Collaboration Summit 2015

The Linux perf tools continuously grows in scope as an observability toolchest, with more features being added. This talk showcases the 'probe' and 'trace' tools: one allows adding dynamic probes in arbitrary points in the kernel and in userspace libs and programs, collecting global and local variables and callchains, that can then be used in conjunction with other perf tools: record, top, trace, script. The later, trace, started as a super strace, one that allows stracing not just threads, but other targets such as the whole system, sets of CPUs. It also has a lower overhead as it doesn't use the ptrace syscall. Support for tracing page faults and other events and collecting callchains is planned and may be ready by Collab'15. An example of integration using 'probe', capturing syscall arguments for pretty printing, showcasing 'candidate tracepoints' will also be presented.