1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * Copyright Intel Corporation, 2023
4 *
5 * Author: Chao Peng <chao.p.peng@linux.intel.com>
6 */
7 #include <stdlib.h>
8 #include <string.h>
9 #include <unistd.h>
10 #include <errno.h>
11 #include <stdio.h>
12 #include <fcntl.h>
13
14 #include <linux/bitmap.h>
15 #include <linux/falloc.h>
16 #include <linux/sizes.h>
17 #include <sys/mman.h>
18 #include <sys/types.h>
19 #include <sys/stat.h>
20
21 #include "kvm_util.h"
22 #include "numaif.h"
23 #include "test_util.h"
24 #include "ucall_common.h"
25
26 static size_t page_size;
27
test_file_read_write(int fd,size_t total_size)28 static void test_file_read_write(int fd, size_t total_size)
29 {
30 char buf[64];
31
32 TEST_ASSERT(read(fd, buf, sizeof(buf)) < 0,
33 "read on a guest_mem fd should fail");
34 TEST_ASSERT(write(fd, buf, sizeof(buf)) < 0,
35 "write on a guest_mem fd should fail");
36 TEST_ASSERT(pread(fd, buf, sizeof(buf), 0) < 0,
37 "pread on a guest_mem fd should fail");
38 TEST_ASSERT(pwrite(fd, buf, sizeof(buf), 0) < 0,
39 "pwrite on a guest_mem fd should fail");
40 }
41
test_mmap_cow(int fd,size_t size)42 static void test_mmap_cow(int fd, size_t size)
43 {
44 void *mem;
45
46 mem = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
47 TEST_ASSERT(mem == MAP_FAILED, "Copy-on-write not allowed by guest_memfd.");
48 }
49
test_mmap_supported(int fd,size_t total_size)50 static void test_mmap_supported(int fd, size_t total_size)
51 {
52 const char val = 0xaa;
53 char *mem;
54 size_t i;
55 int ret;
56
57 mem = kvm_mmap(total_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd);
58
59 memset(mem, val, total_size);
60 for (i = 0; i < total_size; i++)
61 TEST_ASSERT_EQ(READ_ONCE(mem[i]), val);
62
63 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE, 0,
64 page_size);
65 TEST_ASSERT(!ret, "fallocate the first page should succeed.");
66
67 for (i = 0; i < page_size; i++)
68 TEST_ASSERT_EQ(READ_ONCE(mem[i]), 0x00);
69 for (; i < total_size; i++)
70 TEST_ASSERT_EQ(READ_ONCE(mem[i]), val);
71
72 memset(mem, val, page_size);
73 for (i = 0; i < total_size; i++)
74 TEST_ASSERT_EQ(READ_ONCE(mem[i]), val);
75
76 kvm_munmap(mem, total_size);
77 }
78
test_mbind(int fd,size_t total_size)79 static void test_mbind(int fd, size_t total_size)
80 {
81 const unsigned long nodemask_0 = 1; /* nid: 0 */
82 unsigned long nodemask = 0;
83 unsigned long maxnode = 8;
84 int policy;
85 char *mem;
86 int ret;
87
88 if (!is_multi_numa_node_system())
89 return;
90
91 mem = kvm_mmap(total_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd);
92
93 /* Test MPOL_INTERLEAVE policy */
94 kvm_mbind(mem, page_size * 2, MPOL_INTERLEAVE, &nodemask_0, maxnode, 0);
95 kvm_get_mempolicy(&policy, &nodemask, maxnode, mem, MPOL_F_ADDR);
96 TEST_ASSERT(policy == MPOL_INTERLEAVE && nodemask == nodemask_0,
97 "Wanted MPOL_INTERLEAVE (%u) and nodemask 0x%lx, got %u and 0x%lx",
98 MPOL_INTERLEAVE, nodemask_0, policy, nodemask);
99
100 /* Test basic MPOL_BIND policy */
101 kvm_mbind(mem + page_size * 2, page_size * 2, MPOL_BIND, &nodemask_0, maxnode, 0);
102 kvm_get_mempolicy(&policy, &nodemask, maxnode, mem + page_size * 2, MPOL_F_ADDR);
103 TEST_ASSERT(policy == MPOL_BIND && nodemask == nodemask_0,
104 "Wanted MPOL_BIND (%u) and nodemask 0x%lx, got %u and 0x%lx",
105 MPOL_BIND, nodemask_0, policy, nodemask);
106
107 /* Test MPOL_DEFAULT policy */
108 kvm_mbind(mem, total_size, MPOL_DEFAULT, NULL, 0, 0);
109 kvm_get_mempolicy(&policy, &nodemask, maxnode, mem, MPOL_F_ADDR);
110 TEST_ASSERT(policy == MPOL_DEFAULT && !nodemask,
111 "Wanted MPOL_DEFAULT (%u) and nodemask 0x0, got %u and 0x%lx",
112 MPOL_DEFAULT, policy, nodemask);
113
114 /* Test with invalid policy */
115 ret = mbind(mem, page_size, 999, &nodemask_0, maxnode, 0);
116 TEST_ASSERT(ret == -1 && errno == EINVAL,
117 "mbind with invalid policy should fail with EINVAL");
118
119 kvm_munmap(mem, total_size);
120 }
121
test_numa_allocation(int fd,size_t total_size)122 static void test_numa_allocation(int fd, size_t total_size)
123 {
124 unsigned long node0_mask = 1; /* Node 0 */
125 unsigned long node1_mask = 2; /* Node 1 */
126 unsigned long maxnode = 8;
127 void *pages[4];
128 int status[4];
129 char *mem;
130 int i;
131
132 if (!is_multi_numa_node_system())
133 return;
134
135 mem = kvm_mmap(total_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd);
136
137 for (i = 0; i < 4; i++)
138 pages[i] = (char *)mem + page_size * i;
139
140 /* Set NUMA policy after allocation */
141 memset(mem, 0xaa, page_size);
142 kvm_mbind(pages[0], page_size, MPOL_BIND, &node0_mask, maxnode, 0);
143 kvm_fallocate(fd, FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE, 0, page_size);
144
145 /* Set NUMA policy before allocation */
146 kvm_mbind(pages[0], page_size * 2, MPOL_BIND, &node1_mask, maxnode, 0);
147 kvm_mbind(pages[2], page_size * 2, MPOL_BIND, &node0_mask, maxnode, 0);
148 memset(mem, 0xaa, total_size);
149
150 /* Validate if pages are allocated on specified NUMA nodes */
151 kvm_move_pages(0, 4, pages, NULL, status, 0);
152 TEST_ASSERT(status[0] == 1, "Expected page 0 on node 1, got it on node %d", status[0]);
153 TEST_ASSERT(status[1] == 1, "Expected page 1 on node 1, got it on node %d", status[1]);
154 TEST_ASSERT(status[2] == 0, "Expected page 2 on node 0, got it on node %d", status[2]);
155 TEST_ASSERT(status[3] == 0, "Expected page 3 on node 0, got it on node %d", status[3]);
156
157 /* Punch hole for all pages */
158 kvm_fallocate(fd, FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE, 0, total_size);
159
160 /* Change NUMA policy nodes and reallocate */
161 kvm_mbind(pages[0], page_size * 2, MPOL_BIND, &node0_mask, maxnode, 0);
162 kvm_mbind(pages[2], page_size * 2, MPOL_BIND, &node1_mask, maxnode, 0);
163 memset(mem, 0xaa, total_size);
164
165 kvm_move_pages(0, 4, pages, NULL, status, 0);
166 TEST_ASSERT(status[0] == 0, "Expected page 0 on node 0, got it on node %d", status[0]);
167 TEST_ASSERT(status[1] == 0, "Expected page 1 on node 0, got it on node %d", status[1]);
168 TEST_ASSERT(status[2] == 1, "Expected page 2 on node 1, got it on node %d", status[2]);
169 TEST_ASSERT(status[3] == 1, "Expected page 3 on node 1, got it on node %d", status[3]);
170
171 kvm_munmap(mem, total_size);
172 }
173
test_fault_sigbus(int fd,size_t accessible_size,size_t map_size)174 static void test_fault_sigbus(int fd, size_t accessible_size, size_t map_size)
175 {
176 const char val = 0xaa;
177 char *mem;
178 size_t i;
179
180 mem = kvm_mmap(map_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd);
181
182 TEST_EXPECT_SIGBUS(memset(mem, val, map_size));
183 TEST_EXPECT_SIGBUS((void)READ_ONCE(mem[accessible_size]));
184
185 for (i = 0; i < accessible_size; i++)
186 TEST_ASSERT_EQ(READ_ONCE(mem[i]), val);
187
188 kvm_munmap(mem, map_size);
189 }
190
test_fault_overflow(int fd,size_t total_size)191 static void test_fault_overflow(int fd, size_t total_size)
192 {
193 test_fault_sigbus(fd, total_size, total_size * 4);
194 }
195
test_fault_private(int fd,size_t total_size)196 static void test_fault_private(int fd, size_t total_size)
197 {
198 test_fault_sigbus(fd, 0, total_size);
199 }
200
test_mmap_not_supported(int fd,size_t total_size)201 static void test_mmap_not_supported(int fd, size_t total_size)
202 {
203 char *mem;
204
205 mem = mmap(NULL, page_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
206 TEST_ASSERT_EQ(mem, MAP_FAILED);
207
208 mem = mmap(NULL, total_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
209 TEST_ASSERT_EQ(mem, MAP_FAILED);
210 }
211
test_file_size(int fd,size_t total_size)212 static void test_file_size(int fd, size_t total_size)
213 {
214 struct stat sb;
215 int ret;
216
217 ret = fstat(fd, &sb);
218 TEST_ASSERT(!ret, "fstat should succeed");
219 TEST_ASSERT_EQ(sb.st_size, total_size);
220 TEST_ASSERT_EQ(sb.st_blksize, page_size);
221 }
222
test_fallocate(int fd,size_t total_size)223 static void test_fallocate(int fd, size_t total_size)
224 {
225 int ret;
226
227 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE, 0, total_size);
228 TEST_ASSERT(!ret, "fallocate with aligned offset and size should succeed");
229
230 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
231 page_size - 1, page_size);
232 TEST_ASSERT(ret, "fallocate with unaligned offset should fail");
233
234 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE, total_size, page_size);
235 TEST_ASSERT(ret, "fallocate beginning at total_size should fail");
236
237 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE, total_size + page_size, page_size);
238 TEST_ASSERT(ret, "fallocate beginning after total_size should fail");
239
240 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
241 total_size, page_size);
242 TEST_ASSERT(!ret, "fallocate(PUNCH_HOLE) at total_size should succeed");
243
244 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
245 total_size + page_size, page_size);
246 TEST_ASSERT(!ret, "fallocate(PUNCH_HOLE) after total_size should succeed");
247
248 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
249 page_size, page_size - 1);
250 TEST_ASSERT(ret, "fallocate with unaligned size should fail");
251
252 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
253 page_size, page_size);
254 TEST_ASSERT(!ret, "fallocate(PUNCH_HOLE) with aligned offset and size should succeed");
255
256 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE, page_size, page_size);
257 TEST_ASSERT(!ret, "fallocate to restore punched hole should succeed");
258 }
259
test_invalid_punch_hole(int fd,size_t total_size)260 static void test_invalid_punch_hole(int fd, size_t total_size)
261 {
262 struct {
263 off_t offset;
264 off_t len;
265 } testcases[] = {
266 {0, 1},
267 {0, page_size - 1},
268 {0, page_size + 1},
269
270 {1, 1},
271 {1, page_size - 1},
272 {1, page_size},
273 {1, page_size + 1},
274
275 {page_size, 1},
276 {page_size, page_size - 1},
277 {page_size, page_size + 1},
278 };
279 int ret, i;
280
281 for (i = 0; i < ARRAY_SIZE(testcases); i++) {
282 ret = fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
283 testcases[i].offset, testcases[i].len);
284 TEST_ASSERT(ret == -1 && errno == EINVAL,
285 "PUNCH_HOLE with !PAGE_SIZE offset (%lx) and/or length (%lx) should fail",
286 testcases[i].offset, testcases[i].len);
287 }
288 }
289
test_create_guest_memfd_invalid_sizes(struct kvm_vm * vm,uint64_t guest_memfd_flags)290 static void test_create_guest_memfd_invalid_sizes(struct kvm_vm *vm,
291 uint64_t guest_memfd_flags)
292 {
293 size_t size;
294 int fd;
295
296 for (size = 1; size < page_size; size++) {
297 fd = __vm_create_guest_memfd(vm, size, guest_memfd_flags);
298 TEST_ASSERT(fd < 0 && errno == EINVAL,
299 "guest_memfd() with non-page-aligned page size '0x%lx' should fail with EINVAL",
300 size);
301 }
302 }
303
test_create_guest_memfd_multiple(struct kvm_vm * vm)304 static void test_create_guest_memfd_multiple(struct kvm_vm *vm)
305 {
306 int fd1, fd2, ret;
307 struct stat st1, st2;
308
309 fd1 = __vm_create_guest_memfd(vm, page_size, 0);
310 TEST_ASSERT(fd1 != -1, "memfd creation should succeed");
311
312 ret = fstat(fd1, &st1);
313 TEST_ASSERT(ret != -1, "memfd fstat should succeed");
314 TEST_ASSERT(st1.st_size == page_size, "memfd st_size should match requested size");
315
316 fd2 = __vm_create_guest_memfd(vm, page_size * 2, 0);
317 TEST_ASSERT(fd2 != -1, "memfd creation should succeed");
318
319 ret = fstat(fd2, &st2);
320 TEST_ASSERT(ret != -1, "memfd fstat should succeed");
321 TEST_ASSERT(st2.st_size == page_size * 2, "second memfd st_size should match requested size");
322
323 ret = fstat(fd1, &st1);
324 TEST_ASSERT(ret != -1, "memfd fstat should succeed");
325 TEST_ASSERT(st1.st_size == page_size, "first memfd st_size should still match requested size");
326 TEST_ASSERT(st1.st_ino != st2.st_ino, "different memfd should have different inode numbers");
327
328 close(fd2);
329 close(fd1);
330 }
331
test_guest_memfd_flags(struct kvm_vm * vm)332 static void test_guest_memfd_flags(struct kvm_vm *vm)
333 {
334 uint64_t valid_flags = vm_check_cap(vm, KVM_CAP_GUEST_MEMFD_FLAGS);
335 uint64_t flag;
336 int fd;
337
338 for (flag = BIT(0); flag; flag <<= 1) {
339 fd = __vm_create_guest_memfd(vm, page_size, flag);
340 if (flag & valid_flags) {
341 TEST_ASSERT(fd >= 0,
342 "guest_memfd() with flag '0x%lx' should succeed",
343 flag);
344 close(fd);
345 } else {
346 TEST_ASSERT(fd < 0 && errno == EINVAL,
347 "guest_memfd() with flag '0x%lx' should fail with EINVAL",
348 flag);
349 }
350 }
351 }
352
353 #define gmem_test(__test, __vm, __flags) \
354 do { \
355 int fd = vm_create_guest_memfd(__vm, page_size * 4, __flags); \
356 \
357 test_##__test(fd, page_size * 4); \
358 close(fd); \
359 } while (0)
360
__test_guest_memfd(struct kvm_vm * vm,uint64_t flags)361 static void __test_guest_memfd(struct kvm_vm *vm, uint64_t flags)
362 {
363 test_create_guest_memfd_multiple(vm);
364 test_create_guest_memfd_invalid_sizes(vm, flags);
365
366 gmem_test(file_read_write, vm, flags);
367
368 if (flags & GUEST_MEMFD_FLAG_MMAP) {
369 if (flags & GUEST_MEMFD_FLAG_INIT_SHARED) {
370 gmem_test(mmap_supported, vm, flags);
371 gmem_test(fault_overflow, vm, flags);
372 gmem_test(numa_allocation, vm, flags);
373 } else {
374 gmem_test(fault_private, vm, flags);
375 }
376
377 gmem_test(mmap_cow, vm, flags);
378 gmem_test(mbind, vm, flags);
379 } else {
380 gmem_test(mmap_not_supported, vm, flags);
381 }
382
383 gmem_test(file_size, vm, flags);
384 gmem_test(fallocate, vm, flags);
385 gmem_test(invalid_punch_hole, vm, flags);
386 }
387
test_guest_memfd(unsigned long vm_type)388 static void test_guest_memfd(unsigned long vm_type)
389 {
390 struct kvm_vm *vm = vm_create_barebones_type(vm_type);
391 uint64_t flags;
392
393 test_guest_memfd_flags(vm);
394
395 __test_guest_memfd(vm, 0);
396
397 flags = vm_check_cap(vm, KVM_CAP_GUEST_MEMFD_FLAGS);
398 if (flags & GUEST_MEMFD_FLAG_MMAP)
399 __test_guest_memfd(vm, GUEST_MEMFD_FLAG_MMAP);
400
401 /* MMAP should always be supported if INIT_SHARED is supported. */
402 if (flags & GUEST_MEMFD_FLAG_INIT_SHARED)
403 __test_guest_memfd(vm, GUEST_MEMFD_FLAG_MMAP |
404 GUEST_MEMFD_FLAG_INIT_SHARED);
405
406 kvm_vm_free(vm);
407 }
408
guest_code(uint8_t * mem,uint64_t size)409 static void guest_code(uint8_t *mem, uint64_t size)
410 {
411 size_t i;
412
413 for (i = 0; i < size; i++)
414 __GUEST_ASSERT(mem[i] == 0xaa,
415 "Guest expected 0xaa at offset %lu, got 0x%x", i, mem[i]);
416
417 memset(mem, 0xff, size);
418 GUEST_DONE();
419 }
420
test_guest_memfd_guest(void)421 static void test_guest_memfd_guest(void)
422 {
423 /*
424 * Skip the first 4gb and slot0. slot0 maps <1gb and is used to back
425 * the guest's code, stack, and page tables, and low memory contains
426 * the PCI hole and other MMIO regions that need to be avoided.
427 */
428 const uint64_t gpa = SZ_4G;
429 const int slot = 1;
430
431 struct kvm_vcpu *vcpu;
432 struct kvm_vm *vm;
433 uint8_t *mem;
434 size_t size;
435 int fd, i;
436
437 if (!kvm_check_cap(KVM_CAP_GUEST_MEMFD_FLAGS))
438 return;
439
440 vm = __vm_create_shape_with_one_vcpu(VM_SHAPE_DEFAULT, &vcpu, 1, guest_code);
441
442 TEST_ASSERT(vm_check_cap(vm, KVM_CAP_GUEST_MEMFD_FLAGS) & GUEST_MEMFD_FLAG_MMAP,
443 "Default VM type should support MMAP, supported flags = 0x%x",
444 vm_check_cap(vm, KVM_CAP_GUEST_MEMFD_FLAGS));
445 TEST_ASSERT(vm_check_cap(vm, KVM_CAP_GUEST_MEMFD_FLAGS) & GUEST_MEMFD_FLAG_INIT_SHARED,
446 "Default VM type should support INIT_SHARED, supported flags = 0x%x",
447 vm_check_cap(vm, KVM_CAP_GUEST_MEMFD_FLAGS));
448
449 size = vm->page_size;
450 fd = vm_create_guest_memfd(vm, size, GUEST_MEMFD_FLAG_MMAP |
451 GUEST_MEMFD_FLAG_INIT_SHARED);
452 vm_set_user_memory_region2(vm, slot, KVM_MEM_GUEST_MEMFD, gpa, size, NULL, fd, 0);
453
454 mem = kvm_mmap(size, PROT_READ | PROT_WRITE, MAP_SHARED, fd);
455 memset(mem, 0xaa, size);
456 kvm_munmap(mem, size);
457
458 virt_pg_map(vm, gpa, gpa);
459 vcpu_args_set(vcpu, 2, gpa, size);
460 vcpu_run(vcpu);
461
462 TEST_ASSERT_EQ(get_ucall(vcpu, NULL), UCALL_DONE);
463
464 mem = kvm_mmap(size, PROT_READ | PROT_WRITE, MAP_SHARED, fd);
465 for (i = 0; i < size; i++)
466 TEST_ASSERT_EQ(mem[i], 0xff);
467
468 close(fd);
469 kvm_vm_free(vm);
470 }
471
main(int argc,char * argv[])472 int main(int argc, char *argv[])
473 {
474 unsigned long vm_types, vm_type;
475
476 TEST_REQUIRE(kvm_has_cap(KVM_CAP_GUEST_MEMFD));
477
478 page_size = getpagesize();
479
480 /*
481 * Not all architectures support KVM_CAP_VM_TYPES. However, those that
482 * support guest_memfd have that support for the default VM type.
483 */
484 vm_types = kvm_check_cap(KVM_CAP_VM_TYPES);
485 if (!vm_types)
486 vm_types = BIT(VM_TYPE_DEFAULT);
487
488 for_each_set_bit(vm_type, &vm_types, BITS_PER_TYPE(vm_types))
489 test_guest_memfd(vm_type);
490
491 test_guest_memfd_guest();
492 }
493