Messages in this thread | | | Date | Sat, 11 Oct 2008 19:50:33 -0400 (EDT) | From | Steven Rostedt <> | Subject | Re: [PATCH 4/5] tracing/fastboot: Add a time field on the sched_switch entry |
| |
On Sat, 11 Oct 2008, Fr?d?ric Weisbecker wrote:
> 2008/10/11 Steven Rostedt <rostedt@goodmis.org>: > >> + > >> +static enum print_line_t boot_trace_print_line(struct trace_iterator *iter) > >> +{ > >> + struct trace_entry *entry = iter->ent; > >> + > >> + if (entry->type == TRACE_BOOT) > >> + return initcall_print_line(iter); > >> + if (entry->type == TRACE_CTX || entry->type == TRACE_WAKE) > >> + return sched_switch_print_line(iter); > >> return TRACE_TYPE_UNHANDLED; > >> } > >> > >> @@ -100,7 +159,7 @@ struct tracer boot_tracer __read_mostly = > >> .init = boot_trace_init, > >> .reset = reset_boot_trace, > >> .ctrl_update = boot_trace_ctrl_update, > >> - .print_line = initcall_print_line, > >> + .print_line = boot_trace_print_line, > >> }; > >> > >> void trace_boot(struct boot_trace *it, initcall_t fn) > >> > >> > > > > Yes but this one doesn't seem to start at zero at the boot time. > There is a big difference between those two ...
Even so, I disagree having the context switch tracer hold another 8 byte timestamp. If you want to add something to the boot tracer, than fine. But not to something that other tracers use.
You can tap into the context switch trace_point too, and have that record a special trace entry for you instead.
-- Steve
| |