GitLab

Remember to initialise stp->bitmap.  Avoids a potential segfault with +RTS -c.

Initialise n_large_blocks and n_scavenged_large_blocks.

Oops, only ASCII alphanumeric characters are allowed to be used
unencoded in C identifiers, but Char.isAlphaNum includes ISO-8851
alphanumeric characters.

Disable debugging machinery which skeaked in in the last commit.

C-side FFI support for Byte/Ptr arrays.

Change "dunno, mate" into something a bit more helpful.

Add support for passing ptr/byte arrays to C.

Fix a bug in gcStablePtrTable().

Disable use of finalisers attached to UnlinkedBCOs, since finalisers
attached to non-atomic objects may run too early :-(

Attach finaliser for malloc'd blocks to the UnlinkedBCOs, not to
linked really-really-really BCOs.  This is because an unlinked BCO
may be copied many times to generated LinkedBCOs before it dies.
Attaching finalisers to linked BCOs could mean multiple free()s on
the same address.

Define allocaBytes in terms of the new pinned ByteArray primitives.
This version is over 6 times faster than the old version using
malloc/free.

update expected output

Remove parse error.  This test actually doesn't work!

Fix parse error and remove ^Ms (an untested test?)

update expected output

semantics of take, drop and splitAt have changed

update expected output

Flag tweaks: +RTS -c now means "enable compaction all the time"
(previously there was no way to do this *and* run without a maximum
heap size).

The heuristics for determining the generation sizes are also slightly
better now.

Use the bytecode generator's monad to keep track of the malloc'd blocks
created for each BCO.  Eventually use this info to generate a finaliser
which is tied to the real, linked BCO

wibble

oops, forgot to thread the link and global_link fields of a large
TSO.  Fixes a potential crash with the compacting collector.

update expected output

lib specifications aren't allowed any more

[ merge into new test suite ]

Remove -H80M options introduced a week ago.

Change some section ids so they aren't as likely to clash with the User Guide.

- Rearrange the FFI documentation: put it all in chapter 8.

- Add a section "Using the FFI with GHC", and add some information about
  the stub files generated by GHC for foreign export dynamic.

Instead of copying the assemblies we need to this directory, write an
app.exe.config file for each program we run.

Remove the last use of unsavouryPerformIO in this module.  What a lot
of hassle.  Gimme a Von Neumann machine any day.

oops, don't set the -M size to the -H size if -M is zero (i.e. turned
off).  Should fix the breakage encountered by last night's nightly build.

"Greetings, earthlings.  Take us to your mutable variables."
^A^K^K
Build the bytecode generator's monad on top of IO, and as a result get
rid of various unsafePerformIOs.

Had a brainwave on the way to work this morning, and realised that the
garbage collector can handle "pinned objects" as long as they don't
contain any pointers.

This is absolutely ideal for doing temporary allocation in the FFI,
because what we really want to do is allocate a pinned ByteArray and
let the GC clean it up later.  So this set of changes adds the
required framework.

There are two new primops:

 newPinnedByteArray# :: Int# -> State# s -> (# State# s, MutByteArr# s #)
 byteArrayContents#  :: ByteArr# -> Addr#

obviously byteArrayContents# is highly unsafe.

Allocating a pinned ByteArr# isn't the default, because a pinned
ByteArr# will hold an entire block (currently 4k) live until it is
garbage collected (that doesn't mean each pinned ByteArr# requires
4k of storage, just that if a block contains a single live pinned
ByteArray, the whole block must be retained).

update expected output

As discussed at HIM in Cambridge, the GHC Commentary is now located in the
main repository.  The idea is that any developer who wants to share his wisdom
with the world-at-large can alter the document.  Obviously, it would be nice
to keep the structure and style of the document as I have written it so far
(admittedly not terribly much).  In particular, please add links to the files
in CVS where appropriate.  Moreover, I have now included a version number in
the title, which should be pumped whenever there are substantial changes;
check the file index.html.

I am planing to set up a cron job that updates the version of the Commentary
on my Web page with the CVS version every night (this is also the version
linked from GHC's documentation page).

Nuke the "Compilation had errors" message again.

Back up previous change, which was not really a fix of any bug, let alone
the bug it seemed to have fixed.

Change the number of threads to spawn in this test from 1000000 to 10000000,
to (more likely) trigger the Alpha StgRun() bug that I just fixed.

Fixed (I hope, and have reason to believe) mysterious segfaulting problem
on the Alpha.

The problem (I hope, and have reason to believe) was that, during a few
instructions, the register saving/restoring code in StgRun() places data
below the current SP value.  (I.e., it should first reserve stack space,
then put data in said space, rather than first put data in, then reserve
the space.)

   "The SP value might be used by the hardware when raising exceptions and
    asynchronous interrupts. It must be assumed that the contents of the stack
    below the current SP value and within the stack for the current thread are
    continually and unpredictably modified, as specified in the _Alpha
    Architecture Reference Manual_, and as a result of asynchronous software
    actions."

   -- Compaq Computer Corporation, Houston. Tru64 UNIX Calling Standard for
      Alpha Systems, 5.1 edition, August 2000, section 3.2.1.  http://www.
      tru64unix.compaq.com/docs/base_doc/DOCUMENTATION/V51_PDF/ARH9MBTE.PDF

Fix 32-bit assumptions in debugging fprintf's.

For now, let mkMarshalCode_wrk = undefined for non-i386 architectures.

Rewrite the machinery for pushing args to CCalls so that it can
suitably mangle those :: ForeignObj# and ByteArray# and PtrArray#.