Jockey: Extensible user-space library for record-replay debugging
Next:
1. Getting started
Jockey: Extensible user-space library for record-replay debugging
Yasushi Saito
Hewlett-Packard Labs (then) / Google Inc (now)
1. Getting started
1.1 Recording program execution
1.2 Replaying program execution
2. Limitations
3. Basic uses of Jockey
3.1 Recording and replaying program execution
3.2 Checkpointing and restoring the program state
3.3 Changing the event-log directory and debug message output
3.4 Handling fork and exec
3.5 Esoteric options
3.6 Using Jockey manually
3.6.1 Linking libjockey.so to the target program
3.6.2 Controlling Jockey through environment variables and
--jockey
command-line option
3.7 Multithreading
3.8 Debugging programs using Jockey
4. Jockey internals
4.1 Instruction patching
4.2 Segregating resource usage
4.2.1 Heap
4.2.2 Stack
4.2.3 File descriptors
4.3 Checkpointing
4.3.1 Preventing brain damage to the dynamic linker
4.3.2 Exec shield
4.4 Handling signals
4.5 Reducing logging overhead for I/O system calls
4.6 Handling memory-mapped I/Os
5. Jockey utility programs
5.1 jockeylog: Displaying event log contents
5.2 jockeyctl: Changing a parameter of a running program
6. Files used or created by Jockey
7. Programming Jockey
7.1 User-defined invariant checker
7.2 Interposer library
8. Reporting bugs
Index
About this document ...
Jockey: Extensible user-space library for record-replay debugging
Next:
1. Getting started