* wip: geth connection test
* set port until disabling discovery sets port
* ignore test until it is complete
* check that first geth event is SessionEstablished
* fix endpoint string for the geth instance
* force an incoming connection
* we should also test establishing a session on an outgoing connections
* use NetworkEventStream helper in connect test
* create outgoing test
* make geth tests use different ports and datadirs
* update geth options
* cargo fmt
* s/geth_socket/reth_socket
* cargo fmt
* fix disc port collisions and re-add incoming test
* should add test utilities for getting unused disc/p2p ports
Co-authored-by: Matthias Seitz <matthias.seitz@outlook.de>
* move hello to separate file
* cargo fmt
* wip: actual fuzz test
* should probably also take advantage of test-fuzz to generate
benchmarks like impl_fuzzer_with_input
* impl generic roundtrip method
* generate test with macro
* change testname to fuzzname
* add reth-eth-wire to fuzz in ci
* add other message types to fuzz
* remove unused_crate_dependencies
* was causing test issues, may want to revisit whether or not we can
include this warning and still use test_fuzz
* more afl debugging ci
* use more explicit imports in fuzz_rlp
* impl Default for types and fuzz ping/pong
* Default is necessary for test-fuzz to auto generate a corpus for each
type we are fuzz testing
* enable AFL_I_DONT_CARE_ABOUT_MISSING_CRASHES
* not sure if we should do this in the workflow instead:
echo core >/proc/sys/kernel/core_pattern
we may miss crashes if we keep this enabled?
* remove reth-interfaces from fuzzing
* add secp256k1 to reth-db dev deps
* wip: start adding serde to stuff
* add serde traits to NewBlockHashes
* derive serde traits for the rest of the types
* derive serde traits for p2p subprotocol messages
* cargo fmt
* refactor(net): use shared objects on a per peer basis
* feat(net): send full transactions to fraction of all peers
Co-authored-by: Georgios Konstantopoulos <me@gakonst.com>
* fix disconnect reason encoding
* move disconnect to own file
* add encoding test for 0x80 in list
* rlp([0u8]) is a good test case to have because it is the RLP encoding
of a DisconnectRequested p2p message
* fix disconnect encoding and decoding
* directly decode the disconnect id to handle the 0x80 case of
DisconnectRequested
* the previous manual snappy encoding / decoding was incorrect - a
snappy encoded disconnect message is 4 bytes, not 3 bytes. The tests
and implementation are changed to reflect this.
* fix p2p message decoding
* had a similar issue to disconnect reason decoding where it would not
handle a 0x80 message id
* make invalid rlp header case more explicit
* cargo fmt
