[Haskell-cafe] Porting GHC to OSX86?
Reilly Hayes
rfh at reillyhayes.com
Sat May 6 14:05:04 EDT 2006
You'll get a better response to this on the glasgow-haskell-users
list. I'm cross-posting my reply.
I'm brand new to hacking on GHC, but I've been working on this in my
pitifully meagre spare time. The actual expert is Wolfgang Thaller,
but he doesn't seem to be around the lists lately. I was able to
generate hc files on a 386 linux box (actually on a Parallels virtual
machine on my mac, as my linux boxen all run 64 bit linuces). I'll
share my findings so far:
1) STABLE does not have the appropriate code in the Mangler to deal
with Darwin/86. I've been playing with various versions of HEAD.
2) HEAD has gone through a major revision to the directory
structure. The documentation and some of the build processes have
not caught up. Simon Marlow sent a helpful e-mail to the ghc users
list a few days ago that you should look at.
3) Building the .hc files mostly requires the appropriate settings in
mk/build.mk on the Host (linux) machine. I'll include my build.mk
below. There is a target in the top level makefile called hc-file-
bundle (which needs to be invoked with the parameter
ProjectNameShort=ghc in HEAD and Project=ghc in STABLE). This target
packages up the .hc files, but does not build them. Some of the .hc
files in utils (genapply, genprimopcode, & ghc-pkg) don't get built.
I just cd to the directories and make them by hand (be sure to use
the in-place compiler). A GhcUtilsHcOpts variable in the make
structure would be nice (in order to pass -keep-hc-files to ghc when
building these on the host).
4) I have been working with Registerised hc files. This may have
been a mistake, as registerised code seems to present some unique
challenges on Darwin/86. See the items below for a discussion.
5) If I understand correctly (somebody with better knowledge please
correct me), there is a register allocation conflict between ghc and
relocatable code generated by gcc on the 386 (gcc flag -fPIC). This
limits ghc to producing static binaries. The gcc in Xcode builds
relocatable code by default and requires -static to build static
binaries.
I think this conflict is limited to code that goes through the
Mangler (registerized code).
6) If I understand correctly, there is code in the RTS that cannot be
built using the native code generator. Which suggests that we're
stuck with static binaries. There is a ticket to fix this in HEAD.
7) Mac OS X really doesn't like static binaries. In fact, in order
to link a static binary, you have to go to opendarwin and download
the Csu package to build crt0.o. It's not included in any of the
development tools. Apple warns that static binaries are likely to
fail to operate in O/S version changes.
Curently I'm fighting with the Makefiles to figure out how to get the
-static flag stuffed into all of the invocations of gcc. Some of the
invocations in rts components don't seem to obey the normal variables
used in the makefile structure. I haven't had time to puzzle this
out and won't for a few days.
mk/build.mk used to generate hc files:
GhcLibHcOpts = -O -fasm -keep-hc-files
GhcRtsHcOpts = -fasm -keep-hc-files
GhcWithInterpreter = NO
GhcStage1HcOpts = -O
GhcStage2HcOpts = -O -fasm -keep-hc-files
SRC_HC_OPTS += -H32m
-reilly hayes
On May 5, 2006, at 7:34 PM, Scott Weeks wrote:
> Hi All,
>
> Does anyone know if there's been any headway on this? If there's not
> a port available, where do I go about finding the hc files? Could I
> compile on a windows or linux x86 box and use the generated hc files
> to bootstrap?
>
> Cheers,
> Scott
>
> On 22/03/2006, at 7:09 AM, Deling Ren wrote:
>
>> Hi there,
>>
>> Has anyone made any attempt to port GHC to Mac OS X on x86?
>> Wolfgang Thaller’s binary package runs over Rosetta but slow (not
>> surprising). It can not be used to compile a native version either
>> (I got some errors related to machine registers).
>>
>> I tried to do a bootstrap but can't find the ".HC" files mentioned
>> in the manual. They don't seem to be on the download page of GHC.
>> Any ideas?
>>
>> Thanks.
>>
>> Deling_______________________________________________
>> Haskell-Cafe mailing list
>> Haskell-Cafe at haskell.org
>> http://www.haskell.org/mailman/listinfo/haskell-cafe
>>
>>
> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe at haskell.org
> http://www.haskell.org/mailman/listinfo/haskell-cafe
Reilly Hayes
rfh at reillyhayes.com
More information about the Haskell-Cafe
mailing list