Previously, gVisor did not represent loopback devices as an ethernet
device as Linux does. To maintain Linux API compatibility for packet
sockets, a workaround was used to add an ethernet header if a link
header was not already present in the packet buffer delivered to a
packet endpoint.
However, this workaround is a bug for non-ethernet based interfaces; not
all links use an ethernet header (e.g. pure L3/TUN interfaces).
As of 3b4bb94751, gVisor represents
loopback devices as an ethernet-based device so this workaround can
now be removed.
BUG: https://fxbug.dev/81592
Updates #6530, #6531.
PiperOrigin-RevId: 395819151
This change removes NetworkDispatcher.DeliverOutboundPacket.
Since all packet writes go through the NIC (the only NetworkDispatcher),
we can deliver outgoing packets to interested packet endpoints before
writing the packet to the link endpoint as the stack expects that all
packets that get delivered to a link endpoint are transmitted on the
wire. That is, link endpoints no longer need to let the stack know when
it writes a packet as the stack already knows about the packet it writes
through a link endpoint.
PiperOrigin-RevId: 395761629
NewTempAbsPathInDir("/tmp") prevents the generated socket address from
exceeding sizeof(addr.sun_path). However, existing systems that are built with
the ANDROID configuration have their temp directory in a different location.
This change allows those systems to run tests that depend on UniqueUnixAddr.
PiperOrigin-RevId: 395336483
This defines common infrastructure for dynamically-configured security checks,
including an example usage in the clone(2) path.
PiperOrigin-RevId: 394797270
If we want to take two endpoint locks, we need to be sure that we always
take them in the same order.
Accept() locks the listening endpoint to work with acceptedChan and then
it calls GetLocalAddress that locks an accepted endpoint. Actually, we
can release the listening endpoint lock before calling GetLocalAddress.
Reported-by: syzbot+f52bd603f51a4ae91054@syzkaller.appspotmail.com
PiperOrigin-RevId: 394553823
...through the loopback interface, only.
This change only supports sending on packet sockets through the loopback
interface as the loopback interface is the only interface used in packet
socket syscall tests - the other link endpoints are not excercised with
the existing test infrastructure.
Support for sending on packet sockets through the other interfaces will
be added as needed.
BUG: https://fxbug.dev/81592
PiperOrigin-RevId: 394368899
For a small receive buffer the first out-of-order segment will get accepted and
fill up the receive buffer today. This change now includes the size of the
out-of-order segment when checking whether to queue the out of order segment or
not.
PiperOrigin-RevId: 394351309
...from the UDP endpoint.
Datagram-based transport endpoints (e.g. UDP, RAW IP) can share a lot
of their write path due to the datagram-based nature of these endpoints.
Extract the common facilities from UDP so they can be shared with other
transport endpoints (in a later change).
Test: UDP syscall tests.
PiperOrigin-RevId: 394347774
reflect.ValueOf takes an interface{}, so when passed a slice the compiler emits
a call to runtime.convTslice to heap-allocate a copy of the slice header.
PiperOrigin-RevId: 394310052
Fix syzcaller panic SIGBUS on error handling. Done by
adding an interface, errors.GuestError, which errors can
implement in order to be compared against each other.
PiperOrigin-RevId: 393867554
Remove freestanding functions that convert time values to raw integers;
centralize time->uint32 logic in methods on tcp.endpoint. Importantly,
the knowledge that TSVal is in milliseconds now lives in adjacent
functions rather than being spread around various files.
Incidental cleanup:
- Remove unused constant
- Remove redundant conversion
- Remove redundant parentheses
- Add missing error check
PiperOrigin-RevId: 393184768
Add logs to big query for benchmarks. This is to debug the actual
sent data from BQ so it can be compared to the data produced by
the benchmark run.
Current logs end with "Sending benchmarks", which is ambiguous.
Log success to make sure they are sent.
PiperOrigin-RevId: 392944104
Some tcp unit tests are affected by this change:
- Some retransmission tests assumed RTO=1s when connection is established. This
is no longer true because minRTO was set to 3s in tests so now RTO becomes 3s
after the first updateRTO call. Set minRTO=1s for these tests.
- Some RACK enabled tests are affected because now that RTT is initialized, and
the estimated RTT is quite small, spurious TLP might be sent out and causing
flakes, introduce an artificial delay for these tests so that the estimated
RTT is larger.
PiperOrigin-RevId: 392768725