overflow support
cover overflow support
This series adds more support overflow conditions, including testing what happens when an overflow CQE is not able to be allocated.
Patches 1,2,4:
- clean up existing documentation to account for updated kernel behaviour
Patch 3:
- exposes some API to allow applications to inspect overflow state
Patch 5:
- test new overflow API and new kernel error signalling
Dylan Yudaken (5):
fix documentation shortenings
update io_uring_enter.2 docs for IORING_FEAT_NODROP expose CQ ring overflow state
add docs for overflow lost errors
overflow: add tests
man/io_uring_enter.2 | 24 +++- man/io_uring_setup.2 | 11 +- man/io_uring_wait_cqe.3 | 2 +- man/io_uring_wait_cqe_nr.3 | 2 +- man/io_uring_wait_cqe_timeout.3 | 2 +- man/io_uring_wait_cqes.3 | 2 +- src/include/liburing.h | 10 ++ src/queue.c | 31 +++-- test/cq-overflow.c | 240 +++++++++++++++++++++++++++++++- 9 files changed, 298 insertions(+), 26 deletions(-)
base-commit: b7d8dd8bbf5b8550c8a0c1ed70431cd8050709f0
Signature · 1 line
2.30.2
PATCH 1/5 fix documentation shortenings
'parm' probably was intended to be param, but parameter is more explicit, so change it.
--- man/io_uring_wait_cqe.3 | 2 +- man/io_uring_wait_cqe_nr.3 | 2 +- man/io_uring_wait_cqe_timeout.3 | 2 +- man/io_uring_wait_cqes.3 | 2 +- 4 files changed, 4 insertions(+), 4 deletions(-)
man/io_uring_wait_cqe.3 +1 -1
@@ -30,6 +30,6 @@ the application can retrieve the completion with .SH RETURN VALUE On success .BR io_uring_wait_cqe (3) -returns 0 and the cqe_ptr parm is filled in. On failure it returns -errno. +returns 0 and the cqe_ptr parameter is filled in. On failure it returns -errno. .SH SEE ALSO .BR io_uring_submit (3), io_uring_wait_cqes(3)
man/io_uring_wait_cqe_nr.3 +1 -1
@@ -33,6 +33,6 @@ the application can retrieve the completion with .SH RETURN VALUE On success .BR io_uring_wait_cqe_nr (3) -returns 0 and the cqe_ptr parm is filled in. On failure it returns -errno. +returns 0 and the cqe_ptr parameter is filled in. On failure it returns -errno. .SH SEE ALSO .BR io_uring_submit (3), io_uring_wait_cqes (3)
man/io_uring_wait_cqe_timeout.3 +1 -1
@@ -34,6 +34,6 @@ calling io_uring_wait_cqes (3). .SH RETURN VALUE On success .BR io_uring_wait_cqes (3) -returns 0 and the cqe_ptr parm is filled in. On failure it returns -errno. +returns 0 and the cqe_ptr parameter is filled in. On failure it returns -errno. .SH SEE ALSO .BR io_uring_submit (3), io_uring_wait_cqe_timeout (3), io_uring_wait_cqe(3).
man/io_uring_wait_cqes.3 +1 -1
@@ -41,6 +41,6 @@ calling io_uring_wait_cqes (3). .SH RETURN VALUE On success .BR io_uring_wait_cqes (3) -returns 0 and the cqe_ptr parm is filled in. On failure it returns -errno. +returns 0 and the cqe_ptr parameter is filled in. On failure it returns -errno. .SH SEE ALSO .BR io_uring_submit (3), io_uring_wait_cqe_timeout (3), io_uring_wait_cqe(3).
Signature · 1 line
2.30.2
PATCH 2/5 update io_uring_enter.2 docs for IORING_FEAT_NODROP
In reply to Dylan YudakenThe EBUSY docs are out of date, so update them for the IORING_FEAT_NODROP feature flag
--- man/io_uring_enter.2 | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-)
man/io_uring_enter.2 +12 -1
@@ -1327,7 +1327,18 @@ is a valid file descriptor, but the io_uring ring is not in the right state for details on how to enable the ring. .TP .B EBUSY -The application is attempting to overcommit the number of requests it can have +If the +.B IORING_FEAT_NODROP +feature flag is set, then +.B EBUSY +will be returned if there were overflow entries, +.B IORING_ENTER_GETEVENTS +flag is set and not all of the overflow entries were able to be flushed to +the CQ ring. + +Without +.B IORING_FEAT_NODROP +the application is attempting to overcommit the number of requests it can have pending. The application should wait for some completions and try again. May occur if the application tries to queue more requests than we have room for in the CQ ring, or if the application attempts to wait for more events without
Signature · 1 line
2.30.2
PATCH 3/5 expose CQ ring overflow state
In reply to Dylan YudakenAllow the application to easily view if the CQ ring is in overflow state, and also a simple method to flush overflow entries on to the CQ ring. Explicit flushing can be useful for applications that prefer to have reduced latency on CQs than to process as many as possible.
--- src/include/liburing.h | 10 ++++++++++ src/queue.c | 31 +++++++++++++++++++------------ 2 files changed, 29 insertions(+), 12 deletions(-)
src/include/liburing.h +10 -0
@@ -135,6 +135,7 @@ int io_uring_submit_and_wait_timeout(struct io_uring *ring,unsigned wait_nr, struct __kernel_timespec *ts, sigset_t *sigmask); +int io_uring_flush_overflow (struct io_uring *ring);int io_uring_register_buffers (struct io_uring *ring,const struct iovec *iovecs,unsigned nr_iovecs);
@@ -781,6 +782,15 @@ static inline unsigned io_uring_cq_ready(const struct io_uring *ring)return io_uring_smp_load_acquire (ring->cq.ktail) - *ring->cq.khead; } +/* +* Returns true if there are overflow entries waiting to be flushed onto +* the CQ ring +*/ +static inline bool io_uring_cq_has_overflow (const struct io_uring *ring) +{ +return IO_URING_READ_ONCE (*ring->sq.kflags) &IORING_SQ_CQ_OVERFLOW ; +} +/* * Returns true if the eventfd notification is currently enabled */
src/queue.c +19 -12
@@ -33,14 +33,10 @@ static inline bool sq_ring_needs_enter(struct io_uring *ring, unsigned *flags)return false; } -static inline bool cq_ring_needs_flush (struct io_uring *ring) -{ -return IO_URING_READ_ONCE (*ring->sq.kflags) &IORING_SQ_CQ_OVERFLOW ; -} -static inline bool cq_ring_needs_enter (struct io_uring *ring) { -return (ring->flags &IORING_SETUP_IOPOLL ) ||cq_ring_needs_flush (ring); +return (ring->flags &IORING_SETUP_IOPOLL ) || +io_uring_cq_has_overflow (ring); }struct get_data {
@@ -123,6 +119,21 @@ int __io_uring_get_cqe(struct io_uring *ring, struct io_uring_cqe **cqe_ptr,return _io_uring_get_cqe (ring, cqe_ptr, &data); } +static int io_uring_get_events (struct io_uring *ring) +{ +int flags =IORING_ENTER_GETEVENTS ; + +if (ring->int_flags &INT_FLAG_REG_RING ) + flags |=IORING_ENTER_REGISTERED_RING ; +return ____sys_io_uring_enter (ring->enter_ring_fd,0 ,0 , flags,NULL ); +} + +int io_uring_flush_overflow (struct io_uring *ring) +{ +return io_uring_get_events (ring); +} + +/* * Fill in an array of IO completions up to count, if any are available. * Returns the amount of IO completions filled.
@@ -152,12 +163,8 @@ again:if (overflow_checked) goto done; -if (cq_ring_needs_flush (ring)) { -int flags =IORING_ENTER_GETEVENTS ; - -if (ring->int_flags &INT_FLAG_REG_RING ) - flags |=IORING_ENTER_REGISTERED_RING ; -____sys_io_uring_enter (ring->enter_ring_fd,0 ,0 , flags,NULL ); +if (io_uring_cq_has_overflow (ring)) { +io_uring_get_events (ring); overflow_checked = true; goto again; }
Signature · 1 line
2.30.2
PATCH 4/5 add docs for overflow lost errors
In reply to Dylan YudakenAdd man docs for return values indicating a CQE was lost.
--- man/io_uring_enter.2 | 11 +++++++++++ man/io_uring_setup.2 | 11 +++++++++-- 2 files changed, 20 insertions(+), 2 deletions(-)
man/io_uring_enter.2 +11 -0
@@ -1326,6 +1326,17 @@ is a valid file descriptor, but the io_uring ring is not in the right state .BR io_uring_register (2) for details on how to enable the ring. .TP +.B EBADR +At least one CQE was dropped even with the +.B IORING_FEAT_NODROP +feature, and there are no otherwise available CQEs. This clears the error state +and so with no other changes the next call to +.BR io_uring_setup (2) +will not have this error. This error should be extremely rare and indicates the +machine is running critically low on memory and. It may be reasonable for the +application to terminate running unless it is able to safely handle any CQE +being lost. +.TP .B EBUSY If the .B IORING_FEAT_NODROP
man/io_uring_setup.2 +9 -2
@@ -218,7 +218,7 @@ call. The SQEs must still be allocated separately. This brings the necessary calls down from three to two. Available since kernel 5.4. .TP .B IORING_FEAT_NODROP -If this flag is set, io_uring supports never dropping completion events. +If this flag is set, io_uring supports almost never dropping completion events. If a completion event occurs and the CQ ring is full, the kernel stores the event internally until such a time that the CQ ring has room for more entries. If this overflow condition is entered, attempting to submit more
@@ -226,7 +226,14 @@ IO will fail with the .B -EBUSY error value, if it can't flush the overflown events to the CQ ring. If this happens, the application must reap events from the CQ ring and attempt the -submit again. Available since kernel 5.5. +submit again. If the kernel has no free memory to store the event internally +it will be visible by an increase in the overflow value on the cqring. +Available since kernel 5.5. Additionally +.BR io_uring_enter (2) +will return +.B -EBADR +the next time it would otherwise sleep waiting for completions (since kernel 5.19). + .TP .B IORING_FEAT_SUBMIT_STABLE If this flag is set, applications can be certain that any data for
Signature · 1 line
2.30.2
PATCH 5/5 overflow: add tests
In reply to Dylan YudakenAdd tests that verify that overflow conditions behave appropriately. Specifically:
* if overflow is continually flushed, then CQEs should arrive mostly in order to prevent starvation of some completions
* if CQEs are dropped due to GFP_ATOMIC allocation failures it is possible to terminate cleanly. This is not tested by default as it requires debug kernel config, and also has system-wide effects
--- test/cq-overflow.c | 240 +++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 233 insertions(+), 7 deletions(-)
test/cq-overflow.c +233 -7
@@ -9,6 +9,7 @@#include <stdlib.h> #include <string.h> #include <fcntl.h> +#include <assert.h> #include "helpers.h" #include "liburing.h"
@@ -21,6 +22,32 @@ static struct iovec *vecs;#define ENTRIES 8 +/* +* io_uring has rare cases where CQEs are lost. +* This happens when there is no space in the CQ ring, and also there is no +* GFP_ATOMIC memory available. In reality this probably means that the process +* is about to be killed as many other things might start failing, but we still +* want to test that liburing and the kernel deal with this properly. The fault +* injection framework allows us to test this scenario. Unfortunately this +* requires some system wide changes and so we do not enable this by default. +* The tests in this file should work in both cases (where overflows are queued +* and where they are dropped) on recent kernels. +* +* In order to test dropped CQEs you should enable fault injection in the kernel +* config: +* +* CONFIG_FAULT_INJECTION=y +* CONFIG_FAILSLAB=y +* CONFIG_FAULT_INJECTION_DEBUG_FS=y +* +* and then run the test as follows: +* echo Y > /sys/kernel/debug/failslab/task-filter +* echo 100 > /sys/kernel/debug/failslab/probability +* echo 0 > /sys/kernel/debug/failslab/verbose +* echo 100000 > /sys/kernel/debug/failslab/times +* bash -c "echo 1 > /proc/self/make-it-fail && exec ./cq-overflow.t" +*/ +static int test_io (const char *file,unsigned long usecs,unsigned *drops,int fault) {struct io_uring_sqe *sqe;
@@ -29,6 +56,7 @@ static int test_io(const char *file, unsigned long usecs, unsigned *drops, int funsigned reaped, total;struct io_uring ring;int nodrop, i, fd, ret; +bool cqe_dropped = false; fd =open (file,O_RDONLY |O_DIRECT );if (fd <0 ) {
@@ -103,8 +131,8 @@ static int test_io(const char *file, unsigned long usecs, unsigned *drops, int f reap_it: reaped =0 ; do { - if (nodrop) { - /* nodrop should never lose events */ + if (nodrop && !cqe_dropped) { + /* nodrop should never lose events unless cqe_dropped */ if (reaped == total) break; } else {
@@ -112,7 +140,10 @@ reap_it:break ; } ret =io_uring_wait_cqe (&ring, &cqe); -if (ret) { +if (nodrop && ret == -EBADR ) { + cqe_dropped = true; +continue ; + }else if (ret) {fprintf (stderr,"wait_cqe=%d\n" , ret); goto err; }
@@ -132,7 +163,7 @@ reap_it: goto err; } -if (!nodrop) { +if (!nodrop || cqe_dropped) { *drops = *ring.cq.koverflow; }else if (*ring.cq.koverflow) {fprintf (stderr,"Found %u overflows\n" , *ring.cq.koverflow);
@@ -153,18 +184,31 @@ static int reap_events(struct io_uring *ring, unsigned nr_events, int do_wait) {struct io_uring_cqe *cqe;int i, ret =0 , seq =0 ; +unsigned int start_overflow = *ring->cq.koverflow; +unsigned int drop_count =0 ; +bool dropped = false;for (i =0 ; i < nr_events; i++) {if (do_wait) ret =io_uring_wait_cqe (ring, &cqe);else ret =io_uring_peek_cqe (ring, &cqe); -if (ret) { +if (do_wait && ret == -EBADR ) { +unsigned int this_drop = *ring->cq.koverflow - + start_overflow; + + dropped = true; + drop_count += this_drop; + start_overflow = *ring->cq.koverflow; +assert (this_drop >0 ); + i += (this_drop -1 ); +continue ; + }else if (ret) {if (ret != -EAGAIN )fprintf (stderr,"cqe peek failed: %d\n" , ret);break ; } -if (cqe->user_data != seq) { +if (!dropped && cqe->user_data != seq) {fprintf (stderr,"cqe sequence out-of-order\n" );fprintf (stderr,"got %d, wanted %d\n" , (int ) cqe->user_data, seq);
@@ -241,19 +285,201 @@ err:return 1 ; } + +static void submit_one_nop (struct io_uring *ring,int ud) +{ +struct io_uring_sqe *sqe; +int ret; + + sqe =io_uring_get_sqe (ring); +assert (sqe); +io_uring_prep_nop (sqe); + sqe->user_data = ud; + ret =io_uring_submit (ring); +assert (ret ==1 ); +} + +/* +* Create an overflow condition and ensure that SQEs are still processed +*/ +static int test_overflow_handling ( +bool batch, +int cqe_multiple, +bool poll) +{ +struct io_uring ring; +struct io_uring_params p; +int ret, i, j, ud, cqe_count; +unsigned int count; +int const N =8 ; +int const LOOPS =128 ; +int const QUEUE_LENGTH =1024 ; +int completions[N ]; +int queue[QUEUE_LENGTH ]; +int queued =0 ; +int outstanding =0 ; +bool cqe_dropped = false; + +memset (&completions,0 ,sizeof (int ) *N ); +memset (&p,0 ,sizeof (p)); + p.cq_entries =2 * cqe_multiple; + p.flags |=IORING_SETUP_CQSIZE ; + +if (poll) + p.flags |=IORING_SETUP_IOPOLL ; + + ret =io_uring_queue_init_params (2 , &ring, &p); +if (ret) { +fprintf (stderr,"io_uring_queue_init failed %d\n" , ret); +return 1 ; + } + +assert (p.cq_entries <N ); +/* submit N SQEs, some should overflow */ +for (i =0 ; i <N ; i++) { +submit_one_nop (&ring, i); + outstanding++; + } + +for (i =0 ; i <LOOPS ; i++) { +struct io_uring_cqe *cqes[N ]; + +if (io_uring_cq_has_overflow (&ring)) { +/* +* Flush any overflowed CQEs and process those. Actively +* flush these to make sure CQEs arrive in vague order +* of being sent. +*/ + ret =io_uring_flush_overflow (&ring); +if (ret !=0 ) { +fprintf (stderr, +"io_uring_flush_overflow returned %d\n" , + ret); + goto err; + } + }else if (!cqe_dropped) { +for (j =0 ; j < queued; j++) { +submit_one_nop (&ring, queue[j]); + outstanding++; + } + queued =0 ; + } + +/* We have lost some random cqes, stop if no remaining. */ +if (cqe_dropped && outstanding == *ring.cq.koverflow) +break ; + + ret =io_uring_wait_cqe (&ring, &cqes[0 ]); +if (ret == -EBADR ) { + cqe_dropped = true; +fprintf (stderr,"CQE dropped\n" ); +continue ; + }else if (ret !=0 ) { +fprintf (stderr,"io_uring_wait_cqes failed %d\n" , ret); + goto err; + } + cqe_count =1 ; +if (batch) { + ret =io_uring_peek_batch_cqe (&ring, &cqes[0 ],2 ); +if (ret <0 ) { +fprintf (stderr, +"io_uring_peek_batch_cqe failed %d\n" , + ret); + goto err; + } + cqe_count = ret; + } +for (j =0 ; j < cqe_count; j++) { +assert (cqes[j]->user_data <N ); + ud = cqes[j]->user_data; + completions[ud]++; +assert (queued <QUEUE_LENGTH ); + queue[queued++] = (int )ud; + } +io_uring_cq_advance (&ring, cqe_count); + outstanding -= cqe_count; + } + +/* See if there were any drops by flushing the CQ ring *and* overflow */ +do { +struct io_uring_cqe *cqe; + + ret =io_uring_flush_overflow (&ring); +if (ret <0 ) { +if (ret == -EBADR ) { +fprintf (stderr,"CQE dropped\n" ); + cqe_dropped = true; +break ; + } + goto err; + } +if (outstanding && !io_uring_cq_ready (&ring)) + ret =io_uring_wait_cqe_timeout (&ring, &cqe,NULL ); + +if (ret && ret != -ETIME ) { +if (ret == -EBADR ) { +fprintf (stderr,"CQE dropped\n" ); + cqe_dropped = true; +break ; + } +fprintf (stderr,"wait_cqe_timeout = %d\n" , ret); + goto err; + } + count =io_uring_cq_ready (&ring); +io_uring_cq_advance (&ring, count); + outstanding -= count; + }while (count); + +io_uring_queue_exit (&ring); + +/* Make sure that completions come back in the same order they were +* sent. If they come back unfairly then this will concentrate on a +* couple of indices. +*/ +for (i =1 ; !cqe_dropped && i <N ; i++) { +if (abs (completions[i] - completions[i -1 ]) >1 ) { +fprintf ( + stderr, +"bad completion size %d %d\n" , + completions[i], + completions[i -1 ]); + goto err; + } + } +return 0 ; +err: +io_uring_queue_exit (&ring); +return 1 ; +} +int main (int argc,char *argv[]) {const char *fname =".cq-overflow" ;unsigned iters, drops;unsigned long usecs;int ret; +int i;if (argc >1 )return 0 ; +for (i =0 ; i <8 ; i++) { +bool batch = i &1 ; +int mult = (i &2 ) ?1 :2 ; +bool poll = i &4 ; + + ret =test_overflow_handling (batch, mult, poll); +if (ret) { +fprintf (stderr,"test_overflow_handling(" +"batch=%d, mult=%d, poll=%d) failed\n" , + batch, mult, poll); + goto err; + } + } + ret =test_overflow ();if (ret) { -printf ("test_overflow failed\n" ); +fprintf (stderr,"test_overflow failed\n" );return ret; }
Signature · 1 line
2.30.2
Re: [PATCH liburing 5/5] overflow: add tests
Quoted 10 lines · Dylan Yudaken
Add tests that verify that overflow conditions behave appropriately. Specifically:
* if overflow is continually flushed, then CQEs should arrive mostly in order to prevent starvation of some completions
* if CQEs are dropped due toGFP_ATOMICallocation failures it is possible to terminate cleanly. This is not tested by default as it requires debug kernel config, and also has system-wide effectsSigned-off-by Dylan Yudaken <[email protected]>---
Dylan, this breaks -Werror build with clang-15.
```
cq-overflow.c:188:15: error: variable 'drop_count' set but not used [-Werror,-Wunused-but-set-variable]
unsigned int drop_count = 0;
^
1 error generated.
make[1]: *** [Makefile:210: cq-overflow.t] Error 1
make[1]: *** Waiting for unfinished jobs....
```Maybe you miss something that you forgot to use the value of @drop_count?
Signature · 1 line
Ammar Faizi
Re: [PATCH liburing 5/5] overflow: add tests
Quoted 27 lines · Ammar Faizi
Quoted 11 lines · Dylan Yudaken
Add tests that verify that overflow conditions behave appropriately.
Specifically:
* if overflow is continually flushed, then CQEs should arrive mostly in
order to prevent starvation of some completions
* if CQEs are dropped due toGFP_ATOMICallocation failures it is possible to terminate cleanly. This is not tested by default as it
requires debug kernel config, and also has system-wide effectsQuoted 2 lines
Signed-off-by Dylan Yudaken <[email protected]>---
Dylan, this breaks -Werror build with clang-15.
``` cq-overflow.c:188:15: error: variable 'drop_count' set but not used [-Werror,-Wunused-but-set-variable] unsigned int drop_count = 0; ^ 1 error generated. make[1]: *** [Makefile:210: cq-overflow.t] Error 1 make[1]: *** Waiting for unfinished jobs.... ```
Quoted 3 lines
Maybe you miss something that you forgot to use the value of @drop_count?
Ah - I had some debug statements that were using it but did not remove it when removing those (the versions of gcc/clang I have do not complain). I will clean it up in v2.
Thanks for spotting it!
Jens Axboe
maintainerRe: [PATCH liburing 5/5] overflow: add tests
In reply to Dylan YudakenQuoted 7 lines · Dylan Yudaken
Add tests that verify that overflow conditions behave appropriately. Specifically:
* if overflow is continually flushed, then CQEs should arrive mostly in order to prevent starvation of some completions
* if CQEs are dropped due toGFP_ATOMICallocation failures it is possible to terminate cleanly. This is not tested by default as it requires debug kernel config, and also has system-wide effects
Since you're doing a v2 of this anyway, can we change:
test/cq-overflow.c err: · line 309
Quoted 8 lines
test/cq-overflow.c +8
-0+/* +* Create an overflow condition and ensure that SQEs are still processed +*/ +static int test_overflow_handling ( +bool batch, +int cqe_multiple, +bool poll) +{
to follow the normal stye of:
static int test_overflow_handling(bool batch, int cqe_multiple, bool poll)
{
instead?
Signature · 1 line
Jens Axboe