rr-debugger

Deterministic debugging with rr record-replay records program execution to replay crashes identically. You record the program with rr, replay the recording, and use gdb reverse execution commands like reverse-next to step backwards and find the root cause.

Reverse execution in gdb allows you to step backwards through a recorded program execution. It helps crash analysis by letting you trace from the exact crash point back to the originating memory fault or ASAN error using commands like reverse-continue and reverse-stepi.

Yes, rr record-replay supports ASAN faults. The automation script scripts/crash_trace.py automatically extracts execution traces from rr recordings of both regular crashes and ASAN crashes to identify the root cause.

Yes, basic gdb familiarity is required to navigate reverse workflows. You need to use standard gdb commands to inspect variables and memory, alongside rr-specific reverse execution commands like reverse-next and reverse-stepi during replay.

The best way to automate crash trace extraction is using the provided scripts/crash_trace.py. It automatically extracts execution traces from your rr recordings, streamlining the deterministic debugging workflow for regular crashes and ASAN reports.

Use deterministic debugging when standard gdb fails to catch intermittent crashes or ASAN faults due to non-deterministic execution. rr record-replay guarantees identical replay, ensuring reverse execution reliably reaches the memory fault root cause.