GitLab

More Mac OS X-specific gcc 3.3 fixes

merge to stable

Warning police

Warning police

Added prototypes for __hscore_[gs]et_saved_termios to make Linker.c
happy and avoid warnings in RtsStartup.c

Saving/restoring termios is only done for GHC

Update Linker following changes to saved_termios.

Better error trace when cat_evals crashes

To fix dynamic linking issues with Hugs, access saved_termios via
getter/setter.

Add saved_termios symbol

remove docs for -fwith option

The concensus seems to be that 'with' should go away now, after its
customary period of deprecation.  Hugs has already removed it, so
we're following suit.

Move forkOS_createThread into the RTS so its implementation can be
dependent on RTS_SUPPORTS_THREADS, which means we can provide a stub
implementation in the !RTS_SUPPORTS_THREADS case, and hence not depend
on pthread_create, which requires -lpthread.  The upshot is that GHCi
now works again when !RTS_SUPPORTS_THREADS.

If we change the terminal settings as a result of hSetBuffering or
hSetEcho, then restore them again in hs_exit().  This is just good
citizenship on Unixy platforms.

We *don't* just automatically save the terminal settings and restore
them at the end, because that would prevent implementing stty-like
programs in Haskell.  This scheme is a compromise that hopefully DTRT
in most cases.

wibble

If the user explicitly did an instances-only import, i.e.,
"import Foo ()", don't emit an unused-import warning.

Mac OS X uses .dylib instead of .so, recognize that properly.

Change some statics to dynamics, and remove a couple of obsolete options.

Some wibbles to the optimisation section.

Add a BF_PINNED block flag, and attach it to blocks containing pinned
objects (in addition to the usual BF_LARGE).

In heapCensus, we now ignore blocks containing pinned objects, because
they might contain gaps, and in any case it isn't clear that we want
to include the whole block in a heap census, because much of it might
well be dead.  Ignoring it isn't right either, though, so this patch
just fixes the crash and leaves a ToDo.

Fix for clearing the evacuated flag on a block: don't throw away other
flag settings in the process.

--------------------------
         Much grunting about let-floating
	   --------------------------

We want to avoid putting bindings between the '=' of a defn and a '\':
	let { f = let ... in \y-> ... } in ...

Reason: float-in often follows float-out, and it may then add yte
more bindings there, some of which may be strict.  But f may by
not be marked as not-demanded (for other reasons: see the call to
zapDemandInfo in Simplify.completeLazyBind); and now the strict binding
may not be able to float out again.  (Well, it triggers the ASSERT in
simplLazyBind.)

So this commit adds FloatOut.floatNonRecRhs (to complement floatRhs) which
is a big more vigorous about floating out.

But that in turn showed up a pile of gore to do with unlifted bindings.
We can't have them showing up at top level.  After thrashing in the swamp
for a while, I eventually arranged that
	let x# = e in b
(where x# has an unlifted type) is treated exactly like
	case e of x# -> b
That is, it is never floated.  Yes, we lose opportunities to float some
(very cheap!  unlifted let-bindings are always cheap) out of a lambda,
but we're missing much bigger opportunities already.  For example:
	\x -> f (h y)

where h :: Int -> Int# is expensive. We'd like to float the (h y) outside
the \x, but we don't because it's unboxed.  Possible solution: box it.
Anyway, that's for the future.

--------------------------
     Move demand-zapping code to where it belongs
	   --------------------------

A rather subtle point in the simplifier concerns the zapping of demand-info
when the RHS of a binding is a value.  This used to be tucked away inside
IdInfo where it was hard to find.  This commit moves the code to Simplify,
so it occurs where you'd look for it.  Along with copious comments.

See the zapDemandInfo in Simplify.completeLazyBind

Add more informative assertion

Comment only

Remove redundant where

--------------------------
     Make MachString literals a bit bigger
	   --------------------------

Up to now, unboxed string literals of up to 3 characters had
size 1, which means they are inlined in place of a variable.
That seems over-eager (duplication), so I've upped their size a bit.

Fixes for the DEBUG case.

- Convert many of the optimisation options into dynamic options (that is,
  they can be mentioned in {-# OPTIONS #-} pragmas).

- Add a new way to specify constructor-field unboxing on a selective
  basis.  To tell the compiler to unbox a constructor field, do this:

      data T = T !!Int

  and GHC will store that field unboxed if possible.  If it isn't possible
  (say, because the field has a sum type) then the annotation is ignored.

  The -funbox-strict-fields flag is now a dynamic flag, and has the same
  effect as replacing all the '!' annotations with '!!'.

Allow .so or .dll libraries to be named explicitly on the command line.

Bound Threads
=============

Introduce a way to use foreign libraries that rely on thread local state
from multiple threads (mainly affects the threaded RTS).

See the file threads.tex in CVS at haskell-report/ffi/threads.tex
(not entirely finished yet) for a definition of this extension. A less formal
description is also found in the documentation of Control.Concurrent.

The changes mostly affect the THREADED_RTS (./configure --enable-threaded-rts),
except for saving & restoring errno on a per-TSO basis, which is also necessary
for the non-threaded RTS (a bugfix).

Detailed list of changes
------------------------

- errno is saved in the TSO object and restored when necessary:
ghc/includes/TSO.h, ghc/rts/Interpreter.c, ghc/rts/Schedule.c

- rts_mainLazyIO is no longer needed, main is no special case anymore
ghc/includes/RtsAPI.h, ghc/rts/RtsAPI.c, ghc/rts/Main.c, ghc/rts/Weak.c

- passCapability: a new function that releases the capability and "passes"
  it to a specific OS thread:
ghc/rts/Capability.h ghc/rts/Capability.c

- waitThread(), scheduleWaitThread() and schedule() get an optional
  Capability *initialCapability passed as an argument:
ghc/includes/SchedAPI.h, ghc/rts/Schedule.c, ghc/rts/RtsAPI.c

- Bound Thread scheduling (that's what this is all about):
ghc/rts/Schedule.h, ghc/rts/Schedule.c

- new Primop isCurrentThreadBound#:
ghc/compiler/prelude/primops.txt.pp, ghc/includes/PrimOps.h, ghc/rts/PrimOps.hc,
ghc/rts/Schedule.h, ghc/rts/Schedule.c

- a simple function, rtsSupportsBoundThreads, that returns true if THREADED_RTS
  is defined:
ghc/rts/Schedule.h, ghc/rts/Schedule.c

- a new implementation of forkProcess (the old implementation stays in place
  for the non-threaded case). Partially broken; works for the standard
  fork-and-exec case, but not for much else. A proper forkProcess is
  really next to impossible to implement:
ghc/rts/Schedule.c

- Library support for bound threads:
    Control.Concurrent.
      rtsSupportsBoundThreads, isCurrentThreadBound, forkOS,
      runInBoundThread, runInUnboundThread
libraries/base/Control/Concurrent.hs, libraries/base/Makefile,
libraries/base/include/HsBase.h, libraries/base/cbits/forkOS.c (new file)

Generalise the x86-64 hack to all 64-bit arches.

Alpha fixes.

Change IF_OS_darwin to IF_ARCH_powerpc so addImportNat is imported in
exactly the case in which it is used.

Add cases for:

arm-unknown-linux
hppa-unknown-linux
m68k-unknown-linux
mips-unknown-linux
s390-ibm-linux

Re-arrange the interface to TcMatches to allow typechecking of case
commands (part of arrow notation):

* replace the export of the internal tcGRHSs with a more specific
  tcGRHSsPat for checking PatMonoBinds.

* generalize match contexts in the same way as stmt contexts, to include
  a typechecker for the bodies of alts.

This should probably be reviewed, but I hope it can make it into STABLE
after a while.

fixes to desugaring of arrow notation:

* fix free variable calculation for if's

* various fixes for case

please merge to STABLE

POSIX threads should be pthread_detach-ed after they are created.
Otherwise they'll stick around waiting for a pthread_join.

(This wasn't really a problem because our worker threads never exit anyway).

The whole family of index<blah>OffForeignPtr# primops has been broken
for some time now.  This commit fixes them.

I'm tempted to get rid of them altogether, since arguably the
withForeignPtr interface subsumes all this.  But if you want to index
off a ForeignPtr in pure code, you have to use unsafePerformIO, which
is hard to optimise away (GHC doesn't do it), so these primops still
have their uses.

Be a bit more honest about how easy it is to port the linker.

Add irix_TARGET_OS support.