_test.rs): - Test 1: O_TRUNC properly frees blocks (not just clears size) - Test 2: Multi-file corruption regression (reproduces original bug) - Test 3: Block reuse after truncate These tests provide regression safety for the s_first_data_block offset bugs in allocate_block() and free_block(), and the block leak bug in truncate_file(). Co-Authored-By: Claude Opus 4.5 <[email protected]>
This file was referenced in Cargo.toml but not committed, causing CI failure. simple_exit.rs is a minimal test binary that exits with code 42, used for testing exec from ext2 filesystem. Co-Authored-By: Claude Opus 4.5 <[email protected]>
The load_elf_from_ext2() function was logging 10+ messages per exec syscall, adding significant overhead from serial I/O. This caused timing issues in CI's slower environment, breaking TCP tests. Since most binaries fall back to the test disk, every exec was: 1. Logging "loading '/bin/program_name'" 2. Scanning ext2 filesystem 3. Logging the failure 4. Falling back to test disk Removed all logging from the exec path. Debug issues with GDB, not logs. Co-Authored-By: Claude Opus 4.5 <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
load_elf_from_ext2()helper function that loads ELF binaries from the ext2 filesystem with proper permission checkssys_execv_with_frame()to try ext2 filesystem first, with test disk fallback for backward compatibilityChanges
New helper function
load_elf_from_ext2():Modified exec syscall behavior:
/(e.g.,/bin/ls): load directly from ext2 filesystemls): try/bin/<name>on ext2 first, fall back to test diskTest plan
/bindirectory (requires ext2 tools)Next steps (Phase 2b)
🤖 Generated with Claude Code