Age | Commit message (Collapse) | Author |
|
When recovering from a zero-receive-window situation, and asked to
send out an ACK, ensure that we apply SWS avoidance in our window
updates.
Fixes #5984
PiperOrigin-RevId: 373689578
|
|
On receiving an ICMP error during handshake, the error is propagated
by reading `endpoint.lastError`. This can race with the socket layer
invoking getsockopt() with SO_ERROR where the same value is read and
cleared, causing the handshake to bail out with a non-error state.
Fix the race by checking for lastError state and failing the
handshake with ErrConnectionAborted if the lastError was read and
cleared by say SO_ERROR.
The race mentioned in the bug, is caught only with the newly added
tcp_test unit test, where we have control over stopping/resuming
protocol loop. Adding a packetimpact test as well for sanity testing
of ICMP error handling during handshake.
Fixes #5922
PiperOrigin-RevId: 372135662
|
|
PiperOrigin-RevId: 369505182
|
|
- Change the accept queue full condition for a listening endpoint
to only honor completed (and delivered) connections.
- Use syncookies if the number of incomplete connections is beyond
listen backlog. This also cleans up the SynThreshold option code
as that is no longer used with this change.
- Added a new stack option to unconditionally generate syncookies.
Similar to sysctl -w net.ipv4.tcp_syncookies=2 on Linux.
- Enable keeping of incomplete connections beyond listen backlog.
- Drop incoming SYNs only if the accept queue is filled up.
- Drop incoming ACKs that complete handshakes when accept queue is full
- Enable the stack to accept one more connection than programmed by
listen backlog.
- Handle backlog argument being zero, negative for listen, as Linux.
- Add syscall and packetimpact tests to reflect the changes above.
- Remove TCPConnectBacklog test which is polling for completed
connections on the client side which is not reflective of whether
the accept queue is filled up by the test. The modified syscall test
in this CL addresses testing of connecting sockets.
Fixes #3153
PiperOrigin-RevId: 366935921
|
|
Split usermem package to help remove syserror dependency in go_marshal.
New hostarch package contains code not dependent on syserror.
PiperOrigin-RevId: 365651233
|
|
The tests were not using the correct windowSize so the testing segments were
actually within the window for seqNumOffset=0 tests. The issue is already fixed
by #5674.
PiperOrigin-RevId: 364252630
|
|
TCP, in CLOSING state, MUST send an ACK with next expected SEQ number after
receiving any segment with OTW SEQ number and remain in the same state.
While I am here, I also changed shutdown to behave the same as other calls
in posix_server.
PiperOrigin-RevId: 362976955
|
|
The test queries for RTO via TCP_INFO and applies that to the
rest of the test. The RTO is estimated by processing incoming ACK.
There is a race in the test where we may query for RTO before the
incoming ACK was processed. Fix the race in the test by letting the
DUT complete a payload receive, thus estimating RTO before proceeding
to query the RTO. Bump up the time correction to reduce flakes.
PiperOrigin-RevId: 362865904
|
|
- Implement Stringer for it so that we can improve error messages.
- Use TCPFlags through the code base. There used to be a mixed usage of byte,
uint8 and int as TCP flags.
PiperOrigin-RevId: 361940150
|
|
Fix a race where the DUT could send out test data before it received the
peer window advertisement. Such a race results in the DUT taking longer
time to retransmit zero window probe, thus causing the test to fail
receiving the last expected probe.
To ensure this ordering, piggyback a non-zero payload with the zero
window advertisement and let the DUT receive that, before continuing
with the test.
PiperOrigin-RevId: 361640241
|
|
The syscall package has been deprecated in favor of golang.org/x/sys.
Note that syscall is still used in some places because the following don't seem
to have an equivalent in unix package:
- syscall.SysProcIDMap
- syscall.Credential
Updates #214
PiperOrigin-RevId: 361332034
|
|
One of the preparation to decouple underlying buffer implementation.
There are still some methods that tie to VectorisedView, and they will be
changed gradually in later CLs.
This CL also introduce a new ICMPv6ChecksumParams to replace long list of
parameters when calling ICMPv6Checksum, aiming to be more descriptive.
PiperOrigin-RevId: 360778149
|
|
- Removed (*testbench.Connection)(&conn) like casts
- Removed redundant definition of Drain, Close and ExpectFrame
PiperOrigin-RevId: 360727788
|
|
Without this change, the error produced is quite useless:
--- FAIL: TestZeroWindowProbeRetransmit (11.44s)
tcp_zero_window_probe_retransmit_test.go:81: expected a probe with sequence number 824638527212: loop 5
FAIL
PiperOrigin-RevId: 359796370
|
|
- open flags can be different on different OSs, by putting SetNonblocking into
the posix_server rather than the testbench, we can always get the right value
for O_NONBLOCK
- merged the tcp_queue_{send,receive}_in_syn_sent into a single file
PiperOrigin-RevId: 359620630
|
|
Remove unused argument while I'm here and avoid returning
syscall.Errno(0) which should rather be a nil error.
PiperOrigin-RevId: 358227396
|
|
Previously, we make two connect attempts. If the first attempt is still on
going when the second attempt is made, the test will fail. This change deflakes
the situation by not making the second attempt, instead, we poll for the first
attempt's completion and read the errno from SO_ERROR.
PiperOrigin-RevId: 358104769
|
|
- TCP_INFO is used to get the RTO instead of calculating it manually.
PiperOrigin-RevId: 358032487
|
|
- Wrap comments at 80 columns
- Avoid duplicating the number of retries (5)
- Reduce indentation
- Use (*testing.T).Fatal rather than (*testing.T).Fatalf
PiperOrigin-RevId: 358017412
|
|
Remove flaky tcp_reordering_test as it does not check reordering. We have
added new reorder tests in tcp_rack_test.go
PiperOrigin-RevId: 357278769
|
|
Deflake this test by retransmitting the ACK and retrying RST
expectation after the supposed state transition to CLOSED.
This gives time for the state transition to complete.
Without such a retransmit from the test, the ACK could get silently
dropped by the listener when the passively connecting endpoint
has not yet completely updated the state (in gVisor this would be
endpoint state and decrement of synRcvdCount).
PiperOrigin-RevId: 356825562
|
|
The test will sometimes fail on Bind calls using the old RPCTimeout.
PiperOrigin-RevId: 356646668
|
|
PiperOrigin-RevId: 355645297
|
|
Netstack today will send dupACK's with no rate limit for incoming out of
window segments. This can result in ACK loops for example if a TCP socket
connects to itself (actually permitted by TCP). Where the ACK sent in
response to packets being out of order itself gets considered as an out
of window segment resulting in another ACK being generated.
PiperOrigin-RevId: 355206877
|
|
PiperOrigin-RevId: 355067082
|
|
...to prevent panicking in case of failure.
PiperOrigin-RevId: 354970257
|
|
When these specific tests were first added, they would fail when ran by
kokoro during the presubmit tests (but they always passed locally).
These tests are now passing, so they can be re-enabled. Unclear what
changed, one hypothesis is the move from kokoro to buildkite.
Fixes #4971
PiperOrigin-RevId: 354425395
|
|
There used to be a race condition where we may call Close before the connection
is established. Adding poll support so that we can eliminate this kind of race.
Startblock:
has LGTM from iyerm
and then
add reviewer tamird
PiperOrigin-RevId: 354369130
|
|
Previously, sending on an unconnected UDP socket would ignore the
SO_BINDTODEVICE option. Send on the configured interface when an UDP socket
is bound to an interface through setsockop SO_BINDTODEVICE.
Add packetimpact tests exercising UDP reads and writes with every combination
of bound/unbound, broadcast/multicast/unicast destination, and bound/not-bound
to device.
PiperOrigin-RevId: 354299670
|
|
This CL adds support for the following fields:
- RTT, RTTVar, RTO
- send congestion window (sndCwnd) and send slow start threshold (sndSsthresh)
- congestion control state(CaState)
- ReorderSeen
PiperOrigin-RevId: 354195361
|
|
Fix the test to rely on more deterministic retransmission interval
computations by skipping the initial probe transmission time as that
can be non-deterministic given arbitrary time taken for the DUT to
receive a send command and initiate a send.
Fixes #5080
PiperOrigin-RevId: 354146256
|
|
- Added packetimpact tests for RACK.
PiperOrigin-RevId: 353282342
|
|
These are primarily simplification and lint mistakes. However, minor
fixes are also included and tests added where appropriate.
PiperOrigin-RevId: 351425971
|
|
When the scaled receive window size > 65535 (max uint16), we advertise
the scaled value as 65535, but are not adjusting the saved receive
window value when doing so. This would keep our current window
calculation logic to be incorrect, as the saved receive window value
is different from what was advertised.
Fixes #4903
PiperOrigin-RevId: 347771340
|
|
Without this change, this test produces quite opaque errors as valid TCP
segments are not logged; what we see in the end is any random traffic that
happened *after* the final TCP segment.
PiperOrigin-RevId: 346864169
|
|
PiperOrigin-RevId: 346852543
|
|
Recursive make is difficult to follow and debug. Drop this by using
internal functions, which, while difficult, are easier than trying to
following recursive invokations.
Further simplify the Makefile by collapsing the image bits and removing
the tools/vm directory, which is effectively unused.
Fixes #4952
PiperOrigin-RevId: 346569133
|
|
With the recent changes db36d948fa63ce950d94a5e8e9ebc37956543661, we try
to balance the receive window advertisements between payload lengths vs
segment overhead length. This works fine when segment size are much
higher than the overhead, but not otherwise. In cases where the segment
length is smaller than the segment overhead, we may end up not
advertising zero receive window for long time and end up tail-dropping
segments. This is especially pronounced when application socket reads
are slow or stopped. In this change we do not grow the right edge of
the receive window for smaller segment sizes similar to Linux.
Also, we keep track of the socket buffer usage and let the window grow
if the application is actively reading data.
Fixes #4903
PiperOrigin-RevId: 345832012
|
|
However, receiving duplicated fragments will not cause reassembly to
fail. This is what Linux does too:
https://github.com/torvalds/linux/blob/38525c6/net/ipv4/inet_fragment.c#L355
PiperOrigin-RevId: 345309546
|
|
To create DUTs in parallel, we need to create goroutines to do the setup. The
old code base has a lot of t.Fatal(f) usage in those setup functions which is
not great for this change: "FailNow must be called from the goroutine running
the test or benchmark function, not from other goroutines created during the
test" (https://golang.org/pkg/testing/#T.FailNow).
- Cleanup all t.Fatal(f) usage in DUT.Prepare()
- use goroutines to create DUTs in parallel
PiperOrigin-RevId: 344275809
|
|
Added a new flag num_duts to the test runner to create multiple DUTs for the
testbench can connect to.
PiperOrigin-RevId: 344195435
|
|
Summary of the approach: the test runner will set up a few DUTs according to
a flag and pass all the test networks to the testbench. The testbench will only
reside in a single container. The testbench will put all the test networks into
a buffered channel which served as a semaphore and now the user can freely use
t.Parallel() in (sub)tests and the true parallelism will be determined by how
many DUTs are configured. Creating DUTs on demand is not supported yet, the
test author should determine the number of DUTs to be used statically.
Specifically in this change:
- Don't export any global variables about the test network in testbench.
- Sniffer only binds on the local interface because it will be possible to have
multiple interfaces to multiple DUTs in a single testbench container.
- Migrate existing tests to stop using global variables.
PiperOrigin-RevId: 343965962
|
|
Updates #4427
PiperOrigin-RevId: 342703931
|
|
Teach ICMP.Parser/ToBytes to handle some non echo ICMP packets.
Teach IPv4.Parser that fragments only have a payload, not an upper layer.
Fix IPv4 and IPv6 reassembly tests to handle the change.
Fixes #4758
PiperOrigin-RevId: 341549665
|
|
The IPv6 reassembly test was also refactored to be easily extended with
more cases.
PiperOrigin-RevId: 339768605
|
|
This makes handling inbound fragmented packets easier, because a fragmented
packet might not have an actual ICMP header but only a payload. After this
change, the ICMPv4 is the last layer you can get because the payload is
embedded in it.
Note that this makes it consistent with the ICMPv6 implementation.
While I'm here, I've also added the Ident and Sequence fields on the ICMPv4
type. Defaults are still zero.
PiperOrigin-RevId: 339577094
|
|
By exposing an ALL_TESTS list in defs.bzl we can make sure all packetimpact
users get to agree on the list of all tests. A defect in this approach is that
we have to keep a list of packetimpact_testbench rules in the BUILD file. An
helper validate_all_tests has been added to help keep BUILD and .bzl files in
sync.
PiperOrigin-RevId: 337411839
|
|
PiperOrigin-RevId: 336350318
|
|
Most of the IPv4 fragmentation code was moved in the fragmentation
package and it is reused by IPv6 fragmentation.
Test:
- pkg/tcpip/network/ipv4:ipv4_test
- pkg/tcpip/network/ipv6:ipv6_test
- pkg/tcpip/network/fragmentation:fragmentation_test
Fixes #4389
PiperOrigin-RevId: 335714280
|
|
When the socket is set with SO_LINGER and close()'d in the initial state, it
should not linger and return immediately.
PiperOrigin-RevId: 334263149
|