update

tests/TC108-lzma-large-file.md

@@ -0,0 +1,49 @@
+# Test Case: LZMA Large File Decompression
+
+**Test ID**: TC108-lzma-large-file  
+
+**G5 fixture contract**: Use [G5-MANUAL-SETUP.md](G5-MANUAL-SETUP.md).
+Its generated paths, source comparisons, structured corruption offsets, and
+cleanup rules supersede placeholder examples in this file.
+**Category**: Compression  
+**Priority**: High  
+**Regression**: None
+
+## Objective
+Verify LZMA decompression works correctly for large files (100MB+).
+
+## Preconditions
+- `images/lzma-large.erofs`, generated with LZMA level 6
+- `/large.bin` and `sources/lzma-large/large.bin`, exactly 104857601 bytes
+- FreeBSD 13.0+ with liblzma
+
+## Test Steps
+1. Create image with large LZMA-compressed file
+2. Attach and mount: `unit=$(mdconfig -a -t vnode -f test.img); mount -t erofs -o ro /dev/${unit} /mnt`
+3. Compare complete source and guest SHA256, then run
+   `timeout 1800 cmp SOURCE /mnt/large.bin`.
+4. Compare 4096 bytes at offsets 0, 52428800, and 103809024 with the source.
+5. Record `vmstat -m` while mounted and after cleanup; active EROFS
+   allocations must return to zero.
+
+## Expected Results
+- Large file reads successfully
+- Checksum matches original
+- Random access works at any offset
+- Memory usage < 256MB during decompression
+
+## Verification Method
+- SHA256 checksum verification
+- Random read correctness at offsets: 0, 50MB, 99MB
+- Memory usage monitored via top
+- No kernel panics or OOM
+
+## Cleanup
+```sh
+umount /mnt
+mdconfig -d -u "${unit#md}"
+rm -f /tmp/out
+```
+
+## Notes
+Tests feature 60: LZMA decompression with large files.