* [PATCH 1/5] io_uring: check non-sync defer_list carefully
2020-04-30 19:31 [PATCH 0/5] timeout fixes Pavel Begunkov
@ 2020-04-30 19:31 ` Pavel Begunkov
2020-04-30 19:31 ` [PATCH 2/5] io_uring: pass nxt from sync_file_range() Pavel Begunkov
` (4 subsequent siblings)
5 siblings, 0 replies; 10+ messages in thread
From: Pavel Begunkov @ 2020-04-30 19:31 UTC (permalink / raw)
To: Jens Axboe, io-uring, linux-kernel
io_req_defer() do double-checked locking. Use proper helpers for that,
i.e. list_empty_careful().
Signed-off-by: Pavel Begunkov <[email protected]>
---
fs/io_uring.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 8ee7b4f72b8f..6b4d3d8a6941 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -4974,7 +4974,7 @@ static int io_req_defer(struct io_kiocb *req, const struct io_uring_sqe *sqe)
int ret;
/* Still need defer if there is pending req in defer list. */
- if (!req_need_defer(req) && list_empty(&ctx->defer_list))
+ if (!req_need_defer(req) && list_empty_careful(&ctx->defer_list))
return 0;
if (!req->io && io_alloc_async_ctx(req))
--
2.24.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 2/5] io_uring: pass nxt from sync_file_range()
2020-04-30 19:31 [PATCH 0/5] timeout fixes Pavel Begunkov
2020-04-30 19:31 ` [PATCH 1/5] io_uring: check non-sync defer_list carefully Pavel Begunkov
@ 2020-04-30 19:31 ` Pavel Begunkov
2020-04-30 19:31 ` [PATCH 3/5] io_uring: trigger timeout after any sqe->off CQEs Pavel Begunkov
` (3 subsequent siblings)
5 siblings, 0 replies; 10+ messages in thread
From: Pavel Begunkov @ 2020-04-30 19:31 UTC (permalink / raw)
To: Jens Axboe, io-uring, linux-kernel
Make io_sync_file_range_finish() use io_steal_work()
to pass its nxt work.
Signed-off-by: Pavel Begunkov <[email protected]>
---
fs/io_uring.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 6b4d3d8a6941..8fff427345d5 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -3502,7 +3502,7 @@ static void io_sync_file_range_finish(struct io_wq_work **workptr)
if (io_req_cancelled(req))
return;
__io_sync_file_range(req);
- io_put_req(req); /* put submission ref */
+ io_steal_work(req, workptr);
}
static int io_sync_file_range(struct io_kiocb *req, bool force_nonblock)
--
2.24.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 3/5] io_uring: trigger timeout after any sqe->off CQEs
2020-04-30 19:31 [PATCH 0/5] timeout fixes Pavel Begunkov
2020-04-30 19:31 ` [PATCH 1/5] io_uring: check non-sync defer_list carefully Pavel Begunkov
2020-04-30 19:31 ` [PATCH 2/5] io_uring: pass nxt from sync_file_range() Pavel Begunkov
@ 2020-04-30 19:31 ` Pavel Begunkov
2020-04-30 19:31 ` [PATCH 4/5] io_uring: don't trigger timeout with another t-out Pavel Begunkov
` (2 subsequent siblings)
5 siblings, 0 replies; 10+ messages in thread
From: Pavel Begunkov @ 2020-04-30 19:31 UTC (permalink / raw)
To: Jens Axboe, io-uring, linux-kernel
sequence mode timeouts wait not for sqe->off CQEs, but rather
sqe->off + number of prior inflight requests with a quirk ignoring other
timeouts completions. Wait exactly for sqe->off using completion count
(tail) for accounting.
Reported-by: Jens Axboe <[email protected]>
Signed-off-by: Pavel Begunkov <[email protected]>
---
fs/io_uring.c | 120 +++++++++++++++++++-------------------------------
1 file changed, 46 insertions(+), 74 deletions(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 8fff427345d5..006ac57af842 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -384,7 +384,8 @@ struct io_timeout {
struct file *file;
u64 addr;
int flags;
- u32 count;
+ u32 off;
+ u32 target_seq;
};
struct io_rw {
@@ -982,23 +983,6 @@ static struct io_kiocb *io_get_deferred_req(struct io_ring_ctx *ctx)
return NULL;
}
-static struct io_kiocb *io_get_timeout_req(struct io_ring_ctx *ctx)
-{
- struct io_kiocb *req;
-
- req = list_first_entry_or_null(&ctx->timeout_list, struct io_kiocb, list);
- if (req) {
- if (req->flags & REQ_F_TIMEOUT_NOSEQ)
- return NULL;
- if (!__req_need_defer(req)) {
- list_del_init(&req->list);
- return req;
- }
- }
-
- return NULL;
-}
-
static void __io_commit_cqring(struct io_ring_ctx *ctx)
{
struct io_rings *rings = ctx->rings;
@@ -1114,12 +1098,42 @@ static void io_kill_timeouts(struct io_ring_ctx *ctx)
spin_unlock_irq(&ctx->completion_lock);
}
+static inline bool io_check_in_range(u32 pos, u32 start, u32 end)
+{
+ /* if @end < @start, check for [end, MAX_UINT] + [MAX_UINT, start] */
+ return (pos - start) <= (end - start);
+}
+
+static void __io_flush_timeouts(struct io_ring_ctx *ctx)
+{
+ u32 end, start;
+
+ start = end = ctx->cached_cq_tail;
+ do {
+ struct io_kiocb *req = list_first_entry(&ctx->timeout_list,
+ struct io_kiocb, list);
+
+ if (req->flags & REQ_F_TIMEOUT_NOSEQ)
+ break;
+ /*
+ * multiple timeouts may have the same target,
+ * check that @req is in [first_tail, cur_tail]
+ */
+ if (!io_check_in_range(req->timeout.target_seq, start, end))
+ break;
+
+ list_del_init(&req->list);
+ io_kill_timeout(req);
+ end = ctx->cached_cq_tail;
+ } while (!list_empty(&ctx->timeout_list));
+}
+
static void io_commit_cqring(struct io_ring_ctx *ctx)
{
struct io_kiocb *req;
- while ((req = io_get_timeout_req(ctx)) != NULL)
- io_kill_timeout(req);
+ if (!list_empty(&ctx->timeout_list))
+ __io_flush_timeouts(ctx);
__io_commit_cqring(ctx);
@@ -4540,20 +4554,8 @@ static enum hrtimer_restart io_timeout_fn(struct hrtimer *timer)
* We could be racing with timeout deletion. If the list is empty,
* then timeout lookup already found it and will be handling it.
*/
- if (!list_empty(&req->list)) {
- struct io_kiocb *prev;
-
- /*
- * Adjust the reqs sequence before the current one because it
- * will consume a slot in the cq_ring and the cq_tail
- * pointer will be increased, otherwise other timeout reqs may
- * return in advance without waiting for enough wait_nr.
- */
- prev = req;
- list_for_each_entry_continue_reverse(prev, &ctx->timeout_list, list)
- prev->sequence++;
+ if (!list_empty(&req->list))
list_del_init(&req->list);
- }
io_cqring_fill_event(req, -ETIME);
io_commit_cqring(ctx);
@@ -4633,18 +4635,19 @@ static int io_timeout_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe,
{
struct io_timeout_data *data;
unsigned flags;
+ u32 off = READ_ONCE(sqe->off);
if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
return -EINVAL;
if (sqe->ioprio || sqe->buf_index || sqe->len != 1)
return -EINVAL;
- if (sqe->off && is_timeout_link)
+ if (off && is_timeout_link)
return -EINVAL;
flags = READ_ONCE(sqe->timeout_flags);
if (flags & ~IORING_TIMEOUT_ABS)
return -EINVAL;
- req->timeout.count = READ_ONCE(sqe->off);
+ req->timeout.off = off;
if (!req->io && io_alloc_async_ctx(req))
return -ENOMEM;
@@ -4668,68 +4671,37 @@ static int io_timeout_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe,
static int io_timeout(struct io_kiocb *req)
{
struct io_ring_ctx *ctx = req->ctx;
- struct io_timeout_data *data;
+ struct io_timeout_data *data = &req->io->timeout;
struct list_head *entry;
- unsigned span = 0;
- u32 count = req->timeout.count;
- u32 seq = req->sequence;
+ u32 tail, off = req->timeout.off;
- data = &req->io->timeout;
+ spin_lock_irq(&ctx->completion_lock);
/*
* sqe->off holds how many events that need to occur for this
* timeout event to be satisfied. If it isn't set, then this is
* a pure timeout request, sequence isn't used.
*/
- if (!count) {
+ if (!off) {
req->flags |= REQ_F_TIMEOUT_NOSEQ;
- spin_lock_irq(&ctx->completion_lock);
entry = ctx->timeout_list.prev;
goto add;
}
- req->sequence = seq + count;
+ tail = ctx->cached_cq_tail;
+ req->timeout.target_seq = tail + off;
/*
* Insertion sort, ensuring the first entry in the list is always
* the one we need first.
*/
- spin_lock_irq(&ctx->completion_lock);
list_for_each_prev(entry, &ctx->timeout_list) {
struct io_kiocb *nxt = list_entry(entry, struct io_kiocb, list);
- unsigned nxt_seq;
- long long tmp, tmp_nxt;
- u32 nxt_offset = nxt->timeout.count;
-
- if (nxt->flags & REQ_F_TIMEOUT_NOSEQ)
- continue;
-
- /*
- * Since seq + count can overflow, use type long
- * long to store it.
- */
- tmp = (long long)seq + count;
- nxt_seq = nxt->sequence - nxt_offset;
- tmp_nxt = (long long)nxt_seq + nxt_offset;
+ u32 nxt_off = nxt->timeout.target_seq - tail;
- /*
- * cached_sq_head may overflow, and it will never overflow twice
- * once there is some timeout req still be valid.
- */
- if (seq < nxt_seq)
- tmp += UINT_MAX;
-
- if (tmp > tmp_nxt)
+ if (!(nxt->flags & REQ_F_TIMEOUT_NOSEQ) && (off >= nxt_off))
break;
-
- /*
- * Sequence of reqs after the insert one and itself should
- * be adjusted because each timeout req consumes a slot.
- */
- span++;
- nxt->sequence++;
}
- req->sequence -= span;
add:
list_add(&req->list, entry);
data->timer.function = io_timeout_fn;
--
2.24.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 4/5] io_uring: don't trigger timeout with another t-out
2020-04-30 19:31 [PATCH 0/5] timeout fixes Pavel Begunkov
` (2 preceding siblings ...)
2020-04-30 19:31 ` [PATCH 3/5] io_uring: trigger timeout after any sqe->off CQEs Pavel Begunkov
@ 2020-04-30 19:31 ` Pavel Begunkov
2020-04-30 19:31 ` [PATCH 5/5] io_uring: fix timeout offset with batch CQ commit Pavel Begunkov
2020-05-01 8:21 ` [PATCH 0/5] timeout fixes Pavel Begunkov
5 siblings, 0 replies; 10+ messages in thread
From: Pavel Begunkov @ 2020-04-30 19:31 UTC (permalink / raw)
To: Jens Axboe, io-uring, linux-kernel
When deciding whether to fire a timeout basing on number of completions,
ignore CQEs emitted by other timeouts.
Signed-off-by: Pavel Begunkov <[email protected]>
---
fs/io_uring.c | 19 +++----------------
1 file changed, 3 insertions(+), 16 deletions(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 006ac57af842..fb8ec4b00375 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -1098,33 +1098,20 @@ static void io_kill_timeouts(struct io_ring_ctx *ctx)
spin_unlock_irq(&ctx->completion_lock);
}
-static inline bool io_check_in_range(u32 pos, u32 start, u32 end)
-{
- /* if @end < @start, check for [end, MAX_UINT] + [MAX_UINT, start] */
- return (pos - start) <= (end - start);
-}
-
static void __io_flush_timeouts(struct io_ring_ctx *ctx)
{
- u32 end, start;
-
- start = end = ctx->cached_cq_tail;
do {
struct io_kiocb *req = list_first_entry(&ctx->timeout_list,
struct io_kiocb, list);
if (req->flags & REQ_F_TIMEOUT_NOSEQ)
break;
- /*
- * multiple timeouts may have the same target,
- * check that @req is in [first_tail, cur_tail]
- */
- if (!io_check_in_range(req->timeout.target_seq, start, end))
+ if (req->timeout.target_seq != ctx->cached_cq_tail
+ - atomic_read(&ctx->cq_timeouts))
break;
list_del_init(&req->list);
io_kill_timeout(req);
- end = ctx->cached_cq_tail;
} while (!list_empty(&ctx->timeout_list));
}
@@ -4688,7 +4675,7 @@ static int io_timeout(struct io_kiocb *req)
goto add;
}
- tail = ctx->cached_cq_tail;
+ tail = ctx->cached_cq_tail - atomic_read(&ctx->cq_timeouts);
req->timeout.target_seq = tail + off;
/*
--
2.24.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 5/5] io_uring: fix timeout offset with batch CQ commit
2020-04-30 19:31 [PATCH 0/5] timeout fixes Pavel Begunkov
` (3 preceding siblings ...)
2020-04-30 19:31 ` [PATCH 4/5] io_uring: don't trigger timeout with another t-out Pavel Begunkov
@ 2020-04-30 19:31 ` Pavel Begunkov
2020-05-01 8:21 ` [PATCH 0/5] timeout fixes Pavel Begunkov
5 siblings, 0 replies; 10+ messages in thread
From: Pavel Begunkov @ 2020-04-30 19:31 UTC (permalink / raw)
To: Jens Axboe, io-uring, linux-kernel
Completions may be done in batches, where io_commit_cqring() is called
only once at the end. It means, that timeout sequence checks are done
only once and don't consider events in between, so potentially failing
to trigger some timeouts.
Do a separate CQ sequence accounting in u64. On timeout sequence
checking look up to UINT_MAX sequence behind, which it could have
missed. It's safe to do, because sqe->off is u32 and so can't wrap
around to used [seq - UINT_MAX, seq] window.
It's also necessary to decouple CQ timeout sequences from
ctx->cq_cached_tail for implementing "single CQE per link" feature and
others.
Signed-off-by: Pavel Begunkov <[email protected]>
---
fs/io_uring.c | 33 ++++++++++++++++++++++++++-------
1 file changed, 26 insertions(+), 7 deletions(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index fb8ec4b00375..f09c1d3a7e63 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -298,6 +298,7 @@ struct io_ring_ctx {
unsigned cq_entries;
unsigned cq_mask;
atomic_t cq_timeouts;
+ u64 cq_seq;
unsigned long cq_check_overflow;
struct wait_queue_head cq_wait;
struct fasync_struct *cq_fasync;
@@ -385,7 +386,7 @@ struct io_timeout {
u64 addr;
int flags;
u32 off;
- u32 target_seq;
+ u64 target_seq;
};
struct io_rw {
@@ -1081,6 +1082,7 @@ static void io_kill_timeout(struct io_kiocb *req)
ret = hrtimer_try_to_cancel(&req->io->timeout.timer);
if (ret != -1) {
atomic_inc(&req->ctx->cq_timeouts);
+ req->ctx->cq_seq--;
list_del_init(&req->list);
req->flags |= REQ_F_COMP_LOCKED;
io_cqring_fill_event(req, 0);
@@ -1098,16 +1100,31 @@ static void io_kill_timeouts(struct io_ring_ctx *ctx)
spin_unlock_irq(&ctx->completion_lock);
}
+static inline bool io_check_in_range(u64 pos, u64 start, u64 end)
+{
+ /* if @end < @start, check for [end, MAX_U64] + [MAX_U64, start] */
+ return (pos - start) <= (end - start);
+}
+
static void __io_flush_timeouts(struct io_ring_ctx *ctx)
{
+ u64 start_seq = ctx->cq_seq;
+
+
+ /*
+ * Batched CQ commit may have left some pending timeouts sequences
+ * behind @cq_sqe. Look back to find them. Note, that sqe->off is u32,
+ * and it uses u64 to not falsely trigger timeouts with large off.
+ */
+ start_seq -= UINT_MAX;
do {
struct io_kiocb *req = list_first_entry(&ctx->timeout_list,
struct io_kiocb, list);
if (req->flags & REQ_F_TIMEOUT_NOSEQ)
break;
- if (req->timeout.target_seq != ctx->cached_cq_tail
- - atomic_read(&ctx->cq_timeouts))
+ if (!io_check_in_range(req->timeout.target_seq, start_seq,
+ ctx->cq_seq))
break;
list_del_init(&req->list);
@@ -1143,6 +1160,7 @@ static struct io_uring_cqe *io_get_cqring(struct io_ring_ctx *ctx)
return NULL;
ctx->cached_cq_tail++;
+ ctx->cq_seq++;
return &rings->cqes[tail & ctx->cq_mask];
}
@@ -4537,6 +4555,8 @@ static enum hrtimer_restart io_timeout_fn(struct hrtimer *timer)
atomic_inc(&ctx->cq_timeouts);
spin_lock_irqsave(&ctx->completion_lock, flags);
+ ctx->cq_seq--;
+
/*
* We could be racing with timeout deletion. If the list is empty,
* then timeout lookup already found it and will be handling it.
@@ -4660,7 +4680,7 @@ static int io_timeout(struct io_kiocb *req)
struct io_ring_ctx *ctx = req->ctx;
struct io_timeout_data *data = &req->io->timeout;
struct list_head *entry;
- u32 tail, off = req->timeout.off;
+ u32 off = req->timeout.off;
spin_lock_irq(&ctx->completion_lock);
@@ -4675,8 +4695,7 @@ static int io_timeout(struct io_kiocb *req)
goto add;
}
- tail = ctx->cached_cq_tail - atomic_read(&ctx->cq_timeouts);
- req->timeout.target_seq = tail + off;
+ req->timeout.target_seq = ctx->cq_seq + off;
/*
* Insertion sort, ensuring the first entry in the list is always
@@ -4684,7 +4703,7 @@ static int io_timeout(struct io_kiocb *req)
*/
list_for_each_prev(entry, &ctx->timeout_list) {
struct io_kiocb *nxt = list_entry(entry, struct io_kiocb, list);
- u32 nxt_off = nxt->timeout.target_seq - tail;
+ u32 nxt_off = (u32)(nxt->timeout.target_seq - ctx->cq_seq);
if (!(nxt->flags & REQ_F_TIMEOUT_NOSEQ) && (off >= nxt_off))
break;
--
2.24.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH 0/5] timeout fixes
2020-04-30 19:31 [PATCH 0/5] timeout fixes Pavel Begunkov
` (4 preceding siblings ...)
2020-04-30 19:31 ` [PATCH 5/5] io_uring: fix timeout offset with batch CQ commit Pavel Begunkov
@ 2020-05-01 8:21 ` Pavel Begunkov
2020-05-01 9:38 ` Pavel Begunkov
5 siblings, 1 reply; 10+ messages in thread
From: Pavel Begunkov @ 2020-05-01 8:21 UTC (permalink / raw)
To: Jens Axboe, io-uring, linux-kernel
On 30/04/2020 22:31, Pavel Begunkov wrote:
> [1,2] are small random patches.
> [3,4] are the last 2 timeout patches, but with 1 var renamed.
> [5] fixes a timeout problem related to batched CQ commits. From
> what I see, this should be the last fixing timeouts.
Something gone wrong with testing or rebasing. Never mind this.
>
> Pavel Begunkov (5):
> io_uring: check non-sync defer_list carefully
> io_uring: pass nxt from sync_file_range()
> io_uring: trigger timeout after any sqe->off CQEs
> io_uring: don't trigger timeout with another t-out
> io_uring: fix timeout offset with batch CQ commit
>
> fs/io_uring.c | 130 +++++++++++++++++++++-----------------------------
> 1 file changed, 54 insertions(+), 76 deletions(-)
>
--
Pavel Begunkov
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 0/5] timeout fixes
2020-05-01 8:21 ` [PATCH 0/5] timeout fixes Pavel Begunkov
@ 2020-05-01 9:38 ` Pavel Begunkov
2020-05-01 4:26 ` Jens Axboe
0 siblings, 1 reply; 10+ messages in thread
From: Pavel Begunkov @ 2020-05-01 9:38 UTC (permalink / raw)
To: Jens Axboe, io-uring, linux-kernel
On 01/05/2020 11:21, Pavel Begunkov wrote:
> On 30/04/2020 22:31, Pavel Begunkov wrote:
>> [1,2] are small random patches.
>> [3,4] are the last 2 timeout patches, but with 1 var renamed.
>> [5] fixes a timeout problem related to batched CQ commits. From
>> what I see, this should be the last fixing timeouts.
>
> Something gone wrong with testing or rebasing. Never mind this.
io_uring-5.7 hangs the first test in link_timeout.c. I'll debug it today,
but by any chance, does anyone happen to know something?
>
>>
>> Pavel Begunkov (5):
>> io_uring: check non-sync defer_list carefully
>> io_uring: pass nxt from sync_file_range()
>> io_uring: trigger timeout after any sqe->off CQEs
>> io_uring: don't trigger timeout with another t-out
>> io_uring: fix timeout offset with batch CQ commit
>>
>> fs/io_uring.c | 130 +++++++++++++++++++++-----------------------------
>> 1 file changed, 54 insertions(+), 76 deletions(-)
>>
>
--
Pavel Begunkov
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 0/5] timeout fixes
2020-05-01 9:38 ` Pavel Begunkov
@ 2020-05-01 4:26 ` Jens Axboe
2020-05-01 13:55 ` Pavel Begunkov
0 siblings, 1 reply; 10+ messages in thread
From: Jens Axboe @ 2020-05-01 4:26 UTC (permalink / raw)
To: Pavel Begunkov, io-uring, linux-kernel
On 5/1/20 3:38 AM, Pavel Begunkov wrote:
> On 01/05/2020 11:21, Pavel Begunkov wrote:
>> On 30/04/2020 22:31, Pavel Begunkov wrote:
>>> [1,2] are small random patches.
>>> [3,4] are the last 2 timeout patches, but with 1 var renamed.
>>> [5] fixes a timeout problem related to batched CQ commits. From
>>> what I see, this should be the last fixing timeouts.
>>
>> Something gone wrong with testing or rebasing. Never mind this.
>
> io_uring-5.7 hangs the first test in link_timeout.c. I'll debug it today,
> but by any chance, does anyone happen to know something?
That's not your stuff, see:
https://lore.kernel.org/linux-fsdevel/[email protected]/T/#ma61d47f59eaaa7f04ae686c117fab69c957e0d7d
which then just turned into a modification to a patch in io_uring-5.7
instead. Just force rebase that branch and it should work fine.
--
Jens Axboe
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 0/5] timeout fixes
2020-05-01 4:26 ` Jens Axboe
@ 2020-05-01 13:55 ` Pavel Begunkov
0 siblings, 0 replies; 10+ messages in thread
From: Pavel Begunkov @ 2020-05-01 13:55 UTC (permalink / raw)
To: Jens Axboe, io-uring, linux-kernel
On 01/05/2020 07:26, Jens Axboe wrote:
> On 5/1/20 3:38 AM, Pavel Begunkov wrote:
>> On 01/05/2020 11:21, Pavel Begunkov wrote:
>>> On 30/04/2020 22:31, Pavel Begunkov wrote:
>>>> [1,2] are small random patches.
>>>> [3,4] are the last 2 timeout patches, but with 1 var renamed.
>>>> [5] fixes a timeout problem related to batched CQ commits. From
>>>> what I see, this should be the last fixing timeouts.
>>>
>>> Something gone wrong with testing or rebasing. Never mind this.
>>
>> io_uring-5.7 hangs the first test in link_timeout.c. I'll debug it today,
>> but by any chance, does anyone happen to know something?
>
Yeah, just found the culprit myself
> That's not your stuff, see:
>
> https://lore.kernel.org/linux-fsdevel/[email protected]/T/#ma61d47f59eaaa7f04ae686c117fab69c957e0d7d
>
> which then just turned into a modification to a patch in io_uring-5.7
> instead. Just force rebase that branch and it should work fine.
Got it, thanks
--
Pavel Begunkov
^ permalink raw reply [flat|nested] 10+ messages in thread