X-Git-Url: https://git.kernel.dk/?p=fio.git;a=blobdiff_plain;f=t%2Fio_uring.c;h=da3b4d1f4a8f16bfb23ed3c37ac9a33699ffcf1a;hp=0461329b64d2267ae240fca1545d605338b73a0e;hb=919850d217e139b4f82f9798893cc816a7435e44;hpb=a7abc9fb769596d3bbf6d779e99d1cb8c1fcd49b

diff --git a/t/io_uring.c b/t/io_uring.c
index 0461329b..da3b4d1f 100644
--- a/t/io_uring.c
+++ b/t/io_uring.c
@@ -22,9 +22,7 @@
 
 #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))
 
@@ -47,8 +45,8 @@ struct io_cq_ring {
 
 #define DEPTH			128
 
-#define BATCH_SUBMIT		64
-#define BATCH_COMPLETE		64
+#define BATCH_SUBMIT		32
+#define BATCH_COMPLETE		32
 
 #define BS			4096
 
@@ -59,7 +57,8 @@ static unsigned sq_ring_mask, cq_ring_mask;
 struct file {
 	unsigned long max_blocks;
 	unsigned pending_ios;
-	int fd;
+	int real_fd;
+	int fixed_fd;
 };
 
 struct submitter {
@@ -77,6 +76,8 @@ struct submitter {
 	unsigned long cachehit, cachemiss;
 	volatile int finish;
 
+	__s32 *fds;
+
 	struct file files[MAX_FDS];
 	unsigned nr_files;
 	unsigned cur_file;
@@ -87,35 +88,36 @@ static volatile int finish;
 
 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)
 {
-	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, &reg);
+			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, &reg);
-	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_sys_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)
@@ -147,6 +149,11 @@ 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 {
@@ -155,6 +162,7 @@ static void init_io(struct submitter *s, unsigned index)
 			s->cur_file++;
 			if (s->cur_file == s->nr_files)
 				s->cur_file = 0;
+			f = &s->files[s->cur_file];
 		}
 	}
 	f->pending_ios++;
@@ -162,22 +170,27 @@ static void init_io(struct submitter *s, unsigned index)
 	lrand48_r(&s->rand, &r);
 	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->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 {
-		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)
@@ -188,7 +201,7 @@ 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;
 
@@ -201,9 +214,9 @@ static int prep_more_ios(struct submitter *s, int max_ios)
 
 	if (*ring->tail != tail) {
 		/* order tail store with writes to sqes above */
-		barrier();
+		write_barrier();
 		*ring->tail = tail;
-		barrier();
+		write_barrier();
 	}
 	return prepped;
 }
@@ -212,12 +225,12 @@ static int get_file_size(struct file *f)
 {
 	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;
@@ -240,15 +253,17 @@ static int reap_events(struct submitter *s)
 	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);
+				return -1;
+			}
 		}
 		if (cqe->flags & IOCQE_FLAG_CACHEHIT)
 			s->cachehit++;
@@ -260,7 +275,7 @@ static int reap_events(struct submitter *s)
 
 	s->inflight -= reaped;
 	*ring->head = head;
-	barrier();
+	write_barrier();
 	return reaped;
 }
 
@@ -286,7 +301,7 @@ static void *submitter_fn(void *data)
 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);
@@ -296,8 +311,11 @@ submit:
 		 * 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;
+			ret = io_uring_enter(s, to_submit, to_wait, flags);
 			s->calls++;
 		}
 
@@ -310,9 +328,10 @@ submit:
 		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;
@@ -378,12 +397,14 @@ static int setup_ring(struct submitter *s)
 
 	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);
@@ -401,10 +422,12 @@ static int setup_ring(struct submitter *s)
 		}
 	}
 
-	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),
@@ -437,15 +460,33 @@ static int setup_ring(struct submitter *s)
 	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;
 	}
@@ -455,7 +496,7 @@ int main(int argc, char *argv[])
 		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);
@@ -463,7 +504,7 @@ int main(int argc, char *argv[])
 			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;
@@ -479,11 +520,15 @@ int main(int argc, char *argv[])
 		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();
@@ -509,6 +554,7 @@ int main(int argc, char *argv[])
 
 	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;
@@ -538,9 +584,10 @@ int main(int argc, char *argv[])
 			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), 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;
@@ -550,5 +597,6 @@ int main(int argc, char *argv[])
 
 	pthread_join(s->thread, &ret);
 	close(s->ring_fd);
+	free(fdepths);
 	return 0;
 }