From 29b94c09d49ea65c64b8061e80016e45828cca52 Mon Sep 17 00:00:00 2001
From: Graham McIntire <graham@mcintire.me>
Date: Tue, 10 Feb 2026 14:14:07 -0600
Subject: [PATCH] Fix segfault from Alpine version mismatch, add crash handler

The builder (rust:1.93-alpine) uses Alpine 3.23 but the runtime used
Alpine 3.19. The net-snmp-dev headers from 3.23 don't match
net-snmp-libs from 3.19, causing a segfault on the first SNMP call.

Also adds a SIGSEGV/SIGBUS/SIGABRT signal handler that prints a
diagnostic message instead of silently exiting with code 139.
---
 Dockerfile  |  5 +++--
 src/main.rs | 33 +++++++++++++++++++++++++++++++++
 2 files changed, 36 insertions(+), 2 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index 9a3e070..6aca1c5 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -52,8 +52,9 @@ RUN --mount=type=cache,id=cargo-registry-${TARGETARCH},target=/usr/local/cargo/r
     BUILD_VERSION="$VERSION" cargo build --release --target "$RUST_TARGET" && \
     cp "target/$RUST_TARGET/release/towerops-agent" /tmp/towerops-agent
 
-# Runtime stage
-FROM alpine:3.19
+# Runtime stage - must match builder's Alpine version for ABI compatibility
+# (rust:1.93-alpine uses Alpine 3.23)
+FROM alpine:3.23
 
 # Install runtime dependencies
 # iputils provides ping with setuid root (doesn't require CAP_NET_RAW)
diff --git a/src/main.rs b/src/main.rs
index 71097b3..ab4741f 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -116,7 +116,40 @@ struct Args {
     snmpv3_priv_pass: String,
 }
 
+fn install_crash_handler() {
+    unsafe {
+        // Install a signal handler for SIGSEGV/SIGBUS/SIGABRT so we get
+        // diagnostic output instead of a silent exit code 139.
+        extern "C" fn crash_handler(sig: libc::c_int) {
+            let name = match sig {
+                libc::SIGSEGV => "SIGSEGV (segmentation fault)",
+                libc::SIGBUS => "SIGBUS (bus error)",
+                libc::SIGABRT => "SIGABRT (abort)",
+                _ => "unknown signal",
+            };
+            let msg = format!(
+                "\n*** FATAL: {} (signal {})\n\
+                 *** This is likely a bug in C FFI code (libnetsnmp).\n\
+                 *** Set RUST_BACKTRACE=1 for more info.\n",
+                name, sig
+            );
+            unsafe {
+                libc::write(libc::STDERR_FILENO, msg.as_ptr() as _, msg.len());
+                // Re-raise with default handler to get the correct exit code
+                libc::signal(sig, libc::SIG_DFL);
+                libc::raise(sig);
+            }
+        }
+
+        libc::signal(libc::SIGSEGV, crash_handler as libc::sighandler_t);
+        libc::signal(libc::SIGBUS, crash_handler as libc::sighandler_t);
+        libc::signal(libc::SIGABRT, crash_handler as libc::sighandler_t);
+    }
+}
+
 fn main() {
+    install_crash_handler();
+
     // Install ring as the default TLS crypto provider. Required because both
     // ring and aws-lc-rs features are enabled transitively, so rustls can't
     // auto-detect which one to use.