segment_queue today has its own standalone limit of MaxUnprocessedSegments but
this can be a problem in UnlockUser() we do not release the lock till there are
segments to be processed. What can happen is as handleSegments dequeues packets
more keep getting queued and we will never release the lock. This can keep
happening even if the receive buffer is full because nothing can read() till we
release the lock.
Further having a separate limit for pending segments makes it harder to track
memory usage etc. Unifying the limits makes it easier to reason about memory in
use and makes the overall buffer behaviour more consistent.
PiperOrigin-RevId: 333508122
RACK requires the segments to be in the order of their transmission
or retransmission times. This cl creates a new list and moves the
retransmitted segments to the end of the list.
PiperOrigin-RevId: 327325153
RACK (Recent Acknowledgement) is a new loss detection
algorithm in TCP. These are the fields which should be
stored on connections to implement RACK algorithm.
PiperOrigin-RevId: 324948703
TCP now tracks the overhead of the segment structure itself in it's out-of-order
queue (pending). This is required to ensure that a malicious sender sending 1
byte out-of-order segments cannot queue like 1000's of segments which bloat up
memory usage.
We also reduce the default receive window to 32KB. With TCP moderation there is
no need to keep this window at 1MB which means that for new connections the
default out-of-order queue will be small unless the application actually reads
the data that is being sent. This prevents a sender from just maliciously
filling up pending buf with lots of tiny out-of-order segments.
PiperOrigin-RevId: 323450913
Avoid a race where an arbitrary goroutine scheduling delay can cause the
processor to miss events and hang indefinitely.
Reduce allocations by storing processors by-value in the dispatcher, and
by using a single WaitGroup rather than one per processor.
PiperOrigin-RevId: 319665861
Connection tracking is used to track packets in prerouting and
output hooks of iptables. The NAT rules modify the tuples in
connections. The connection tracking code modifies the packets by
looking at the modified tuples.
This change adds a layer of abstraction around the internal Docker APIs,
and eliminates all direct dependencies on Dockerfiles in the infrastructure.
A subsequent change will automated the generation of local images (with
efficient caching). Note that this change drops the use of bazel container
rules, as that experiment does not seem to be viable.
PiperOrigin-RevId: 308095430
Tests now use a MinRTO of 3s instead of default 200ms. This reduced flakiness in
a lot of the congestion control/recovery tests which were flaky due to
retransmit timer firing too early in case the test executors were overloaded.
This change also bumps some of the timeouts in tests which were too sensitive to
timer variations and reduces the number of slow start iterations which can
make the tests run for too long and also trigger retansmit timeouts etc if
the executor is overloaded.
PiperOrigin-RevId: 306562645
This is a precursor to be being able to build an intrusive list
of PacketBuffers for use in queuing disciplines being implemented.
Updates #2214
PiperOrigin-RevId: 302677662
By putting slices into the pool, the slice header escapes. This can be avoided
by not putting the slice header into the pool.
This removes an allocation from the TCP segment send path.
PiperOrigin-RevId: 299215480
When sending a RST on shutdown we need to double check the
state after acquiring the work mutex as the endpoint could
have transitioned out of a connected state from the time
we checked it and we acquired the workMutex.
I added two tests but sadly neither reproduce the panic. I am
going to leave the tests in as they are good to have anyway.
PiperOrigin-RevId: 292393800
All inbound segments for connections in ESTABLISHED state are delivered to the
endpoint's queue but for every segment delivered we also queue the endpoint for
processing to a selected processor. This ensures that when there are a large
number of connections in ESTABLISHED state the inbound packets are all handled
by a small number of goroutines and significantly reduces the amount of work the
goscheduler has to perform.
We let connections in other states follow the current path where the
endpoint's goroutine directly handles the segments.
Updates #231
PiperOrigin-RevId: 289728325
* Rename syncutil to sync.
* Add aliases to sync types.
* Replace existing usage of standard library sync package.
This will make it easier to swap out synchronization primitives. For example,
this will allow us to use primitives from github.com/sasha-s/go-deadlock to
check for lock ordering violations.
Updates #1472
PiperOrigin-RevId: 289033387
The implementation follows the linux behavior where specifying
a TCP_USER_TIMEOUT will cause the resend timer to honor the
user specified timeout rather than the default rto based timeout.
Further it alters when connections are timedout due to keepalive
failures. It does not alter the behavior of when keepalives are
sent. This is as per the linux behavior.
PiperOrigin-RevId: 285099795
This change adds explicit support for honoring the 2MSL timeout
for sockets in TIME_WAIT state. It also adds support for the
TCP_LINGER2 option that allows modification of the FIN_WAIT2
state timeout duration for a given socket.
It also adds an option to modify the Stack wide TIME_WAIT timeout
but this is only for testing. On Linux this is fixed at 60s.
Further, we also now correctly process RST's in CLOSE_WAIT and
close the socket similar to linux without moving it to error
state.
We also now handle SYN in ESTABLISHED state as per
RFC5961#section-4.1. Earlier we would just drop these SYNs.
Which can result in some tests that pass on linux to fail on
gVisor.
Netstack now honors TIME_WAIT correctly as well as handles the
following cases correctly.
- TCP RSTs in TIME_WAIT are ignored.
- A duplicate TCP FIN during TIME_WAIT extends the TIME_WAIT
and a dup ACK is sent in response to the FIN as the dup FIN
indicates potential loss of the original final ACK.
- An out of order segment during TIME_WAIT generates a dup ACK.
- A new SYN w/ a sequence number > the highest sequence number
in the previous connection closes the TIME_WAIT early and
opens a new connection.
Further to make the SYN case work correctly the ISN (Initial
Sequence Number) generation for Netstack has been updated to
be as per RFC. Its not a pure random number anymore and follows
the recommendation in https://tools.ietf.org/html/rfc6528#page-3.
The current hash used is not a cryptographically secure hash
function. A separate change will update the hash function used
to Siphash similar to what is used in Linux.
PiperOrigin-RevId: 279106406
Right now, we send each tcp packet separately, we call one system
call per-packet. This patch allows to generate multiple tcp packets
and send them by sendmmsg.
The arguable part of this CL is a way how to handle multiple headers.
This CL adds the next field to the Prepandable buffer.
Nginx test results:
Server Software: nginx/1.15.9
Server Hostname: 10.138.0.2
Server Port: 8080
Document Path: /10m.txt
Document Length: 10485760 bytes
w/o gso:
Concurrency Level: 5
Time taken for tests: 5.491 seconds
Complete requests: 100
Failed requests: 0
Total transferred: 1048600200 bytes
HTML transferred: 1048576000 bytes
Requests per second: 18.21 [#/sec] (mean)
Time per request: 274.525 [ms] (mean)
Time per request: 54.905 [ms] (mean, across all concurrent requests)
Transfer rate: 186508.03 [Kbytes/sec] received
sw-gso:
Concurrency Level: 5
Time taken for tests: 3.852 seconds
Complete requests: 100
Failed requests: 0
Total transferred: 1048600200 bytes
HTML transferred: 1048576000 bytes
Requests per second: 25.96 [#/sec] (mean)
Time per request: 192.576 [ms] (mean)
Time per request: 38.515 [ms] (mean, across all concurrent requests)
Transfer rate: 265874.92 [Kbytes/sec] received
w/o gso:
$ ./tcp_benchmark --client --duration 15 --ideal
[SUM] 0.0-15.1 sec 2.20 GBytes 1.25 Gbits/sec
software gso:
$ tcp_benchmark --client --duration 15 --ideal --gso $((1<<16)) --swgso
[SUM] 0.0-15.1 sec 3.99 GBytes 2.26 Gbits/sec
PiperOrigin-RevId: 276112677
Netstack always picks a random start point everytime PickEphemeralPort
is called. While this is required for UDP so that DNS requests go
out through a randomized set of ports it is not required for TCP. Infact
Linux explicitly hashes the (srcip, dstip, dstport) and a one time secret
initialized at start of the application to get a random offset. But to
ensure it doesn't start from the same point on every scan it uses a static
hint that is incremented by 2 in every call to pick ephemeral ports.
The reason for 2 is Linux seems to split the port ranges where active connects
seem to use even ones while odd ones are used by listening sockets.
This CL implements a similar strategy where we use a hash + hint to generate
the offset to start the search for a free Ephemeral port.
This ensures that we cycle through the available port space in order for
repeated connects to the same destination and significantly reduces the
chance of picking a recently released port.
PiperOrigin-RevId: 272058370
This CL also cleans up the error returned for setting congestion
control which was incorrectly returning EINVAL instead of ENOENT.
PiperOrigin-RevId: 252889093
Nothing reads them and they can simply get stale.
Generated with:
$ sed -i "s/licenses(\(.*\)).*/licenses(\1)/" **/BUILD
PiperOrigin-RevId: 231818945
Change-Id: Ibc3f9838546b7e94f13f217060d31f4ada9d4bf0
Furthermore, allow for the specification of an ElementMapper. This allows a
single "Element" type to exist on multiple inline lists, and work without
having to embed the entry type.
This is a requisite change for supporting a per-Inode list of Dirents.
PiperOrigin-RevId: 211467497
Change-Id: If2768999b43e03fdaecf8ed15f435fe37518d163
This CL implements CUBIC as described in https://tools.ietf.org/html/rfc8312.
PiperOrigin-RevId: 207353142
Change-Id: I329cbf3277f91127e99e488f07d906f6779c6603
We have been unnecessarily creating too many savable types implicitly.
PiperOrigin-RevId: 206334201
Change-Id: Idc5a3a14bfb7ee125c4f2bb2b1c53164e46f29a8
This CL also puts the congestion control logic behind an
interface so that we can easily swap it out for say CUBIC
in the future.
PiperOrigin-RevId: 205732848
Change-Id: I891cdfd17d4d126b658b5faa0c6bd6083187944b
Bhasker Hariharan
Ghanan Gowripalan
Nayana Bidari
Ting-Yu Wang
Tamir Duberstein
Adin Scannell
Eyal Soha
Jay Zhuang
Ian Gudger
Andrei Vagin
Kevin Krakauer
Michael Pratt
Nicolas Lacasse
Fabricio Voznika
Zhaozhong Ni