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
|
--- b/tools/testing/selftests/vm/userfaultfd.c
+++ b/tools/testing/selftests/vm/userfaultfd.c
@@ -82,9 +82,11 @@
static char *huge_fd_off0;
static unsigned long long *count_verify;
static int uffd, uffd_flags, finished, *pipefd;
+static volatile bool ready_for_fork;
static char *area_src, *area_src_alias, *area_dst, *area_dst_alias;
static char *zeropage;
pthread_attr_t attr;
+pthread_key_t long_jmp_key;
/* pthread_mutex_t starts at page offset 0 */
#define area_mutex(___area, ___nr) \
@@ -139,8 +141,11 @@
static void anon_allocate_area(void **alloc_area)
{
- if (posix_memalign(alloc_area, page_size, nr_pages * page_size)) {
- fprintf(stderr, "out of memory\n");
+ // We can't use posix_memalign due to pointer-tagging used in bionic.
+ *alloc_area = mmap(NULL, nr_pages * page_size, PROT_READ | PROT_WRITE,
+ MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+ if (*alloc_area == MAP_FAILED) {
+ fprintf(stderr, "anon memory mmap failed\n");
*alloc_area = NULL;
}
}
@@ -284,23 +289,11 @@
static void *locking_thread(void *arg)
{
unsigned long cpu = (unsigned long) arg;
- struct random_data rand;
unsigned long page_nr = *(&(page_nr)); /* uninitialized warning */
- int32_t rand_nr;
unsigned long long count;
- char randstate[64];
- unsigned int seed;
time_t start;
- if (bounces & BOUNCE_RANDOM) {
- seed = (unsigned int) time(NULL) - bounces;
- if (!(bounces & BOUNCE_RACINGFAULTS))
- seed += cpu;
- bzero(&rand, sizeof(rand));
- bzero(&randstate, sizeof(randstate));
- if (initstate_r(seed, randstate, sizeof(randstate), &rand))
- fprintf(stderr, "srandom_r error\n"), exit(1);
- } else {
+ if (!(bounces & BOUNCE_RANDOM)) {
page_nr = -bounces;
if (!(bounces & BOUNCE_RACINGFAULTS))
page_nr += cpu * nr_pages_per_cpu;
@@ -308,13 +301,9 @@
while (!finished) {
if (bounces & BOUNCE_RANDOM) {
- if (random_r(&rand, &rand_nr))
- fprintf(stderr, "random_r 1 error\n"), exit(1);
- page_nr = rand_nr;
- if (sizeof(page_nr) > sizeof(rand_nr)) {
- if (random_r(&rand, &rand_nr))
- fprintf(stderr, "random_r 2 error\n"), exit(1);
- page_nr |= (((unsigned long) rand_nr) << 16) <<
+ page_nr = random();
+ if (sizeof(page_nr) > sizeof(uint32_t)) {
+ page_nr |= (((unsigned long) random()) << 16) <<
16;
}
} else
@@ -501,6 +490,9 @@
pollfd[1].fd = pipefd[cpu*2];
pollfd[1].events = POLLIN;
+ // Notify the main thread that it can now fork.
+ ready_for_fork = true;
+
for (;;) {
ret = poll(pollfd, 2, -1);
if (!ret)
@@ -548,18 +540,31 @@
pthread_mutex_t uffd_read_mutex = PTHREAD_MUTEX_INITIALIZER;
+static void sigusr1_handler(int signum, siginfo_t *siginfo, void *ptr)
+{
+ jmp_buf *env;
+ env = pthread_getspecific(long_jmp_key);
+ longjmp(*env, 1);
+}
+
static void *uffd_read_thread(void *arg)
{
unsigned long *this_cpu_userfaults;
struct uffd_msg msg;
+ jmp_buf env;
+ int setjmp_ret;
+
+ pthread_setspecific(long_jmp_key, &env);
this_cpu_userfaults = (unsigned long *) arg;
*this_cpu_userfaults = 0;
pthread_mutex_unlock(&uffd_read_mutex);
- /* from here cancellation is ok */
- for (;;) {
+ // One first return setjmp return 0. On second (fake) return from
+ // longjmp() it returns the provided value, which will be 1 in our case.
+ setjmp_ret = setjmp(env);
+ while (!setjmp_ret) {
if (uffd_read_msg(uffd, &msg))
continue;
(*this_cpu_userfaults) += uffd_handle_page_fault(&msg);
@@ -608,6 +613,7 @@
background_thread, (void *)cpu))
return 1;
}
+
for (cpu = 0; cpu < nr_cpus; cpu++)
if (pthread_join(background_threads[cpu], NULL))
return 1;
@@ -640,7 +646,7 @@
if (pthread_join(uffd_threads[cpu], &_userfaults[cpu]))
return 1;
} else {
- if (pthread_cancel(uffd_threads[cpu]))
+ if (pthread_kill(uffd_threads[cpu], SIGUSR1))
return 1;
if (pthread_join(uffd_threads[cpu], NULL))
return 1;
@@ -654,7 +660,7 @@
{
struct uffdio_api uffdio_api;
- uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK);
+ uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK | UFFD_USER_MODE_ONLY);
if (uffd < 0) {
fprintf(stderr,
"userfaultfd syscall not available in this kernel\n");
@@ -914,6 +920,10 @@
pid_t pid;
char c;
+ // All the syscalls below up to pthread_create will ensure that this
+ // write is completed before, the uffd_thread sets it to true.
+ ready_for_fork = false;
+
printf("testing events (fork, remap, remove): ");
fflush(stdout);
@@ -942,6 +952,11 @@
if (pthread_create(&uffd_mon, &attr, uffd_poll_thread, NULL))
perror("uffd_poll_thread create"), exit(1);
+ // Wait for the poll_thread to start executing before forking. This is
+ // required to avoid a deadlock, which can happen if poll_thread doesn't
+ // start getting executed by the time fork is invoked.
+ while (!ready_for_fork);
+
pid = fork();
if (pid < 0)
perror("fork"), exit(1);
@@ -974,6 +989,10 @@
pid_t pid;
char c;
+ // All the syscalls below up to pthread_create will ensure that this
+ // write is completed before, the uffd_thread sets it to true.
+ ready_for_fork = false;
+
printf("testing signal delivery: ");
fflush(stdout);
@@ -1007,6 +1026,11 @@
if (pthread_create(&uffd_mon, &attr, uffd_poll_thread, NULL))
perror("uffd_poll_thread create"), exit(1);
+ // Wait for the poll_thread to start executing before forking. This is
+ // required to avoid a deadlock, which can happen if poll_thread doesn't
+ // start getting executed by the time fork is invoked.
+ while (!ready_for_fork);
+
pid = fork();
if (pid < 0)
perror("fork"), exit(1);
@@ -1036,6 +1060,7 @@
char *tmp_area;
unsigned long nr;
struct uffdio_register uffdio_register;
+ struct sigaction act;
unsigned long cpu;
int err;
unsigned long userfaults[nr_cpus];
@@ -1094,6 +1119,17 @@
pthread_attr_init(&attr);
pthread_attr_setstacksize(&attr, 16*1024*1024);
+ // For handling thread termination of read thread in the absense of
+ // pthread_cancel().
+ pthread_key_create(&long_jmp_key, NULL);
+ memset(&act, 0, sizeof(act));
+ act.sa_sigaction = sigusr1_handler;
+ act.sa_flags = SA_SIGINFO;
+ if (sigaction(SIGUSR1, &act, 0)) {
+ perror("sigaction");
+ return 1;
+ }
+
err = 0;
while (bounces--) {
unsigned long expected_ioctls;
@@ -1215,6 +1251,8 @@
printf("\n");
}
+ pthread_key_delete(long_jmp_key);
+
if (err)
return err;
@@ -1291,6 +1329,9 @@
int main(int argc, char **argv)
{
+ char randstate[64];
+ unsigned int seed;
+
if (argc < 4)
usage();
@@ -1332,6 +1373,12 @@
}
printf("nr_pages: %lu, nr_pages_per_cpu: %lu\n",
nr_pages, nr_pages_per_cpu);
+
+ seed = (unsigned int) time(NULL);
+ bzero(&randstate, sizeof(randstate));
+ if (!initstate(seed, randstate, sizeof(randstate)))
+ fprintf(stderr, "srandom_r error\n"), exit(1);
+
return userfaultfd_stress();
}
--- a/tools/testing/selftests/vm/userfaultfd.c
+++ b/tools/testing/selftests/vm/userfaultfd.c
@@ -662,8 +662,13 @@ static int userfaultfd_open(int features)
uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK | UFFD_USER_MODE_ONLY);
if (uffd < 0) {
+ if (errno == ENOSYS) {
+ fprintf(stderr,
+ "userfaultfd syscall not available in this kernel\n");
+ exit(KSFT_PASS);
+ }
fprintf(stderr,
- "userfaultfd syscall not available in this kernel\n");
+ "userfaultfd syscall failed with errno: %d\n", errno);
return 1;
}
uffd_flags = fcntl(uffd, F_GETFD, NULL);
|
