commit | d35aa624a5e3ac65ed33f1a874d98663219a5ea4 | [log] [tgz] |
---|---|---|
author | Ryan Savitski <rsavitski@google.com> | Wed Nov 24 15:12:32 2021 +0000 |
committer | Ryan Savitski <rsavitski@google.com> | Wed Nov 24 15:24:40 2021 +0000 |
tree | 0d10c16d799176597cf17182aebf2b87ddbc9b1f | |
parent | 31c1534a991af2be98b94ba1984a2950a3770c5e [diff] |
traced_perf: support explicit clockid for the sample timestamps Still defaulting to MONOTONIC_RAW if unset, not trying to be clever about detecting the situations when it's safe to "upgrade" to the boot clock. The default could be the base monotonic clock, but I kept monotonic_raw for consistency with the previous implementation. I no longer remember whether there were important differences between the two in practice. If the config asks for a clock that is incompatible with the event, the syscall will fail and the data source won't be started, as with other kinds of misconfiguration. Hand-tested with strace and traceconv, verifying that the captured syscall parameter reflects the config, and that the trace packet defaults has the correct clockid as well. Change-Id: If38d5fb901c32ff7f48e18ed79a24f6c79b2ec52
Perfetto is a production-grade open-source stack for performance instrumentation and trace analysis. It offers services and libraries and for recording system-level and app-level traces, native + java heap profiling, a library for analyzing traces using SQL and a web-based UI to visualize and explore multi-GB traces.
See https://perfetto.dev/docs or the /docs/ directory for documentation.