Out-of-bounds Iterator Access
Out-of-bounds iterator access is an undefined behavior, which might give a segfault, or give a garbage value.
1
2
3
4
5
6
7
8
9
10
11
12
#include <vector>
#include <iostream>
int main() {
std::vector<int> v = {10, 20, 30};
auto iter_first = v.begin();
auto bad_iter = iter_first + 5;
// Dereference — UB, likely a segfault
std::cout << *bad_iter << "\n";
return 0;
}
Workflow - Coredump
On Linux, debugging a segfault usually means inspecting a core dump, which is a snapshot of a process’s memory at the moment it crashed. Whether a core dump is produced — and where it ends up — depends on a few layers of configuration.
ulimit controls per-shell resource limits for processes you start from that shell. In particular, ulimit -c sets the maximum size of a core dump; if it is 0, no core dump will be generated at all. Setting it to unlimited enables core dump generation for that session and its child processes.
On most modern distributions, core dumps are captured by systemd-coredump rather than written directly to files. In this setup, crashes are recorded by systemd and accessed via coredumpctl, which can list crashes, show metadata, dump the core to a file, or launch gdb directly on the crashing binary.
Optionally, the kernel’s core_pattern can be changed to bypass systemd and write core dumps as plain files instead. This is a system-wide setting and should be used with care, but can be useful in environments without systemd (e.g. minimal containers) or when you explicitly want a local core file.
1
2
3
4
5
6
7
8
9
10
11
12
13
ulimit -c unlimited
// systemd-coredump
// If you see something piped into systemd (often contains `systemd-coredump`), then `coredumpctl` is the right tool.
cat /proc/sys/kernel/core_pattern
// ### (Optional) Check coredump storage policy
cat /etc/systemd/coredump.conf
// dump core into a plain file.
// It affects the whole machine**, not just your ROS 2 node.
sudo sh -c 'echo "core_backend_test.%p" > /proc/sys/kernel/core_pattern'
coredumpctlis a command-line tool that comes with systemd. On many distros, systemd captures crashes and stores the “core dump” in the journal or on disk.coredumpctllets you: - list recorded crashes. dump the core to a file, launch gdb directly on them
