Bug 94791
| Summary: | syscall fanotify_mark overflow. | ||
|---|---|---|---|
| Product: | File System | Reporter: | Guowei Du (du.guowei2) |
| Component: | ext4 | Assignee: | fs_ext4 (fs_ext4) |
| Status: | NEW --- | ||
| Severity: | normal | CC: | du.guowei2, eparis, szg00000, tim.pickersgill |
| Priority: | P1 | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Kernel Version: | 4.x | Subsystem: | |
| Regression: | No | Bisected commit-id: | |
A __u64 value is an unsigned long long, a 64-bit integer, regardless of system bitness. Not sure how/why this would overflow on a 32-bit system. What symptoms are you seeing? We're using fanotify_mark() on 32-bit systems without seeing any issue here. Not an ext4 issue, I'd suggest that this be reclassified as "Other" and assigned to "fs_other@kernel-bugs.osdl.org" or Eric Paris (eparis@redhat.com) Tinkering with Eric Paris's fanotify-example (http://git.infradead.org/users/eparis/fanotify-example.git), to forcibly set higher bits in the mask passed to fanotify_mark(), the call correctly returns -EINVAL on both 32- and 64-bit kernels. Suggest this bug should be rejected unless a failing example can be provided. |
file: fanotify_usr.c: syscall function fanotify_mark,below code will overflow on 32-bit system. /* we only use the lower 32 bits as of right now. */ if ( mask & ((__u64)0xffffffff << 32)) return -EINVAL; include/linux/syscalls.h: asmlinkage long sys_fanotify_mark(int fanotify_fd, unsigned int flags, u64 mask, int fd, const char __usr *pathname); u64 mask ,should convert to unsigned int mask???