#include "../arch/arch.h"
#include "../lib/types.h"
-#include "../os/io_uring.h"
+#include "../os/linux/io_uring.h"
#define barrier() __asm__ __volatile__("": : :"memory")
unsigned *tail;
unsigned *ring_mask;
unsigned *ring_entries;
+ unsigned *flags;
unsigned *array;
};
#define DEPTH 128
-#define BATCH_SUBMIT 64
-#define BATCH_COMPLETE 64
+#define BATCH_SUBMIT 32
+#define BATCH_COMPLETE 32
#define BS 4096
struct file {
unsigned long max_blocks;
unsigned pending_ios;
- int fd;
+ int real_fd;
+ int fixed_fd;
};
struct submitter {
unsigned long cachehit, cachemiss;
volatile int finish;
+ __s32 *fds;
+
struct file files[MAX_FDS];
unsigned nr_files;
unsigned cur_file;
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 io_uring_register_buffers(struct submitter *s)
{
.nr_iovecs = DEPTH
};
+ if (do_nop)
+ return 0;
+
return syscall(__NR_sys_io_uring_register, s->ring_fd,
IORING_REGISTER_BUFFERS, ®);
}
+static int io_uring_register_files(struct submitter *s)
+{
+ struct io_uring_register_files reg;
+ 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;
+ }
+ reg.fds = s->fds;
+ reg.nr_fds = s->nr_files;
+
+ return syscall(__NR_sys_io_uring_register, s->ring_fd,
+ IORING_REGISTER_FILES, ®);
+}
+
static int io_uring_setup(unsigned entries, struct io_uring_params *p)
{
return syscall(__NR_sys_io_uring_setup, entries, p);
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;
- sqe->flags = 0;
- 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->opcode = IORING_OP_READ_FIXED;
sqe->addr = s->iovecs[index].iov_base;
sqe->len = BS;
sqe->buf_index = index;
- sqe->flags |= IOSQE_FIXED_BUFFER;
} else {
+ sqe->opcode = IORING_OP_READV;
sqe->addr = &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)
{
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;
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);
+ return -1;
+ }
}
if (cqe->flags & IOCQE_FLAG_CACHEHIT)
s->cachehit++;
static void *submitter_fn(void *data)
{
struct submitter *s = data;
+ struct io_sq_ring *ring = &s->sq_ring;
int ret, prepped;
printf("submitter=%d\n", gettid());
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);
- ret = io_uring_enter(s, to_submit, to_wait,
- IORING_ENTER_GETEVENTS);
- s->calls++;
+ /*
+ * 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)) {
+ unsigned flags = 0;
+
+ if (to_wait)
+ flags = IORING_ENTER_GETEVENTS;
+ ret = io_uring_enter(s, to_submit, to_wait, flags);
+ s->calls++;
+ }
- this_reap = reap_events(s);
- if (this_reap == -1)
- break;
+ /*
+ * For non SQ thread poll, we already got the events we needed
+ * through the io_uring_enter() above. For SQ thread poll, we
+ * need to loop here until we find enough events.
+ */
+ this_reap = 0;
+ do {
+ int r;
+ r = reap_events(s);
+ if (r == -1) {
+ s->finish = 1;
+ break;
+ } else if (r > 0)
+ this_reap += r;
+ } while (sq_thread_poll && this_reap < to_wait);
s->reaps += this_reap;
if (ret >= 0) {
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);
if (fixedbufs) {
ret = io_uring_register_buffers(s);
if (ret < 0) {
- perror("io_uring_register");
+ perror("io_uring_register_buffers");
+ return 1;
+ }
+ }
+
+ if (register_files) {
+ ret = io_uring_register_files(s);
+ if (ret < 0) {
+ perror("io_uring_register_files");
return 1;
}
}
sring->tail = ptr + p.sq_off.tail;
sring->ring_mask = ptr + p.sq_off.ring_mask;
sring->ring_entries = ptr + p.sq_off.ring_entries;
+ sring->flags = ptr + p.sq_off.flags;
sring->array = ptr + p.sq_off.array;
sq_ring_mask = *sring->ring_mask;
return 0;
}
+static void file_depths(char *buf)
+{
+ struct submitter *s = &submitters[0];
+ 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);
+ }
+}
+
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;
+ char *fdepths;
void *ret;
- if (argc < 2) {
+ if (!do_nop && argc < 2) {
printf("%s: filename\n", argv[0]);
return 1;
}
flags |= O_DIRECT;
i = 1;
- while (i < argc) {
+ while (!do_nop && i < argc) {
struct file *f = &s->files[s->nr_files];
fd = open(argv[i], flags);
perror("open");
return 1;
}
- f->fd = fd;
+ 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();
pthread_create(&s->thread, NULL, submitter_fn, s);
+ fdepths = malloc(8 * s->nr_files);
cache_hit = cache_miss = reap = calls = done = 0;
do {
unsigned long this_done = 0;
if (this_call - calls) {
rpc = (this_done - done) / (this_call - calls);
ipc = (this_reap - reap) / (this_call - calls);
- }
- printf("IOPS=%lu, IOS/call=%lu/%lu, inflight=%u (head=%u tail=%u), Cachehit=%0.2f%%\n",
+ } else
+ rpc = ipc = -1;
+ file_depths(fdepths);
+ printf("IOPS=%lu, IOS/call=%ld/%ld, inflight=%u (%s), Cachehit=%0.2f%%\n",
this_done - done, rpc, ipc, s->inflight,
- *s->cq_ring.head, *s->cq_ring.tail, hit);
+ fdepths, hit);
done = this_done;
calls = this_call;
reap = this_reap;
pthread_join(s->thread, &ret);
close(s->ring_fd);
+ free(fdepths);
return 0;
}