smite-scenarios: ignore unknown odd-type messages

[morehouse]

BOLT 1 requires nodes to ignore unknown message types if the type is odd. In particular, Eclair sends various custom informational messages (e.g., recommended_feerates as type 39409) that interfere with message processing if we fail to ignore them.

This change is required to actually extract any accept_channel response from Eclair.

Ref: #5 (Milestone 1)

[ekzyis]

LGTM, just one question:

In particular, Eclair sends various custom informational messages (e.g., recommended_feerates as type 39409) that interfere with message processing if we fail to ignore them.

This change is required to actually extract any accept_channel response from Eclair.

I wanted to reproduce this with a open_channel scenario prototype, but I'm getting this error when running the existing init scenario with eclair:

[-] The program took more than 1000 ms to process one of the initial test cases.
    This is bad news; raising the limit with the -t option is possible, but
    will probably make the fuzzing process extremely slow.

    If this test case is just a fluke, the other option is to just avoid it
    altogether, and find one that is less of a CPU hog.

[-] PROGRAM ABORT : Test case 'id:000000,time:0,execs:0,orig:seed1' results in a timeout
         Location : perform_dry_run(), src/afl-fuzz-init.c:1014

The encrypted_bytes scenario has a warning about slow exec/s (less than 90/sec).

Are these different issues? Is the init scenario with eclair working for you?

[morehouse]

LGTM, just one question:

In particular, Eclair sends various custom informational messages (e.g., recommended_feerates as type 39409) that interfere with message processing if we fail to ignore them.
This change is required to actually extract any accept_channel response from Eclair.

I wanted to reproduce this with a open_channel scenario prototype, but I'm getting this error when running the existing init scenario with eclair:

[-] The program took more than 1000 ms to process one of the initial test cases.
    This is bad news; raising the limit with the -t option is possible, but
    will probably make the fuzzing process extremely slow.

    If this test case is just a fluke, the other option is to just avoid it
    altogether, and find one that is less of a CPU hog.

[-] PROGRAM ABORT : Test case 'id:000000,time:0,execs:0,orig:seed1' results in a timeout
         Location : perform_dry_run(), src/afl-fuzz-init.c:1014

The encrypted_bytes scenario has a warning about slow exec/s (less than 90/sec).

Are these different issues? Is the init scenario with eclair working for you?

Yes, eclair is really slow. But I think the main issue you're hitting is ACINQ/eclair#3271.

That issue was fixed upstream but hasn't been release yet, so I haven't updated our workflow to use it. This is the patch I use locally:

diff --git a/smite-scenarios/src/targets/eclair.rs b/smite-scenarios/src/targets/eclair.rs
index bfe97a3..1c345f7 100644
--- a/smite-scenarios/src/targets/eclair.rs
+++ b/smite-scenarios/src/targets/eclair.rs
@@ -119,6 +119,7 @@ impl EclairTarget {
         // which inherits our crash handler wrapper and could trigger a false
         // crash report on exit().
         cmd.arg("-no-version-check")
+            .arg("-Declair.allow-unsafe-startup=true")
             .arg(format!("-Declair.datadir={}", eclair_dir.display()))
             .stdout(Stdio::null())
             .stderr(Stdio::null());
diff --git a/workloads/eclair/Dockerfile b/workloads/eclair/Dockerfile
index 4027dce..4b08940 100644
--- a/workloads/eclair/Dockerfile
+++ b/workloads/eclair/Dockerfile
@@ -2,8 +2,8 @@
 FROM eclipse-temurin:21 AS builder
 
 # Build arguments
-ARG ECLAIR_VERSION=v0.13.1
-ARG BITCOIN_VERSION=29.2
+ARG ECLAIR_COMMIT=fde8de65ffd984f071ea1cc9657dbf710f52994f
+ARG BITCOIN_VERSION=30.2
 ARG SCENARIO
 
 # Install build dependencies.
@@ -28,9 +28,10 @@ RUN wget https://bitcoincore.org/bin/bitcoin-core-${BITCOIN_VERSION}/bitcoin-${B
     mv bitcoin-${BITCOIN_VERSION}/bin/bitcoin-cli /usr/local/bin/bitcoin-cli && \
     rm -rf bitcoin-${BITCOIN_VERSION}*
 
-# Clone and build Eclair.
-RUN git clone --depth 1 --branch ${ECLAIR_VERSION} https://github.com/ACINQ/eclair.git /eclair-src
+# Clone Eclair at a specific commit and apply the deserialization fix patch.
+RUN git clone https://github.com/ACINQ/eclair.git /eclair-src
 WORKDIR /eclair-src
+RUN git checkout ${ECLAIR_COMMIT}
 RUN ./mvnw package -DskipTests -pl eclair-node -am
 
 # Unzip the Eclair distribution to /opt/eclair/.
diff --git a/workloads/eclair/Dockerfile.coverage b/workloads/eclair/Dockerfile.coverage
index ff9b83d..8ea338a 100644
--- a/workloads/eclair/Dockerfile.coverage
+++ b/workloads/eclair/Dockerfile.coverage
@@ -2,8 +2,8 @@
 FROM eclipse-temurin:21 AS builder
 
 # Build arguments
-ARG ECLAIR_VERSION=v0.13.1
-ARG BITCOIN_VERSION=29.2
+ARG ECLAIR_COMMIT=fde8de65ffd984f071ea1cc9657dbf710f52994f
+ARG BITCOIN_VERSION=30.2
 ARG SCENARIO
 ARG JACOCO_VERSION=0.8.14
 
@@ -35,8 +35,9 @@ RUN wget https://github.com/jacoco/jacoco/releases/download/v${JACOCO_VERSION}/j
     rm jacoco-${JACOCO_VERSION}.zip
 
 # Clone and build Eclair.
-RUN git clone --depth 1 --branch ${ECLAIR_VERSION} https://github.com/ACINQ/eclair.git /eclair-src
+RUN git clone https://github.com/ACINQ/eclair.git /eclair-src
 WORKDIR /eclair-src
+RUN git checkout ${ECLAIR_COMMIT}
 RUN ./mvnw package -DskipTests -pl eclair-node -am
 
 # Unzip the Eclair distribution to /opt/eclair/.