GitLab

To improve performance of StablePtr.

We were doing it in two different ways and asserting that the results
were the same.  In most cases they were, but I found one case where
they weren't: the GC itself allocates some memory for running
finalizers, and this memory was accounted for one way but not the
other.

It was simpler to remove the old way of counting allocation that to
try to fix it up, so I did that.

An earlier attempt at cleaning things up had made the debug way
compile with optimisation due to flag ordering issues.

ioManagerCapabilitiesChanged now queries getNumCapabilities for the current number of enabled capabilities.

This enables the IO manager to change the number of IO loops it uses (usually one per capability).

Fixes the following crash:

  internal error: threadStackUnderflow: not enough space for return values

when using STM.

MSDN says of Sleep:
    If dwMilliseconds is greater than one tick but less than two, the
    wait can be anywhere between one and two ticks, and so on.
so we need to add (milliseconds-per-tick - 1) to the amount of time we
sleep for.

We were using "us" elsewhere, so this was inconsistent.

Vector values are now always passed on the stack. This isn't particularly
efficient, but it will have to do for now.

Previously, threads blocked on an STM retry would be sent a wakeup
message each time an unpark was requested. This could result in the
accumulation of a large number of wake-up messages, which would slow
wake-up once the sleeping thread is finally scheduled.

Here, we introduce a new closure type, STM_AWOKEN, which marks a TSO
which has been sent a wake-up message, allowing us to send only one
wakeup.

The bug where TSOs were unconditionally kept on the mutable list was #1589
which was fixed in 04cddd33.
Curiously enough, the commit that changed this comment
0417404f occurred *after* this
change was made; I can only assume Simon Marlow accidentally forgot
that he had fixed this bug. :-)
Signed-off-by: Edward Z. Yang <ezyang@mit.edu>

Submitted by: Karel Gardas <karel.gardas@centrum.cz>

Submitted by: Karel Gardas <karel.gardas@centrum.cz>

Reordering of includes in GC.c broke on OS X because gctKey is
declared in Task.h and is needed in the storage manager.  This is
really the wrong place for it anyway, so I've moved the gctKey pieces
to where they should be.

Fixes signals004(threaded1,threaded2) on OS X 32.

Fixes T3807 on OS X 32.

This adds some new functions: peekRunQueue, promoteInRunQueue,
singletonRunQueue and truncateRunQueue which help abstract away
manual linked list manipulation, making it easier to swap in
a new queue implementation.
Signed-off-by: Edward Z. Yang <ezyang@mit.edu>

This fixes dynamic library resolution when --enable-new-dtags is used
(#7062).

When --enable-new-dtags is used when linking an executable, a RUNPATH as
well as RPATH is set. The linker then ignores RPATH, and RUNPATH is only
used for directly (not transitively) needed libraries. As the program
doesn't directly need libffi, it isn't found.

This reverts commit f184d9ca.

The next commit will fix it in a better way.

Exposed by the following snippet, courtesy of Bas van Dijk and Patrick
Palka on libraries@haskell.org:

import Control.Concurrent.STM
main = do
  x <- atomically $ do
         t <- newTVar 1
         writeTVar t 2
         ((readTVar t >> retry) `orElse` return ()) `orElse` return ()
         readTVar t
  print x

Based on a patch from Thorkil Naur.

Based on patch from Peter Trommler:

    From 293495d40f62e691520331a41c6d85d82e120169 Mon Sep 17 00:00:00 2001
    From: Peter Trommler <ptrommler@acm.org>
    Date: Sun, 21 Oct 2012 18:47:01 +0200
    Subject: [PATCH] Add configure option to use system provided libffi This
     fixes track # 5743 and #4496.