Skip to contentSkip to conversation

Add a sysctl to allow TCP window shrinking in order to honor memory limits

PATCH In review Mike Freemon 5 participants 9 messages Jun 5, 2023 – Jun 6, 2023

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Add a sysctl to allow TCP window shrinking in order to honor memory limits

From: "[email protected]" <[email protected]>

Under certain circumstances, the tcp receive buffer memory limit set by autotuning is ignored, and the receive buffer can grow unrestrained until it reaches tcp_rmem[2].

To reproduce: Connect a TCP session with the receiver doing nothing and the sender sending small packets (an infinite loop of socket send() with 4 bytes of payload with a sleep of 1 ms in between each send()). This will fill the tcp receive buffer all the way to tcp_rmem[2], ignoring the autotuning limit (sk_rcvbuf).

As a result, a host can have individual tcp sessions with receive buffers of size tcp_rmem[2], and the host itself can reach tcp_mem limits, causing the host to go into tcp memory pressure mode.

The fundamental issue is the relationship between the granularity of the window scaling factor and the number of byte ACKed back to the sender. This problem has previously been identified in RFC 7323, appendix F [1].

The Linux kernel currently adheres to never shrinking the window.

In addition to the overallocation of memory mentioned above, this is also functionally incorrect, because once tcp_rmem[2] is reached, the receiver will drop in-window packets resulting in retransmissions and an eventual timeout of the tcp session. A receive buffer full condition should instead result in a zero window and an indefinite wait.

In practice, this problem is largely hidden for most flows. It is not applicable to mice flows. Elephant flows can send data fast enough to "overrun" the sk_rcvbuf limit (in a single ACK), triggering a zero window.

But this problem does show up for other types of flows. A good example are websockets and other type of flows that send small amounts of data spaced apart slightly in time. In these cases, we directly encounter the problem described in [1].

RFC 7323, section 2.4 [2], says there are instances when a retracted window can be offered, and that TCP implementations MUST ensure that they handle a shrinking window, as specified in RFC 1122, section 4.2.2.16 [3]. All prior RFCs on the topic of tcp window management have made clear that sender must accept a shrunk window from the receiver, including RFC 793 [4] and RFC 1323 [5].

This patch implements the functionality to shrink the tcp window when necessary to keep the right edge within the memory limit by autotuning (sk_rcvbuf). This new functionality is enabled with the following sysctl:

sysctl: net.ipv4.tcp_shrink_window

This sysctl changes how the TCP window is calculated.

If sysctl tcp_shrink_window is zero (the default value), then the window is never shrunk.

If sysctl tcp_shrink_window is non-zero, then the memory limit set by autotuning is honored. This requires that the TCP window be shrunk ("retracted") as described in RFC 1122.

[1] https://www.rfc-editor.org/rfc/rfc7323#appendix-F
[2] https://www.rfc-editor.org/rfc/rfc7323#section-2.4
[3] https://www.rfc-editor.org/rfc/rfc1122#page-91
[4] https://www.rfc-editor.org/rfc/rfc793
[5] https://www.rfc-editor.org/rfc/rfc1323

---
 Documentation/networking/ip-sysctl.rst | 14 ++++++
 include/net/netns/ipv4.h               |  1 +
 net/ipv4/sysctl_net_ipv4.c             |  9 ++++
 net/ipv4/tcp_ipv4.c                    |  2 +
 net/ipv4/tcp_output.c                  | 59 +++++++++++++++++++-------
 5 files changed, 70 insertions(+), 15 deletions(-)
Documentation/networking/ip-sysctl.rst +14 -0
index 6ec06a33688a..6c713d3a97e5 100644
@@ -970,6 +970,20 @@ tcp_tw_reuse - INTEGER
 tcp_window_scaling - BOOLEAN
 	Enable window scaling as defined in RFC1323.
 
+tcp_shrink_window - BOOLEAN
+	This changes how the TCP receive window is calculated when window
+	scaling is in effect.
+
+	RFC 7323, section 2.4, says there are instances when a retracted
+	window can be offered, and that TCP implementations MUST ensure
+	that they handle a shrinking window, as specified in RFC 1122.
+
+	- 0 - Disabled.	The window is never shrunk.
+	- 1 - Enabled.	The window is shrunk when necessary to remain within
+			the memory limit set by autotuning (sk_rcvbuf).
+
+	Default: 0
+
 tcp_wmem - vector of 3 INTEGERs: min, default, max
 	min: Amount of memory reserved for send buffers for TCP sockets.
 	Each TCP socket has rights to use it due to fact of its birth.
include/net/netns/ipv4.h +1 -0
index db762e35aca9..e30442bdc9c0 100644
@@ -65,6 +65,7 @@ struct netns_ipv4 {
 #endif
 	bool			fib_has_custom_local_routes;
 	bool			fib_offload_disabled;
+	u8			sysctl_tcp_shrink_window;
 #ifdef CONFIG_IP_ROUTE_CLASSID
 	atomic_t		fib_num_tclassid_users;
 #endif
net/ipv4/sysctl_net_ipv4.c +9 -0
index 40fe70fc2015..12470660744d 100644
@@ -1470,6 +1470,15 @@ static struct ctl_table ipv4_net_table[] = {
 		.extra1         = SYSCTL_ZERO,
 		.extra2         = &tcp_plb_max_cong_thresh,
 	},
+	{
+		.procname	= "tcp_shrink_window",
+		.data		= &init_net.ipv4.sysctl_tcp_shrink_window,
+		.maxlen		= sizeof(u8),
+		.mode		= 0644,
+		.proc_handler	= proc_dou8vec_minmax,
+		.extra1		= SYSCTL_ZERO,
+		.extra2		= SYSCTL_ONE,
+	},
 	{ }
 };
 
net/ipv4/tcp_ipv4.c +2 -0
index 06d2573685ca..ed0ce4b4b96d 100644
@@ -3276,6 +3276,8 @@ static int __net_init tcp_sk_init(struct net *net)
 	else
 		net->ipv4.tcp_congestion_control = &tcp_reno;
 
+	net->ipv4.sysctl_tcp_shrink_window = 0;
+
 	return 0;
 }
 
net/ipv4/tcp_output.c +44 -15

106 lines not on this page · raw message

Signature · 1 line
2.40.0
#1 lore raw
Quoted 64 lines · On Mon, 5 Jun 2023 15:38:57 -0500 Mike Freemon <[email protected]> wrote:

From: "[email protected]" <[email protected]>

Under certain circumstances, the tcp receive buffer memory limit set by autotuning is ignored, and the receive buffer can grow unrestrained until it reaches tcp_rmem[2].

To reproduce: Connect a TCP session with the receiver doing nothing and the sender sending small packets (an infinite loop of socket send() with 4 bytes of payload with a sleep of 1 ms in between each send()). This will fill the tcp receive buffer all the way to tcp_rmem[2], ignoring the autotuning limit (sk_rcvbuf).

As a result, a host can have individual tcp sessions with receive buffers of size tcp_rmem[2], and the host itself can reach tcp_mem limits, causing the host to go into tcp memory pressure mode.

The fundamental issue is the relationship between the granularity of the window scaling factor and the number of byte ACKed back to the sender. This problem has previously been identified in RFC 7323, appendix F [1].

The Linux kernel currently adheres to never shrinking the window.

In addition to the overallocation of memory mentioned above, this is also functionally incorrect, because once tcp_rmem[2] is reached, the receiver will drop in-window packets resulting in retransmissions and an eventual timeout of the tcp session. A receive buffer full condition should instead result in a zero window and an indefinite wait.

In practice, this problem is largely hidden for most flows. It is not applicable to mice flows. Elephant flows can send data fast enough to "overrun" the sk_rcvbuf limit (in a single ACK), triggering a zero window.

But this problem does show up for other types of flows. A good example are websockets and other type of flows that send small amounts of data spaced apart slightly in time. In these cases, we directly encounter the problem described in [1].

RFC 7323, section 2.4 [2], says there are instances when a retracted window can be offered, and that TCP implementations MUST ensure that they handle a shrinking window, as specified in RFC 1122, section 4.2.2.16 [3]. All prior RFCs on the topic of tcp window management have made clear that sender must accept a shrunk window from the receiver, including RFC 793 [4] and RFC 1323 [5].

This patch implements the functionality to shrink the tcp window when necessary to keep the right edge within the memory limit by autotuning (sk_rcvbuf). This new functionality is enabled with the following sysctl:

sysctl: net.ipv4.tcp_shrink_window

This sysctl changes how the TCP window is calculated.

If sysctl tcp_shrink_window is zero (the default value), then the window is never shrunk.

If sysctl tcp_shrink_window is non-zero, then the memory limit set by autotuning is honored. This requires that the TCP window be shrunk ("retracted") as described in RFC 1122.

Quoted 7 lines

[1] https://www.rfc-editor.org/rfc/rfc7323#appendix-F
[2] https://www.rfc-editor.org/rfc/rfc7323#section-2.4
[3] https://www.rfc-editor.org/rfc/rfc1122#page-91
[4] https://www.rfc-editor.org/rfc/rfc793
[5] https://www.rfc-editor.org/rfc/rfc1323

Does Linux TCP really need another tuning parameter? Will tests get run with both feature on and off? What default will distributions ship with?

Sounds like unbounded receive window growth is always a bad idea and a latent bug.

#2 lore raw
Quoted 17 lines · On Mon, 5 Jun 2023 15:42:29 -0700 Stephen Hemminger…
Quoted 17 lines

sysctl: net.ipv4.tcp_shrink_window

This sysctl changes how the TCP window is calculated.

If sysctl tcp_shrink_window is zero (the default value), then the window is never shrunk.

If sysctl tcp_shrink_window is non-zero, then the memory limit set by autotuning is honored. This requires that the TCP window be shrunk ("retracted") as described in RFC 1122.

[1] https://www.rfc-editor.org/rfc/rfc7323#appendix-F
[2] https://www.rfc-editor.org/rfc/rfc7323#section-2.4
[3] https://www.rfc-editor.org/rfc/rfc1122#page-91
[4] https://www.rfc-editor.org/rfc/rfc793
[5] https://www.rfc-editor.org/rfc/rfc1323

Quoted 8 lines
Quoted 1 line

Does Linux TCP really need another tuning parameter? Will tests get run with both feature on and off? What default will distributions ship with?

Sounds like unbounded receive window growth is always a bad idea and a latent bug.

FYI - I worked in an environment where every bug fix had to have a tuning parameter to turn it off. It was a bad idea, driven by management problems with updating. The number of knobs lead to confusion and geometric growth in possible code paths.

#3 lore raw
Quoted 27 lines · Stephen Hemminger
Quoted 23 lines · On Mon, 5 Jun 2023 15:42:29 -0700 Stephen Hemminger…
Quoted 11 lines

sysctl: net.ipv4.tcp_shrink_window

This sysctl changes how the TCP window is calculated.

If sysctl tcp_shrink_window is zero (the default value), then the window is never shrunk.

If sysctl tcp_shrink_window is non-zero, then the memory limit set by autotuning is honored. This requires that the TCP window be shrunk ("retracted") as described in RFC 1122.

Quoted 7 lines

[1] https://www.rfc-editor.org/rfc/rfc7323#appendix-F
[2] https://www.rfc-editor.org/rfc/rfc7323#section-2.4
[3] https://www.rfc-editor.org/rfc/rfc1122#page-91
[4] https://www.rfc-editor.org/rfc/rfc793
[5] https://www.rfc-editor.org/rfc/rfc1323

Does Linux TCP really need another tuning parameter? Will tests get run with both feature on and off? What default will distributions ship with?

Quoted 8 lines
Quoted 2 lines

Sounds like unbounded receive window growth is always a bad idea and a latent bug.

FYI - I worked in an environment where every bug fix had to have a tuning parameter to turn it off. It was a bad idea, driven by management problems with updating. The number of knobs lead to confusion and geometric growth in possible code paths.

I agree. More than this, shrinking window prohibited in those classic RFCs could cause unexpected/unwanted behaviour.

CC Eric and Neal

Thanks,
Jason

#4 lore raw
Quoted 32 lines · Jason Xing
Quoted 30 lines · Stephen Hemminger
Quoted 25 lines · On Mon, 5 Jun 2023 15:42:29 -0700 Stephen Hemminger…
Quoted 11 lines

sysctl: net.ipv4.tcp_shrink_window

This sysctl changes how the TCP window is calculated.

If sysctl tcp_shrink_window is zero (the default value), then the window is never shrunk.

If sysctl tcp_shrink_window is non-zero, then the memory limit set by autotuning is honored. This requires that the TCP window be shrunk ("retracted") as described in RFC 1122.

Quoted 7 lines

[1] https://www.rfc-editor.org/rfc/rfc7323#appendix-F
[2] https://www.rfc-editor.org/rfc/rfc7323#section-2.4
[3] https://www.rfc-editor.org/rfc/rfc1122#page-91
[4] https://www.rfc-editor.org/rfc/rfc793
[5] https://www.rfc-editor.org/rfc/rfc1323

Does Linux TCP really need another tuning parameter? Will tests get run with both feature on and off? What default will distributions ship with?

Sounds like unbounded receive window growth is always a bad idea and a latent bug.

Quoted 8 lines
Quoted 5 lines

FYI - I worked in an environment where every bug fix had to have a tuning parameter to turn it off. It was a bad idea, driven by management problems with updating. The number of knobs lead to confusion and geometric growth in possible code paths.

I agree. More than this, shrinking window prohibited in those classic RFCs could cause unexpected/unwanted behaviour.

I discuss the RFCs in more detail in my blog post here:
https://blog.cloudflare.com/unbounded-memory-usage-by-tcp-for-receive-buffers-and-how-we-fixed-it/

#5 lore raw
Quoted 41 lines · Mike Freemon
Quoted 38 lines · Jason Xing
Quoted 35 lines · Stephen Hemminger
Quoted 17 lines · On Mon, 5 Jun 2023 15:42:29 -0700 Stephen Hemminger…
Quoted 17 lines

sysctl: net.ipv4.tcp_shrink_window

This sysctl changes how the TCP window is calculated.

If sysctl tcp_shrink_window is zero (the default value), then the window is never shrunk.

If sysctl tcp_shrink_window is non-zero, then the memory limit set by autotuning is honored. This requires that the TCP window be shrunk ("retracted") as described in RFC 1122.

[1] https://www.rfc-editor.org/rfc/rfc7323#appendix-F
[2] https://www.rfc-editor.org/rfc/rfc7323#section-2.4
[3] https://www.rfc-editor.org/rfc/rfc1122#page-91
[4] https://www.rfc-editor.org/rfc/rfc793
[5] https://www.rfc-editor.org/rfc/rfc1323

Quoted 8 lines
Quoted 1 line

Does Linux TCP really need another tuning parameter? Will tests get run with both feature on and off? What default will distributions ship with?

Sounds like unbounded receive window growth is always a bad idea and a latent bug.

FYI - I worked in an environment where every bug fix had to have a tuning parameter to turn it off. It was a bad idea, driven by management problems with updating. The number of knobs lead to confusion and geometric growth in possible code paths.

Quoted 5 lines
Quoted 2 lines

I agree. More than this, shrinking window prohibited in those classic RFCs could cause unexpected/unwanted behaviour.

I discuss the RFCs in more detail in my blog post here:
https://blog.cloudflare.com/unbounded-memory-usage-by-tcp-for-receive-buffers-and-how-we-fixed-it/

Mike, the usual process to push linux patches is to make them self contained.

changelog should be enough, no need to read a lengthy blog post.

Also, I did not receive a copy of the patch, which is unfortunate, given I am the linux TCP maintainer.

Next time you post it, make sure to CC me.

Thanks.

maintainer
#6 lore raw
Quoted 2 lines · Eric Dumazet

On Tue, Jun 6, 2023 at 5:17 PM Mike Freemon <[email protected]> wrote:

...

Quoted 5 lines
Quoted 2 lines

I discuss the RFCs in more detail in my blog post here:
https://blog.cloudflare.com/unbounded-memory-usage-by-tcp-for-receive-buffers-and-how-we-fixed-it/

Mike, the usual process to push linux patches is to make them self contained.

Quoted 8 lines

changelog should be enough, no need to read a lengthy blog post.

Also, I did not receive a copy of the patch, which is unfortunate, given I am the linux TCP maintainer.

Next time you post it, make sure to CC me.

Thanks.

Also, before the repost, can you please rebase on to the latest net-next branch? The patch does not currently apply to the latest net-next tree. Thanks!

neal

#7 lore raw
Quoted 14 lines · Neal Cardwell
Quoted 2 lines · Eric Dumazet

On Tue, Jun 6, 2023 at 5:17 PM Mike Freemon <[email protected]> wrote:

...

Quoted 10 lines
Quoted 2 lines

I discuss the RFCs in more detail in my blog post here:
https://blog.cloudflare.com/unbounded-memory-usage-by-tcp-for-receive-buffers-and-how-we-fixed-it/

Mike, the usual process to push linux patches is to make them self contained.

changelog should be enough, no need to read a lengthy blog post.

Also, I did not receive a copy of the patch, which is unfortunate, given I am the linux TCP maintainer.

Quoted 7 lines
Quoted 3 lines

Next time you post it, make sure to CC me.

Thanks.

Also, before the repost, can you please rebase on to the latest net-next branch? The patch does not currently apply to the latest net-next tree. Thanks!

I just posted "v2", which applies successfully on net-next.

#8 lore raw
In reply to Stephen Hemminger
Quoted 73 lines · Stephen Hemminger
Quoted 70 lines · On Mon, 5 Jun 2023 15:38:57 -0500 Mike Freemon <[email protected]> wrote:

From: "[email protected]" <[email protected]>

Under certain circumstances, the tcp receive buffer memory limit set by autotuning is ignored, and the receive buffer can grow unrestrained until it reaches tcp_rmem[2].

To reproduce: Connect a TCP session with the receiver doing nothing and the sender sending small packets (an infinite loop of socket send() with 4 bytes of payload with a sleep of 1 ms in between each send()). This will fill the tcp receive buffer all the way to tcp_rmem[2], ignoring the autotuning limit (sk_rcvbuf).

As a result, a host can have individual tcp sessions with receive buffers of size tcp_rmem[2], and the host itself can reach tcp_mem limits, causing the host to go into tcp memory pressure mode.

The fundamental issue is the relationship between the granularity of the window scaling factor and the number of byte ACKed back to the sender. This problem has previously been identified in RFC 7323, appendix F [1].

The Linux kernel currently adheres to never shrinking the window.

In addition to the overallocation of memory mentioned above, this is also functionally incorrect, because once tcp_rmem[2] is reached, the receiver will drop in-window packets resulting in retransmissions and an eventual timeout of the tcp session. A receive buffer full condition should instead result in a zero window and an indefinite wait.

In practice, this problem is largely hidden for most flows. It is not applicable to mice flows. Elephant flows can send data fast enough to "overrun" the sk_rcvbuf limit (in a single ACK), triggering a zero window.

But this problem does show up for other types of flows. A good example are websockets and other type of flows that send small amounts of data spaced apart slightly in time. In these cases, we directly encounter the problem described in [1].

RFC 7323, section 2.4 [2], says there are instances when a retracted window can be offered, and that TCP implementations MUST ensure that they handle a shrinking window, as specified in RFC 1122, section 4.2.2.16 [3]. All prior RFCs on the topic of tcp window management have made clear that sender must accept a shrunk window from the receiver, including RFC 793 [4] and RFC 1323 [5].

This patch implements the functionality to shrink the tcp window when necessary to keep the right edge within the memory limit by autotuning (sk_rcvbuf). This new functionality is enabled with the following sysctl:

sysctl: net.ipv4.tcp_shrink_window

This sysctl changes how the TCP window is calculated.

If sysctl tcp_shrink_window is zero (the default value), then the window is never shrunk.

If sysctl tcp_shrink_window is non-zero, then the memory limit set by autotuning is honored. This requires that the TCP window be shrunk ("retracted") as described in RFC 1122.

[1] https://www.rfc-editor.org/rfc/rfc7323#appendix-F
[2] https://www.rfc-editor.org/rfc/rfc7323#section-2.4
[3] https://www.rfc-editor.org/rfc/rfc1122#page-91
[4] https://www.rfc-editor.org/rfc/rfc793
[5] https://www.rfc-editor.org/rfc/rfc1323

Quoted 3 lines
Quoted 1 line

Does Linux TCP really need another tuning parameter?

It's useful to make testing faster, i.e. comparing enabled vs disabled. It could also be useful as a quick diagnostic test, i.e. someone is having a problem and they want to quickly eliminate this patch as a cause.

But I left it in mainly as a risk response. This patch requires that the receiving TCP implementation handle the shrinking window correctly. This patch has been deployed at Cloudflare and we have not discovered any cases where the peer TCP fails to be RFC compliant. But we cannot rule out the possibility completely. The concern is what if someone is running some old software on a non-public network and their software does not handle a shrinking window. Simply disabling this feature via a sysctl parameter seems like a good solution for that situation.

If the consensus is to not have a sysctl parameter, I am happy to remove it.

A related question: If we leave it in, what do we think the default value should be? It's disabled by default right now, but that is just me being conservative. If we are comfortable enabling this by default, I'm happy to do that too.

Quoted 1 line

Will tests get run with both feature on and off?

More background and details about the patch is here, including the test results you're looking for:
https://blog.cloudflare.com/unbounded-memory-usage-by-tcp-for-receive-buffers-and-how-we-fixed-it/

Quoted 1 line

What default will distributions ship with?

I'm not sure how to answer this. Isn't that up to the distributions to decide?