Skip to contentSkip to conversation

io_uring: fix skipping of old timeout events

PATCH In review Marcelo Diop-Gonzalez 3 participants 7 messages Jan 15, 2021

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v4 cover io_uring: fix skipping of old timeout events

This patch tries to fix a problem with IORING_OP_TIMEOUT events not being flushed if they should already have expired. The test below hangs before this change (unless you run with $ ./a.out ~/somefile 1):

#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

#include <liburing.h>

int main(int argc, char **argv) {
	if (argc < 2)
		return 1;

	int fd = open(argv[1], O_RDONLY);
	if (fd < 0) {
		perror("open");
		return 1;
	}

	struct io_uring ring;
	io_uring_queue_init(4, &ring, 0);

	struct io_uring_sqe *sqe = io_uring_get_sqe(&ring);

	struct __kernel_timespec ts = { .tv_sec = 9999999 };
	io_uring_prep_timeout(sqe, &ts, 1, 0);
	sqe->user_data = 123;
	int ret = io_uring_submit(&ring);
	if (ret < 0) {
		fprintf(stderr, "submit(timeout_sqe): %d\n", ret);
		return 1;
	}

	int n = 2;
	if (argc > 2)
		n = atoi(argv[2]);

	char buf;
	for (int i = 0; i < n; i++) {
		sqe = io_uring_get_sqe(&ring);
		if (!sqe) {
			fprintf(stderr, "too many\n");
			exit(1);
		}
		io_uring_prep_read(sqe, fd, &buf, 1, 0);
	}
	ret = io_uring_submit(&ring);
	if (ret < 0) {
		fprintf(stderr, "submit(read_sqe): %d\n", ret);
		exit(1);
	}

	struct io_uring_cqe *cqe;
	for (int i = 0; i < n+1; i++) {
		struct io_uring_cqe *cqe;
		int ret = io_uring_wait_cqe(&ring, &cqe);
		if (ret < 0) {
			fprintf(stderr, "wait_cqe(): %d\n", ret);
			return 1;
		}
		if (cqe->user_data == 123)
			printf("timeout found\n");
		io_uring_cqe_seen(&ring, cqe);
	}
}

v4: Reorder ->cq_timeouts read, and add some comments
v3: Add ->last_flush member to handle more overflow issues
v2: Properly handle u32 overflow issues

Marcelo Diop-Gonzalez (1):
io_uring: flush timeouts that should already have expired

 fs/io_uring.c | 34 ++++++++++++++++++++++++++++++----
 1 file changed, 30 insertions(+), 4 deletions(-)
Signature · 1 line
2.20.1
#1 lore raw

PATCH v4 1/1 io_uring: flush timeouts that should already have expired

Right now io_flush_timeouts() checks if the current number of events is equal to ->timeout.target_seq, but this will miss some timeouts if there have been more than 1 event added since the last time they were flushed (possible in io_submit_flush_completions(), for example). Fix it by recording the last sequence at which timeouts were flushed so that the number of events seen can be compared to the number of events needed without overflow.

---
 fs/io_uring.c | 34 ++++++++++++++++++++++++++++++----
 1 file changed, 30 insertions(+), 4 deletions(-)
fs/io_uring.c +30 -4
index 372be9caf340..06cc79d39586 100644
@@ -354,6 +354,7 @@ struct io_ring_ctx {
 		unsigned		cq_entries;
 		unsigned		cq_mask;
 		atomic_t		cq_timeouts;
+		unsigned		cq_last_tm_flush;
 		unsigned long		cq_check_overflow;
 		struct wait_queue_head	cq_wait;
 		struct fasync_struct	*cq_fasync;
@@ -1639,19 +1640,38 @@ static void __io_queue_deferred(struct io_ring_ctx *ctx)
 
 static void io_flush_timeouts(struct io_ring_ctx *ctx)
 {
-	while (!list_empty(&ctx->timeout_list)) {
+	u32 seq;
+
+	if (list_empty(&ctx->timeout_list))
+		return;
+
+	seq = ctx->cached_cq_tail - atomic_read(&ctx->cq_timeouts);
+
+	do {
+		u32 events_needed, events_got;
 		struct io_kiocb *req = list_first_entry(&ctx->timeout_list,
 						struct io_kiocb, timeout.list);
 
 		if (io_is_timeout_noseq(req))
 			break;
-		if (req->timeout.target_seq != ctx->cached_cq_tail
-					- atomic_read(&ctx->cq_timeouts))
+
+		/*
+		 * Since seq can easily wrap around over time, subtract
+		 * the last seq at which timeouts were flushed before comparing.
+		 * Assuming not more than 2^31-1 events have happened since,
+		 * these subtractions won't have wrapped, so we can check if
+		 * target is in [last_seq, current_seq] by comparing the two.
+		 */
+		events_needed = req->timeout.target_seq - ctx->cq_last_tm_flush;
+		events_got = seq - ctx->cq_last_tm_flush;
+		if (events_got < events_needed)
 			break;
 
 		list_del_init(&req->timeout.list);
 		io_kill_timeout(req);
-	}
+	} while (!list_empty(&ctx->timeout_list));
+
+	ctx->cq_last_tm_flush = seq;
 }
 
 static void io_commit_cqring(struct io_ring_ctx *ctx)
@@ -5837,6 +5857,12 @@ static int io_timeout(struct io_kiocb *req)
 	tail = ctx->cached_cq_tail - atomic_read(&ctx->cq_timeouts);
 	req->timeout.target_seq = tail + off;
 
+	/* Update the last seq here in case io_flush_timeouts() hasn't.
+	 * This is safe because ->completion_lock is held, and submissions
+	 * and completions are never mixed in the same ->completion_lock section.
+	 */
+	ctx->cq_last_tm_flush = tail;
+
 	/*
 	 * Insertion sort, ensuring the first entry in the list is always
 	 * the one we need first.
Signature · 1 line
2.20.1
maintainer
#2 lore raw
In reply to Marcelo Diop-Gonzalez
Quoted 60 lines · Marcelo Diop-Gonzalez

This patch tries to fix a problem with IORING_OP_TIMEOUT events not being flushed if they should already have expired. The test below hangs before this change (unless you run with $ ./a.out ~/somefile 1):

#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

#include <liburing.h>

int main(int argc, char **argv) {
	if (argc < 2)
		return 1;

	int fd = open(argv[1], O_RDONLY);
	if (fd < 0) {
		perror("open");
		return 1;
	}

	struct io_uring ring;
	io_uring_queue_init(4, &ring, 0);

	struct io_uring_sqe *sqe = io_uring_get_sqe(&ring);

	struct __kernel_timespec ts = { .tv_sec = 9999999 };
	io_uring_prep_timeout(sqe, &ts, 1, 0);
	sqe->user_data = 123;
	int ret = io_uring_submit(&ring);
	if (ret < 0) {
		fprintf(stderr, "submit(timeout_sqe): %d\n", ret);
		return 1;
	}

	int n = 2;
	if (argc > 2)
		n = atoi(argv[2]);

	char buf;
	for (int i = 0; i < n; i++) {
		sqe = io_uring_get_sqe(&ring);
		if (!sqe) {
			fprintf(stderr, "too many\n");
			exit(1);
		}
		io_uring_prep_read(sqe, fd, &buf, 1, 0);
	}
	ret = io_uring_submit(&ring);
	if (ret < 0) {
		fprintf(stderr, "submit(read_sqe): %d\n", ret);
		exit(1);
	}

	struct io_uring_cqe *cqe;
	for (int i = 0; i < n+1; i++) {
		struct io_uring_cqe *cqe;
		int ret = io_uring_wait_cqe(&ring, &cqe);
		if (ret < 0) {
Quoted 8 lines
			fprintf(stderr, "wait_cqe(): %d\n", ret);
			return 1;
		}
		if (cqe->user_data == 123)
			printf("timeout found\n");
		io_uring_cqe_seen(&ring, cqe);
	}
}

Can you turn this into a test case for liburing? I'll apply the associated patch, thanks (and to Pavel for review as well).

Signature · 1 line
Jens Axboe
#3 lore raw
Quoted 70 lines · Jens Axboe
Quoted 60 lines · Marcelo Diop-Gonzalez

This patch tries to fix a problem with IORING_OP_TIMEOUT events not being flushed if they should already have expired. The test below hangs before this change (unless you run with $ ./a.out ~/somefile 1):

#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

#include <liburing.h>

int main(int argc, char **argv) {
	if (argc < 2)
		return 1;

	int fd = open(argv[1], O_RDONLY);
	if (fd < 0) {
		perror("open");
		return 1;
	}

	struct io_uring ring;
	io_uring_queue_init(4, &ring, 0);

	struct io_uring_sqe *sqe = io_uring_get_sqe(&ring);

	struct __kernel_timespec ts = { .tv_sec = 9999999 };
	io_uring_prep_timeout(sqe, &ts, 1, 0);
	sqe->user_data = 123;
	int ret = io_uring_submit(&ring);
	if (ret < 0) {
		fprintf(stderr, "submit(timeout_sqe): %d\n", ret);
		return 1;
	}

	int n = 2;
	if (argc > 2)
		n = atoi(argv[2]);

	char buf;
	for (int i = 0; i < n; i++) {
		sqe = io_uring_get_sqe(&ring);
		if (!sqe) {
			fprintf(stderr, "too many\n");
			exit(1);
		}
		io_uring_prep_read(sqe, fd, &buf, 1, 0);
	}
	ret = io_uring_submit(&ring);
	if (ret < 0) {
		fprintf(stderr, "submit(read_sqe): %d\n", ret);
		exit(1);
	}

	struct io_uring_cqe *cqe;
	for (int i = 0; i < n+1; i++) {
		struct io_uring_cqe *cqe;
		int ret = io_uring_wait_cqe(&ring, &cqe);
		if (ret < 0) {
Quoted 8 lines
			fprintf(stderr, "wait_cqe(): %d\n", ret);
			return 1;
		}
		if (cqe->user_data == 123)
			printf("timeout found\n");
		io_uring_cqe_seen(&ring, cqe);
	}
}
Quoted 2 lines

Can you turn this into a test case for liburing? I'll apply the associated patch, thanks (and to Pavel for review as well).

Yup, can do. I'll try to clean it up some first (especially so it doesn't just hang when it fails :/)

Quoted 4 lines
Signature · 2 lines
-- 
Jens Axboe
maintainer
#4 lore raw
Quoted 71 lines · Marcelo Diop-Gonzalez
Quoted 70 lines · Jens Axboe
Quoted 68 lines · Marcelo Diop-Gonzalez

This patch tries to fix a problem with IORING_OP_TIMEOUT events not being flushed if they should already have expired. The test below hangs before this change (unless you run with $ ./a.out ~/somefile 1):

#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

#include <liburing.h>

int main(int argc, char **argv) {
	if (argc < 2)
		return 1;

	int fd = open(argv[1], O_RDONLY);
	if (fd < 0) {
		perror("open");
		return 1;
	}

	struct io_uring ring;
	io_uring_queue_init(4, &ring, 0);

	struct io_uring_sqe *sqe = io_uring_get_sqe(&ring);

	struct __kernel_timespec ts = { .tv_sec = 9999999 };
	io_uring_prep_timeout(sqe, &ts, 1, 0);
	sqe->user_data = 123;
	int ret = io_uring_submit(&ring);
	if (ret < 0) {
		fprintf(stderr, "submit(timeout_sqe): %d\n", ret);
		return 1;
	}

	int n = 2;
	if (argc > 2)
		n = atoi(argv[2]);

	char buf;
	for (int i = 0; i < n; i++) {
		sqe = io_uring_get_sqe(&ring);
		if (!sqe) {
			fprintf(stderr, "too many\n");
			exit(1);
		}
		io_uring_prep_read(sqe, fd, &buf, 1, 0);
	}
	ret = io_uring_submit(&ring);
	if (ret < 0) {
		fprintf(stderr, "submit(read_sqe): %d\n", ret);
		exit(1);
	}

	struct io_uring_cqe *cqe;
	for (int i = 0; i < n+1; i++) {
		struct io_uring_cqe *cqe;
		int ret = io_uring_wait_cqe(&ring, &cqe);
		if (ret < 0) {
			fprintf(stderr, "wait_cqe(): %d\n", ret);
			return 1;
		}
		if (cqe->user_data == 123)
			printf("timeout found\n");
		io_uring_cqe_seen(&ring, cqe);
	}
}
Quoted 5 lines
Quoted 2 lines

Can you turn this into a test case for liburing? I'll apply the associated patch, thanks (and to Pavel for review as well).

Yup, can do. I'll try to clean it up some first (especially so it doesn't just hang when it fails :/)

That'd of course be nice, but not a hard requirement. A lot of the regressions tests will crash a broken kernel, so...

Signature · 1 line
Jens Axboe
#5 lore raw
Quoted 7 lines · Jens Axboe
Quoted 6 lines · Marcelo Diop-Gonzalez
Quoted 5 lines · Jens Axboe
Quoted 3 lines · Marcelo Diop-Gonzalez

This patch tries to fix a problem with IORING_OP_TIMEOUT events not being flushed if they should already have expired. The test below hangs before this change (unless you run with $ ./a.out ~/somefile 1):

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

Can you turn this into a test case for liburing? I'll apply the associated patch, thanks (and to Pavel for review as well).

Yup, can do. I'll try to clean it up some first (especially so it doesn't just hang when it fails :/)

That'd of course be nice, but not a hard requirement. A lot of the regressions tests will crash a broken kernel, so...

Ha, they definitely will.

Marcelo, replacing reads with nop requests should trigger it as well, it's probably easier and even more reliable as we always complete them inline (if not linked or IOSQE_ASYNC).

Signature · 1 line
Pavel Begunkov
#6 lore raw
Quoted 14 lines · Pavel Begunkov
Quoted 13 lines · Jens Axboe
Quoted 11 lines · Marcelo Diop-Gonzalez
Quoted 7 lines · Jens Axboe
Quoted 3 lines · Marcelo Diop-Gonzalez

This patch tries to fix a problem with IORING_OP_TIMEOUT events not being flushed if they should already have expired. The test below hangs before this change (unless you run with $ ./a.out ~/somefile 1):

Can you turn this into a test case for liburing? I'll apply the associated patch, thanks (and to Pavel for review as well).

Yup, can do. I'll try to clean it up some first (especially so it doesn't just hang when it fails :/)

Quoted 8 lines
Quoted 2 lines

That'd of course be nice, but not a hard requirement. A lot of the regressions tests will crash a broken kernel, so...

Ha, they definitely will.

Marcelo, replacing reads with nop requests should trigger it as well, it's probably easier and even more reliable as we always complete them inline (if not linked or IOSQE_ASYNC).

Oh good idea, yeah that's better for sure. Didn't even know that existed :D

Quoted 3 lines
Signature · 2 lines
-- 
Pavel Begunkov