On Wed, Mar 24, 2021 at 2:52 AM Steven Rostedt <rostedt@xxxxxxxxxxx> wrote: > > On Mon, 22 Mar 2021 11:59:30 +0200 > "Tzvetomir Stoyanov (VMware)" <tz.stoyanov@xxxxxxxxx> wrote: > > > Trace clock, used for curent trace sessions is saved in the trace.dat > > file. This information is used when displaying timestamps of the events. > > However, only the selected clock is needed. > > A new API is added: > > tracecmd_set_out_clock(); > > that can be used to set the trace clock on the output handle. > > > > There's no reason not to have the other clocks in the file. I would > rather keep them as it doesn't hurt to have them right? > > The reason for keeping them is when a customer sends you a trace.dat > file (from commands you ask them to run), you can use the trace.dat > file to see what other clocks they have configured in case you want to > have them try another clock in their next recording. > > Extra data about the system the trace.dat file was run on is always a > plus. I'll extend the TRACECMD_OPTION_TRACECLOCK option with all clocks and will keep the changes from this patch. That way there will be information for all available clocks from the trace session and the old versions of trace-cmd will still work. > > -- Steve -- Tzvetomir (Ceco) Stoyanov VMware Open Source Technology Center
|