#include "../arch/arch.h"
#include "../lib/types.h"
-#include "../os/io_uring.h"
-
-#define barrier() __asm__ __volatile__("": : :"memory")
+#include "../os/linux/io_uring.h"
#define min(a, b) ((a < b) ? (a) : (b))
};
#define DEPTH 128
-
-#define BATCH_SUBMIT 64
-#define BATCH_COMPLETE 64
-
+#define BATCH_SUBMIT 32
+#define BATCH_COMPLETE 32
#define BS 4096
#define MAX_FDS 16
struct file {
unsigned long max_blocks;
unsigned pending_ios;
- int fd;
+ int real_fd;
+ int fixed_fd;
};
struct submitter {
pthread_t thread;
int ring_fd;
- struct drand48_data rand;
struct io_sq_ring sq_ring;
struct io_uring_sqe *sqes;
- struct iovec iovecs[DEPTH];
struct io_cq_ring cq_ring;
int inflight;
unsigned long reaps;
unsigned long done;
unsigned long calls;
- unsigned long cachehit, cachemiss;
volatile int finish;
+ __s32 *fds;
+
struct file files[MAX_FDS];
unsigned nr_files;
unsigned cur_file;
+ struct iovec iovecs[];
};
-static struct submitter submitters[1];
+static struct submitter *submitter;
static volatile int finish;
+static int depth = DEPTH;
+static int batch_submit = BATCH_SUBMIT;
+static int batch_complete = BATCH_COMPLETE;
+static int bs = BS;
static int polled = 1; /* use IO polling */
static int fixedbufs = 1; /* use fixed user buffers */
+static int register_files = 1; /* use fixed files */
static int buffered = 0; /* use buffered IO, not O_DIRECT */
static int sq_thread_poll = 0; /* use kernel submission/poller thread */
static int sq_thread_cpu = -1; /* pin above thread to this CPU */
+static int do_nop = 0; /* no-op SQ ring commands */
+
+static int vectored = 1;
static int io_uring_register_buffers(struct submitter *s)
{
- struct io_uring_register_buffers reg = {
- .iovecs = s->iovecs,
- .nr_iovecs = DEPTH
- };
+ if (do_nop)
+ return 0;
- return syscall(__NR_sys_io_uring_register, s->ring_fd,
- IORING_REGISTER_BUFFERS, ®);
+ return syscall(__NR_io_uring_register, s->ring_fd,
+ IORING_REGISTER_BUFFERS, s->iovecs, depth);
}
static int io_uring_register_files(struct submitter *s)
{
- struct io_uring_register_files reg;
- int i, ret;
-
- reg.fds = calloc(s->nr_files, sizeof(int));
- for (i = 0; i < s->nr_files; i++)
- reg.fds[i] = s->files[i].fd;
- reg.nr_fds = s->nr_files;
-
- ret = syscall(__NR_sys_io_uring_register, s->ring_fd,
- IORING_REGISTER_FILES, ®);
- free(reg.fds);
- return ret;
+ int i;
+
+ if (do_nop)
+ return 0;
+
+ s->fds = calloc(s->nr_files, sizeof(__s32));
+ for (i = 0; i < s->nr_files; i++) {
+ s->fds[i] = s->files[i].real_fd;
+ s->files[i].fixed_fd = i;
+ }
+
+ return syscall(__NR_io_uring_register, s->ring_fd,
+ IORING_REGISTER_FILES, s->fds, s->nr_files);
}
static int io_uring_setup(unsigned entries, struct io_uring_params *p)
{
- return syscall(__NR_sys_io_uring_setup, entries, p);
+ return syscall(__NR_io_uring_setup, entries, p);
+}
+
+static void io_uring_probe(int fd)
+{
+ struct io_uring_probe *p;
+ int ret;
+
+ p = malloc(sizeof(*p) + 256 * sizeof(struct io_uring_probe_op));
+ if (!p)
+ return;
+
+ memset(p, 0, sizeof(*p) + 256 * sizeof(struct io_uring_probe_op));
+ ret = syscall(__NR_io_uring_register, fd, IORING_REGISTER_PROBE, p, 256);
+ if (ret < 0)
+ goto out;
+
+ if (IORING_OP_READ > p->ops_len)
+ goto out;
+
+ if ((p->ops[IORING_OP_READ].flags & IO_URING_OP_SUPPORTED))
+ vectored = 0;
+out:
+ free(p);
}
static int io_uring_enter(struct submitter *s, unsigned int to_submit,
unsigned int min_complete, unsigned int flags)
{
- return syscall(__NR_sys_io_uring_enter, s->ring_fd, to_submit,
- min_complete, flags);
+ return syscall(__NR_io_uring_enter, s->ring_fd, to_submit, min_complete,
+ flags, NULL, 0);
}
+#ifndef CONFIG_HAVE_GETTID
static int gettid(void)
{
return syscall(__NR_gettid);
}
+#endif
static unsigned file_depth(struct submitter *s)
{
- return (DEPTH + s->nr_files - 1) / s->nr_files;
+ return (depth + s->nr_files - 1) / s->nr_files;
}
static void init_io(struct submitter *s, unsigned index)
struct file *f;
long r;
+ if (do_nop) {
+ sqe->opcode = IORING_OP_NOP;
+ return;
+ }
+
if (s->nr_files == 1) {
f = &s->files[0];
} else {
s->cur_file++;
if (s->cur_file == s->nr_files)
s->cur_file = 0;
+ f = &s->files[s->cur_file];
}
}
f->pending_ios++;
- lrand48_r(&s->rand, &r);
- offset = (r % (f->max_blocks - 1)) * BS;
+ r = lrand48();
+ offset = (r % (f->max_blocks - 1)) * bs;
- sqe->flags = IOSQE_FIXED_FILE;
- sqe->opcode = IORING_OP_READV;
+ if (register_files) {
+ sqe->flags = IOSQE_FIXED_FILE;
+ sqe->fd = f->fixed_fd;
+ } else {
+ sqe->flags = 0;
+ sqe->fd = f->real_fd;
+ }
if (fixedbufs) {
- sqe->addr = s->iovecs[index].iov_base;
- sqe->len = BS;
+ sqe->opcode = IORING_OP_READ_FIXED;
+ sqe->addr = (unsigned long) s->iovecs[index].iov_base;
+ sqe->len = bs;
sqe->buf_index = index;
- sqe->flags |= IOSQE_FIXED_BUFFER;
+ } else if (!vectored) {
+ sqe->opcode = IORING_OP_READ;
+ sqe->addr = (unsigned long) s->iovecs[index].iov_base;
+ sqe->len = bs;
+ sqe->buf_index = 0;
} else {
- sqe->addr = &s->iovecs[index];
+ sqe->opcode = IORING_OP_READV;
+ sqe->addr = (unsigned long) &s->iovecs[index];
sqe->len = 1;
sqe->buf_index = 0;
}
sqe->ioprio = 0;
- sqe->fd = f->fd;
sqe->off = offset;
- sqe->data = (unsigned long) f;
+ sqe->user_data = (unsigned long) f;
}
static int prep_more_ios(struct submitter *s, int max_ios)
next_tail = tail = *ring->tail;
do {
next_tail++;
- barrier();
+ read_barrier();
if (next_tail == *ring->head)
break;
} while (prepped < max_ios);
if (*ring->tail != tail) {
- /* order tail store with writes to sqes above */
- barrier();
*ring->tail = tail;
- barrier();
+ write_barrier();
}
return prepped;
}
{
struct stat st;
- if (fstat(f->fd, &st) < 0)
+ if (fstat(f->real_fd, &st) < 0)
return -1;
if (S_ISBLK(st.st_mode)) {
unsigned long long bytes;
- if (ioctl(f->fd, BLKGETSIZE64, &bytes) != 0)
+ if (ioctl(f->real_fd, BLKGETSIZE64, &bytes) != 0)
return -1;
- f->max_blocks = bytes / BS;
+ f->max_blocks = bytes / bs;
return 0;
} else if (S_ISREG(st.st_mode)) {
- f->max_blocks = st.st_size / BS;
+ f->max_blocks = st.st_size / bs;
return 0;
}
do {
struct file *f;
- barrier();
+ read_barrier();
if (head == *ring->tail)
break;
cqe = &ring->cqes[head & cq_ring_mask];
- f = (struct file *) cqe->data;
- f->pending_ios--;
- if (cqe->res != BS) {
- printf("io: unexpected ret=%d\n", cqe->res);
- return -1;
+ if (!do_nop) {
+ f = (struct file *) (uintptr_t) cqe->user_data;
+ f->pending_ios--;
+ if (cqe->res != bs) {
+ printf("io: unexpected ret=%d\n", cqe->res);
+ if (polled && cqe->res == -EOPNOTSUPP)
+ printf("Your filesystem/driver/kernel doesn't support polled IO\n");
+ return -1;
+ }
}
- if (cqe->flags & IOCQE_FLAG_CACHEHIT)
- s->cachehit++;
- else
- s->cachemiss++;
reaped++;
head++;
} while (1);
s->inflight -= reaped;
*ring->head = head;
- barrier();
+ write_barrier();
return reaped;
}
printf("submitter=%d\n", gettid());
- srand48_r(pthread_self(), &s->rand);
+ srand48(pthread_self());
prepped = 0;
do {
int to_wait, to_submit, this_reap, to_prep;
- if (!prepped && s->inflight < DEPTH) {
- to_prep = min(DEPTH - s->inflight, BATCH_SUBMIT);
+ if (!prepped && s->inflight < depth) {
+ to_prep = min(depth - s->inflight, batch_submit);
prepped = prep_more_ios(s, to_prep);
}
s->inflight += prepped;
submit_more:
to_submit = prepped;
submit:
- if (s->inflight + BATCH_SUBMIT < DEPTH)
+ if (to_submit && (s->inflight + to_submit <= depth))
to_wait = 0;
else
- to_wait = min(s->inflight + to_submit, BATCH_COMPLETE);
+ to_wait = min(s->inflight + to_submit, batch_complete);
/*
* Only need to call io_uring_enter if we're not using SQ thread
* poll, or if IORING_SQ_NEED_WAKEUP is set.
*/
if (!sq_thread_poll || (*ring->flags & IORING_SQ_NEED_WAKEUP)) {
- ret = io_uring_enter(s, to_submit, to_wait,
- IORING_ENTER_GETEVENTS);
+ unsigned flags = 0;
+
+ if (to_wait)
+ flags = IORING_ENTER_GETEVENTS;
+ if ((*ring->flags & IORING_SQ_NEED_WAKEUP))
+ flags |= IORING_ENTER_SQ_WAKEUP;
+ ret = io_uring_enter(s, to_submit, to_wait, flags);
s->calls++;
}
do {
int r;
r = reap_events(s);
- if (r == -1)
+ if (r == -1) {
+ s->finish = 1;
break;
- else if (r > 0)
+ } else if (r > 0)
this_reap += r;
} while (sq_thread_poll && this_reap < to_wait);
s->reaps += this_reap;
static void sig_int(int sig)
{
printf("Exiting on signal %d\n", sig);
- submitters[0].finish = 1;
+ submitter->finish = 1;
finish = 1;
}
memset(&p, 0, sizeof(p));
- if (polled)
+ if (polled && !do_nop)
p.flags |= IORING_SETUP_IOPOLL;
if (sq_thread_poll) {
p.flags |= IORING_SETUP_SQPOLL;
- if (sq_thread_cpu != -1)
+ if (sq_thread_cpu != -1) {
p.flags |= IORING_SETUP_SQ_AFF;
+ p.sq_thread_cpu = sq_thread_cpu;
+ }
}
- fd = io_uring_setup(DEPTH, &p);
+ fd = io_uring_setup(depth, &p);
if (fd < 0) {
perror("io_uring_setup");
return 1;
}
s->ring_fd = fd;
+ io_uring_probe(fd);
+
if (fixedbufs) {
ret = io_uring_register_buffers(s);
if (ret < 0) {
}
}
- ret = io_uring_register_files(s);
- if (ret < 0) {
- perror("io_uring_register_files");
- return 1;
+ if (register_files) {
+ ret = io_uring_register_files(s);
+ if (ret < 0) {
+ perror("io_uring_register_files");
+ return 1;
+ }
}
ptr = mmap(0, p.sq_off.array + p.sq_entries * sizeof(__u32),
return 0;
}
+static void file_depths(char *buf)
+{
+ struct submitter *s = submitter;
+ char *p;
+ int i;
+
+ buf[0] = '\0';
+ p = buf;
+ for (i = 0; i < s->nr_files; i++) {
+ struct file *f = &s->files[i];
+
+ if (i + 1 == s->nr_files)
+ p += sprintf(p, "%d", f->pending_ios);
+ else
+ p += sprintf(p, "%d, ", f->pending_ios);
+ }
+}
+
+static void usage(char *argv)
+{
+ printf("%s [options] -- [filenames]\n"
+ " -d <int> : IO Depth, default %d\n"
+ " -s <int> : Batch submit, default %d\n"
+ " -c <int> : Batch complete, default %d\n"
+ " -b <int> : Block size, default %d\n"
+ " -p <bool> : Polled IO, default %d\n",
+ argv, DEPTH, BATCH_SUBMIT, BATCH_COMPLETE, BS, polled);
+ exit(0);
+}
+
int main(int argc, char *argv[])
{
- struct submitter *s = &submitters[0];
- unsigned long done, calls, reap, cache_hit, cache_miss;
- int err, i, flags, fd;
- struct rlimit rlim;
+ struct submitter *s;
+ unsigned long done, calls, reap;
+ int err, i, flags, fd, opt;
+ char *fdepths;
void *ret;
- if (argc < 2) {
- printf("%s: filename\n", argv[0]);
+ if (!do_nop && argc < 2) {
+ printf("%s: filename [options]\n", argv[0]);
return 1;
}
+ while ((opt = getopt(argc, argv, "d:s:c:b:p:B:F:h?")) != -1) {
+ switch (opt) {
+ case 'd':
+ depth = atoi(optarg);
+ break;
+ case 's':
+ batch_submit = atoi(optarg);
+ break;
+ case 'c':
+ batch_complete = atoi(optarg);
+ break;
+ case 'b':
+ bs = atoi(optarg);
+ break;
+ case 'p':
+ polled = !!atoi(optarg);
+ break;
+ case 'B':
+ fixedbufs = !!atoi(optarg);
+ break;
+ case 'F':
+ register_files = !!atoi(optarg);
+ break;
+ case 'h':
+ case '?':
+ default:
+ usage(argv[0]);
+ break;
+ }
+ }
+
+ submitter = malloc(sizeof(*submitter) + depth * sizeof(struct iovec));
+ memset(submitter, 0, sizeof(*submitter) + depth * sizeof(struct iovec));
+ s = submitter;
+
flags = O_RDONLY | O_NOATIME;
if (!buffered)
flags |= O_DIRECT;
- i = 1;
- while (i < argc) {
- struct file *f = &s->files[s->nr_files];
+ i = optind;
+ while (!do_nop && i < argc) {
+ struct file *f;
+ if (s->nr_files == MAX_FDS) {
+ printf("Max number of files (%d) reached\n", MAX_FDS);
+ break;
+ }
fd = open(argv[i], flags);
if (fd < 0) {
perror("open");
return 1;
}
- f->fd = fd;
+
+ f = &s->files[s->nr_files];
+ f->real_fd = fd;
if (get_file_size(f)) {
printf("failed getting size of device/file\n");
return 1;
i++;
}
- rlim.rlim_cur = RLIM_INFINITY;
- rlim.rlim_max = RLIM_INFINITY;
- if (setrlimit(RLIMIT_MEMLOCK, &rlim) < 0) {
- perror("setrlimit");
- return 1;
+ if (fixedbufs) {
+ struct rlimit rlim;
+
+ rlim.rlim_cur = RLIM_INFINITY;
+ rlim.rlim_max = RLIM_INFINITY;
+ if (setrlimit(RLIMIT_MEMLOCK, &rlim) < 0) {
+ perror("setrlimit");
+ return 1;
+ }
}
arm_sig_int();
- for (i = 0; i < DEPTH; i++) {
+ for (i = 0; i < depth; i++) {
void *buf;
- if (posix_memalign(&buf, BS, BS)) {
+ if (posix_memalign(&buf, bs, bs)) {
printf("failed alloc\n");
return 1;
}
s->iovecs[i].iov_base = buf;
- s->iovecs[i].iov_len = BS;
+ s->iovecs[i].iov_len = bs;
}
err = setup_ring(s);
printf("ring setup failed: %s, %d\n", strerror(errno), err);
return 1;
}
- printf("polled=%d, fixedbufs=%d, buffered=%d", polled, fixedbufs, buffered);
- printf(" QD=%d, sq_ring=%d, cq_ring=%d\n", DEPTH, *s->sq_ring.ring_entries, *s->cq_ring.ring_entries);
+ printf("polled=%d, fixedbufs=%d, register_files=%d, buffered=%d", polled, fixedbufs, register_files, buffered);
+ printf(" QD=%d, sq_ring=%d, cq_ring=%d\n", depth, *s->sq_ring.ring_entries, *s->cq_ring.ring_entries);
pthread_create(&s->thread, NULL, submitter_fn, s);
- cache_hit = cache_miss = reap = calls = done = 0;
+ fdepths = malloc(8 * s->nr_files);
+ reap = calls = done = 0;
do {
unsigned long this_done = 0;
unsigned long this_reap = 0;
unsigned long this_call = 0;
- unsigned long this_cache_hit = 0;
- unsigned long this_cache_miss = 0;
unsigned long rpc = 0, ipc = 0;
- double hit = 0.0;
sleep(1);
this_done += s->done;
this_call += s->calls;
this_reap += s->reaps;
- this_cache_hit += s->cachehit;
- this_cache_miss += s->cachemiss;
- if (this_cache_hit && this_cache_miss) {
- unsigned long hits, total;
-
- hits = this_cache_hit - cache_hit;
- total = hits + this_cache_miss - cache_miss;
- hit = (double) hits / (double) total;
- hit *= 100.0;
- }
if (this_call - calls) {
rpc = (this_done - done) / (this_call - calls);
ipc = (this_reap - reap) / (this_call - calls);
} else
rpc = ipc = -1;
- printf("IOPS=%lu, IOS/call=%ld/%ld, inflight=%u (head=%u tail=%u), Cachehit=%0.2f%%\n",
+ file_depths(fdepths);
+ printf("IOPS=%lu, IOS/call=%ld/%ld, inflight=%u (%s)\n",
this_done - done, rpc, ipc, s->inflight,
- *s->cq_ring.head, *s->cq_ring.tail, hit);
+ fdepths);
done = this_done;
calls = this_call;
reap = this_reap;
- cache_hit = s->cachehit;
- cache_miss = s->cachemiss;
} while (!finish);
pthread_join(s->thread, &ret);
close(s->ring_fd);
+ free(fdepths);
return 0;
}