- ghc-devs - Haskell.org

Vector registers assumed to be caller or callee-saved?
by Stefan Schulze Frielinghaus 29 Jun '19

29 Jun '19

Hi all, I'm wondering what GHC assumes about vector registers XMMi, YMMi, and ZMMi used by the STG machine: are those assumed to be caller or callee-saved? Only for the x86-64 architecture there exist macro definitions like CALLER_SAVES_XMM1 in includes/stg/MachRegs.h. However, I cannot find any other place where those macros are used. AFAIK most C ABIs assume that vector registers are call clobbered. Is this also the case for GHC? Many thanks in advance, Stefan

1 0

Re: Guarded Impredicativity
by Ryan Scott 28 Jun '19

28 Jun '19

Would this permit explicit impredicativity as described in [1]? Simon mentions in [1] that explicit impredicativity is easier to implement than guarded impredicativity, although it's not clear to me if the latter would imply the former. Ryan S. ----- [1] https://gitlab.haskell.org/ghc/ghc/issues/14859

2 1

The future of the SPARC NCG backend
by Ben Gamari 27 Jun '19

27 Jun '19

Hello everyone, If you use (or think you might use in the future) GHC's SPARC NCG backend please do leave a note on #16882 [1]. My impression is that it has no users and no plausible means of testing. Consequently I am suggesting that we remove it in GHC 8.12. Cheers, - Ben [1] https://gitlab.haskell.org/ghc/ghc/issues/16882

1 0

Preemption a haskell thread inside a Transaction
by Rodrigo Duarte 26 Jun '19

26 Jun '19

Hi I am trying to modify the GHC rts so that every time a transaction fails to commit, instead of automatically restart it, I want to preempted and add it to the end of the task queue of its HEC. The problem is, when I try to call yield at the point of the ATOMICALLY_FRAME code where the transaction is restarted, I get this error message: >>Main: Control.Concurrent.STM.atomically was nested >>Main: thread blocked indefinitely in an MVar operation The point in the code of PrimOps.cmm I try to put my code is as below. PrimOps.cmm 1074 /* Transaction was not valid: try again */ 1075 ("ptr" trec) = ccall stmStartTransaction(MyCapability() "ptr", 1076 NO_TREC "ptr"); 1077 StgTSO_trec(CurrentTSO) = trec; 1078 1079 /* call yield function here */ Capability_context_switch(MyCapability()) = 1 :: CInt; jump stg_yield_noregs(); /* */ 1080 jump stg_ap_v_fast 1081 // push the StgAtomicallyFrame again: the code generator is 1082 // clever enough to only assign the fields that have changed. 1083 (ATOMICALLY_FRAME_FIELDS(,,info_ptr,p1,p2, 1084 code,frame_result)) 1085 (code); My doubts are: Is it possible to make a preemption at that point in the ATOMICALLY_FRAME code?. If it is possible, how do I proceed? Thanks for the attention. Best regards. -- *Rodrigo M. Duarte* Bacharelado em Engenharia de Computação LUPS - Laboratory of Ubiquitous and Parallel Systems Computação de Alto Desempenho Sustentável Universidade Federal de Pelotas - UFPel [image: image_png_part?hl=pt&part=2&thumb=1] *Antes de imprimir este e-mail pense em sua responsabilidade e compromisso com o Meio Ambiente. **Só imprima se for realmente necessário. * *Before printing this e-mail consider its responsibility and commitment to the environment. Only print if it is really necessary. *

1 0

Cloning
by Simon Peyton Jones 25 Jun '19

25 Jun '19

What is the right procedure for cloning from someone else's repo. For example, I wanted to see what was going wrong for Tobias in #16615, so after some detective work I tried git clone --recursive git@gitlab.haskell.org:tdammers/ghc.git tdammers but I got lots of errors: Cloning into '/home/simonpj/code/tdammers/libffi-tarballs'... setsockopt IPV6_TCLASS 8: Operation not permitted: > GitLab: The project you were looking for could not be found. fatal: Could not read from remote repository. Please make sure you have the correct access rights and the repository exists. fatal: clone of 'git@gitlab.haskell.org:tdammers/libffi-tarballs.git' into submodule path '/home/simonpj/code/tdammers/libffi-tarballs' failed Failed to clone 'libffi-tarballs'. Retry scheduled Cloning into '/home/simonpj/code/tdammers/libraries/Cabal'... setsockopt IPV6_TCLASS 8: Operation not permitted: > GitLab: The project you were looking for could not be found. fatal: Could not read from remote repository. Please make sure you have the correct access rights and the repository exists. fatal: clone of 'git@gitlab.haskell.org:tdammers/packages/Cabal.git' into submodule path etc etc. So: * What is a reliable way to get a MR onto my machine to test? * If I make a fix, what is the workflow to push that fix back to the author -- presumably I can't push to their repo? Thanks Simon

3 4

StgRun and reserved C stack space
by Stefan Schulze Frielinghaus 24 Jun '19

24 Jun '19

Hi all, I'm trying to understand how StgRun/StgReturn are supposed to be implemented. My current understanding is that StgRun is called from C land in order to save all callee-saved registers, allocate some stack space, and then finally jump to an STG function. StgReturn is the counterpart of StgRun and is called from STG land in order to restore all callee-saved registers, free the previously allocated stack space, and finally return to C land. Assuming that my intuition is correct I still have some concerns regarding the stack space which is allocated by StgRun. StgRun allocates RESERVED_C_STACK_BYTES many bytes on the C stack. However, I cannot see how an STG function is supposed to access this reserved area since no pointer is derived or whatsoever. Does an STG function make use of the C stack pointer and assume that the reserved area is at the very bottom of the stack? Lets make an example. Assume the C ABI demands that every function allocates a Register Save Area (RSA for short) at the bottom of a stack frame which may be used by a subsequent C function. That means a caller allocates some stack space for its callee. This area must be below the local variable and parameter area which means it is at the very bottom of a stack frame. A C stack frame then looks as follows: . . . . | previous frame | +---------------------+ | local variables | +---------------------+ | parameter area | +---------------------+ | register save area | +---------------------+ <---+ SP Thus StgRun must also allocate a register save area which is used by subsequent C functions. An example stack while calling StgRun could look as follows: . . . . | | +---------------------+ RSA1 | register save area | | used by StgRun | +---------------------+ xxx | | xx | | xx | | xx | | xx | | xxxxxx RESERVED_C_STACK_BYTES | | xx | | xx | | xx | | xx +---------------------+ xxx RSA2 | register save area | | for further C calls | +---------------------+ <---+ SP RSA1 is allocated by the caller of StgRun and is used by StgRun in order to save all callee-saved registers. StgRun itself then allocates RESERVED_C_STACK_BYTES many bytes and RSA2. RSA2 will be used by subsequent C functions which are called from STG land. Thus coming back to my initial question how is an STG function supposed to know where the reserved area starts? One way would be to use SP + len(RSA2) but this would mean an STG function is aware of the register in which the C stack pointer is hold and the length of a register save area. I didn't find anything which suggests this while skimming over the code. Many thanks and kind regards, Stefan

2 1

gitlab.haskell.org down?
by chessai . 24 Jun '19

24 Jun '19

hi devs, I tried doing a fresh clone of ghc with git clone --recursive https://gitlab.haskell.org/ghc/ghc and got Cloning into 'ghc'... warning: redirecting to https://gitlab.haskell.org/ghc/ghc.git/ remote: Enumerating objects: 521560, done. remote: Counting objects: 100% (521560/521560), done. remote: Compressing objects: 100% (107771/107771), done. fatal: the remote end hung up unexpectedly0.39 MiB | 9.50 MiB/s fatal: early EOF fatal: index-pack failed Trying again, with the same command, within a minute afterward, I got: Cloning into 'ghc'... fatal: unable to access 'https://gitlab.haskell.org/ghc/ghc/': The requested URL returned error: 502 I also get a 502 when going to the haskell gitlab page.

3 2

Modify quickstart page to use hadrian rather than make?
by Matthew Pickering 24 Jun '19

24 Jun '19

Hi all, I propose that we modify the quickstart page so that the instructions use hadrian rather than make. At this point I don't see why a newcomer should have to use the make build system. https://gitlab.haskell.org/ghc/ghc/wikis/building/quick-start Cheers, Matt

2 1

RE: Hadrian: Suddenly a full build fails with: Rules may not be recursive
by Andrey Mokhov 23 Jun '19

23 Jun '19

Hi all, > I think this is because `--configure` is currently broken > https://gitlab.haskell.org/ghc/ghc/issues/16809. You can work around by > running bootstrap and configure manually for now. Here is a merge request with a fix: https://gitlab.haskell.org/ghc/ghc/merge_requests/1255 Cheers, Andrey

1 0

Hadrian: Suddenly a full build fails with: Rules may not be recursive
by Roland Senn 21 Jun '19

21 Jun '19

Hi, Suddenly I'm unable to build! As usual, I did: git pull upstream master --recurse-submodules=yes git submodule update --init --recursive $ hadrian/build.sh clean Up to date | Removing Hadrian files... | Remove directory _build/stage0 | Remove directory _build/stage1 | Remove directory _build/stage2 | Remove directory _build/stage3 | Remove directory sdistprep | Remove directory _build (after build completes) | Done. ... + statistics $ hadrian/build.sh -c --flavour=devel2 -j2 Up to date | Running boot... Error when running Shake build system: at action, called at src/Rules.hs:68:19 in main:Rules at apply1, called at src/Development/Shake/Internal/Rules/Oracle.hs:159:32 in shake-0.18.2- fca156899176fbf7a9b05f30b3ec991784b3e60af38a503ea1cb277fd96f971e:Develo pment.Shake.Internal.Rules.Oracle * Depends on: OracleQ (KeyValue ("hadrian/cfg/system.config","host- os")) at need, called at src/Hadrian/Oracles/TextFile.hs:90:9 in main:Hadrian.Oracles.TextFile * Depends on: hadrian/cfg/system.config at &%>, called at src/Rules/Configure.hs:21:5 in main:Rules.Configure * Depends on: hadrian/cfg/system.config mk/config.h compiler/ghc.cabal rts/rts.cabal at apply1, called at src/Development/Shake/Internal/Rules/Oracle.hs:159:32 in shake-0.18.2- fca156899176fbf7a9b05f30b3ec991784b3e60af38a503ea1cb277fd96f971e:Develo pment.Shake.Internal.Rules.Oracle * Depends on: OracleQ (KeyValue ("hadrian/cfg/system.config","host- os")) * Raised the exception: Build system error - recursion detected: Key type: OracleQ KeyValue Key value: OracleQ (KeyValue ("hadrian/cfg/system.config","host- os")) Rules may not be recursive OS is Linux Debian 9 (x86_64) Any ideas what's wrong?? Many thanks and kind regards Roland

4 4