GitLab

Noticed by Denys Rtveliashvili <rtvd@mac.com>, see #4004

Our END block was calling system, which alters $?. So now we save and
restore it.

This patch also fixes ullong_format_string (renamed to showStgWord64)
so that it works with values outside the 32bit range (trac #3979), and
simplifies the without-commas case.

Avoid using seq, as FreeBSD has jot instead.

The tso->block_info field is now overwritten by pushOnRunQueue(), but
stg_block_async_info was assuming that it still held a pointer to the
StgAsyncIOResult.  We must therefore save this value somewhere safe
before putting the TSO on the run queue.

I once saw a failure that I think was due to a race on
io_manager_event, this should fix it.

Some of the problems highlighted in trac #3968.

I was assuming that the unifer only unified types of the 
same kind, but now we can "defer" unsolved constraints that
invariant no longer holds.  Or at least is's more complicated
to ensure.  

This patch takes the path of not assuming the invariant, which
is simpler and more robust.  See
Note [Mismatched type lists and application decomposition]

In fixing this I did the usual little bit of refactoring

This stops us having to worry about sh/sed/... portability.

They are now guaranteed to be YES

IND_STATIC used to be an error, but at the moment it can happen
as isAlive doesn't look through IND_STATIC as it ignores static
closures. See trac #3956 for a program that hit this error.

The instances (and deriving declarations) have been taken from the ghc-syb
package.

These are no longer used: once upon a time they used to have different
layout from IND and IND_PERM respectively, but that is no longer the
case since we changed the remembered set to be an array of addresses
instead of a linked list of closures.

The list of threads blocked on an MVar is now represented as a list of
separately allocated objects rather than being linked through the TSOs
themselves.  This lets us remove a TSO from the list in O(1) time
rather than O(n) time, by marking the list object.  Removing this
linear component fixes some pathalogical performance cases where many
threads were blocked on an MVar and became unreachable simultaneously
(nofib/smp/threads007), or when sending an asynchronous exception to a
TSO in a long list of thread blocked on an MVar.

MVar performance has actually improved by a few percent as a result of
this change, slightly to my surprise.

This is the final cleanup in the sequence, which let me remove the old
way of waking up threads (unblockOne(), MSG_WAKEUP) in favour of the
new way (tryWakeupThread and MSG_TRY_WAKEUP, which is idempotent).  It
is now the case that only the Capability that owns a TSO may modify
its state (well, almost), and this simplifies various things.  More of
the RTS is based on message-passing between Capabilities now.

Avoids screeds of "event buffer overflowed; event dropped" in
conc059(threaded1).

This fixes #3838, and was made possible by the new BLACKHOLE
infrastructure.  To allow reording of the run queue I had to make it
doubly-linked, which entails some extra trickiness with regard to
GC write barriers and suchlike.