Messages in this thread | | | Date | Tue, 17 Mar 2015 20:55:55 -0400 | From | Steven Rostedt <> | Subject | Re: [PATCH] tracing: add trace event for memory-failure |
| |
On Tue, 17 Mar 2015 18:47:40 +0800 Xie XiuQi <xiexiuqi@huawei.com> wrote:
> I'm not clearly why we need a hard coded here. As the strings or "result" have > defined in mm/memory-failure.c, so passing "action_name[result]" would be more > clean and more flexible here?
The TP_printk() is what will be shown in the print format of the event "format" file, and is what trace-cmd and perf use to parse the data and know what to print. If you use "action_name[result]" that will be what the user space tools see, and will have no idea what to do with "action_name[result]". The hard coded output is a bit more explicit in how to interpret the raw data.
Another way around this is to create a "plugin" that can be loaded and will override the TP_printk() parsing.
-- Steve
| |