Suppose you want to start contributing to GHC: what should you read by way of background? Here is an annotated list. Please add to it as you come across useful material. When you do so, please consider adding a link to a place where you are reasonably confident the resource will be available in 10 or 20 years. doi purports to enable such links.
Modular type inference with local assumptionsdoi link, Simon Peyton Jones, Dimitrios Vytiniotis, Tom Schrijvers, Martin Sulzmann, Journal of Functional Programming, 2011. This epic 83-page JFP paper brings together, in a single uniform framework, a series of our earlier papers on type inference for type systems involving local constraints, including GADTs and indexed type families.
System FC with Explicit Kind Equality. Stephanie Weirich, Justin Hsu, Richard A. Eisenberg. ICFP '13. doi Merges types with kinds, allowing promotion of GADTs and type families. Implementation not yet merged (July 2015).
Equality proofs and deferred type errors, Simon Peyton Jones, Dimitrios Vytiniotis and Pedro Magalhaes (ICFP 2012). An exploration of what happens when you take equality proofs seriously in a compiler. doipdf
Giving Haskell a promotion, Brent Yorgey, Stepanie Weirich, Julien Cretin, Simon Peyton Jones, and Dimitrios Vytiniotis (TLDI 2012). How to (a) add kind polymorphism and (b) promote data types to become data kinds. doipdf
Evidence Normalization in System FC. Dimitrios Vytiniotis, Simon Peyton Jones. RTA '13. doipdf Explains the coercion optimizer.
System F with Type Equality Coercions, Martin Sulzmann, Manuel Chakravarty, and Simon Peyton Jones (TLDI 2007). The first paper about System FC. doiextended pdf
Associated Types with Class. Manuel M. T. Chakravarty, Gabriele Keller, Simon Peyton Jones, Simon Marlow. POPL '05. doi Introduces associated data families.
Associated Type Synonyms. Manuel M. T. Chakravarty, Gabriele Keller, Simon Peyton Jones. ICFP '05. doi Introduces associated type families.
Secrets of the GHC inliner, Simon Peyton Jones and Simon Marlow, Journal of Functional Programming 12(4), July 2002, pp393-434. Describes how the Simplifier does inlining.
A short cut to deforestation, A Gill, SL Peyton Jones, J Launchbury, Proc Functional Programming Languages and Computer Architecture (FPCA'93), Copenhagen, June 1993, pp223-232. The famous foldr/build rule. Andy's PhD thesis has more.
Constructed Product Result Analysis for Haskell, Clem Baker-Finch, Kevin Glynn, and Simon Peyton Jones, Journal of Functional Programming 14(2), 211–245, March 2004. Describes optimisation that allows to return tuple components in registers (for functions that return tuples).
Demand analysis, Simon Peyton Jones, Peter Sestoft, and John Hughes, draft, 2006.
Arity Analysis, Dana N. Xu and Simon Peyton Jones, Working Notes, 2006
Data Parallel Haskell: a status report, Manuel M. T. Chakravarty, Roman Leshchinskiy, Simon Peyton Jones, Gabriele Keller, and Simon Marlow. , DAMP 2007: Workshop on Declarative Aspects of Multicore Programming, 2007
Vectorisation Avoidance, Gabriele Keller, Manuel M. T. Chakravarty, Roman Leshchinskiy, Ben Lippmeier, and Simon Peyton Jones, Proceedings of ACM SIGPLAN Haskell Symposium 2012, ACM Press, 2012.
Work Efficient Higher-Order Vectorisation, Ben Lippmeier, Manuel M. T. Chakravarty, Gabriele Keller, Roman Leshchinskiy, and Simon Peyton Jones, The 17th ACM SIGPLAN International Conference on Functional Programming, ACM Press, 2012
Runtime Support for Multicore Haskell (Simon Marlow, Simon Peyton Jones, Satnam Singh) In ICFP '09: Proceeding of the 14th ACM SIGPLAN International Conference on Functional Programming, Edinburgh, Scotland, August 2009
Concurrent Haskell, Simon Peyton Jones, Andrew Gordon, Sigbjorn Finne. Deals with the various concurrency constructs in GHC and the Haskell language. E.g., MVars.
Composable Memory Transactions, Tim Harris, Simon Marlow, Simon Peyton-Jones, and Maurice Herlihy. In Proceedings of the tenth ACM SIGPLAN symposium on Principles and practice of parallel programming (PPoPP '05)
Intermediate Representation of GHC (Core & Related)
An External Representation for the GHC Core Language Gives an overview of the semantics and syntax of Core, GHC's internal intermediate representation for Haskell that most of the optimisation work is done on. A good language to understand when starting with GHC.
Unboxed Values as First-Class Citizens, Simon L Peyton Jones and John Launchbury, Conference on Functional Programming Languages and Computer Architecture, September 1991. Describe the design of GHC language and internals for handling machine values and boxing / unboxing them as lazy values.
Faster laziness using dynamic pointer tagging (Simon Marlow, Alexey Rodriguez Yakushev, Simon Peyton Jones) In ICFP '07: Proceedings of the ACM SIGPLAN international conference on Functional programming, Freiburg, Germany, ACM Press, October 2007
Haskell on a Shared-Memory Multiprocessor (Tim Harris, Simon Marlow, Simon Peyton Jones) In Haskell '05: Proceedings of the 2005 ACM SIGPLAN workshop on Haskell, pages 49--61, Tallinn, Estonia, ACM Press, September 2005