Messages in this thread | | | Date | Mon, 22 Feb 2016 22:23:08 -0500 | From | Steven Rostedt <> | Subject | Re: [PATCH] tracing: don't macro-expand arguments before stringification in TP_printk |
| |
On Wed, 16 Dec 2015 01:28:12 +0100 Rasmus Villemoes <linux@rasmusvillemoes.dk> wrote:
> Bah, just found 0462b5664b (ftrace: Output REC->var instead of > __entry->var for trace format). There's some magic here I don't > understand, but I'm wondering if '__entry' wouldn't do just as well as > 'REC' for the tools that try to parse these strings. > >
Sorry for the late reply, but I just found this in my INBOX :-)
Several tools already exist that use REC to parse. It would be ABI breakage to suddenly convert it to __entry.
Not to mention, even in your example of:
((unsigned int) ((REC->dev) >> 20)), ((unsigned int) ((REC->dev) & ((1U << 20) - 1)))
which came from
MAJOR(__entry->dev), MINOR(__entry->dev),
The parsing tools don't know how to parse MAJOR or MINOR but they do know how to do the above macro expansions.
Just an FYI,
-- Steve
| |