TraceCompass为什么同时能解析LTTng和Perf的cft Log

问题的引出和现在的上下文时空

就要准备生成metadata 为我的程序。
一旦开动，就很难转向了。
所以，开始之前，有许多问题在困扰着我，其中一个是为什么LTTng的meta与perf的meta定义并不同，
lttng是compass的亲兄弟，而不是perf.
所以，开始之前，我需要将这件事搞清楚。

问题的引出

被lttng和perf抓取的cft信息，都能被 trace compass解析。

TraceCompass

LTTng的图形

下图是Lttng
### Perf 的图形
下图是Perf
说明：可以看到，二者都能正常工作。

再来看Meta

二者的Meta define

LTTng meta

here is lttng metadata defination of sched_switch

event {
	name = "sched_switch";
	id = 1290;
	stream_id = 0;
	fields := struct {
		integer { size = 8; align = 8; signed = 1; encoding = UTF8; base = 10; } _prev_comm[16];
		integer { size = 32; align = 8; signed = 1; encoding = none; base = 10; } _prev_tid;
		integer { size = 32; align = 8; signed = 1; encoding = none; base = 10; } _prev_prio;
		integer { size = 64; align = 8; signed = 1; encoding = none; base = 10; } _prev_state;
		integer { size = 8; align = 8; signed = 1; encoding = UTF8; base = 10; } _next_comm[16];
		integer { size = 32; align = 8; signed = 1; encoding = none; base = 10; } _next_tid;
		integer { size = 32; align = 8; signed = 1; encoding = none; base = 10; } _next_prio;
	};
};

Perf meta

here is perf’s metadata of sched_switch

event {
	id = 3;
	name = "sched:sched_switch";
	stream_id = 0;
	fields := struct {
		integer { size = 64; align = 1; signed = false; encoding = none; base = hexadecimal; byte_order = le; } perf_ip;
		integer { size = 32; align = 1; signed = true; encoding = none; base = decimal; byte_order = le; } perf_tid;
		integer { size = 32; align = 1; signed = true; encoding = none; base = decimal; byte_order = le; } perf_pid;
		integer { size = 64; align = 1; signed = false; encoding = none; base = decimal; byte_order = le; } perf_id;
		integer { size = 64; align = 1; signed = false; encoding = none; base = decimal; byte_order = le; } perf_period;
		integer { size = 64; align = 1; signed = false; encoding = none; base = decimal; byte_order = le; } perf_data_src;
		integer { size = 32; align = 1; signed = false; encoding = none; base = decimal; byte_order = le; } perf_callchain_size;
		integer { size = 64; align = 1; signed = false; encoding = none; base = hexadecimal; byte_order = le; } perf_callchain[perf_callchain_size];
		integer { size = 32; align = 1; signed = false; encoding = none; base = decimal; byte_order = le; } common_type;
		integer { size = 32; align = 1; signed = false; encoding = none; base = decimal; byte_order = le; } common_flags;
		integer { size = 32; align = 1; signed = false; encoding = none; base = decimal; byte_order = le; } common_preempt_count;
		integer { size = 32; align = 1; signed = true; encoding = none; base = decimal; byte_order = le; } common_pid;
		integer { size = 32; align = 1; signed = false; encoding = none; base = decimal; byte_order = le; } common_migrate_disable;
		integer { size = 32; align = 1; signed = false; encoding = none; base = decimal; byte_order = le; } common_padding;
		string { encoding = UTF8; } prev_comm;
		integer { size = 32; align = 1; signed = true; encoding = none; base = decimal; byte_order = le; } prev_pid;
		integer { size = 32; align = 1; signed = true; encoding = none; base = decimal; byte_order = le; } prev_prio;
		integer { size = 64; align = 1; signed = true; encoding = none; base = decimal; byte_order = le; } prev_state;
		string { encoding = UTF8; } next_comm;
		integer { size = 32; align = 1; signed = true; encoding = none; base = decimal; byte_order = le; } next_pid;
		integer { size = 32; align = 1; signed = true; encoding = none; base = decimal; byte_order = le; } next_prio;
	} align(8);
};

问题的进一步解释

注意上图的右角的可视化部分，即两种ctf文件，因为不同的meta都能绘制出图形。
所以，本文的标题的主语是TraceCompass，
而不是Perf或LTTng。
因为这些图形，是TraceCompass根据几个内嵌的事件，而实现的。
“sched:sched_switch”;
“sched:sched_waking”;
“sched:sched_process_exit”;
“sched:sched_process_fork”;
这些事件能被抓到，是因为kernel，提供了这些probe.
lttng的meta中这些内定的事件，为如下名字：