Nothing is using it yet, but this will allow us to delay the open-time checks to use time, without breaking the normal UNIX permission semantics where permissions are determined by the opener (and the file descriptor can then be passed to a different process, or the process can drop capabilities).
Showing the changes to kernel (1 of 2 files). Show all
Unified · Split
| @@ -392,6 +392,30 @@ bool ns_capable(struct user_namespace *ns, int cap) · 391 unmodified lines | ||
| 392 | 392 | } |
| 393 | 393 | EXPORT_SYMBOL(ns_capable); |
| 394 | 394 | |
| 395 | +/** | |
| 396 | + * file_ns_capable - Determine if the file's opener had a capability in effect | |
| 397 | + * @file: The file we want to check | |
| 398 | + * @ns: The usernamespace we want the capability in | |
| 399 | + * @cap: The capability to be tested for | |
| 400 | + * | |
| 401 | + * Return true if task that opened the file had a capability in effect | |
| 402 | + * when the file was opened. | |
| 403 | + * | |
| 404 | + * This does not set PF_SUPERPRIV because the caller may not | |
| 405 | + * actually be privileged. | |
| 406 | + */ | |
| 407 | +bool file_ns_capable(const struct file *file, struct user_namespace *ns, int cap) | |
| 408 | +{ | |
| 409 | + if (WARN_ON_ONCE(!cap_valid(cap))) | |
| 410 | + return false; | |
| 411 | + | |
| 412 | + if (security_capable(file->f_cred, ns, cap) == 0) | |
| 413 | + return true; | |
| 414 | + | |
| 415 | + return false; | |
| 416 | +} | |
| 417 | +EXPORT_SYMBOL(file_ns_capable); | |
| 418 | + | |
| 395 | 419 | /** |
| 396 | 420 | * capable - Determine if the current task has a superior capability in effect |
| 397 | 421 | * @cap: The capability to be tested for |