A common pattern for syscall extensions is increasing the size of a struct passed from userspace, such that the zero-value of the new fields result in the old kernel behaviour (allowing for a mix of userspace and kernel vintages to operate on one another in most cases). While this interface exists for communication in both directions, only one interface is straightforward to have reasonable semantics for (userspace passing a struct to the kernel). For kernel returns to userspace, what the correct semantics are (whether there should be an error if userspace is unaware of a new extension) is very syscall-dependent and thus probably cannot be unified between syscalls (a good example of this problem is [1]). Previously there was no common lib/ function that implemented the necessary extension-checking semantics (and different syscalls implemented them slightly differently or incompletely[2]). Future patches replace common uses of this pattern to make use of copy_struct_from_user(). Some in-kernel selftests that insure that the handling of alignment and various byte patterns are all handled identically to memchr_inv() usage. [1]: commit 1251201c0d34 ("sched/core: Fix uclamp ABI bug, clean up and robustify sched_read_attr() ABI logic and code") [2]: For instance {sched_setattr,perf_event_open,clone3}(2) all do do similar checks to copy_struct_from_user() while rt_sigprocmask(2) always rejects differently-sized struct arguments.
Showing the changes to include/linux (2 of 5 files). Show all
Unified · Split
| @@ -4,6 +4,13 @@ · 3 unmodified lines | |||
| 4 | #include <asm/types.h> | 4 | #include <asm/types.h> |
| 5 | #include <linux/bits.h> | 5 | #include <linux/bits.h> |
| 6 | 6 | ||
| 7 | /* Set bits in the first 'n' bytes when loaded from memory */ | ||
| 8 | #ifdef __LITTLE_ENDIAN | ||
| 9 | # define aligned_byte_mask(n) ((1UL << 8*(n))-1) | ||
| 10 | #else | ||
| 11 | # define aligned_byte_mask(n) (~0xffUL << (BITS_PER_LONG - 8 - 8*(n))) | ||
| 12 | #endif | ||
| 13 | |||
| 7 | #define BITS_PER_TYPE(type) (sizeof(type) * BITS_PER_BYTE) | 14 | #define BITS_PER_TYPE(type) (sizeof(type) * BITS_PER_BYTE) |
| 8 | #define BITS_TO_LONGS(nr) DIV_ROUND_UP(nr, BITS_PER_TYPE(long)) | 15 | #define BITS_TO_LONGS(nr) DIV_ROUND_UP(nr, BITS_PER_TYPE(long)) |
| 9 | 16 | ||
| @@ -231,6 +231,76 @@ __copy_from_user_inatomic_nocache(void *to, const void __user *from, · 230 unmodified lines | |||
| 231 | 231 | ||
| 232 | #endif /* ARCH_HAS_NOCACHE_UACCESS */ | 232 | #endif /* ARCH_HAS_NOCACHE_UACCESS */ |
| 233 | 233 | ||
| 234 | extern __must_check int check_zeroed_user(const void __user *from, size_t size); | ||
| 235 | |||
| 236 | /** | ||
| 237 | * copy_struct_from_user: copy a struct from userspace | ||
| 238 | * @dst: Destination address, in kernel space. This buffer must be @ksize | ||
| 239 | * bytes long. | ||
| 240 | * @ksize: Size of @dst struct. | ||
| 241 | * @src: Source address, in userspace. | ||
| 242 | * @usize: (Alleged) size of @src struct. | ||
| 243 | * | ||
| 244 | * Copies a struct from userspace to kernel space, in a way that guarantees | ||
| 245 | * backwards-compatibility for struct syscall arguments (as long as future | ||
| 246 | * struct extensions are made such that all new fields are *appended* to the | ||
| 247 | * old struct, and zeroed-out new fields have the same meaning as the old | ||
| 248 | * struct). | ||
| 249 | * | ||
| 250 | * @ksize is just sizeof(*dst), and @usize should've been passed by userspace. | ||
| 251 | * The recommended usage is something like the following: | ||
| 252 | * | ||
| 253 | * SYSCALL_DEFINE2(foobar, const struct foo __user *, uarg, size_t, usize) | ||
| 254 | * { | ||
| 255 | * int err; | ||
| 256 | * struct foo karg = {}; | ||
| 257 | * | ||
| 258 | * if (usize > PAGE_SIZE) | ||
| 259 | * return -E2BIG; | ||
| 260 | * if (usize < FOO_SIZE_VER0) | ||
| 261 | * return -EINVAL; | ||
| 262 | * | ||
| 263 | * err = copy_struct_from_user(&karg, sizeof(karg), uarg, usize); | ||
| 264 | * if (err) | ||
| 265 | * return err; | ||
| 266 | * | ||
| 267 | * // ... | ||
| 268 | * } | ||
| 269 | * | ||
| 270 | * There are three cases to consider: | ||
| 271 | * * If @usize == @ksize, then it's copied verbatim. | ||
| 272 | * * If @usize < @ksize, then the userspace has passed an old struct to a | ||
| 273 | * newer kernel. The rest of the trailing bytes in @dst (@ksize - @usize) | ||
| 274 | * are to be zero-filled. | ||
| 275 | * * If @usize > @ksize, then the userspace has passed a new struct to an | ||
| 276 | * older kernel. The trailing bytes unknown to the kernel (@usize - @ksize) | ||
| 277 | * are checked to ensure they are zeroed, otherwise -E2BIG is returned. | ||
| 278 | * | ||
| 279 | * Returns (in all cases, some data may have been copied): | ||
| 280 | * * -E2BIG: (@usize > @ksize) and there are non-zero trailing bytes in @src. | ||
| 281 | * * -EFAULT: access to userspace failed. | ||
| 282 | */ | ||
| 283 | static __always_inline __must_check int | ||
| 284 | copy_struct_from_user(void *dst, size_t ksize, const void __user *src, | ||
| 285 | size_t usize) | ||
| 286 | { | ||
| 287 | size_t size = min(ksize, usize); | ||
| 288 | size_t rest = max(ksize, usize) - size; | ||
| 289 | |||
| 290 | /* Deal with trailing bytes. */ | ||
| 291 | if (usize < ksize) { | ||
| 292 | memset(dst + size, 0, rest); | ||
| 293 | } else if (usize > ksize) { | ||
| 294 | int ret = check_zeroed_user(src + size, rest); | ||
| 295 | if (ret <= 0) | ||
| 296 | return ret ?: -E2BIG; | ||
| 297 | } | ||
| 298 | /* Copy the interoperable parts of the struct. */ | ||
| 299 | if (copy_from_user(dst, src, size)) | ||
| 300 | return -EFAULT; | ||
| 301 | return 0; | ||
| 302 | } | ||
| 303 | |||
| 234 | /* | 304 | /* |
| 235 | * probe_kernel_read(): safely attempt to read from a location | 305 | * probe_kernel_read(): safely attempt to read from a location |
| 236 | * @dst: pointer to the buffer that shall take the data | 306 | * @dst: pointer to the buffer that shall take the data |