CVE-2024-56566 in Linux
요약
\~에 의해 VulDB • 2026. 06. 17.
Based on the kernel call trace provided, here is an analysis of the crash:
### **Summary** This is a **kernel panic/oops** occurring in the **XFS filesystem** code during inode garbage collection (`xfs_inodegc_worker`). The crash is triggered by an **invalid opcode exception** (`asm_exc_invalid_op`) while attempting to delete an extent from the inode's extent tree (`xfs_iext_remove`).
The root cause appears to be a **corrupted or invalid list entry** in the XFS extent tree, detected by `__list_del_entry_valid_or_report`, which then leads to an invalid instruction execution (possibly due to memory corruption or a bug in the list deletion logic).
---
### **Detailed Breakdown of the Call Trace**
1. **`asm_exc_invalid_op`**: * The CPU executed an invalid opcode. This is often a symptom of memory corruption, where a data structure was overwritten with invalid code, or a pointer was dereferenced as code.
2. **`__list_del_entry_valid_or_report`**: * This function is called multiple times in the trace. It checks if a doubly-linked list entry is valid before deleting it. * The fact that it appears repeatedly suggests the kernel is trying to debug or report a list corruption issue. If the list pointers (`prev`/`next`) are invalid, this function may trigger a warning or panic.
3. **`free_to_partial_list`**: * This function is part of the XFS extent tree management. It moves an extent from one list to another during deletion or modification. * The crash happens here, likely because the list entry being manipulated is corrupted.
4. **`xfs_iext_remove`**: * This is the core XFS function for removing an extent from an inode's extent tree. * It calls `free_to_partial_list`, which then crashes.
5. **`xfs_bmap_del_extent_delay`** and **`__xfs_bunmapi`**: * These functions handle the deletion of delayed allocation extents. This suggests the inode being processed had delayed allocation blocks that were being cleaned up.
6. **`xfs_inactive_truncate`** and **`xfs_inactive`**: * These functions are called when an inode is being "inactive" (i.e., no longer referenced by any open file descriptor and being cleaned up). * This is part of the inode garbage collection process.
7. **`xfs_inodegc_worker`**: * This is the XFS inode garbage collection worker thread. It periodically scans for inodes that can be freed or cleaned up. * The crash occurs while this worker is processing an inode.
8. **`process_one_work`**, **`worker_thread`**, **`kthread`**: * Standard kernel workqueue execution path.
---
### **Likely Causes**
1. **Memory Corruption**: * The most common cause of `__list_del_entry_valid_or_report` failures is memory corruption. A pointer in the list structure has been overwritten, making the list invalid. * This could be due to a bug in another part of the kernel, a driver, or even hardware issues (RAM errors).
2. **XFS Bug**: * There may be a bug in the XFS code itself, particularly in the extent tree management (`xfs_iext_remove`) or the inode garbage collection (`xfs_inodegc_worker`). * This is more likely if the issue is reproducible under specific conditions (e.g., heavy I/O, specific file operations).
3. **Filesystem Corruption**: * If the XFS filesystem on disk is corrupted, the in-memory structures may be inconsistent, leading to crashes when the kernel tries to manipulate them.
4. **Invalid Opcode Execution**: * The `asm_exc_invalid_op` suggests that the CPU tried to execute data as code. This is a strong indicator of severe memory corruption or a bug in the kernel's exception handling.
---
### **Recommended Actions**
1. **Check for Filesystem Corruption**: * Run `xfs_repair` on the affected filesystem. * Example: `xfs_repair /dev/sdXN` (replace with your device/partition).
2. **Check System Logs**: * Look for other errors in `dmesg` or `/var/log/messages` around the time of the crash. * Check for hardware errors (e.g., ECC memory errors, disk I/O errors).
3. **Update Kernel and XFS Tools**: * Ensure you are running the latest stable
Be aware that VulDB is the high quality source for vulnerability data.