1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533
|
/* SPDX-License-Identifier: MIT */
/*
* io_uring_register.c
*
* Description: Unit tests for the io_uring_register system call.
*
* Copyright 2019, Red Hat, Inc.
* Author: Jeff Moyer <jmoyer@redhat.com>
*/
#include <stdio.h>
#include <fcntl.h>
#include <string.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
#include <sys/sysinfo.h>
#include <poll.h>
#include <assert.h>
#include <sys/uio.h>
#include <sys/mman.h>
#include <linux/mman.h>
#include <sys/time.h>
#include <sys/resource.h>
#include <sys/vfs.h>
#include <limits.h>
#include "helpers.h"
#include "liburing.h"
#include "../src/syscall.h"
static int pagesize;
static rlim_t mlock_limit;
static int devnull;
static int expect_fail(int fd, unsigned int opcode, void *arg,
unsigned int nr_args, int error, int error2)
{
int ret;
ret = io_uring_register(fd, opcode, arg, nr_args);
if (ret >= 0) {
int ret2 = 0;
fprintf(stderr, "expected %s, but call succeeded\n", strerror(error));
if (opcode == IORING_REGISTER_BUFFERS) {
ret2 = io_uring_register(fd, IORING_UNREGISTER_BUFFERS,
0, 0);
} else if (opcode == IORING_REGISTER_FILES) {
ret2 = io_uring_register(fd, IORING_UNREGISTER_FILES, 0,
0);
}
if (ret2) {
fprintf(stderr, "internal error: failed to unregister\n");
exit(1);
}
return 1;
}
if (ret != error && (error2 && ret != error2)) {
fprintf(stderr, "expected %d/%d, got %d\n", error, error2, ret);
return 1;
}
return 0;
}
static int new_io_uring(int entries, struct io_uring_params *p)
{
int fd;
fd = io_uring_setup(entries, p);
if (fd < 0) {
perror("io_uring_setup");
exit(1);
}
return fd;
}
#define MAXFDS (UINT_MAX * sizeof(int))
#define OFS_MAGIC 0x794c7630
#define TMPFS_MAGIC 0x01021994
#define RAMFS_MAGIC 0x858458f6
static void *map_filebacked(size_t size)
{
struct statfs buf;
int fd, ret;
void *addr;
char template[32] = "io_uring_register-test-XXXXXXXX";
fd = mkstemp(template);
if (fd < 0) {
perror("mkstemp");
return NULL;
}
if (statfs(template, &buf) < 0) {
perror("statfs");
unlink(template);
close(fd);
return NULL;
}
unlink(template);
/* virtual file systems may not present as file mapped */
if (buf.f_type == OFS_MAGIC || buf.f_type == RAMFS_MAGIC ||
buf.f_type == TMPFS_MAGIC) {
close(fd);
return NULL;
}
ret = ftruncate(fd, size);
if (ret < 0) {
perror("ftruncate");
close(fd);
return NULL;
}
addr = mmap(NULL, size, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0);
if (addr == MAP_FAILED) {
perror("mmap");
close(fd);
return NULL;
}
close(fd);
return addr;
}
/*
* NOTE: this is now limited by SCM_MAX_FD (253). Keep the code for now,
* but probably should augment it to test 253 and 254, specifically.
*/
static int test_max_fds(int uring_fd)
{
int status = 1;
int ret;
void *fd_as; /* file descriptor address space */
int fdtable_fd; /* fd for the file that will be mapped over and over */
int io_fd; /* the valid fd for I/O -- /dev/null */
int *fds; /* used to map the file into the address space */
char template[32] = "io_uring_register-test-XXXXXXXX";
unsigned long long i, nr_maps, nr_fds;
/*
* First, mmap anonymous the full size. That will guarantee the
* mapping will fit in the memory area selected by mmap. Then,
* over-write that mapping using a file-backed mapping, 128MiB at
* a time using MAP_FIXED.
*/
fd_as = mmap(NULL, UINT_MAX * sizeof(int), PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
if (fd_as == MAP_FAILED) {
if (errno == ENOMEM)
return 0;
perror("mmap fd_as");
exit(1);
}
fdtable_fd = mkstemp(template);
if (fdtable_fd < 0) {
perror("mkstemp");
exit(1);
}
unlink(template);
ret = ftruncate(fdtable_fd, 128*1024*1024);
if (ret < 0) {
perror("ftruncate");
exit(1);
}
io_fd = open("/dev/null", O_RDWR);
if (io_fd < 0) {
perror("open /dev/null");
exit(1);
}
fds = mmap(fd_as, 128*1024*1024, PROT_READ|PROT_WRITE,
MAP_SHARED|MAP_FIXED, fdtable_fd, 0);
if (fds == MAP_FAILED) {
perror("mmap fdtable");
exit(1);
}
/* fill the fd table */
nr_fds = 128*1024*1024 / sizeof(int);
for (i = 0; i < nr_fds; i++)
fds[i] = io_fd;
/* map the file through the rest of the address space */
nr_maps = (UINT_MAX * sizeof(int)) / (128*1024*1024);
for (i = 0; i < nr_maps; i++) {
fds = &fds[nr_fds]; /* advance fds by 128MiB */
fds = mmap(fds, 128*1024*1024, PROT_READ|PROT_WRITE,
MAP_SHARED|MAP_FIXED, fdtable_fd, 0);
if (fds == MAP_FAILED) {
fprintf(stderr, "mmap failed at offset %lu\n",
(unsigned long)((char *)fd_as - (char *)fds));
exit(1);
}
}
/* Now fd_as points to the file descriptor array. */
/*
* We may not be able to map all of these files. Let's back off
* until success.
*/
nr_fds = UINT_MAX;
while (nr_fds) {
ret = io_uring_register(uring_fd, IORING_REGISTER_FILES, fd_as,
nr_fds);
if (ret != 0) {
nr_fds /= 2;
continue;
}
status = 0;
ret = io_uring_register(uring_fd, IORING_UNREGISTER_FILES, 0, 0);
if (ret < 0) {
errno = -ret;
perror("io_uring_register UNREGISTER_FILES");
exit(1);
}
break;
}
close(io_fd);
close(fdtable_fd);
ret = munmap(fd_as, UINT_MAX * sizeof(int));
if (ret != 0) {
fprintf(stderr, "munmap(%zu) failed\n", UINT_MAX * sizeof(int));
exit(1);
}
return status;
}
static int test_memlock_exceeded(int fd)
{
int ret;
void *buf;
struct iovec iov;
/* if limit is larger than 2gb, just skip this test */
if (mlock_limit >= 2 * 1024 * 1024 * 1024ULL)
return 0;
iov.iov_len = mlock_limit * 2;
buf = t_malloc(iov.iov_len);
iov.iov_base = buf;
while (iov.iov_len) {
ret = io_uring_register(fd, IORING_REGISTER_BUFFERS, &iov, 1);
if (ret == -ENOMEM) {
iov.iov_len /= 2;
continue;
} else if (ret == -EFAULT) {
free(buf);
return 0;
} else if (ret) {
fprintf(stderr, "expected success or EFAULT, got %d\n", ret);
free(buf);
return 1;
}
ret = io_uring_register(fd, IORING_UNREGISTER_BUFFERS, NULL, 0);
if (ret != 0) {
fprintf(stderr, "error: unregister failed with %d\n", ret);
free(buf);
return 1;
}
break;
}
if (!iov.iov_len)
printf("Unable to register buffers. Check memlock rlimit.\n");
free(buf);
return 0;
}
static int test_iovec_nr(int fd)
{
int i, ret, status = 0;
unsigned int nr = 1000000;
struct iovec *iovs;
void *buf;
iovs = malloc(nr * sizeof(struct iovec));
if (!iovs) {
fprintf(stdout, "can't allocate iovecs, skip\n");
return 0;
}
buf = t_malloc(pagesize);
for (i = 0; i < nr; i++) {
iovs[i].iov_base = buf;
iovs[i].iov_len = pagesize;
}
status |= expect_fail(fd, IORING_REGISTER_BUFFERS, iovs, nr, -EINVAL, 0);
/* reduce to UIO_MAXIOV */
nr = UIO_MAXIOV;
ret = io_uring_register(fd, IORING_REGISTER_BUFFERS, iovs, nr);
if ((ret == -ENOMEM || ret == -EPERM) && geteuid()) {
fprintf(stderr, "can't register large iovec for regular users, skip\n");
} else if (ret != 0) {
fprintf(stderr, "expected success, got %d\n", ret);
status = 1;
} else {
io_uring_register(fd, IORING_UNREGISTER_BUFFERS, 0, 0);
}
free(buf);
free(iovs);
return status;
}
/*
* io_uring limit is 1G. iov_len limit is ~OUL, I think
*/
static int test_iovec_size(int fd)
{
unsigned int status = 0;
int ret;
struct iovec iov;
void *buf;
/* NULL pointer for base */
iov.iov_base = 0;
iov.iov_len = 4096;
status |= expect_fail(fd, IORING_REGISTER_BUFFERS, &iov, 1, -EFAULT, 0);
/* valid base, 0 length */
iov.iov_base = &buf;
iov.iov_len = 0;
status |= expect_fail(fd, IORING_REGISTER_BUFFERS, &iov, 1, -EFAULT, 0);
/* valid base, length exceeds size */
/* this requires an unampped page directly after buf */
buf = mmap(NULL, 2 * pagesize, PROT_READ|PROT_WRITE,
MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
assert(buf != MAP_FAILED);
ret = munmap(buf + pagesize, pagesize);
assert(ret == 0);
iov.iov_base = buf;
iov.iov_len = 2 * pagesize;
status |= expect_fail(fd, IORING_REGISTER_BUFFERS, &iov, 1, -EFAULT, 0);
munmap(buf, pagesize);
/* huge page */
buf = mmap(NULL, 2*1024*1024, PROT_READ|PROT_WRITE,
MAP_PRIVATE | MAP_HUGETLB | MAP_HUGE_2MB | MAP_ANONYMOUS,
-1, 0);
if (buf == MAP_FAILED) {
printf("Unable to map a huge page. Try increasing "
"/proc/sys/vm/nr_hugepages by at least 1.\n");
printf("Skipping the hugepage test\n");
} else {
/*
* This should succeed, so long as RLIMIT_MEMLOCK is
* not exceeded
*/
iov.iov_base = buf;
iov.iov_len = 2*1024*1024;
ret = io_uring_register(fd, IORING_REGISTER_BUFFERS, &iov, 1);
if (ret < 0) {
if (ret == -ENOMEM)
printf("Unable to test registering of a huge "
"page. Try increasing the "
"RLIMIT_MEMLOCK resource limit by at "
"least 2MB.");
else {
fprintf(stderr, "expected success, got %d\n", ret);
status = 1;
}
} else {
ret = io_uring_register(fd, IORING_UNREGISTER_BUFFERS,
0, 0);
if (ret < 0) {
fprintf(stderr, "io_uring_unregister: %s\n",
strerror(-ret));
status = 1;
}
}
}
ret = munmap(iov.iov_base, iov.iov_len);
assert(ret == 0);
/* file-backed buffers -- not supported */
buf = map_filebacked(2*1024*1024);
if (buf) {
iov.iov_base = buf;
iov.iov_len = 2*1024*1024;
status |= expect_fail(fd, IORING_REGISTER_BUFFERS, &iov, 1, -EFAULT, -EOPNOTSUPP);
munmap(buf, 2*1024*1024);
}
/* bump up against the soft limit and make sure we get EFAULT
* or whatever we're supposed to get. NOTE: this requires
* running the test as non-root. */
if (getuid() != 0)
status |= test_memlock_exceeded(fd);
return status;
}
static int ioring_poll(struct io_uring *ring, int fd, int fixed)
{
int ret;
struct io_uring_sqe *sqe;
struct io_uring_cqe *cqe;
sqe = io_uring_get_sqe(ring);
memset(sqe, 0, sizeof(*sqe));
sqe->opcode = IORING_OP_POLL_ADD;
if (fixed)
sqe->flags = IOSQE_FIXED_FILE;
sqe->fd = fd;
sqe->poll_events = POLLIN|POLLOUT;
ret = io_uring_submit(ring);
if (ret != 1) {
fprintf(stderr, "failed to submit poll sqe: %d.\n", ret);
return 1;
}
ret = io_uring_wait_cqe(ring, &cqe);
if (ret < 0) {
fprintf(stderr, "io_uring_wait_cqe failed with %d\n", ret);
return 1;
}
ret = 0;
if (!(cqe->res & POLLOUT)) {
fprintf(stderr, "io_uring_wait_cqe: expected 0x%.8x, got 0x%.8x\n",
POLLOUT, cqe->res);
ret = 1;
}
io_uring_cqe_seen(ring, cqe);
return ret;
}
static int __test_poll_ringfd(int ring_flags)
{
int status = 0;
int ret;
int fd;
struct io_uring ring;
ret = io_uring_queue_init(2, &ring, ring_flags);
if (ret) {
perror("io_uring_queue_init");
return 1;
}
fd = ring.ring_fd;
/* try polling the ring fd */
status = ioring_poll(&ring, fd, 0);
/*
* now register the ring fd, and try the poll again. This should
* fail, because the kernel does not allow registering of the
* ring_fd.
*/
status |= expect_fail(fd, IORING_REGISTER_FILES, &fd, 1, -EBADF, 0);
/* tear down queue */
io_uring_queue_exit(&ring);
return status;
}
static int test_poll_ringfd(void)
{
int ret;
ret = __test_poll_ringfd(0);
if (ret)
return ret;
return __test_poll_ringfd(IORING_SETUP_SQPOLL);
}
int main(int argc, char **argv)
{
int fd, ret;
unsigned int status = 0;
struct io_uring_params p;
struct rlimit rlim;
if (argc > 1)
return T_EXIT_SKIP;
/* setup globals */
pagesize = getpagesize();
ret = getrlimit(RLIMIT_MEMLOCK, &rlim);
if (ret < 0) {
perror("getrlimit");
return T_EXIT_PASS;
}
mlock_limit = rlim.rlim_cur;
devnull = open("/dev/null", O_RDWR);
if (devnull < 0) {
perror("open /dev/null");
exit(T_EXIT_FAIL);
}
/* invalid fd */
status |= expect_fail(-1, 0, NULL, 0, -EBADF, 0);
/* valid fd that is not an io_uring fd */
status |= expect_fail(devnull, 0, NULL, 0, -EOPNOTSUPP, 0);
/* invalid opcode */
memset(&p, 0, sizeof(p));
fd = new_io_uring(1, &p);
ret = expect_fail(fd, ~0U, NULL, 0, -EINVAL, 0);
if (ret) {
/* if this succeeds, tear down the io_uring instance
* and start clean for the next test. */
close(fd);
fd = new_io_uring(1, &p);
}
/* IORING_REGISTER_BUFFERS */
status |= test_iovec_size(fd);
status |= test_iovec_nr(fd);
/* IORING_REGISTER_FILES */
status |= test_max_fds(fd);
close(fd);
/* uring poll on the uring fd */
status |= test_poll_ringfd();
if (status)
fprintf(stderr, "FAIL\n");
return status;
}
|