Messages in this thread | | | Date | Mon, 14 Jan 2002 12:56:15 -0500 (EST) | From | Alexander Viro <> | Subject | [RFLART] kdev_t in ioctls |
| |
Linus, at least some ioctls (e.g. lvm ones) pass kdev_t from/to userland. While the common policy with ioctls is "anything goes", this kind of abuse is IMNSHO over the top.
Example: ioctl(fd, VG_CREATE, ptr) expects the following: at ptr - struct { /* bunch of sane fields */ struct proc_dir_entry *proc; /* ignored */ pv_t *pv[ABS_MAX_PV + 1]; lv_t *lv[ABS_MAX_LV + 1]; /* bunch of stuff */ }
and pointers in the second array are to the following: struct { /* lots of stuff */ kdev_t lv_dev; /* lots of other stuff */ }
They _are_ dereferenced and values of ptr->lv[i]->lv_dev are stored in kernel data structures. And used afterwards. As kdev_t.
The same goes for the rest of LVM ioctls - pretty much all of them pull such stunts. I'm not going to comment on harmless gross indecencies like struct proc_dir_entry * passed from the userland (and fortunately ignored), but kdev_t instances are _not_ harmless.
Public statement along the lines "any API that passes kdev_t values across the kernel boundary is unacceptable" would be a nice thing...
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |