A segmentation fault, often called a segfault, is a crash that happens when a program tries to read or write memory it is not permitted to access. The operating system stops the program to protect the rest of the system. On Linux this appears as signal 11, known as SIGSEGV, and logs often show a process that exited on signal 11.
Most segmentation faults are ordinary software bugs, such as a faulty extension, a bad library, running out of resources or a corrupted file. They are common in native code written in languages without built-in memory safety. A single crash on its own is rarely evidence of an attack.
Repeated crashes can still be worth attention. Memory-corruption vulnerabilities, including out-of-bounds writes, sometimes cause a crash when an attacker's first attempts do not work as intended, so a pattern of segfaults in a web-facing process around the time of unusual requests may be a clue. That is only a lead, since the same pattern is also produced by harmless bugs, and a successful exploit may not crash anything at all.
To investigate, check the system and service logs for which process crashed and when, compare the timing with web server and application logs, and note any recent updates or configuration changes. Keep the software patched, and if crashes continue, identify the failing component rather than assuming either an attack or a harmless bug. Treat core dumps carefully, because they can contain sensitive data from memory.