udp listener fuzzer

[DavidKorczynski]

Signed-off-by: davkor [email protected]

Commit Message: This adds a fuzz test that targets the UdpListener code.
The fuzzer has been tested over a 30 minute experiment and runs without issues. It currently hits code in /source/common/network/udp_listener_impl.cc, /source/common/network/udp_listener_impl.cc, and source/common/network/io_socket_handle_impl.cc
Additional Description: Cross referencing #14889 CC @asraa
Risk Level: Zero. This adds a fuzz test and does not modify anything in the Envoy proxy itself.
Testing: Compiled with OSS-Fuzz to verify fuzzer runs in OSS-Fuzz environment.
Docs Changes: N/A
Release Notes:
Platform Specific Features:
[Optional Runtime guard:]
[Optional Fixes #Issue]
[Optional Deprecated:]
[Optional API Considerations:]

[repokitteh-read-only]

Hi @DavidKorczynski, welcome and thank you for your contribution.

We will try to review your Pull Request as quickly as possible.

In the meantime, please take a look at the contribution guidelines if you have not done so already.

🐱

Caused by: #15974 was opened by DavidKorczynski.

see: more, trace.

[asraa]

Thanks!! You may need to run fix format to help CI
out https://github.com/envoyproxy/envoy/blob/1d1b708c7bf6efa02c41d9ce22cbf1e4a1aeec2c/support/README.md#fixing-format-problems

Couple of initial nits, after CI passes it should generate a fuzz coverage report

[asraa]

Thanks! The fuzz coverage reports were generated, this got added the wanted coverage here https://storage.googleapis.com/envoy-pr/bebe107/fuzz_coverage/source/common/network/utility.cc.gcov.html
(compare to https://storage.googleapis.com/envoy-postsubmit/master/fuzz_coverage/source/common/network/utility.cc.gcov.html)

Assuming oss-fuzz runners support UDP_GRO, maybe we can also fuzz that portion?

[DavidKorczynski]

Thanks! The fuzz coverage reports were generated, this got added the wanted coverage here https://storage.googleapis.com/envoy-pr/bebe107/fuzz_coverage/source/common/network/utility.cc.gcov.html
(compare to https://storage.googleapis.com/envoy-postsubmit/master/fuzz_coverage/source/common/network/utility.cc.gcov.html)

Assuming oss-fuzz runners support UDP_GRO, maybe we can also fuzz that portion?

Sounds good - fixed it up and you can see the coverage report here: https://storage.googleapis.com/envoy-pr/056fbe5/fuzz_coverage/source/common/network/utility.cc.gcov.html

[asraa]

LGTM!

FWIW (for documentation later, not blocking addition to OSS-Fuzz) could you check if this line crashes

envoy/source/common/network/utility.cc

Lines 664 to 665 in a12869f

	ENVOY_LOG_MISC(debug, "Receive a packet with {} bytes from {}", msg_len,
	output.msg_[i].peer_address_->asString());

if you revert the conditions about output.msg_[i].truncated_and_dropped_ from https://github.com/envoyproxy/envoy/pull/14122/files

[DavidKorczynski]

LGTM!

FWIW (for documentation later, not blocking addition to OSS-Fuzz) could you check if this line crashes

envoy/source/common/network/utility.cc

Lines 664 to 665 in a12869f

	ENVOY_LOG_MISC(debug, "Receive a packet with {} bytes from {}", msg_len,
	output.msg_[i].peer_address_->asString());

if you revert the conditions about output.msg_[i].truncated_and_dropped_ from https://github.com/envoyproxy/envoy/pull/14122/files

Sure thing. Just to clarify here, is it the logging line or the assert right before the logging line that should crash?

[asraa]

It's the logging line actually :P since peer_address_ will be nullptr

[DavidKorczynski]

It's the logging line actually :P since peer_address_ will be nullptr

Ahh :) Thanks for the clarification, will update here with documentation shortly.