Part 1 of this series introduced C-Reduce and showed how it combines a domain-independent core with a large collection of domain-specific passes in order to create a highly effective test-case reducer for C and C++ code. This part tells the rest of the story and concludes. Parallel Test-Case Reduction C-Reduce’s second research contribution is to… Continue reading Design and Evolution of C-Reduce (Part 2)
Category: Computer Science
Design and Evolution of C-Reduce (Part 1)
[This piece is posted in parallel on the IEEE Software blog. Karim Ali copyedited.] Since 2008, my colleagues and I have developed and maintained C-Reduce, a tool for programmatically reducing the size of C and C++ files that trigger compiler bugs. C-Reduce also usually does a credible job reducing test cases in languages other than… Continue reading Design and Evolution of C-Reduce (Part 1)
It’s Time for a Modern Synthesis Kernel
Alexia Massalin’s 1992 PhD thesis has long been one of my favorites. It promotes the view that operating systems can be much more efficient than then-current operating systems via runtime code generation, lock-free synchronization, and fine-grained scheduling. In this piece we’ll only look at runtime code generation, which can be cleanly separated from the other… Continue reading It’s Time for a Modern Synthesis Kernel
Verifying Popcount
Popcount is the function that returns the number of set bits in its argument. Showing that a popcount implementation does what it claims to do has become one of my favorite examples to use when I need to quickly show students how we can reason about programs mathematically. Something like a selection sort is probably… Continue reading Verifying Popcount
Walking or Biking to NSF
Since the National Science Foundation funds a large fraction of academic computer science research in the USA, we often end up traveling to Washington to visit the NSF. This post is just to say that if you are traveling light, if you need some exercise, if you have a bit of free time, and if the… Continue reading Walking or Biking to NSF
Synthesizing Constants
(See this blog post for a short introduction to synthesis, or this paper for a long one.) In this piece I want to discuss an aspect of program synthesis that sounds like it should be easy, but isn’t: synthesizing constant values. For example, consider trying to synthesize an optimized x86-64 implementation of this code: The… Continue reading Synthesizing Constants
Learning When Values are Changed by Implicit Integer Casts
C and C++ perform implicit casts when, for example, you pass an integer-typed variable to a function that expects a different type. When the target type is wider, there’s no problem, but when the target type is narrower or when it is the same size and the other signedness, integer values may silently change when… Continue reading Learning When Values are Changed by Implicit Integer Casts
What’s the difference between an integer and a pointer?
(This piece is an alternate introduction and advertisement for a soon-to-be-published research paper.) In an assembly language we typically don’t have to worry very much about the distinction between pointers and integers. Some instructions happen to generate addresses whereas others behave arithmetically, but underneath there’s a single data type: bitvectors. At the opposite end of… Continue reading What’s the difference between an integer and a pointer?
Future Directions for Optimizing Compilers
I wanted to write a manifesto-ish sort of piece about what compilers are supposed to look like in the future. Nuno was the obvious coauthor since I’ve talked to him about this topic so much that I’m overall not really sure which parts started out as his ideas and which were mine. The article didn’t… Continue reading Future Directions for Optimizing Compilers
How LLVM Optimizes a Function
An optimizing, ahead-of-time compiler is usually structured as: A frontend that converts source code into an intermediate representation (IR). A target-independent optimization pipeline: a sequence of passes that successively rewrite the IR to eliminate inefficiencies and forms that cannot be readily translated into machine code. Sometimes called the “middle end.” A target-dependent backend that generates… Continue reading How LLVM Optimizes a Function