July 2013 - ghc-devs - Haskell.org

foreign import from RTS into GHC source
by Nicolas Frisby 13 Jul '13

13 Jul '13

I'm adding a foreign import in a GHC module from the RTS. I'm using a CPP directive to avoid the import in the stage1 compiler, since the RTS function I need doesn't necessarily exist in that version of the RTS. > #if STAGE < 2 > … -- make do without the import > #else > … -- do the import and use it > #endif Do you have any criticism of this overall design? EG Is there a preference to use Config.cStage instead of CPP and the STAGE symbol? Thanks. P.S. — for some context, this is for Solution 1 at http://ghc.haskell.org/trac/ghc/wiki/Plugins/ReinitializeGlobals

2 1

Cannot log into darcs.haskell.org
by Edward Z. Yang 12 Jul '13

12 Jul '13

Is anyone else seeing this? It seems to just be rejecting my SSH key. Edward

1 0

How accessible is a dynamically-linked ghc?
by Nicolas Frisby 12 Jul '13

12 Jul '13

1) How much effort does it take for a user to install a dynamically-linked ghc executable on Tier 1 platforms? Just download the source and set DYNAMIC_GHC_PROGRAMS=YES? 2) Are the major GHC distributors planning on distributing dynamically-linked ghc by default? GHC HQ, Haskell Platform, http://www.haskell.org/ghc/distribution_packages? Bonus question: 3) How many platforms support Core plugins but not a dynamically linked ghc? I'm hoping the answer is "zero". The availability of a dynamically-linked ghc looks good, according to the wiki pages I found. It's not so clear to me regarding support on Windows. Thank you for your time. P.S. — Here's the wiki pages I read. Their up-to-dateness is not clear to me. Please let me know if you see that I missed something. === Core plugins http://ghc.haskell.org/trac/ghc/wiki/NewPlugins — nub is 2 years old === building and using shared libraries (aka "dynamic libraries") http://ghc.haskell.org/trac/ghc/wiki/Platforms — very much alive and useful http://ghc.haskell.org/trac/ghc/wiki/Status/May13#Dynamiclinking — summarizes GHCI behavior http://ghc.haskell.org/trac/ghc/wiki/SharedLibraries/PlatformSupport — 4 years old http://ghc.haskell.org/trac/ghc/ticket/5620 — good info in comments; 1.5 years old === default behavior of an entire GHC installation http://ghc.haskell.org/trac/ghc/wiki/DynamicByDefault — 8 months old especially http://ghc.haskell.org/trac/ghc/wiki/DynamicByDefault#Windows

4 5

Some more questions about Cmm pipeline
by Jan Stolarek 12 Jul '13

12 Jul '13

Hi Simon, I have two questions about the Cmm pipeline: 1. I implemented a pass that replaces tail calls with a copy of the entry block we're calling to. This is slightly extended version of loopification which will hopefully enable further optimisations. Then I noticed that loopification pass used to be implemented (cmm/CmmOpt.hs, line 418) but was commented out (d92bd17ffd8715f77fd49de0fed6e39c8d0ec28b). There's also a comment: "XXX: revisit if we actually want to do this". What was the motivation for removing loopification? Is there a reason why we might not want to do it? 2. There is a module cmm/CmmRewriteAssignments.hs, which is not used at the moment. What was the motivation for disabling that pass? Janek

3 6

use UniqSupply in FastString?
by Nicolas Frisby 11 Jul '13

11 Jul '13

Does it sound reasonable to change the FastString module to use a UniqSupply instead of using that Int for generating uniques? I've been trying to let a statically-linked compiler shares its FastString table with plugins. Status, background info, options here: http://hackage.haskell.org/trac/ghc/wiki/Plugins/ReinitializeGlobals (I got a little ahead of myself with Option 2…) Uniques for FastStrings are currently allocated linearly using a global Int variable. Because unsafePerformIO is used, it's difficult to keep the two global Ints in synch (one for the compiler, the other for the plugins). The danger is that the compiler and a plugin might allocate the same unique for distinct FastStrings — that'd break a major invariant. If we used UniqSupply, we'd avoid that danger, just about for free. I'm not sure how robust/speedy UniqSupply is though. Considering its widespread use, I figured it'd be good enough by a pretty wide margin; FastString *creation* seems relatively infrequent. Thanks for your input.

4 8

Feature Request : Qualified module exports
by Erik de Castro Lopo 10 Jul '13

10 Jul '13

Hi all, I've added a feature request ticket to the GHC trac: http://ghc.haskell.org/trac/ghc/ticket/8043 for qualified module exports. In a nutshell this would allow the following 'qualified module' export eg: module MyModule ( qualified module T ) import qualified Data.Text as T so that any module which imports MyModule has access to the symbols of Data.Text under the quaified module name T. This feature actually looks like something which I'm capable of undertaking so if nobody thinks this is a really bad idea, I'll have a crack at this. Cheers, Erik -- ---------------------------------------------------------------------- Erik de Castro Lopo http://www.mega-nerd.com/

6 7

adding avx register support to x86_32 as well as x86_64?
by Carter Schonwald 08 Jul '13

08 Jul '13

Hey All, Is there any reasons *not* to add SSE / AVX register support to the llvm 32 bit x86 ghc calling convention? It looks like (as with x86_64) that adding additional simd registers to the calling convention would be fully backwards compatible with current ghc approach to 32bit function calls, but would allow ghc 32bit (at some future point) to have decent floating point performance when applicable. I'm ok either way, but would be helpful to get some opinions. just to recap, i'm working on getting a patch sorted out for adding AVX support to the ghc x86_64 LLVM calling convention http://ghc.haskell.org/trac/ghc/ticket/8033, and now would be a good time to add anything to the 32bit x86 calling convention thanks -Carter

2 2

Any interest in moving GHC issues / wiki to github?
by Bryan O'Sullivan 06 Jul '13

06 Jul '13

We have a repeated problem with the GHC wiki and issue tracker getting filled with spam, and the server is currently taking about a minute to respond to each attempt to load a web page. I wonder if anyone would be open to us migrating the content to github instead. (Actually, while I'm at it, cloning from github is much faster than from darcs.haskell.org, but that's a fight for another day.)

5 4

Can't build a ghc cloned from github
by Bryan O'Sullivan 05 Jul '13

05 Jul '13

There appears to be a bad ref on the server for the Cabal repo: == running git submodule update error: Unable to find b4d0c0f2846542dc2e2df189fe145a56ac9b30b6 under http://darcs.haskell.org/libraries/Cabal.git Cannot obtain needed object b4d0c0f2846542dc2e2df189fe145a56ac9b30b6 while processing commit e7e7ce1029707a67d26e6dc29de11141734898e3. error: Fetch failed. Unable to fetch in submodule path 'libraries/Cabal' git failed: 256 at ./sync-all line 116.

1 0

Question about the new codegen in GHC
by Jan Stolarek 04 Jul '13

04 Jul '13

Hi Simon, as part of my internship at MSR I'm working on optimizing the new code generator. My starting point is Krzysztof Woś's project "Low-level code optimizations in the Glasgow Haskell Compiler" [1]. Right now I'm trying to figure out which of the described optimisations are currently implemented and which ones are not. Looking at a simple factorial program, which was used throughout Krzysztof's project, I see that the code gets optimized properly. Does this mean that all the optimizations described in that work are implemented? Two things I'm particularily interested in are: * do tail calls always call the entry code, or can it happen that a tail call is done indirectly (e.g. via another label that does immediate jump to the entry code)? * commentary on the wiki says that Common Block Elimination "essentially implements the Adams optimisation, we believe". Is that really the case, or are there situations when two similar blocks could be eliminated by Adams optimization, but will not be eliminated by CBE? Janek [1] http://research.microsoft.com/en-us/um/people/simonpj/tmp/wos-diss-draft.pdf

2 1