sctp: delay calls to sk_data_ready() as much as possible

It wastes space and gets worse as we add new flags, so convert bit-wide flags to a bitfield.

Currently it already saves 4 bytes in sctp_sock, which are left as holes in it for now. The whole struct needs packing, which should be done in another patch.

Note that do_auto_asconf cannot be merged, as explained in the comment before it.

Signed-off-by Marcelo Ricardo Leitner <[email protected]>

---
 include/net/sctp/structs.h | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

include/net/sctp/structs.h +6 -6

index 6df1ce7a411c548bda4163840a90578b6e1b4cfe..1a6a626904bba4223b7921bbb4be41c2550271a7 100644
@@ -210,14 +210,14 @@ struct sctp_sock {
 	int user_frag;
 
 	__u32 autoclose;
-	__u8 nodelay;
-	__u8 disable_fragments;
-	__u8 v4mapped;
-	__u8 frag_interleave;
 	__u32 adaptation_ind;
 	__u32 pd_point;
-	__u8 recvrcvinfo;
-	__u8 recvnxtinfo;
+	__u16	nodelay:1,
+		disable_fragments:1,
+		v4mapped:1,
+		frag_interleave:1,
+		recvrcvinfo:1,
+		recvnxtinfo:1;
 
 	atomic_t pd_mode;
 	/* Receive to here while partial delivery is in effect. */

Signature · 6 lines

2.5.0

--
To unsubscribe from this list: send the line "unsubscribe linux-sctp" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

--
To unsubscribe from this list: send the line "unsubscribe linux-sctp" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no guarantees at the application level regarding how long it takes to signal data readines from the time we get data off the wire, so I would rather see our throughput raised if we can, as thats been sctp's more pressing achilles heel.

Thats not to say I'd like to enable lower latency, but I'd rather have this now, and start pondering how to design that in. Perhaps we can convert the pending flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

Em 14-04-2016 10:03, Neil Horman escreveu:

Quoted 27 lines · David Miller

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently
stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no
guarantees at the application level regarding how long it takes to signal data
readines from the time we get data off the wire, so I would rather see our
throughput raised if we can, as thats been sctp's more pressing achilles heel.

Quoted 6 lines

Thats not to say I'd like to enable lower latency, but I'd rather have this now,
and start pondering how to design that in. Perhaps we can convert the pending
flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

That and also that there is no chance of the application reading the first chunks before all current ToDo's are performed by either the bh or backlog handlers for that packet. Socket lock won't be cycled in between chunks so the application is going to wait all the processing one way or another.

From: Marcelo Ricardo Leitner <[email protected]>
Date: Thu, 14 Apr 2016 14:00:49 -0300

Quoted 35 lines

Em 14-04-2016 10:03, Neil Horman escreveu:

Quoted 27 lines · David Miller

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently
stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no
guarantees at the application level regarding how long it takes to signal data
readines from the time we get data off the wire, so I would rather see our
throughput raised if we can, as thats been sctp's more pressing achilles heel.

Quoted 6 lines

Thats not to say I'd like to enable lower latency, but I'd rather have this now,
and start pondering how to design that in. Perhaps we can convert the pending
flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

Quoted 5 lines

That and also that there is no chance of the application reading the first chunks before all current ToDo's are performed by either the bh or backlog handlers for that packet. Socket lock won't be cycled in between chunks so the application is going to wait all the processing one way or another.

But it takes time to signal the wakeup to the remote cpu the process was running on, schedule out the current process on that cpu (if it has in fact lost it's timeslice), and then finally look at the socket queue.

Of course this is all assuming the process was sleeping in the first place, either in recv or more likely poll.

I really think signalling early helps performance.

From: Marcelo Ricardo Leitner <[email protected]>
Date: Thu, 14 Apr 2016 14:00:49 -0300

Quoted 35 lines

Em 14-04-2016 10:03, Neil Horman escreveu:

Quoted 27 lines · David Miller

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently
stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no
guarantees at the application level regarding how long it takes to signal data
readines from the time we get data off the wire, so I would rather see our
throughput raised if we can, as thats been sctp's more pressing achilles heel.

Quoted 6 lines

Thats not to say I'd like to enable lower latency, but I'd rather have this now,
and start pondering how to design that in. Perhaps we can convert the pending
flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

Quoted 5 lines

That and also that there is no chance of the application reading the first chunks before all current ToDo's are performed by either the bh or backlog handlers for that packet. Socket lock won't be cycled in between chunks so the application is going to wait all the processing one way or another.

But it takes time to signal the wakeup to the remote cpu the process was running on, schedule out the current process on that cpu (if it has in fact lost it's timeslice), and then finally look at the socket queue.

Of course this is all assuming the process was sleeping in the first place, either in recv or more likely poll.

I really think signalling early helps performance.

Quoted 49 lines · David Miller

From: Marcelo Ricardo Leitner <[email protected]>
Date: Thu, 14 Apr 2016 14:00:49 -0300

Quoted 35 lines

Em 14-04-2016 10:03, Neil Horman escreveu:

Quoted 27 lines · David Miller

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently
stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no
guarantees at the application level regarding how long it takes to signal data
readines from the time we get data off the wire, so I would rather see our
throughput raised if we can, as thats been sctp's more pressing achilles heel.

Quoted 6 lines

Thats not to say I'd like to enable lower latency, but I'd rather have this now,
and start pondering how to design that in. Perhaps we can convert the pending
flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

Quoted 5 lines

That and also that there is no chance of the application reading the first chunks before all current ToDo's are performed by either the bh or backlog handlers for that packet. Socket lock won't be cycled in between chunks so the application is going to wait all the processing one way or another.

But it takes time to signal the wakeup to the remote cpu the process was running on, schedule out the current process on that cpu (if it has in fact lost it's timeslice), and then finally look at the socket queue.

Quoted 5 lines

Of course this is all assuming the process was sleeping in the first place, either in recv or more likely poll.

I really think signalling early helps performance.

Early, yes, often, not so much :). Perhaps what would be adventageous would be to signal at the start of a set of enqueues, rather than at the end. That would be equivalent in terms of not signaling more than needed, but would eliminate
the signaling on every chunk. Perhaps what you could do Marcelo would be to
change the sense of the signal_ready flag to be a has_signaled flag. e.g. call sk_data_ready in ulp_event_tail like we used to, but only if the has_signaled flag isn't set, then set the flag, and clear it at the end of the command interpreter.

That would be a best of both worlds solution, as long as theres no chance of race with user space reading from the socket before we were done enqueuing (i.e. you have to guarantee that the socket lock stays held, which I think we do).

Quoted 65 lines · Neil Horman

Quoted 49 lines · David Miller

From: Marcelo Ricardo Leitner <[email protected]>
Date: Thu, 14 Apr 2016 14:00:49 -0300

Quoted 35 lines

Em 14-04-2016 10:03, Neil Horman escreveu:

Quoted 27 lines · David Miller

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently
stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no
guarantees at the application level regarding how long it takes to signal data
readines from the time we get data off the wire, so I would rather see our
throughput raised if we can, as thats been sctp's more pressing achilles heel.

Quoted 6 lines

Thats not to say I'd like to enable lower latency, but I'd rather have this now,
and start pondering how to design that in. Perhaps we can convert the pending
flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

Quoted 5 lines

That and also that there is no chance of the application reading the first chunks before all current ToDo's are performed by either the bh or backlog handlers for that packet. Socket lock won't be cycled in between chunks so the application is going to wait all the processing one way or another.

But it takes time to signal the wakeup to the remote cpu the process was running on, schedule out the current process on that cpu (if it has in fact lost it's timeslice), and then finally look at the socket queue.

Quoted 5 lines

Of course this is all assuming the process was sleeping in the first place, either in recv or more likely poll.

I really think signalling early helps performance.

Early, yes, often, not so much :). Perhaps what would be adventageous would be to signal at the start of a set of enqueues, rather than at the end. That would be equivalent in terms of not signaling more than needed, but would eliminate
the signaling on every chunk. Perhaps what you could do Marcelo would be to
change the sense of the signal_ready flag to be a has_signaled flag. e.g. call sk_data_ready in ulp_event_tail like we used to, but only if the has_signaled flag isn't set, then set the flag, and clear it at the end of the command interpreter.

Quoted 3 lines

That would be a best of both worlds solution, as long as theres no chance of race with user space reading from the socket before we were done enqueuing (i.e. you have to guarantee that the socket lock stays held, which I think we do).

That is my feeling too. Will work on it. Thanks :-)

Quoted 66 lines · On Thu, Apr 14, 2016 at 05:19:00PM -0300, [email protected] wrote:

Quoted 65 lines · Neil Horman

Quoted 49 lines · David Miller

From: Marcelo Ricardo Leitner <[email protected]>
Date: Thu, 14 Apr 2016 14:00:49 -0300

Quoted 35 lines

Em 14-04-2016 10:03, Neil Horman escreveu:

Quoted 27 lines · David Miller

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently
stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no
guarantees at the application level regarding how long it takes to signal data
readines from the time we get data off the wire, so I would rather see our
throughput raised if we can, as thats been sctp's more pressing achilles heel.

Quoted 6 lines

Thats not to say I'd like to enable lower latency, but I'd rather have this now,
and start pondering how to design that in. Perhaps we can convert the pending
flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

Quoted 5 lines

That and also that there is no chance of the application reading the first chunks before all current ToDo's are performed by either the bh or backlog handlers for that packet. Socket lock won't be cycled in between chunks so the application is going to wait all the processing one way or another.

But it takes time to signal the wakeup to the remote cpu the process was running on, schedule out the current process on that cpu (if it has in fact lost it's timeslice), and then finally look at the socket queue.

Quoted 5 lines

Of course this is all assuming the process was sleeping in the first place, either in recv or more likely poll.

I really think signalling early helps performance.

Early, yes, often, not so much :). Perhaps what would be adventageous would be to signal at the start of a set of enqueues, rather than at the end. That would be equivalent in terms of not signaling more than needed, but would eliminate
the signaling on every chunk. Perhaps what you could do Marcelo would be to
change the sense of the signal_ready flag to be a has_signaled flag. e.g. call sk_data_ready in ulp_event_tail like we used to, but only if the has_signaled flag isn't set, then set the flag, and clear it at the end of the command interpreter.

Quoted 5 lines

Quoted 3 lines

That would be a best of both worlds solution, as long as theres no chance of race with user space reading from the socket before we were done enqueuing (i.e. you have to guarantee that the socket lock stays held, which I think we do).

That is my feeling too. Will work on it. Thanks :-)

I did the change and tested it on real machines set all for performance. I couldn't spot any difference between both implementations.

Set RSS and queue irq affinity for a cpu and taskset netperf and another app I wrote to run on another cpu. It hits socket backlog quite often but still do direct processing every now and then.

With current state, netperf, scenario above. Results of perf sched record for the CPUs in use, reported by perf sched latency:

Task | Runtime ms | Switches | Average delay ms |
Maximum delay ms | Maximum delay at |
netserver:3205 | 9999.490 ms | 10 | avg: 0.003 ms |
max: 0.004 ms | max at: 69087.753356 s

another run
netserver:3483 | 9999.412 ms | 15 | avg: 0.003 ms |
max: 0.004 ms | max at: 69194.749814 s

With the patch below, same test:
netserver:2643 | 10000.110 ms | 14 | avg: 0.003 ms |
max: 0.004 ms | max at: 172.006315 s

another run:
netserver:2698 | 10000.049 ms | 15 | avg: 0.003 ms |
max: 0.004 ms | max at: 368.061672 s

I'll be happy to do more tests if you have any suggestions on how/what to test.

---8<---

Quoted 100 lines · On Thu, Apr 28, 2016 at 05:46:59PM -0300, [email protected] wrote:

Quoted 66 lines · On Thu, Apr 14, 2016 at 05:19:00PM -0300, [email protected] wrote:

Quoted 65 lines · Neil Horman

Quoted 49 lines · David Miller

From: Marcelo Ricardo Leitner <[email protected]>
Date: Thu, 14 Apr 2016 14:00:49 -0300

Quoted 35 lines

Em 14-04-2016 10:03, Neil Horman escreveu:

Quoted 27 lines · David Miller

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently
stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no
guarantees at the application level regarding how long it takes to signal data
readines from the time we get data off the wire, so I would rather see our
throughput raised if we can, as thats been sctp's more pressing achilles heel.

Quoted 6 lines

Thats not to say I'd like to enable lower latency, but I'd rather have this now,
and start pondering how to design that in. Perhaps we can convert the pending
flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

Quoted 5 lines

That and also that there is no chance of the application reading the first chunks before all current ToDo's are performed by either the bh or backlog handlers for that packet. Socket lock won't be cycled in between chunks so the application is going to wait all the processing one way or another.

But it takes time to signal the wakeup to the remote cpu the process was running on, schedule out the current process on that cpu (if it has in fact lost it's timeslice), and then finally look at the socket queue.

Quoted 5 lines

Of course this is all assuming the process was sleeping in the first place, either in recv or more likely poll.

I really think signalling early helps performance.

Early, yes, often, not so much :). Perhaps what would be adventageous would be to signal at the start of a set of enqueues, rather than at the end. That would be equivalent in terms of not signaling more than needed, but would eliminate
the signaling on every chunk. Perhaps what you could do Marcelo would be to
change the sense of the signal_ready flag to be a has_signaled flag. e.g. call sk_data_ready in ulp_event_tail like we used to, but only if the has_signaled flag isn't set, then set the flag, and clear it at the end of the command interpreter.

Quoted 5 lines

Quoted 3 lines

That would be a best of both worlds solution, as long as theres no chance of race with user space reading from the socket before we were done enqueuing (i.e. you have to guarantee that the socket lock stays held, which I think we do).

That is my feeling too. Will work on it. Thanks :-)

I did the change and tested it on real machines set all for performance. I couldn't spot any difference between both implementations.

Set RSS and queue irq affinity for a cpu and taskset netperf and another app I wrote to run on another cpu. It hits socket backlog quite often but still do direct processing every now and then.

With current state, netperf, scenario above. Results of perf sched record for the CPUs in use, reported by perf sched latency:

Task | Runtime ms | Switches | Average delay ms |
Maximum delay ms | Maximum delay at |
netserver:3205 | 9999.490 ms | 10 | avg: 0.003 ms |
max: 0.004 ms | max at: 69087.753356 s

another run
netserver:3483 | 9999.412 ms | 15 | avg: 0.003 ms |
max: 0.004 ms | max at: 69194.749814 s

With the patch below, same test:
netserver:2643 | 10000.110 ms | 14 | avg: 0.003 ms |
max: 0.004 ms | max at: 172.006315 s

another run:
netserver:2698 | 10000.049 ms | 15 | avg: 0.003 ms |
max: 0.004 ms | max at: 368.061672 s

Quoted 5 lines

I'll be happy to do more tests if you have any suggestions on how/what to test.

---8<---

I think this looks reasonable, but can you post it properly please, as a patch against the head of teh net-next tree, rather than a diff from your previous work (which wasn't comitted)

Quoted 101 lines · Neil Horman

Quoted 100 lines · On Thu, Apr 28, 2016 at 05:46:59PM -0300, [email protected] wrote:

Quoted 66 lines · On Thu, Apr 14, 2016 at 05:19:00PM -0300, [email protected] wrote:

Quoted 65 lines · Neil Horman

Quoted 49 lines · David Miller

From: Marcelo Ricardo Leitner <[email protected]>
Date: Thu, 14 Apr 2016 14:00:49 -0300

Quoted 35 lines

Em 14-04-2016 10:03, Neil Horman escreveu:

Quoted 27 lines · David Miller

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently
stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no
guarantees at the application level regarding how long it takes to signal data
readines from the time we get data off the wire, so I would rather see our
throughput raised if we can, as thats been sctp's more pressing achilles heel.

Quoted 6 lines

Thats not to say I'd like to enable lower latency, but I'd rather have this now,
and start pondering how to design that in. Perhaps we can convert the pending
flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

Quoted 5 lines

That and also that there is no chance of the application reading the first chunks before all current ToDo's are performed by either the bh or backlog handlers for that packet. Socket lock won't be cycled in between chunks so the application is going to wait all the processing one way or another.

But it takes time to signal the wakeup to the remote cpu the process was running on, schedule out the current process on that cpu (if it has in fact lost it's timeslice), and then finally look at the socket queue.

Quoted 5 lines

Of course this is all assuming the process was sleeping in the first place, either in recv or more likely poll.

I really think signalling early helps performance.

Early, yes, often, not so much :). Perhaps what would be adventageous would be to signal at the start of a set of enqueues, rather than at the end. That would be equivalent in terms of not signaling more than needed, but would eliminate
the signaling on every chunk. Perhaps what you could do Marcelo would be to
change the sense of the signal_ready flag to be a has_signaled flag. e.g. call sk_data_ready in ulp_event_tail like we used to, but only if the has_signaled flag isn't set, then set the flag, and clear it at the end of the command interpreter.

Quoted 5 lines

Quoted 3 lines

That would be a best of both worlds solution, as long as theres no chance of race with user space reading from the socket before we were done enqueuing (i.e. you have to guarantee that the socket lock stays held, which I think we do).

That is my feeling too. Will work on it. Thanks :-)

I did the change and tested it on real machines set all for performance. I couldn't spot any difference between both implementations.

Set RSS and queue irq affinity for a cpu and taskset netperf and another app I wrote to run on another cpu. It hits socket backlog quite often but still do direct processing every now and then.

With current state, netperf, scenario above. Results of perf sched record for the CPUs in use, reported by perf sched latency:

Task | Runtime ms | Switches | Average delay ms |
Maximum delay ms | Maximum delay at |
netserver:3205 | 9999.490 ms | 10 | avg: 0.003 ms |
max: 0.004 ms | max at: 69087.753356 s

another run
netserver:3483 | 9999.412 ms | 15 | avg: 0.003 ms |
max: 0.004 ms | max at: 69194.749814 s

With the patch below, same test:
netserver:2643 | 10000.110 ms | 14 | avg: 0.003 ms |
max: 0.004 ms | max at: 172.006315 s

another run:
netserver:2698 | 10000.049 ms | 15 | avg: 0.003 ms |
max: 0.004 ms | max at: 368.061672 s

Quoted 8 lines

Quoted 5 lines

I'll be happy to do more tests if you have any suggestions on how/what to test.

---8<---

I think this looks reasonable, but can you post it properly please, as a patch against the head of teh net-next tree, rather than a diff from your previous work (which wasn't comitted)

The idea was to not officially post it yet, more just as a reference, because I can't see any gains from it. I'm reluctant just due to that, no strong opinion here on one way or another.

If you think it's better anyway to signal it early, I'll properly repost it.

Thanks,
Marcelo

Quoted 115 lines · On Fri, Apr 29, 2016 at 10:47:25AM -0300, [email protected] wrote:

Quoted 101 lines · Neil Horman

Quoted 100 lines · On Thu, Apr 28, 2016 at 05:46:59PM -0300, [email protected] wrote:

Quoted 66 lines · On Thu, Apr 14, 2016 at 05:19:00PM -0300, [email protected] wrote:

Quoted 65 lines · Neil Horman

Quoted 49 lines · David Miller

From: Marcelo Ricardo Leitner <[email protected]>
Date: Thu, 14 Apr 2016 14:00:49 -0300

Quoted 35 lines

Em 14-04-2016 10:03, Neil Horman escreveu:

Quoted 27 lines · David Miller

Quoted 7 lines

From: Marcelo Ricardo Leitner <[email protected]>
Date: Fri, 8 Apr 2016 16:41:26 -0300

Quoted 4 lines

1st patch is a preparation for the 2nd. The idea is to not call ->sk_data_ready() for every data chunk processed while processing packets but only once before releasing the socket.

Quoted 6 lines

Quoted 2 lines

v2: patchset re-checked, small changelog fixes
v3: on patch 2, make use of local vars to make it more readable

Applied to net-next, but isn't this reduced overhead coming at the expense of latency? What if that lower latency is important to the application and/or consumer?

Thats a fair point, but I'd make the counter argument that, as it currently
stands, any latency introduced (or removed), is an artifact of our implementation rather than a designed feature of it. That is to say, we make no
guarantees at the application level regarding how long it takes to signal data
readines from the time we get data off the wire, so I would rather see our
throughput raised if we can, as thats been sctp's more pressing achilles heel.

Quoted 6 lines

Thats not to say I'd like to enable lower latency, but I'd rather have this now,
and start pondering how to design that in. Perhaps we can convert the pending
flag to a counter to count the number of events we enqueue, and call sk_data_ready every time we reach a sysctl defined threshold.

Quoted 5 lines

That and also that there is no chance of the application reading the first chunks before all current ToDo's are performed by either the bh or backlog handlers for that packet. Socket lock won't be cycled in between chunks so the application is going to wait all the processing one way or another.

But it takes time to signal the wakeup to the remote cpu the process was running on, schedule out the current process on that cpu (if it has in fact lost it's timeslice), and then finally look at the socket queue.

Quoted 5 lines

Of course this is all assuming the process was sleeping in the first place, either in recv or more likely poll.

I really think signalling early helps performance.

Early, yes, often, not so much :). Perhaps what would be adventageous would be to signal at the start of a set of enqueues, rather than at the end. That would be equivalent in terms of not signaling more than needed, but would eliminate
the signaling on every chunk. Perhaps what you could do Marcelo would be to
change the sense of the signal_ready flag to be a has_signaled flag. e.g. call sk_data_ready in ulp_event_tail like we used to, but only if the has_signaled flag isn't set, then set the flag, and clear it at the end of the command interpreter.

Quoted 5 lines

Quoted 3 lines

That would be a best of both worlds solution, as long as theres no chance of race with user space reading from the socket before we were done enqueuing (i.e. you have to guarantee that the socket lock stays held, which I think we do).

That is my feeling too. Will work on it. Thanks :-)

I did the change and tested it on real machines set all for performance. I couldn't spot any difference between both implementations.

Set RSS and queue irq affinity for a cpu and taskset netperf and another app I wrote to run on another cpu. It hits socket backlog quite often but still do direct processing every now and then.

With current state, netperf, scenario above. Results of perf sched record for the CPUs in use, reported by perf sched latency:

Task | Runtime ms | Switches | Average delay ms |
Maximum delay ms | Maximum delay at |
netserver:3205 | 9999.490 ms | 10 | avg: 0.003 ms |
max: 0.004 ms | max at: 69087.753356 s

another run
netserver:3483 | 9999.412 ms | 15 | avg: 0.003 ms |
max: 0.004 ms | max at: 69194.749814 s

With the patch below, same test:
netserver:2643 | 10000.110 ms | 14 | avg: 0.003 ms |
max: 0.004 ms | max at: 172.006315 s

another run:
netserver:2698 | 10000.049 ms | 15 | avg: 0.003 ms |
max: 0.004 ms | max at: 368.061672 s

Quoted 8 lines

Quoted 5 lines

I'll be happy to do more tests if you have any suggestions on how/what to test.

---8<---

I think this looks reasonable, but can you post it properly please, as a patch against the head of teh net-next tree, rather than a diff from your previous work (which wasn't comitted)

The idea was to not officially post it yet, more just as a reference, because I can't see any gains from it. I'm reluctant just due to that, no strong opinion here on one way or another.

Quoted 3 lines

If you think it's better anyway to signal it early, I'll properly repost it.

Yeah, your results seem to me to indicate that for your test at least, signaling early vs. late doesn't make alot of difference, but Dave I think made a point in principle in that allowing processes to wake up when we start enqueuing can be better in some situations. So all other things being equal, I'd say go with the method that you have here.