SAMSUNG SAAT: Reverse Engineering for Perform ance Analysis - - PDF document
SAMSUNG SAAT: Reverse Engineering for Perform ance Analysis 2004.05.25. Seon-Ah Lee, Seung-Mo Cho, Sung-Kwan Heo Mobile Solution Group Software Center Samsung Electronics salee@samsung.com Softw are Center 1 SAMSUNG Agenda Introduction
Softw are Center
1
SAAT: Reverse Engineering for Perform ance Analysis
2004.05.25. Seon-Ah Lee, Seung-Mo Cho, Sung-Kwan Heo Mobile Solution Group Software Center Samsung Electronics salee@samsung.com
2
Softw are Center
Agenda
I.
Introduction & Background
II.
Concept of SAAT
3
Softw are Center
I . I ntroduction & Background
Motivation Challenge Software Performance
Model
Software Reverse
Engineering
4
Softw are Center
For perform ance analysis at our research center
W e hoped to provide the result of perform ance analysis as
quickly as possible to stakeholders
W e decided to autom ate the process in order to shorten the
tim e of the perform ance analysis
Motivation
P e r f
m a n c e A n a l y s t A n a l y s i s R e q u e s t U n d e r s t a n d S t r u c t u r e & B e h a v i
S
t w a r e F i n d t h e b
t l e n e c k
S
t w a r e I d e n t i f y t h e s
u t i
t
h e p r
l e m u s i n g a s i mu l a t i
t
5
Softw are Center
There is no tool show ing softw are execution structure
w ith perform ance inform ation
Performance Analysis Tool ≠
> Software Behavior
Tools such as gprof, Vtune and Quantify show call graphs and
bottleneck candidates, but does not show the execution flow, so we cannot use the call graphs for understanding system’s behavior.
Reverse Engineering Tool ≠
> Performance information
Static analysis tools such as Source Insight and aiCall show a
software structure, but it is different from a dynamic structure.
Imagix show both static structure and call graphs, but it is only the
combination of upper tools.
Challenge
6
Softw are Center
Softw are Perform ance Model
Research trends
Researchers in performance engineering are studying how to
integrate software architecture with performance information
The representative w orks
C.U. Smith and L.G. Williams, Performance Solutions: A Practical
Guide to Creating Responsive, Scalable Software, Addison-Wesley,
Performance model & PASA methodology: extracting architecture
information from developer interviews and work products
Our position
Automation: the interview task An analysis tool
D e v e l
e r s a r e a l w a y s b u s y
7
Softw are Center
Softw are Reverse Engineering
Research Trends
Dynamic reverse engineering to extract software execution models
from existing systems is also being tried
The representative w orks
223, Brisbane, Australia, November 2000.
Shimba tool: producing sequence diagrams using trace information
at runtime (Java)
Our position
Simulation Model: Execution Structure + Performance Information Suitability: Java C
N
t r u c t u r e & P e r f
m a n c e I n f
m a t i
8
Softw are Center
I I . Concept of SAAT
What is SAAT? Trace Data Behavior Model Execution Model Simulation Model Structure of SAAT Using this tool in
performance analysis
9
Softw are Center
W hat is SAAT?
SAAT
A tool generating a sim ulation m odel from source code
autom atically
SW Source Code SW Source Code Probe Code Probe Code I nstrum ent-API Behavior Model Execution Model Sim ulation Model 1 2 3 Creating Execution Tree Constructing SW Structural Graph Creating Sim ulation Model Run
10
Softw are Center
Trace Data
A record of the interaction betw een the softw are
m odules ( Using TAU, generating Softw are Trace Data)
Function Main(parameter seq) { call A(seq); call B(); call C(); } Function A(parameter order) { if (order = first_sequence) { call a(); call b(); } else { call c(); } } Function B() {} Function C() {} A node Time Start/Finish Main start time of Main start flag A start time of A start flag a start time of a start flag a finish time of a finish flag b start time of b start flag b finish time of b finish flag A finish time of A finish flag B start time of B start flag B finish time of B finish flag C start time of C start flag C finish time of C finish flag Main finish time of Main finish flag
Begin Point of A End Point
11
Softw are Center
Behavior Model
A diagram representing Softw are Trace Data as nodes
and edges
Main A() a() B() C() Main A() c() B() C() b() First Sequence Second Sequence Initial Structure
(sequence of functions) Different structure !! => we need more than a sequence, i.e. Graph
12
Softw are Center
Execution Model
A com posite of several Behavior Models to represent the
dynam ic structure of the softw are
Main() A() End Begin a() c() b() B() C() Begin End
First Seq. of A Second Seq. of A
13
Softw are Center
A()
Sim ulation Model
A m odel running in a sim ulation environm ent to
dem onstrate architectural issues of the present system .
Main() End Begin a() c() b() B() C() Begin End A() Transaction 7 8 9 2 3 4 5 6
14
Softw are Center
Structure of SAAT
representing one execution of softw are
representing softw are run-tim e structure
running sim ulation environm ent.
Simulation Models Execution Models Behavior Models Commercial Tool Input: TAU Output: aiSee Output: Workbench BM EM SM Trace Data SAAT Program Tree Reader Writer Handler Graph List Reader Writer Handler Graph List Handler Mapper User Interface
SAAT is related to existing com m ercial tools.
TAU: Generating Trace Data aiSee: Presenting Trees and Graphs W orkbench: Providing Sim ulation Environm ent
15
Softw are Center
W e can insert the statistic data into each node from
execution inform ation of Trace Data
Using this tool in perform ance analysis
16
Softw are Center
I I I . I ssues & Conclusion
Why to drop static
analysis
How to construct
execution model from behavior model
Fail or Succeed Contribution Future Works
17
Softw are Center
W hy to drop static analysis
structural inform ation of existing softw are
inform ation for analyzing the softw are
inform ation
aspect
m anipulation of our tool
18
Softw are Center
Conditions
A function w ith tw o or m ore invocation sequences the sub-sequences of A function are different
the function should have loops or branches
How to construct execution m odel from behavior m odel
A A a c b B e h a v i
M
e l S c a n n i n g H a s h T a b l e
F u n c t i
s A B C a c b I n i t i a l E x e c u t i
M
e l b e g i n a b e n d c + F i n a l E x e c u t i
M
e l b r a n c h a b e n d c b e g i n
19
Softw are Center
Fail or Succeed
This tool m ight fail?
The trace and behavior m odels can get large, so the
com putation of an execution m odel gets com plex
The result to generate sim ulation m odel is rough, so user
m ediate part should be appended to the tool
This tool m ight succeed?
This tool w as used to discover the pattern of softw are
execution and useful to find problem atic part already
This tool is the first trial to extract sim ulation m odel from
the existing softw are autom atically
20
Softw are Center
Contribution
Benefits
Acquiring the m ethodology facilitating the understanding of
existing softw are
Developing a suitable tool to analyze our em bedded
softw are im plem ented in c language
Pioneering the w ay to create a m odel running a sim ulation
environm ent from existing softw are autom atically
Academ ic interest
Bridging the gap betw een perform ance analysis and reverse
engineering.
P e r f
m a n c e M
e l R e v e r s e E n g i n e e r i n g S A A T : O u r R e s e a r c h
21
Softw are Center
Future W orks
Grouping functions
The w ay to bind the corresponding com ponent User intervention parts / Com ponent-declaring parts Precedence: Dali W orkbench tool m ade by Kazm an
More refined Models
Additional rules for converting Behavior Models to an
Execution Model
Options to m odify the Sim ulation m odel for user tastes
Various Model Environm ents
Adapting this tool to several m odeling environm ents.