arrange_profiles.h File Reference

#include <string>
#include <list>
#include <vector>
#include <iosfwd>
#include "image_errors.h"
#include "locate_images.h"

Include dependency graph for arrange_profiles.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Data Structures
struct	merge_option
struct	profile_template
struct	profile_sample_files
struct	profile_dep_set
struct	profile_set
struct	profile_class
struct	profile_classes
struct	image_set
struct	inverted_profile
Typedefs
typedef std::list< image_set >	image_group_set
Enumerations
enum	axis_types {   AXIS_EVENT, AXIS_TGID, AXIS_TID, AXIS_CPU,   AXIS_MAX }
Functions
std::ostream &	operator<< (std::ostream &, profile_sample_files const &)
std::ostream &	operator<< (std::ostream &, profile_dep_set const &)
std::ostream &	operator<< (std::ostream &, profile_set const &)
std::ostream &	operator<< (std::ostream &, profile_template const &)
std::ostream &	operator<< (std::ostream &, profile_class const &)
std::ostream &	operator<< (std::ostream &, profile_classes const &)
profile_classes const	arrange_profiles (std::list< std::string > const &files, merge_option const &merge_by, extra_images const &extra)
std::list< inverted_profile > const	invert_profiles (profile_classes const &classes)

---

Detailed Description

Classify and process a list of candidate sample files into merged sets and classes.

Remarks:

Copyright 2003 OProfile authors

Read the file COPYING

Author:

John Levon

Definition in file arrange_profiles.h.

---

Typedef Documentation

typedef std::list<image_set> image_group_set

Definition at line 195 of file arrange_profiles.h.

---

Enumeration Type Documentation

enum axis_types

The "axis" says what we've used to split the sample files into the classes. Only one is allowed.

Enumerator:

AXIS_EVENT
AXIS_TGID
AXIS_TID
AXIS_CPU
AXIS_MAX

Definition at line 124 of file arrange_profiles.h.

---

Function Documentation

profile_classes const arrange_profiles	(	std::list< std::string > const &	files,
		merge_option const &	merge_by,
		extra_images const &	extra
	)

Take a list of sample filenames, and process them into a set of classes containing profile_sets. Merging is done at this stage as well as attaching dependent profiles to the main image.

The classes correspond to the columns you'll get in opreport: this can be a number of events, or different CPUs, etc.

std::list<inverted_profile> const invert_profiles

(

profile_classes const &

classes

)

Invert the profile set. For opreport -l, opannotate etc., processing the profile_classes directly is slow, because we end up opening BFDs multiple times (for each class, dependent images etc.). This function returns an inverted set of sample files, where the primary sort is on the binary image to open.

Thus each element in the returned list is for exactly one binary file that we're going to bfd_openr(). Attached to that is the actual sample files we need to process for that binary file. In order to get the output right, these have to be marked with the profile class they're from (hence the groups vector), and the app image that owned the sample file, if applicable (hence image_set).

Definition at line 827 of file arrange_profiles.cpp.

References profile_classes::extra_found_images, inverted_profile::groups, and profile_classes::v.

std::ostream& operator<<	(	std::ostream &	,
		profile_classes const &
	)

std::ostream& operator<<	(	std::ostream &	,
		profile_class const &
	)

std::ostream& operator<<	(	std::ostream &	,
		profile_template const &
	)

std::ostream& operator<<	(	std::ostream &	,
		profile_set const &
	)

std::ostream& operator<<	(	std::ostream &	,
		profile_dep_set const &
	)

std::ostream& operator<<	(	std::ostream &	,
		profile_sample_files const &
	)