Skip to content
/ Emma Public

Emma Memory and Mapfile Analyser

License

GPL-3.0, Unknown licenses found

Licenses found

GPL-3.0
LICENSE
Unknown
COPYING
Notifications You must be signed in to change notification settings

bmwcarit/Emma

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Emma

Codacy Badge Codacy Badge Build Status PyPi License: GPL v3 GitHub pages

Emma Memory and Mapfile Analyser (Emma)

Conduct static (i.e. worst case) memory consumption analyses based on arbitrary linker map files. It produces extensive .csv files which are easy to filter and post-process. Optionally .html and markdown reports as well as neat figures help you visualising your results.

Given a map file input (Green Hills map files are the default but others - like GCC - are supported via configuration options; examples are enclosed) Emma maps the addresses of sections (aka images) and/or objects (aka modules) to memory regions (all addresses given via map files must be known during compile time). Those memory regions are classified into two levels of granularity respectively. The first level defines arbitrary groups based on your personal taste (however using names similar to those defined by your microcontroller vendor makes most sense). Later each of those regions (second level) are assigned to one of four generalised predefined memory regions (those are: INT_RAM, INT_FLASH, EXT_RAM, EXT_FLASH). In case of virtual memory objects and sections lying within virtual address spaces (VASes) get translated back into physical memory. This is depicted in the figure above (lower part).

Categorisation can be used to assign consumers (a consumer would usually represent a software component) to each object or section. This is useful for subsequent steps in order to display memory consumption per consumer type. See the upper part of the figure shown above. Mechanisms are provided to batch categorise object and section names. "Objects in sections" provides ways to obtain a finer granularity of the categorisation result. Therefore categorised sections containing (smaller) objects of a different category got split up and result into a more accurate categorisation.

As a result you will get output files in form of a .csv file which sets you up to do later processing on this data easily. In this file additional information is added like:

  • Overlaps (of sections/objects)
  • Containments (e.g. sections containing objects)
  • Duplicates
  • All meta data about the origin of each section/object (mapfile, addess space, ...)
  • ...

Holding the aforementioned augmented data makes it easy to detect issues in linker scripts and get an in-depth understanding of your program's memory consumption. Including a lot of additional and "corrected" data can cause confusion. Thus all original (unmodified) data is preserved in the output files simultaneously.

The Emma visualiser helps you to create nice plots and reports in a .png and .html and markdown file format.

The whole Emma tool suite contains command line options making it convenient to be run on a build server like --Werror (treat all warnings as errors) or --no-prompt (exit and fail on user prompts; user prompts can happen when ambiguous configurations appear such as multiple matches for one configured map files).


Installation

pip3 install pypiemma

Dependencies: Python 3.6 or higher; pip3 install Pygments Markdown matplotlib pandas pypiscout

Optional: Cython For bigger projects escpecially the number of objects will grow. We provide an optional Cython implementation which can speed-up your analysis (you will gain typically about **30 % speed-up**).

For now we do not provide the binaries with Emma, hence you have to compile (make sure a suitable compiler is installed) it yourself (don't worry it is quick and easy):

Install the Cython package (pip install Cython) and (in the Emma top level folder) execute (MSVC is recommended on Windows):

python setup.py build_ext --inplace --compiler=msvc

General Workflow

The following figure shows a possible workflow using Emma:

Emma - as the core component - produces an intermediate .csv file. Inputs are mapfiles and JSON files (for configuration (memory layout, sizes, ...)). From this point you are very flexible to choose your own pipeline. You could

  • use the Emma tools (Emma Visualiser, Emma Deltas, ...) for further processing (data aggregation and analysis),
  • simply your favourite spreadsheet software (like Microsoft Excel or LibreOffice Calc) or
  • use your own tool for the data analysis.

Quick Start Guide

At this point we want to give you a brief overview what to do in the below two scenarios. If you want to play around go to (project files are already present) and use our example projects in ./doc/test_project*.

Example projects (including Emma* outputs/results) can be found in ./doc/test_project*.

Since version 3.1 Emma can be called in two ways (if you want to run it from the installation folder) where the following variant is recommended:

python Emma.py a --project doc/test_project --mapfiles doc/test_project/mapfiles

The following table provides an overview how you call Emma:

Emma module Entry point + <options> (if installed via pip) Top level sub-command (tlsc) (python Emma.py <tlsc>) Module (python -m + <module> <options>)
Analyser emma a Emma.emma
Visualiser emma_vis v Emma.emma_vis
Deltas emma_deltas d Emma.emma_deltas

Project files are already present

Try python Emma.py a --help to see all possible options or refer to the documentation (./doc/*).

  1. Create intermediate .csv from mapfiles with Emma:
python Emma.py a -p .\MyProjectFolder --map .\MyProjectFolder\mapfiles --dir .\MyProjectFolder\analysis --subdir Analysis_1
  1. Generate reports and graphs with Emma Visualiser:
python Emma.py v -p .\MyProjectFolder --dir .\MyProjectFolder\analysis --subdir Analysis_1 -q 

Project files that have to be created

To create a new project, the following files must be created:

  • globalConfig.json
  • budgets.json
  • categories.json
  • categoriesKeywords.json
  • categoriesSections.json
  • categoriesSectionsKeywords.json

You will find example projects in ./doc/test_project*. In-depth documentation can be found in the full documentation (see ./doc/).

A basic configuration can be short per file. For complex systems you can choose from many optional keywords/options that will provide you means to adjust your analysis as fine grained as you wish.

One main concept includes the globalConfig.json. You can see this as meta-config. Each configuration ID (configID) is a separately conducted analysis. Per configID you state individually the configuration files you want to use for this exact analysis. Herewith you can mix and match any combination of subconfigs you prefer.

A globalConfig.json could look like this:

{
    "configID1": {
        "addressSpacesPath": "addressSpaces.json",
        "sectionsPath": "sections.json",
        "patternsPath": "patterns.json"
    }
}

Full documentation

For the full documentation please refer to the ./doc/ directory.

Contribute

We are glad if you want to participate. In ./doc/dev-guide.md you will find a guide telling you everything you need to know including coding conventions and more.

emma-dev (. at) googlegroups.com

Dependencies & Licences

Library (version) pip package name Licence URL
Markdown (v3.0.1+) Markdown BSD-3-Clause https://github.com/Python-Markdown/markdown; https://python-markdown.github.io/
Pandas (v0.23.4+) pandas BSD-3-Clause https://github.com/pandas-dev/pandas/; http://pandas.pydata.org/getpandas.html
Pygments (v2.3.1+) Pygments BSD-2-Clause https://bitbucket.org/birkenfeld/pygments-main/src/default/; http://pygments.org/download/
Matplotlib (v3.0.0+) matplotlib Matplotlib License (BSD compatible) https://matplotlib.org/users/installing.html; https://github.com/matplotlib/matplotlib
SCout (v2.0+) pypiscout MIT https://github.com/holzkohlengrill/SCout
svgwrite (v1.4+) svgwrite MIT License (MIT License) https://github.com/mozman/svgwrite; https://svgwrite.readthedocs.io/en/latest/

Optional dependencies:

Utility scripts in ./doc/ need additional dependencies. As a normal user you can ignore this.

Library (version) pip package name Licence URL
gprof2dot (v2017.9.19+) gprof2dot LGPL-3.0 https://github.com/jrfonseca/gprof2dot
pylint (v2.3.1+) pylint GPL-2.0 https://github.com/PyCQA/pylint
Cython (v0.29.13+) Cython Apache-2.0 https://cython.org/

Please refer to the gprof2dot project site and install its dependencies (this has to be done even if you install Emma via pip).

Note that those modules are invoked via subprocess calls within the ./genDoc/ scripts.

Dependencies used to generate documentation for GitHub pages (separate, independent branch gh-pages):

Utility scripts used to build GitHub pages documentation. As a normal user you can ignore this.

Library (version) pip package name Licence URL
MkDocs (v1.0.4+) mkdocs BSD-3Clause https://github.com/mkdocs/mkdocs
Material for MkDocs (v4.4.1+) mkdocs-material MIT https://github.com/squidfunk/mkdocs-material

Code snippets etc.:

Name (version) Kind Modified? Licence URL
pygmentize (v2.2.0+) Auto-generated .css file Yes BSD-2-Clause http://pygments.org/download/; https://bitbucket.org/birkenfeld/pygments-main/issues/1496/question-licence-of-auto-generated-css
toHumanReadable (--) Code snippet No MIT https://github.com/TeamFlowerPower/kb/wiki/humanReadable

For the full documentation please refer to the ./doc/ directory.