Namely in T13719 and T13701.

When `readMVar` was implemented using `takeMVar` and `putMVar`,
we needed to use `modifyMVarMasked` in `readChan` just in case
the `readMVar` was interrupted between taking and putting. Now
that `readMVar` uses an atomic primop, this is impossible, so we can
safely unmask `readMVar`.

Reviewers: hvr, bgamari, simonmar

Reviewed By: simonmar

Subscribers: rwbarton, thomie, carter

Differential Revision: https://phabricator.haskell.org/D4738

Makes it possible to print STACK (StgStack) objects easily in gdb

Summary:
Another round and attempt at getting these down to 0.

We really should re-enable the CI and not wait for those cloud based ones.

I've disabled the backpack tests on windows as they are too broad, they test
as much the shell as they do the compiler.

The perf tests have been too long to track down. but the numbers are horrible
but I don't see them getting fixed so just have to accept them.

T9293 has new windows specific output because a Dyn way only flag was added.
This will of course not work on non-Dyn way builds.

Test Plan: ./validate

Reviewers: bgamari, hvr, simonmar

Reviewed By: bgamari

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #15107

Differential Revision: https://phabricator.haskell.org/D4668

Summary:
Fix a number of issues that have broken the 32 bit build.
This makes it build again.

Test Plan: ./validate

Reviewers: hvr, goldfire, bgamari, erikd, simonmar

Reviewed By: bgamari

Subscribers: rwbarton, thomie, carter

Differential Revision: https://phabricator.haskell.org/D4691

Make it clear that max_live_bytes is updated after a major GC whereas
live_bytes is updated after all GCs (including minor collections) and
considers data in uncollected generations as live.

Reviewers: bgamari, simonmar, hvr

Reviewed By: bgamari

Subscribers: rwbarton, thomie, carter

Differential Revision: https://phabricator.haskell.org/D4734

This data type witnesses the lifting of a monoid into an applicative
pointwise.

This appears to have been fixed at some point between GHC 8.0 and
8.2.

Commit 433b80de fixed #14172. Let's
add a regression test to ensure that it stays fixed.

Ben Gamari authored 6 years ago and Ömer Sinan Ağacan committed 6 years ago

Test Plan: Validate

Reviewers: int-index, osa1

Reviewed By: osa1

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #15038

Differential Revision: https://phabricator.haskell.org/D4723

This should have been part of D4477, but got missed.

Summary:
(re-applying this patch now that D4659 is committed)

Space leaks in GHCi emerge from time to time and tend to come back again
after they get fixed. This is an attempt to limit regressions by

* adding a reliable detection for some classes of space leaks in GHCi
* turning on leak checking for all GHCi tests in the test suite, so that
  we'll notice if the leak appears again.

The idea for detecting space leaks is quite simple:

* find some data that we expect to be GC'd later, make a weak pointer to it
* when we expect the data to be dead, do a `performGC` and then check
  the status of the weak pointer.

It would be nice to apply this trick to lots of things in GHC,
e.g. ensuring that HsSyn is not retained after the desugarer, or
ensuring that CoreSyn from the previous simplifier pass is not retained.

Test Plan: validate

Reviewers: bgamari, simonpj, erikd, niteria

Subscribers: thomie, carter

GHC Trac Issues: #15111

Summary:
The way we were pretty-printing conflicting data family
instances in an error message was far from ideal:

1. If a data type had no constructors, it would print an equals sign
   with nothing to the right of it.
2. It would try to print GADTs using Haskell98 syntax.
3. It eta-reduced away some type variables from the LHS.

This patch addresses these three issues:

1. We no longer print constructors at all in this error message.
   There's really no reason to do so in the first place, since
   duplicate data family instances always conflict, regardless of
   their constructors.
2. Since we no longer print constructors, we no longer have to
   worry about whether we're using GADT or Haskell98 syntax.
3. I've put in a fix to ensure that type variables are no longer
   eta-reduced away from the LHS.

Test Plan: make test TEST=T14179

Reviewers: goldfire, bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #14179

Differential Revision: https://phabricator.haskell.org/D4711

Summary:
Template Haskell provides a wormhole through which you can
sneak methods that don't belong to a class into an instance for that
class, bypassing the renamer's validity checks. The solution adopted
here is to mirror the treatment for associated type family instances,
which have an additional check in the typechecker which catch
mismatched associated type families that were snuck through using
Template Haskell. I've put a similar check for class methods into
`tcMethods`.

Test Plan: make test TEST=T12387

Reviewers: bgamari, simonpj

Reviewed By: bgamari, simonpj

Subscribers: simonpj, rwbarton, thomie, carter

GHC Trac Issues: #12387

Differential Revision: https://phabricator.haskell.org/D4710

Sadly I can't easily determine the cause of T13701's regression since the tree
was broken.

Ben Gamari authored 6 years ago and Ben Gamari committed 6 years ago

Unfortunately, this optimisation is infeasible on MachO platforms (e.g.
Darwin) due to an object format limitation. Specifically, linking fails
with errors of the form:

     error: unsupported relocation with subtraction expression, symbol
     '_integerzmgmp_GHCziIntegerziType_quotInteger_closure' can not be
     undefined in a subtraction expression

Apparently MachO does not permit relocations' subtraction expressions to
refer to undefined symbols. As far as I can tell this means that it is
essentially impossible to express an offset between symbols living in
different compilation units. This means that we lively can't use this
optimisation on MachO platforms.

Test Plan: Validate on Darwin

Reviewers: simonmar, erikd

Subscribers: rwbarton, thomie, carter, angerman

GHC Trac Issues: #15169

Differential Revision: https://phabricator.haskell.org/D4715

Summary:
I had good intentions, but they were not being followed. In particular,
this comment:

```
---  - we never resolve a reference to a CAF to the contents of its SRT, since
---    the point of SRTs is to keep CAFs alive.
```

was not true, because we updated the srtMap after generating the SRT
for a CAF. Therefore it was possible for another CAF to refer to an
earlier CAF, and the reference to the earlier CAF would be shortcutted
to refer to its SRT instead of pointing to the CAF itself.

The fix is just to not update the srtMap when generating the SRT for a
CAF, but I also refactored the code and comments around this to be a bit
better organised.

Test Plan: Harbourmaster

Reviewers: bgamari, michalt, simonpj, erikd

Subscribers: rwbarton, thomie, carter

GHC Trac Issues: #15173, #15168

Differential Revision: https://phabricator.haskell.org/D4721

Trac #15122 turned out to be interesting.

* Were calling dischargeFmv in three places.

* In all three cases we dealt with the Given case
  separately.

* In two of the three cases the Given code was right,
  (albeit duplicated).

* In the third case (in TcCanonical.canCFunEqCan), we had
     ; case flav of
         Given -> return () -- nothing more to do.
  which was utterly wrong.

The solution is easy: move the Given-case handling into
dischargeFmv (now reenamed dischargeFunEq), and delete it
from the call sites.

Result: less code, easier to understand (dischargeFunEq handles
all three cases, not just two out of three), and Trac #15122 is fixed.

This is very much a corner case, but Trac #15163 showed
that if you have a RULE like
    forall x. f (g x) = ..x..

and g = undefined, then the simplifier is likely to discard
that 'x' argument. It is usually right to do so; but not here
because then x is used on the right but not bound on the left.

The fix is a narrow one, aimed at this rather pathalogical case.
See Note [Do not expose strictness if sm_inline=False] in
SimplUtils.

isReflCoVar_maybe is called, by CoreLint, on all sorts of
Vars (tyvars, term vars, coercion vars).  But it was silently
assuming that it was always called on a CoVar, and as a result
could crash fatally.  This is the immediate cause of the panic
in Trac #15163.

It's easy to fix.

NB: this does not completely fix Trac #15163; more to come

In TcMType.writeMetaTyVarRef we have an assertion
check, level_check_ok, that the type being written
to a unification variable is not deeper than the level
of the unification varaible itself.

This check used to have a special case for fmv/fsk
flatten vars, but this commit changed fmv/fsks to have
an ordinary level number:

    commit 2bbdd00c
    Author: Simon Peyton Jones <simonpj@microsoft.com>
    Date:   Fri May 18 08:43:11 2018 +0100

    Orient TyVar/TyVar equalities with deepest on the left

So we can delete the isFlattenTyVar special case from
the level_check_ok assertion.  Simpler, less ad hoc.

the hot path contained a call to

    v `elemUnVarSet` (neighbors g v)

and creating the set of neighbors just to check if `v` is inside
accounted for half the allocations of the test case of #15164.

By introducing a non-allocating function `hasLoopAt` for this we shave
off half the allocations. This brings the total cost of Call Arity down
to 20% of time and 23% of allocations, according to a profiled run. Not
amazing, but still much better.

Differential Revision: https://phabricator.haskell.org/D4718

This reverts commit 1cdc14f9.

This is causing non-deterministic testsuite output.

This patch adds a check for type families to the instance-decl
termination check.  See Note [Type families in instance contexts]
and Trac #15172.

A Derived CFunEqCan does not "own" its FlatMetaTv (fmv), and should not
update it.  But one caller (canCFunEqCan) was failing to satisfy the
precondition to dischargeFmv, which led to a crash (Trac #15170).

I fixed this by making dischargeFmv handle Deriveds (to avoid forcing
each caller to do so separately).

NB: this does not completely fix the original #15170 bug, but I'll
explain that on the ticket.  The test case for this patch is actually
the program in comment:1.

In the olden days we insisted that only TcTyVars could appear
in a TcType.  But now we are more accommodating; see TcType
  Note [TcTyVars and TyVars in the typechecker]

This patch removes a function that converted a Type to a TcType.
It didn't do anything useful except statisfy an invariant that
we no longer have.  Now it's gone.

Trac #15164 showed that it sometimes really matters to share
sub-proofs when solving constraints.  Without it, we can get
exponentialy bad behaviour.

Fortunately, it's easily solved.
Note [Shortcut try_solve_from_instance] explains.

I did some minor assocaited refactoring.