GitLab

For platforms with no PIC method defined, use
LabelAccessStyle of AccessDirectly, in howToAccessLabel.
This seems to keep the ncg working on OpenBSD.

Is this the correct default, Wolfgang?

Add missing final #else case for howToAccessLabel.

Wibbles to hi-boot files and newtypes

Position Independent Code and Dynamic Linking Support, Part 1

This commit allows generation of position independent code (PIC) that fully supports dynamic linking on Mac OS X and PowerPC Linux.
Other platforms are not yet supported, and there is no support for actually linking or using dynamic libraries - so if you use the -fPIC or -dynamic code generation flags, you have to type your (platform-specific) linker command lines yourself.


nativeGen/PositionIndependentCode.hs:
New file. Look here for some more comments on how this works.

cmm/CLabel.hs:
Add support for DynamicLinkerLabels and PIC base labels - for use inside the NCG.
needsCDecl: Case alternative labels now need C decls, see the codeGen/CgInfoTbls.hs below for details

cmm/Cmm.hs:
Add CmmPicBaseReg (used in NCG),
and CmmLabelDiffOff (used in NCG and for offsets in info tables)

cmm/CmmParse.y:
support offsets in info tables

cmm/PprC.hs:
support CmmLabelDiffOff
Case alternative labels now need C decls (see the codeGen/CgInfoTbls.hs for details), so we need to pprDataExterns for info tables.

cmm/PprCmm.hs:
support CmmLabelDiffOff

codeGen/CgInfoTbls.hs:
no longer store absolute addresses in info tables, instead, we store offsets.
Also, for vectored return points, emit the alternatives _after_ the vector table. This is to work around a limitation in Apple's as, which refuses to handle label differences where one label is at the end of a section. Emitting alternatives after vector info tables makes sure this never happens in GHC generated code. Case alternatives now require prototypes in hc code, though (see changes in PprC.hs, CLabel.hs).

main/CmdLineOpts.lhs:
Add a new option, -fPIC.

main/DriverFlags.hs:
Pass the correct options for PIC to gcc, depending on the platform. Only for powerpc for now.

nativeGen/AsmCodeGen.hs:
Many changes...
Mac OS X-specific management of import stubs is no longer, it's now part of a general mechanism to handle such things for all platforms that need it (Darwin [both ppc and x86], Linux on ppc, and some platforms we don't support).
Move cmmToCmm into its own monad which can accumulate a list of imported symbols. Make it call cmmMakeDynamicReference at the right places.

nativeGen/MachCodeGen.hs:
nativeGen/MachInstrs.hs:
nativeGen/MachRegs.lhs:
nativeGen/PprMach.hs:
nativeGen/RegAllocInfo.hs:
Too many changes to enumerate here, PowerPC specific.

nativeGen/NCGMonad.hs:
NatM still tracks imported symbols, as more labels can be created during code generation (float literals, jump tables; on some platforms all data access has to go through the dynamic linking mechanism).

driver/mangler/ghc-asm.lprl:
Mangle absolute addresses in info tables to offsets.
Correctly pass through GCC-generated PIC for Mac OS X and powerpc linux.

includes/Cmm.h:
includes/InfoTables.h:
includes/Storage.h:
includes/mkDerivedConstants.c:
rts/GC.c:
rts/GCCompact.c:
rts/HeapStackCheck.cmm:
rts/Printer.c:
rts/RetainerProfile.c:
rts/Sanity.c:
Adapt to the fact that info tables now contain offsets.

rts/Linker.c:
Mac-specific: change machoInitSymbolsWithoutUnderscore to support PIC.

Check for shell scripts first - they don't exist on my Windows nightly build.

Need suffix.mk, otherwise .hc bootstrapping dies in ghc/compiler, with
"No rule to make target Foo.o" error.

Use XML mode, not SGML

Use >> append onto build.mk, so that build scripts can add extra vars
to build.mk.

Replace grep \(a\|b\) with [ab]. \| is an extended regex, whereas the
[] class is a basic RE -- so it works with BSD grep.

Document -ferror-spans

Report bogus strictness annotation on newtype

------------------------------------
	Part-fix an awkward interaction
	between case-of-case and GADTs
	------------------------------------

Consider
	data T a where
	  MkT :: a -> b -> T a

	f = /\a. \(w::a).
	   case (case ...) of
		  MkT a' b (p::a') (q::b) -> [p,w]

The danger is that we'll make a join point

	j a' p = [p,w]

and that's ill-typed, because (p::a') but (w::a).

Solution so far: don't abstract over a', because the type refinement
maps [a' -> a] .  Ultimately that won't work when real refinement goes on.

Then we must abstract over any refined free variables.  Hmm.  Maybe we
could just abstract over *all* free variables, thereby lambda-lifting
the join point?   We should try this.

Sort decls in cycle for error message

Better reporting of duplicate top-level defns

Make combineSrcSpans commutative

Improve error locations for binding conflicts

Warning police. Move prototype of genericRaise into RtsExternal.h, so
the libraries can see it.

Mac OS X:
Make the Linker deal properly with object files that lack a symbol table or
a dynamic symbol table.
Objects that don't import any symbols don't get a dynamic symbol table,
which caused GHCi to crash.

MERGE TO STABLE

Warning police (added missing #include)

Warning police (bring 'raise' into scope)

Synched with latest GHC #include changes: Add '.' to the #include path manually.

Warning police

Improved #include path handling:

* Don't use '-I-', it breaks a lot of system headers, e.g.

     #include <GL/glut.h>

  fails (when using freeglut), because /usr/include/GL/glut.h contains

     #include "freeglut_std.h"

  but /usr/include/GL/freeglut_std.h will not be found. It is a bit
  debatable if the header is broken and should use

     #include "GL/freeglut_std.h"

  instead. Anyway, a grep through the SuSE 9.1 system headers shows that
  there seems to be no real common practice, so let's play safe and don't
  use '-I-'.

* Don't use '-I .', #include stub headers "locally" instead, e.g. use

     #include "Concurrent_stub.h"

  instead of

     #include "Control/Concurrent_stub.h"

  Note that "Control" is still in the #include path, because the *.hc file
  is normally in /tmp and the stub header is in the directory where *.hs
  is. We could remove this path element, too, if the stub header would be
  copied to the directory of the *.hc file during compilation. SimonM?

spelling

Make sure genericRaise symbol is defined on all platforms, so Linker.c
won't complain, even if they don't actually use it.

Deal with Apple's broken cpp 3.3 (you know, the one that inserts useless
#pragma lines into it's output) by filtering it's output through
grep -v.
This shouldn't be a problem for other platforms, and it's much simpler
than to make the parsers for ghc-pkg, genprimopcode and happy ignore
those lines.

MERGE TO STABLE

When linking against libpthreads, raise(3) can behave strangely on
some platforms (OpenBSD at least, maybe other BSDs, not Linux). So use
pthread_kill() to generate signals when running the threaded rts,
instead of raise(), when System.Posix.Signals.raiseSignal is called.

raiseSignal will call genericRaise, in the rts, which knows which
function to use.

Layout rule again...

Allow kind signatures in GADT data type declarations

First-cut documentation for GADTs

A bit more about hi-boot

Comments only

fix dummy ASSERTM2 macro

------------------------------------
	Simplify the treatment of newtypes
	Complete hi-boot file consistency checking
	------------------------------------

In the representation of types, newtypes used to have a special constructor
all to themselves, very like TyConApp, called NewTcApp.    The trouble is
that means we have to *know* when a newtype is a newtype, and in an hi-boot
context we may not -- the data type might be declared as
	data T
in the hi-boot file, but as
	newtype T = ...
in the source file.  In GHCi, which accumulates stuff from multiple compiles,
this makes a difference.

So I've nuked NewTcApp.  Newtypes are represented using TyConApps again. This
turned out to reduce the total amount of code, and simplify the Type data type,
which is all to the good.


This commit also fixes a few things in the hi-boot consistency checking
stuff.

Remove bogus defn in hi-boot

-----------------------------------
	Do simple checking on hi-boot files
	-----------------------------------

This commit arranges that, when compiling A.hs, we compare
the types we infer with those in A.hi-boot, if the latter 
exists.  (Or, more accurately, if anything A.hs imports in
turn imports A.hi-boot, directly or indirectly.)

This has been on the to-do list forever.

Generate in the @size suffix for a stdcall symbol properly.

Replace \? by \{0,1\} in the sed regular expressions, as the former
does not work with Mac OS X's (= BSD's) sed.

I'm a bit confused by the situation here:
According to Mac OS X man pages, the latter syntax is part of the
POSIX 1003.2 standard for basic regular expressions, while the former
is not. According to the manpage installed with my Gentoo linux, neither
is in the POSIX standard.

However, GNU sed accepts both \? and \{0,1\}, while Mac OS X's sed accepts
only \{0,1\}.

Stylistic changes

Get all project info from configure.ac

Make it compile with a GHC 6.x. The layout rule seems to be a mystery
even for one of the "fathers" of Haskell...  :-}