1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
|
\section{Introduction}
High-level synthesis (HLS), which refers to the automatic translation of software into hardware, is becoming an increasingly important part of the computing landscape.
It promises to increase the productivity of hardware engineers by raising the abstraction level of their designs, and it promises software engineers the ability to produce application-specific hardware accelerators without having to understand hardware desciption languages (HDL) such as Verilog and VHDL.
It is even being used in high-assurance settings, such as financial services~\cite{hls_fintech}, control systems~\cite{hls_controller}, and real-time object detection~\cite{hls_objdetect}. As such, HLS tools are increasingly relied upon. In this paper, we investigate whether they are trustworthy.
\begin{figure}[t]
\centering
\begin{minted}{c}
unsigned int b = 0x1194D7FF;
int a[6] = {1, 1, 1, 1, 1, 1};
int main() {
for (int c = 0; c < 2; c++)
b = b >> a[c];
return b;
}
\end{minted}
\caption{Miscompilation bug found in Xilinx Vivado HLS 2018.3 and 2019.2. The program returns \code{0x006535FF} but the correct result is \code{0x046535FF}.}
\label{fig:vivado_bug1}
\end{figure}
The approach we take in this paper is \emph{fuzzing}.
%To test the trustworthiness of HLS tools, we need a robust way of generating programs that both have good coverage and also explores various corner cases.
%Therein lies the difficulty in testing HLS tools.
%Human testing may not achieve both these objectives, as HLS tools are often require complex inputs to trigger wrong behaviour.
%In this paper, we employ program fuzzing on HLS tools.
This is an automated testing method in which randomly generated programs are given to compilers to test their robustness~\cite{fuzzing+chen+13+taming,fuzz+sun+16+toward,fuzzing+liang+18+survey,fuzzing+zhang+19,yang11_findin_under_bugs_c_compil,lidbury15_many_core_compil_fuzzin}.
The generated programs are typically large and rather complex, and they often combine language features in ways that are legal but counter-intuitive; hence they can be effective at exercising corner cases missed by human-designed test suites.
Fuzzing has been used extensively to test conventional compilers; for example, Yang \textit{et al.}~\cite{yang11_findin_under_bugs_c_compil} used it to reveal more than three hundred bugs in GCC and Clang. In this paper, we bring fuzzing to the HLS context.
%We specifically target HLS by restricting a fuzzer to generate programs within the subset of C supported by HLS.
% Most fuzzing tools randomly generate random C programs that are then provided to the compiler under test.
% Furthermore, fuzzing tools can be configured by users to generate and avoid particular patterns, which is important since HLS tools typically support a subset of C.
% Compiler Fuzzing is a popular technique to find bugs in programs, and it is especially effective at finding compiler bugs, as This technique can therefore also be used to find bugs in HLS tools. There has also been some prior work in trying to find bugs in HLS tools and also ensuring that synthesis tools output a correct design.
% \NR{add a sentence about fuzzing}
% % Our method is brought over from the compiler testing literature.
% Program fuzzing is bla..
% Fuzzing enables us to overcome
\paragraph{An example of a compiler bug found by fuzzing}
Figure~\ref{fig:vivado_bug1} shows a program that produces the wrong result during RTL simulation in Xilinx Vivado HLS. The bug was initially revealed by a randomly generated program of around 113 lines, which we were able to reduce to the minimal example shown in the figure.
The program repeatedly shifts a large integer value \code{b} right by the values stored in array \code{a}.
Vivado HLS returns \code{0x006535FF}, but the result returned by GCC (and subsequently confirmed manually to be the correct one) is \code{0x046535FF}.
The circumstances in which we found this bug illustrate some of the challenges in testing HLS tools.
For instance, without the for-loop, the bug goes away.
Moreover, the bug only appears if the shift values are accessed from an array.
And -- particularly curiously -- even though the for-loop only has two iterations, the array \code{a} must have at least six elements; if it has fewer than six, the bug disappears.
Even the seemingly random value of \code{b} could not be changed without masking the bug.
It seems unlikely that a manually generated test program would bring together all of the components necessary for exposing this bug.
In contrast, producing counter-intuitive, complex but valid C programs is the cornerstone of fuzzing tools.
For this reason, we find it natural to adopt fuzzing for our HLS testing campaign.
% \NR{Yann, please double check my claims about the bug. I hope I accurately described what we discussed. }\YH{Yes I agree with all that, I think that is a good description of it}
\paragraph{Our contribution}
This paper reports on our campaign to test HLS tools by fuzzing.
\begin{itemize}
\item We use Csmith~\cite{yang11_findin_under_bugs_c_compil} to generate ten thousand valid C programs from within the subset of the C language that is supported by all the HLS tools we test. We augment each program with a random selection of HLS-specific directives.
\item We give these programs to three widely used HLS tools: Xilinx Vivado HLS~\cite{xilinx20_vivad_high_synth}, LegUp HLS~\cite{canis13_legup} and the Intel HLS Compiler, which is also known as i++~\cite{intel20_sdk_openc_applic}. When we find a program that causes an HLS tool to crash, or to generate hardware that produces a different result from GCC, we reduce it to a minimal example with the help of the \creduce{} tool~\cite{creduce}.
\item Our testing campaign revealed that all three tools could be made to crash while compiling or to generate wrong RTL. In total, we found 6 bugs across the three tools.
\item To investigate whether HLS tools are getting more or less reliable over time, we also tested three different versions of Vivado HLS (2018.3, 2019.1, and 2019.2). We found that in general there about half as many failures in versions 2019.1 and 2019.2 compared to 2018.3. However, there were also test-cases that only failed in versions 2019.1 and 2019.2, meaning bugs were probably introduced due to the addition of new features.
\end{itemize}
% we test, and then augment each program with randomly chosen HLS-specific directives. We synthesise each C program to RTL, and use a Verilog simulator to calculate its return value. If synthesis crashes, or if this return value differs from the return value obtained by executing a binary compiled from the C program by GCC, then we have found a candidate bug. We then use trial-and-error to reduce the C program to a minimal version that still triggers a bug.
% We have tested three widely used HLS tools: LegUp~\cite{canis13_legup}, Xilinx Vivado HLS~\cite{xilinx20_vivad_high_synth}, and the Intel HLS Compiler~\cite{?}. For all three tools, we were able to find valid C programs that cause crashes while compiling and valid C programs that cause wrong RTL to be generated. We have submitted a total of \ref{?} bug reports to the developers, \ref{?} of which have been confirmed and \ref{?} of which have now been fixed at the time of writing.
% We hope that our work serves to stimulate efforts to improve the quality of HLS tools.
%%% Local Variables:
%%% mode: latex
%%% TeX-master: "main"
%%% End:
|
