Tools and Models – Lab 2
Completion Date: 14
of December 2018
Value: 12 marks
Explain briefly how the code here
ports/ uses the pl101 DR and FR registers to transmit and receive data over the associated UART peripheral of the Versatilepb board.
Give a brief overview of the clang generated Abstract Syntax Tree (AST) for the
while loop of the attached t2.c file. The attached t2.ast file from line 25 to line 53
shows a dump of the WhileStmt node and its children which corresponds to the
while loop. The t2.ast file was generated using:
clang-check -ast-dump t2.c –extra-arg=”-fno-color-diagnostics” >t2.ast
Clang/LLVM is a framework for the compilation and analysis of source code
written is several different programming languages. Clang provides many tools for the manipulation of its ASTs. See the video linked in this tutorial for details https://jonasdevlieghere.com/understanding-the-clang-ast/.
In this case we use clang to generate an AST for a given simple c function fred().
We use the -ast-dump command line argument to generate a textual
representation of the AST, which is stored in the file t2.ast. Each node in the AST
models a particular language construct. For example, the VarDecl class models a c declaration of a variable.
For instance, the c statement in the file t1.c:
is modeled in the AST as two VarDecl nodes and one IntegerLiteral node which
are in the t2.ast from line 15 to line 18:
-VarDecl 0x1c08820 <col:1, col:5> col:5 used i ‘int’
-VarDecl 0x1c08890 <col:1, col:9> col:7 used j ‘int’ cinit
-IntegerLiteral 0x1c088f0 ‘int’ 0
If a c variable is used in an expression in t2.c, the AST will have a corresponding
DeclRefExpr node to model that use.
For example, the c statement:
j = 0;
will be modeled using the following AST nodes (dumped using the -ast-dump flag
to the compiler):
BinaryOperator 0x1c089f8 <line:4:1, col:3> ‘int’ ‘=’
| |-DeclRefExpr 0x1c089b0 ‘int’ lvalue Var 0x1c08890 ‘j’ ‘int’
| `-IntegerLiteral 0x1c089d8 ‘int’ 0
The attached code (testbench.c and peak.c) is a testbench and a state machine
implementation of the peak constraint
http://sofdem.github.io/gccat/gccat/Cpeak.html. You are to write a c function
which implements the “valley” constraint and a second c function which
implements the “increasing peak” constraint. You are to update the testbench to
test these new constraints. A constraint is just a form of if statement. The valley
and increasing peak constraints are defined here:
The valley constraint is a simple state machine, which is practically identical to
the peak constraint. An implementation of the peak constraint state machine is
given in peak.c. See figure 5.417.3 below for the valley constraint state machine.
The scope of this lab is the design of a hardware accelerator for detecting
anomalies in time series data. These anomalies are defined by the user in terms
of constraints which can be used to detect combinations of spikes and/or troughs
with particular characteristics. Such accelerators could be deployed for example
2 | P a g e
into a nuclear power facility to quickly detect fluctuations in the sensor readings
from the instrumentation control systems of the plant.
The different constraints can be combined to form more complex filters. See for
where a systemc model of an EXOR gate is implemented using four NAND gates.
The code (testbench.c and peak.c) is a high level data flow model/design that
can be synthesised directly onto a FPGA using for example vivado HLS.