[Git][ghc/ghc][wip/tsan/all] 132 commits: Parser regression tests, close #12862 #12446

Ben Gamari gitlab at gitlab.haskell.org
Sun Nov 8 14:43:30 UTC 2020



Ben Gamari pushed to branch wip/tsan/all at Glasgow Haskell Compiler / GHC


Commits:
e2c4a947 by Vladislav Zavialov at 2020-10-21T16:00:30+03:00
Parser regression tests, close #12862 #12446

These issues were fixed by earlier parser changes, most likely related
to whitespace-sensitive parsing.

- - - - -
711929e6 by Simon Peyton Jones at 2020-10-23T02:42:59-04:00
Fix error message location in tcCheckPatSynDecl

Ticket #18856 showed that we were failing to set the right location
for an error message.  Easy to fix, happily.

Turns out that this also improves the error location in test T11010,
which was bogus before but we had never noticed.

- - - - -
730bb590 by Ben Gamari at 2020-10-23T02:43:33-04:00
cmm: Add Note reference to ForeignHint
- - - - -
9ad51bc9 by David Beacham at 2020-10-27T13:59:35-04:00
Fix `instance Bounded a => Bounded (Down a)` (#18716)

* Flip `minBound` and `maxBound` to respect the change in ordering
* Remove awkward `Enum` (and hence `Integral`) instances for
  `Data.Ord.Down`
* Update changelog

- - - - -
eedec53d by Vladislav Zavialov at 2020-10-27T14:00:11-04:00
Version bump: base-4.16 (#18712)

Also bumps upper bounds on base in boot libraries (incl. submodules).

- - - - -
412018c1 by Tamar Christina at 2020-10-27T14:00:49-04:00
winio: simplify logic remove optimization step.

- - - - -
4950dd07 by Ben Gamari at 2020-10-27T14:01:24-04:00
hadrian: Suppress xelatex output unless it fails

As noted in #18835, xelatex produces an absurd amount of output, nearly
all of which is meaningless. Silence this.

Fixes #18835.

- - - - -
f3d8ab2e by Ben Gamari at 2020-10-27T14:02:00-04:00
build system: Clean mingw tarballs

Tamar noticed in !4293 that the build systems fail to clean up the mingw
tarballs directory (`ghc-tarballs`). Fix this in both the make build
system and Hadrian.

- - - - -
0b3d23af by Simon Peyton Jones at 2020-10-27T14:02:34-04:00
Fix two constraint solving problems

This patch fixes two problems in the constraint solver.

* An actual bug #18555: we were floating out a constraint to eagerly,
  and that was ultimately fatal.  It's explained in
  Note [Do not float blocked constraints] in GHC.Core.Constraint.

  This is all very delicate, but it's all going to become irrelevant
  when we stop floating constraints (#17656).

* A major performance infelicity in the flattener.  When flattening
  (ty |> co) we *never* generated Refl, even when there was nothing
  at all to do.  Result: we would gratuitously rewrite the constraint
  to exactly the same thing, wasting work.  Described in #18413, and
  came up again in #18855.

  Solution: exploit the special case by calling the new function
  castCoercionKind1.  See Note [castCoercionKind1] in
  GHC.Core.Coercion

- - - - -
f76c5a08 by Sergei Trofimovich at 2020-10-27T14:03:14-04:00
ghc.mk: amend 'make sdist'

Noticed 'make sdist' failure seen as:

```
"rm" -rf sdistprep/ghc/ghc-9.1.0.20201020/hadrian/_build/ (SRC_DIST_GHC_DIR)/hadrian/dist-newstyle/
/bin/sh: -c: line 0: syntax error near unexpected token `('
```

commit 9657f6f34
("sdist: Include hadrian sources in source distribution")
added a new cleanup path without a variable expantion.

The change adds variable reference. While at it move directory
cleanup to a separate statement.

Amends #18794

Signed-off-by: Sergei Trofimovich <slyfox at gentoo.org>

- - - - -
78b52c88 by David Eichmann at 2020-10-27T14:03:51-04:00
Use config.run_ways for multi_compile_and_run tests

- - - - -
e3fdd419 by Alan Zimmerman at 2020-10-27T14:04:26-04:00
Api Annotations: Introduce AnnPercent for HsExplicitMult

For the case

  foo :: a %p -> b

The location of the '%' is captured, separate from the 'p'

- - - - -
d2a25f42 by Ben Gamari at 2020-10-27T14:05:02-04:00
gitlab-ci: Bump ci-images

Bumps bootstrap compiler to 8.10.1.

- - - - -
28f98b01 by Sebastian Graf at 2020-10-27T14:05:37-04:00
DmdAnal: Kill `is_thunk` case in `splitFV`

The `splitFV` function implements the highly dubious hack
described in `Note [Lazy und unleashable free variables]` in
GHC.Core.Opt.DmdAnal. It arranges it so that demand signatures only
carry strictness info on free variables. Usage info is released through
other means, see the Note. It's purely for analysis performance reasons.

It turns out that `splitFV` has a quite involved case for thunks that
produces slightly different usage signatures and it's not clear why we
need it: `splitFV` is only relevant in the LetDown case and the only
time we call it on thunks is for top-level or local recursive thunks.

Since usage signatures of top-level thunks can only reference other
top-level bindings and we completely discard demand info we have on
top-level things (see the lack of `setIdDemandInfo` in
`dmdAnalTopBind`), the `is_thunk` case is completely irrelevant here.

For local, recursive thunks, the added benefit of the `is_thunk` test
is marginal: We get used-multiple-times in some cases where previously
we had used-once if a recursive thunk has multiple call sites. It's
very unlikely and not a case to optimise for.

So we kill the `is_thunk` case and inline `splitFV` at its call site,
exposing `isWeakDmd` from `GHC.Types.Demand` instead.

The NoFib summary supports this decision:

```
            Min           0.0%     -0.0%
            Max           0.0%     +0.0%
 Geometric Mean          -0.0%     -0.0%
```

- - - - -
60322f93 by Ben Gamari at 2020-10-28T21:11:39-04:00
hadrian: Don't quote metric baseline argument

Previously this was quoted inappropriately.
- - - - -
c85eb372 by Alan Zimmerman at 2020-10-28T21:12:15-04:00
API Annotations: put constructors in alphabetical order

- - - - -
795908dc by John Ericson at 2020-10-29T03:53:14-04:00
Widen acceptance threshold for T10421a

Progress towards #18842. As @sgraf812 points out, widening the window is
dangerous until the exponential described in #17658 is fixed. But this
test has caused enough misery and is low stakes enough that we and
@bgamari think it's worth it in this one case for the time being.

- - - - -
0e9f6def by Sylvain Henry at 2020-10-29T03:53:52-04:00
Split GHC.Driver.Types

I was working on making DynFlags stateless (#17957), especially by
storing loaded plugins into HscEnv instead of DynFlags. It turned out to
be complicated because HscEnv is in GHC.Driver.Types but LoadedPlugin
isn't: it is in GHC.Driver.Plugins which depends on GHC.Driver.Types. I
didn't feel like introducing yet another hs-boot file to break the loop.

Additionally I remember that while we introduced the module hierarchy
(#13009) we talked about splitting GHC.Driver.Types because it contained
various unrelated types and functions, but we never executed. I didn't
feel like making GHC.Driver.Types bigger with more unrelated Plugins
related types, so finally I bit the bullet and split GHC.Driver.Types.

As a consequence this patch moves a lot of things. I've tried to put
them into appropriate modules but nothing is set in stone.

Several other things moved to avoid loops.

* Removed Binary instances from GHC.Utils.Binary for random compiler
  things
* Moved Typeable Binary instances into GHC.Utils.Binary.Typeable: they
  import a lot of things that users of GHC.Utils.Binary don't want to
  depend on.
* put everything related to Units/Modules under GHC.Unit:
  GHC.Unit.Finder, GHC.Unit.Module.{ModGuts,ModIface,Deps,etc.}
* Created several modules under GHC.Types: GHC.Types.Fixity, SourceText,
  etc.
* Split GHC.Utils.Error (into GHC.Types.Error)
* Finally removed GHC.Driver.Types

Note that this patch doesn't put loaded plugins into HscEnv. It's left
for another patch.

Bump haddock submodule

- - - - -
22f5d9a9 by Sylvain Henry at 2020-10-29T03:53:52-04:00
GC: Avoid data race (#18717, #17964)

- - - - -
2ef2fac4 by Ryan Scott at 2020-10-29T04:18:52-04:00
Check for large tuples more thoroughly

This fixes #18723 by:

* Moving the existing `GHC.Tc.Gen.HsType.bigConstraintTuple` validity
  check to `GHC.Rename.Utils.checkCTupSize` for consistency with
  `GHC.Rename.Utils.checkTupSize`, and
* Using `check(C)TupSize` when checking tuple _types_, in addition
  to checking names, expressions, and patterns.

Note that I put as many of these checks as possible in the typechecker so
that GHC can properly distinguish between boxed and constraint tuples. The
exception to this rule is checking names, which I perform in the renamer
(in `GHC.Rename.Env`) so that we can rule out `(,, ... ,,)` and
`''(,, ... ,,)` alike in one fell swoop.

While I was in town, I also removed the `HsConstraintTuple` and
`HsBoxedTuple` constructors of `HsTupleSort`, which are functionally
unused. This requires a `haddock` submodule bump.

- - - - -
7f8be3eb by Richard Eisenberg at 2020-10-29T22:08:13-04:00
Remove unnecessary gender from comments/docs

While, say, alternating "he" and "she" in sequential writing
may be nicer than always using "they", reading code/documentation
is almost never sequential. If this small change makes individuals
feel more welcome in GHC's codebase, that's a good thing.

- - - - -
9902d9ec by Viktor Dukhovni at 2020-10-30T05:28:30-04:00
[skip ci] Fix typo in `callocBytes` haddock.

- - - - -
31fcb55f by Ryan Scott at 2020-10-30T18:52:50-04:00
Split HsConDecl{H98,GADT}Details

Haskell98 and GADT constructors both use `HsConDeclDetails`, which includes
`InfixCon`. But `InfixCon` is never used for GADT constructors, which results
in an awkward unrepresentable state. This removes the unrepresentable state by:

* Renaming the existing `HsConDeclDetails` synonym to `HsConDeclH98Details`,
  which emphasizes the fact that it is now only used for Haskell98-style data
  constructors, and
* Creating a new `HsConDeclGADTDetails` data type with `PrefixConGADT` and
  `RecConGADT` constructors that closely resemble `PrefixCon` and `InfixCon`
  in `HsConDeclH98Details`. The key difference is that `HsConDeclGADTDetails`
  lacks any way to represent infix constructors.

The rest of the patch is refactoring to accommodate the new structure of
`HsConDecl{H98,GADT}Details`. Some highlights:

* The `getConArgs` and `hsConDeclArgTys` functions have been removed, as
  there is no way to implement these functions uniformly for all
  `ConDecl`s. For the most part, their previous call sites now
  pattern match on the `ConDecl`s directly and do different things for
  `ConDeclH98`s and `ConDeclGADT`s.

  I did introduce one new function to make the transition easier:
  `getRecConArgs_maybe`, which extracts the arguments from a `RecCon(GADT)`.
  This is still possible since `RecCon(GADT)`s still use the same representation
  in both `HsConDeclH98Details` and `HsConDeclGADTDetails`, and since the
  pattern that `getRecConArgs_maybe` implements is used in several places,
  I thought it worthwhile to factor it out into its own function.
* Previously, the `con_args` fields in `ConDeclH98` and `ConDeclGADT` were
  both of type `HsConDeclDetails`. Now, the former is of type
  `HsConDeclH98Details`, and the latter is of type `HsConDeclGADTDetails`,
  which are distinct types. As a result, I had to rename the `con_args` field
  in `ConDeclGADT` to `con_g_args` to make it typecheck.

  A consequence of all this is that the `con_args` field is now partial, so
  using `con_args` as a top-level field selector is dangerous. (Indeed, Haddock
  was using `con_args` at the top-level, which caused it to crash at runtime
  before I noticed what was wrong!) I decided to add a disclaimer in the 9.2.1
  release notes to advertise this pitfall.

Fixes #18844. Bumps the `haddock` submodule.

- - - - -
57c3db96 by Ryan Scott at 2020-10-31T02:53:55-04:00
Make typechecker equality consider visibility in ForAllTys

Previously, `can_eq_nc'` would equate `ForAllTy`s regardless of their
`ArgFlag`, including `forall i -> i -> Type` and `forall i. i -> Type`! To fix
this, `can_eq_nc'` now uses the `sameVis` function to first check if the
`ArgFlag`s are equal modulo specificity. I have also updated `tcEqType`'s
implementation to match this behavior. For more explanation on the "modulo
specificity" part, see the new `Note [ForAllTy and typechecker equality]`
in `GHC.Tc.Solver.Canonical`.

While I was in town, I fixed some related documentation issues:

* I added `Note [Typechecker equality]` to `GHC.Tc.Utils.TcType` to describe
  what exactly distinguishes `can_eq_nc'` and `tcEqType` (which implement
  typechecker equality) from `eqType` (which implements definitional equality,
  which does not care about the `ArgFlags` of `ForAllTy`s at all).
* The User's Guide had some outdated prose on the specified/inferred
  distinction being different for types and kinds, a holdover from #15079. This
  is no longer the case on today's GHC, so I removed this prose, added some new
  prose to take its place, and added a regression test for the programs in
  #15079.
* The User's Guide had some _more_ outdated prose on inferred type variables
  not being allowed in `default` type signatures for class methods, which is no
  longer true as of the resolution of #18432.
* The related `Note [Deferred Unification]` was being referenced as
  `Note [Deferred unification]` elsewhere, which made it harder to `grep`
  for. I decided to change the name of the Note to `Deferred unification`
  for consistency with the capitalization style used for most other Notes.

Fixes #18863.

- - - - -
a98593f0 by Sylvain Henry at 2020-10-31T02:54:34-04:00
Refactor numeric constant folding rules

Avoid the use of global pattern synonyms.

1) I think it's going to be helpful to implement constant folding for
   other numeric types, especially Natural which doesn't have a wrapping
   behavior. We'll have to refactor these rules even more so we'd better
   make them less cryptic.

2) It should also be slightly faster because global pattern synonyms
   matched operations for every numeric types instead of the current one:
   e.g., ":**:" pattern was matching multiplication for both Int# and
   Word# types. As we will probably want to implement constant folding
   for other numeric types (Int8#, Int16#, etc.), it is more efficient
   to only match primops for a given type as we do now.

- - - - -
730ef38f by Sylvain Henry at 2020-10-31T02:54:34-04:00
Simplify constant-folding (#18032)

See #18032 for the details.

* Use `Lit (LitNumber _ i)` instead of `isLitValue_maybe` which does
  more work but that is not needed for constant-folding
* Don't export `GHC.Types.Literal.isLitValue_maybe`
* Kill `GHC.Types.Literal.isLitValue` which isn't used

- - - - -
d5a53c1a by Ben Gamari at 2020-10-31T02:55:10-04:00
primops.txt.pp: Move ByteArray# primops to separate file

This file will be generated.

- - - - -
b4278a41 by Ben Gamari at 2020-10-31T02:55:10-04:00
primops: Generate ByteArray# index/read/write primops

Previously these were mostly undocumented and was ripe for potential
inconsistencies.

- - - - -
08e6993a by Sylvain Henry at 2020-10-31T02:55:50-04:00
Move loadDecl into IfaceToCore

- - - - -
cb1f755c by Tamar Christina at 2020-10-31T09:26:56-04:00
winio: Fix unused variables warnings

- - - - -
eb368078 by Andrzej Rybczak at 2020-10-31T09:27:34-04:00
Add testcase for #816

- - - - -
bd4abdc9 by Ben Gamari at 2020-11-01T01:10:31-04:00
testsuite: Add performance test for #18698

- - - - -
dfd27445 by Hécate at 2020-11-01T01:11:09-04:00
Add the proper HLint rules and remove redundant keywords from compiler

- - - - -
ce1bb995 by Hécate at 2020-11-01T08:52:08-05:00
Fix a leak in `transpose`

This patch was authored by David Feuer <david.feuer at gmail.com>

- - - - -
e63db32c by Ben Gamari at 2020-11-01T08:52:44-05:00
Scav: Use bd->gen_no instead of bd->gen->no

This potentially saves a cache miss per scavenge.

- - - - -
4ce2f7d6 by GHC GitLab CI at 2020-11-02T23:45:06-05:00
testsuite: Add --top flag to driver

This allows us to make `config.top` a proper Path. Previously it was a
str, which caused the Ghostscript detection logic to break.

- - - - -
0b772221 by Ben Gamari at 2020-11-02T23:45:42-05:00
Document that ccall convention doesn't support varargs

We do not support foreign "C" imports of varargs functions. While this
works on amd64, in general the platform's calling convention may need
more type information that our Cmm representation can currently provide.
For instance, this is the case with Darwin's AArch64 calling convention.
Document this fact in the users guide and fix T5423 which makes use of a
disallowed foreign import.

Closes #18854.

- - - - -
81006a06 by David Eichmann at 2020-11-02T23:46:19-05:00
RtsAPI: pause and resume the RTS

The `rts_pause` and `rts_resume` functions have been added to `RtsAPI.h` and
allow an external process to completely pause and resume the RTS.

Co-authored-by: Sven Tennie <sven.tennie at gmail.com>
Co-authored-by: Matthew Pickering <matthewtpickering at gmail.com>
Co-authored-by: Ben Gamari <bgamari.foss at gmail.com>

- - - - -
bfb1e272 by Ryan Scott at 2020-11-02T23:46:55-05:00
Display results of GHC.Core.Lint.lint* functions consistently

Previously, the functions in `GHC.Core.Lint` used a patchwork of
different ways to display Core Lint errors:

* `lintPassResult` (which is the source of most Core Lint errors) renders
  Core Lint errors with a distinctive banner (e.g.,
  `*** Core Lint errors : in result of ... ***`) that sets them apart
  from ordinary GHC error messages.
* `lintAxioms`, in contrast, uses a completely different code path that
  displays Core Lint errors in a rather confusing manner. For example,
  the program in #18770 would give these results:

  ```
  Bug.hs:1:1: error:
      Bug.hs:12:1: warning:
          Non-*-like kind when *-like expected: RuntimeRep
          when checking the body of forall: 'TupleRep '[r]
          In the coercion axiom Bug.N:T :: []. Bug.T ~_R Any
          Substitution: [TCvSubst
                           In scope: InScope {r}
                           Type env: [axl :-> r]
                           Co env: []]
    |
  1 | {-# LANGUAGE DataKinds #-}
    | ^
  ```
* Further digging reveals that `GHC.IfaceToCore` displays Core Lint
  errors for iface unfoldings as though they were a GHC panic. See, for
  example, this excerpt from #17723:

  ```
  ghc: panic! (the 'impossible' happened)
    (GHC version 8.8.2 for x86_64-unknown-linux):
          Iface Lint failure
    In interface for Lib
    ...
  ```

This patch makes all of these code paths display Core Lint errors and
warnings consistently. I decided to adopt the conventions that
`lintPassResult` currently uses, as they appear to have been around the
longest (and look the best, in my subjective opinion). We now use the
`displayLintResult` function for all three scenarios mentioned above.
For example, here is what the Core Lint error for the program in #18770 looks
like after this patch:

```
[1 of 1] Compiling Bug              ( Bug.hs, Bug.o )
*** Core Lint errors : in result of TcGblEnv axioms ***
Bug.hs:12:1: warning:
    Non-*-like kind when *-like expected: RuntimeRep
    when checking the body of forall: 'TupleRep '[r_axn]
    In the coercion axiom N:T :: []. T ~_R Any
    Substitution: [TCvSubst
                     In scope: InScope {r_axn}
                     Type env: [axn :-> r_axn]
                     Co env: []]
*** Offending Program ***
axiom N:T :: T = Any -- Defined at Bug.hs:12:1
*** End of Offense ***

<no location info>: error:
Compilation had errors
```

Fixes #18770.

- - - - -
a9e5f52c by Simon Peyton Jones at 2020-11-02T23:47:31-05:00
Expand type synonyms with :kind!

The User's Guide claims that `:kind!` should expand type synonyms,
but GHCi wasn't doing this in practice. Let's just update the implementation
to match the specification in the User's Guide.

Fixes #13795. Fixes #18828.

Co-authored-by: Ryan Scott <ryan.gl.scott at gmail.com>

- - - - -
1370eda7 by Ben Gamari at 2020-11-02T23:48:06-05:00
hadrian: Don't capture RunTest output

There are a few reasons why capturing the output of the RunTest builder
is undesirable:

 * there is a large amount of output which then gets unnecessarily
   duplicated by Hadrian if the builder fails

 * the output may contain codepoints which are unrepresentable in the
   current codepage on Windows, causing Hadrian to crash

 * capturing the output causes the testsuite driver to disable
   its colorisation logic, making the output less legible.

- - - - -
78f2767d by Matthew Pickering at 2020-11-03T17:39:53-05:00
Update inlining flags documentation

- - - - -
14ce454f by Sylvain Henry at 2020-11-03T17:40:34-05:00
Linker: reorganize linker related code

Move linker related code into GHC.Linker. Previously it was scattered
into GHC.Unit.State, GHC.Driver.Pipeline, GHC.Runtime.Linker, etc.

Add documentation in GHC.Linker

- - - - -
616bec0d by Alan Zimmerman at 2020-11-03T17:41:10-05:00
Restrict Linear arrow %1 to exactly literal 1 only

This disallows `a %001 -> b`, and makes sure the type literal is
printed from its SourceText so it is clear why.

Closes #18888

- - - - -
3486ebe6 by Sylvain Henry at 2020-11-03T17:41:48-05:00
Hadrian: don't fail if ghc-tarballs dir doesn't exist

- - - - -
37f0434d by Sylvain Henry at 2020-11-03T17:42:26-05:00
Constant-folding: don't pass through GHC's Int/Word (fix #11704)

Constant-folding rules for integerToWord/integerToInt were performing
the following coercions at compilation time:

    integerToWord: target's Integer -> ghc's Word -> target's Word
    integerToInt : target's Integer -> ghc's Int -> target's Int

1) It was wrong for cross-compilers when GHC's word size is smaller than
   the target one. This patch avoids passing through GHC's word-sized
   types:

    integerToWord: target's Integer -> ghc's Integer -> target's Word
    integerToInt : target's Integer -> ghc's Integer -> target's Int

2) Additionally we didn't wrap the target word/int literal to make it
   fit into the target's range! This broke the invariant of literals
   only containing values in range.

   The existing code is wrong only with a 64-bit cross-compiling GHC,
   targeting a 32-bit platform, and performing constant folding on a
   literal that doesn't fit in a 32-bit word. If GHC was built with
   DEBUG, the assertion in GHC.Types.Literal.mkLitWord would fail.
   Otherwise the bad transformation would go unnoticed.

- - - - -
bff74de7 by Sylvain Henry at 2020-11-03T17:43:03-05:00
Bignum: make GMP's bignat_add not recursive

bignat_add was a loopbreaker with an INLINE pragma (spotted by
@mpickering). This patch makes it non recursive to avoid the issue.

- - - - -
bb100805 by Andreas Klebinger at 2020-11-04T16:47:24-05:00
NCG: Fix 64bit int comparisons on 32bit x86

We no compare these by doing 64bit subtraction and
checking the resulting flags.

We used to do this differently but the old approach was
broken when the high bits compared equal and the comparison
was one of >= or <=.

The new approach should be both correct and faster.

- - - - -
b790b7f9 by Andreas Klebinger at 2020-11-04T16:47:59-05:00
Testsuite: Support for user supplied package dbs

We can now supply additional package dbs to the testsuite.
For make the package db can be supplied by
passing PACKAGE_DB=/path/to/db.

In the testsuite driver it's passed via the --test-package-db
argument.

- - - - -
81560981 by Sylvain Henry at 2020-11-04T16:48:42-05:00
Don't use LEA with 8-bit registers (#18614)

- - - - -
17d5c518 by Viktor Dukhovni at 2020-11-05T00:50:23-05:00
Naming, value types and tests for Addr# atomics

The atomic Exchange and CAS operations on integral types are updated to
take and return more natural `Word#` rather than `Int#` values.  These
are bit-block not arithmetic operations, and the sign bit plays no
special role.

Standardises the names to `atomic<OpType><ValType>Addr#`, where `OpType` is one
of `Cas` or `Exchange` and `ValType` is presently either `Word` or `Addr`.
Eventually, variants for `Word32` and `Word64` can and should be added,
once #11953 and related issues (e.g. #13825) are resolved.

Adds tests for `Addr#` CAS that mirror existing tests for
`MutableByteArray#`.

- - - - -
2125b1d6 by Ryan Scott at 2020-11-05T00:51:01-05:00
Add a regression test for #18920

Commit f594a68a5500696d94ae36425bbf4d4073aca3b2
(`Use level numbers for generalisation`) ended up fixing #18920. Let's add a
regression test to ensure that it stays fixed.

Fixes #18920.

- - - - -
e07e383a by Ryan Scott at 2020-11-06T03:45:28-05:00
Replace HsImplicitBndrs with HsOuterTyVarBndrs

This refactors the GHC AST to remove `HsImplicitBndrs` and replace it with
`HsOuterTyVarBndrs`, a type which records whether the outermost quantification
in a type is explicit (i.e., with an outermost, invisible `forall`) or
implicit. As a result of this refactoring, it is now evident in the AST where
the `forall`-or-nothing rule applies: it's all the places that use
`HsOuterTyVarBndrs`. See the revamped `Note [forall-or-nothing rule]` in
`GHC.Hs.Type` (previously in `GHC.Rename.HsType`).

Moreover, the places where `ScopedTypeVariables` brings lexically scoped type
variables into scope are a subset of the places that adhere to the
`forall`-or-nothing rule, so this also makes places that interact with
`ScopedTypeVariables` easier to find. See the revamped
`Note [Lexically scoped type variables]` in `GHC.Hs.Type` (previously in
`GHC.Tc.Gen.Sig`).

`HsOuterTyVarBndrs` are used in type signatures (see `HsOuterSigTyVarBndrs`)
and type family equations (see `HsOuterFamEqnTyVarBndrs`). The main difference
between the former and the latter is that the former cares about specificity
but the latter does not.

There are a number of knock-on consequences:

* There is now a dedicated `HsSigType` type, which is the combination of
  `HsOuterSigTyVarBndrs` and `HsType`. `LHsSigType` is now an alias for an
  `XRec` of `HsSigType`.
* Working out the details led us to a substantial refactoring of
  the handling of explicit (user-written) and implicit type-variable
  bindings in `GHC.Tc.Gen.HsType`.

  Instead of a confusing family of higher order functions, we now
  have a local data type, `SkolemInfo`, that controls how these
  binders are kind-checked.

  It remains very fiddly, not fully satisfying. But it's better
  than it was.

Fixes #16762. Bumps the Haddock submodule.

Co-authored-by: Simon Peyton Jones <simonpj at microsoft.com>
Co-authored-by: Richard Eisenberg <rae at richarde.dev>
Co-authored-by: Zubin Duggal <zubin at cmi.ac.in>

- - - - -
c85f4928 by Sylvain Henry at 2020-11-06T03:46:08-05:00
Refactor -dynamic-too handling

1) Don't modify DynFlags (too much) for -dynamic-too: now when we
   generate dynamic outputs for "-dynamic-too", we only set "dynamicNow"
   boolean field in DynFlags instead of modifying several other fields.
   These fields now have accessors that take dynamicNow into account.

2) Use DynamicTooState ADT to represent -dynamic-too state. It's much
   clearer than the undocumented "DynamicTooConditional" that was used
   before.

As a result, we can finally remove the hscs_iface_dflags field in
HscRecomp. There was a comment on this field saying:

   "FIXME (osa): I don't understand why this is necessary, but I spent
   almost two days trying to figure this out and I couldn't .. perhaps
   someone who understands this code better will remove this later."

I don't fully understand the details, but it was needed because of the
changes made to the DynFlags for -dynamic-too.

There is still something very dubious in GHC.Iface.Recomp: we have to
disable the "dynamicNow" flag at some point for some Backpack's "heinous
hack" to continue to work. It may be because interfaces for indefinite
units are always non-dynamic, or because we mix and match dynamic and
non-dynamic interfaces (#9176), or something else, who knows?

- - - - -
2cb87909 by Moritz Angermann at 2020-11-06T03:46:44-05:00
[AArch64] Aarch64 Always PIC

- - - - -
b1d2c1f3 by Ben Gamari at 2020-11-06T03:47:19-05:00
rts/Sanity: Avoid nasty race in weak pointer sanity-checking

See Note [Racing weak pointer evacuation] for all of the gory details.

- - - - -
13707ac1 by Ben Gamari at 2020-11-08T09:32:22-05:00
SMP.h: Add C11-style atomic operations

- - - - -
4fdcb23c by Ben Gamari at 2020-11-08T09:40:20-05:00
rts: Infrastructure for testing with ThreadSanitizer

- - - - -
9843e80b by Ben Gamari at 2020-11-08T09:40:20-05:00
rts/CNF: Initialize all bdescrs in group

It seems wise and cheap to ensure that the whole bdescr of all blocks of
a compact group is valid, even if most cases only look at the flags
field.

- - - - -
d05aa054 by Ben Gamari at 2020-11-08T09:40:20-05:00
rts/Capability: Intialize interrupt field

Previously this was left uninitialized.

Also clarify some comments.

- - - - -
3f3fc00c by Ben Gamari at 2020-11-08T09:40:20-05:00
rts/Task: Make comments proper Notes

- - - - -
69fa87bb by Ben Gamari at 2020-11-08T09:40:20-05:00
rts/SpinLock: Move to proper atomics

This is fairly straightforward; we just needed to use relaxed operations
for the PROF_SPIN counters and a release store instead of a write
barrier.

- - - - -
6e1440f8 by Ben Gamari at 2020-11-08T09:40:20-05:00
rts/OSThreads: Fix data race

Previously we would race on the cached processor count. Avoiding this is
straightforward; just use relaxed operations.

- - - - -
662aade5 by Ben Gamari at 2020-11-08T09:40:20-05:00
rts/ClosureMaros: Use relaxed atomics

- - - - -
2f1b08f5 by Ben Gamari at 2020-11-08T09:40:20-05:00
configure: Bump minimum-supported gcc version to 4.7

Since the __atomic_* builtins are not supported until gcc 4.7. Given
that this version was released in 2012 I think this is acceptable.

- - - - -
751d094b by Ben Gamari at 2020-11-08T09:40:20-05:00
testsuite: Fix thread leak in hs_try_putmvar00[13]

- - - - -
5c1e8015 by Ben Gamari at 2020-11-08T09:40:20-05:00
rts: Introduce SET_HDR_RELEASE

Also ensure that we also store the info table pointer last to ensure
that the synchronization covers all stores.

- - - - -
df8c2abb by Ben Gamari at 2020-11-08T09:40:52-05:00
rts: Factor out logic to identify a good capability for running a task

Not only does this make the control flow a bit clearer but it also
allows us to add a TSAN suppression on this logic, which requires
(harmless) data races.

- - - - -
adc9fa09 by Ben Gamari at 2020-11-08T09:40:52-05:00
rts: Annotate benign race in waitForCapability

- - - - -
f77310e3 by Ben Gamari at 2020-11-08T09:40:52-05:00
rts: Clarify locking behavior of releaseCapability_

- - - - -
b6c0e9d1 by Ben Gamari at 2020-11-08T09:40:52-05:00
rts: Add assertions for task ownership of capabilities

- - - - -
d4a858b0 by Ben Gamari at 2020-11-08T09:40:52-05:00
rts: Use relaxed atomics on n_returning_tasks

This mitigates the warning of a benign race on n_returning_tasks in
shouldYieldCapability.

See #17261.

- - - - -
2a0b44ef by Ben Gamari at 2020-11-08T09:40:53-05:00
rts: Mitigate races in capability interruption logic

- - - - -
b099f2dc by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/Capability: Use relaxed operations for last_free_capability

- - - - -
9efa169f by Ben Gamari at 2020-11-08T09:40:53-05:00
rts: Use relaxed operations for cap->running_task (TODO)

This shouldn't be necessary since only the owning thread of the capability
should be touching this.

- - - - -
1bfdc2ad by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/Schedule: Use relaxed operations for sched_state

- - - - -
4ef6b40f by Ben Gamari at 2020-11-08T09:40:53-05:00
rts: Accept data race in work-stealing implementation

This race is okay since the task is owned by the capability pushing it.
By Note [Ownership of Task] this means that the capability is free to
write to `task->cap` without taking `task->lock`.

Fixes #17276.

- - - - -
b403f575 by Ben Gamari at 2020-11-08T09:40:53-05:00
rts: Eliminate data races on pending_sync

- - - - -
f416d1d1 by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/Schedule: Eliminate data races on recent_activity

We cannot safely use relaxed atomics here.

- - - - -
22dd4f3e by Ben Gamari at 2020-11-08T09:40:53-05:00
rts: Avoid data races in message handling

- - - - -
c5154ab1 by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/Messages: Drop incredibly fishy write barrier

executeMessage previously had a write barrier at the beginning of its
loop apparently in an attempt to synchronize with another thread's
writes to the Message. I would guess that the author had intended to use
a load barrier here given that there are no globally-visible writes done
in executeMessage.

I've removed the redundant barrier since the necessary load barrier is
now provided by the ACQUIRE_LOAD.

- - - - -
1c656731 by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/ThreadPaused: Avoid data races

- - - - -
1673f4e5 by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/Schedule: Eliminate data races in run queue management

- - - - -
6207b94b by Ben Gamari at 2020-11-08T09:40:53-05:00
rts: Eliminate shutdown data race on task counters

- - - - -
6d5a70f9 by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/Threads: Avoid data races (TODO)

Replace barriers with appropriate ordering. Drop redundant barrier in
tryWakeupThread (the RELEASE barrier will be provided by sendMessage's
mutex release).

We use relaxed operations on why_blocked and the stack although it's not
clear to me why this is necessary.

- - - - -
d0b5dede by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/Messages: Annotate benign race

- - - - -
f96cf570 by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/RaiseAsync: Synchronize what_next read

- - - - -
35253584 by Ben Gamari at 2020-11-08T09:40:53-05:00
rts/Task: Move debugTrace to avoid data race

Specifically, we need to hold all_tasks_mutex to read taskCount.

- - - - -
a0445cad by Ben Gamari at 2020-11-08T09:40:53-05:00
Disable flawed assertion

- - - - -
cdc05f91 by Ben Gamari at 2020-11-08T09:40:53-05:00
Document schedulePushWork race

- - - - -
e1ae10d4 by Ben Gamari at 2020-11-08T09:40:53-05:00
Capabiliity: Properly fix data race on n_returning_tasks

There is a real data race but can be made safe by using proper atomic
(but relaxed) accesses.

- - - - -
6c3c646c by Ben Gamari at 2020-11-08T09:40:53-05:00
rts: Make write of to_cap->inbox atomic

This is necessary since emptyInbox may read from to_cap->inbox without
taking cap->lock.

- - - - -
40157768 by Ben Gamari at 2020-11-08T09:41:10-05:00
gitlab-ci: Add nightly-x86_64-linux-deb9-tsan job

- - - - -
e990a1f6 by GHC GitLab CI at 2020-11-08T09:41:10-05:00
testsuite: Mark setnumcapabilities001 as broken with TSAN

Due to #18808.

- - - - -
8684e662 by GHC GitLab CI at 2020-11-08T09:41:10-05:00
testsuite: Skip divbyzero and derefnull under TSAN

ThreadSanitizer changes the output of these tests.

- - - - -
3d41230c by Ben Gamari at 2020-11-08T09:41:10-05:00
testsuite: Skip high memory usage tests with TSAN

ThreadSanitizer significantly increases the memory footprint of tests,
so much so that it can send machines into OOM.

- - - - -
a164b12f by Ben Gamari at 2020-11-08T09:41:10-05:00
testsuite: Mark hie002 as high_memory_usage

This test has a peak residency of 1GByte; this is large enough to
classify as "high" in my book.

- - - - -
0b3c2e45 by Ben Gamari at 2020-11-08T09:41:10-05:00
testsuite: Mark T9872[abc] as high_memory_usage

These all have a maximum residency of over 2 GB.

- - - - -
958796a2 by Ben Gamari at 2020-11-08T09:41:10-05:00
gitlab-ci: Disable documentation in TSAN build

Haddock chews through enough memory to cause the CI builders to OOM and
there's frankly no reason to build documentation in this job anyways.

- - - - -
1be57065 by Ben Gamari at 2020-11-08T09:41:10-05:00
TSANUtils: Ensure that C11 atomics are supported

- - - - -
299c516c by Ben Gamari at 2020-11-08T09:41:10-05:00
testsuite: Mark T3807 as broken with TSAN

Due to #18883.

- - - - -
b4acdabc by Ben Gamari at 2020-11-08T09:41:10-05:00
testsuite: Mark T13702 as broken with TSAN due to #18884

- - - - -
42409217 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts/BlockAlloc: Use relaxed operations

- - - - -
c00f55d3 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts: Rework handling of mutlist scavenging statistics

- - - - -
c7097264 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts: Avoid data races in StablePtr implementation

This fixes two potentially problematic data races in the StablePtr
implementation:

 * We would fail to RELEASE the stable pointer table when enlarging it,
   causing other cores to potentially see uninitialized memory.

 * We would fail to ACQUIRE when dereferencing a stable pointer.

- - - - -
0ffcb040 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts/Storage: Use atomics

- - - - -
24865cb4 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts/Updates: Use proper atomic operations

- - - - -
ce181c57 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts/Weak: Eliminate data races

By taking all_tasks_mutex in stat_exit. Also better-document the fact
that the task statistics are protected by all_tasks_mutex.

- - - - -
fd4d4cd1 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts/GC: Use atomics

- - - - -
baa322e4 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts: Use RELEASE ordering in unlockClosure

- - - - -
bd0fb1d2 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts/Storage: Accept races on heap size counters

- - - - -
26e8e670 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts: Join to concurrent mark thread during shutdown

Previously we would take all capabilities but fail to join on the thread
itself, potentially resulting in a leaked thread.

- - - - -
40d7aa8a by GHC GitLab CI at 2020-11-08T09:41:41-05:00
rts: Fix race in GC CPU time accounting

Ensure that the GC leader synchronizes with workers before calling
stat_endGC.

- - - - -
7f1ebbf4 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts/SpinLock: Separate out slow path

Not only is this in general a good idea, but it turns out that GCC
unrolls the retry loop, resulting is massive code bloat in critical
parts of the RTS (e.g. `evacuate`).

- - - - -
688373b0 by Ben Gamari at 2020-11-08T09:41:41-05:00
rts: Use relaxed ordering on spinlock counters

- - - - -
a0d2713c by Ben Gamari at 2020-11-08T09:41:41-05:00
rts: Annotate hopefully "benign" races in freeGroup

- - - - -
6f05d266 by Ben Gamari at 2020-11-08T09:41:41-05:00
Strengthen ordering in releaseGCThreads

- - - - -
06665ca2 by Ben Gamari at 2020-11-08T09:41:48-05:00
rts/WSDeque: Rewrite with proper atomics

After a few attempts at shoring up the previous implementation, I ended
up turning to the literature and now use the proven implementation,

> N.M. Lê, A. Pop, A.Cohen, and F.Z. Nardelli. "Correct and Efficient
> Work-Stealing for Weak Memory Models". PPoPP'13, February 2013,
> ACM 978-1-4503-1922/13/02.

Note only is this approach formally proven correct under C11 semantics
but it is also proved to be a bit faster in practice.

- - - - -
adac9ad3 by Ben Gamari at 2020-11-08T09:41:54-05:00
rts: Use relaxed atomics for whitehole spin stats

- - - - -
296de125 by Ben Gamari at 2020-11-08T09:41:54-05:00
rts: Avoid lock order inversion during fork

Fixes #17275.

- - - - -
5e3d99c2 by GHC GitLab CI at 2020-11-08T09:41:54-05:00
rts: Use proper relaxe operations in getCurrentThreadCPUTime

Here we are doing lazy initialization; it's okay if we do the check more
than once, hence relaxed operation is fine.

- - - - -
1061b8e9 by Ben Gamari at 2020-11-08T09:41:59-05:00
rts/STM: Use atomics

This fixes a potentially harmful race where we failed to synchronize
before looking at a TVar's current_value.

Also did a bit of refactoring to avoid abstract over management of
max_commits.

- - - - -
41a40d25 by Ben Gamari at 2020-11-08T09:41:59-05:00
rts/stm: Strengthen orderings to SEQ_CST instead of volatile

Previously the `current_value`, `first_watch_queue_entry`, and
`num_updates` fields of `StgTVar` were marked as `volatile` in an
attempt to provide strong ordering. Of course, this isn't sufficient.

We now use proper atomic operations. In most of these cases I strengthen
the ordering all the way to SEQ_CST although it's possible that some
could be weakened with some thought.

- - - - -
9c096a37 by Ben Gamari at 2020-11-08T09:42:03-05:00
Mitigate data races in event manager startup/shutdown

- - - - -
06d604a2 by Ben Gamari at 2020-11-08T09:42:12-05:00
rts: Accept benign races in Proftimer

- - - - -
0aeea61f by Ben Gamari at 2020-11-08T09:42:12-05:00
rts: Pause timer while changing capability count

This avoids #17289.

- - - - -
5e333a8e by Ben Gamari at 2020-11-08T09:42:36-05:00
rts: Fix timer initialization

Previously `initScheduler` would attempt to pause the ticker and in so
doing acquire the ticker mutex. However, initTicker, which is
responsible for initializing said mutex, hadn't been called
yet.

- - - - -
07d0e7a9 by Ben Gamari at 2020-11-08T09:42:36-05:00
rts: Fix races in Pthread timer backend shudown

We can generally be pretty relaxed in the barriers here since the timer
thread is a loop.

- - - - -
747aef01 by Ben Gamari at 2020-11-08T09:42:50-05:00
rts/Stats: Hide a few unused unnecessarily global functions

- - - - -
7640556b by Ben Gamari at 2020-11-08T09:42:50-05:00
rts/Stats: Protect with mutex

While on face value this seems a bit heavy, I think it's far better than
enforcing ordering on every access.

- - - - -
05d737a2 by Ben Gamari at 2020-11-08T09:42:50-05:00
rts: Tear down stats_mutex after exitHeapProfiling

Since the latter wants to call getRTSStats.

- - - - -
30476e1b by Ben Gamari at 2020-11-08T09:43:19-05:00
Merge branches 'wip/tsan/sched', 'wip/tsan/ci', 'wip/tsan/storage', 'wip/tsan/wsdeque', 'wip/tsan/misc', 'wip/tsan/stm', 'wip/tsan/event-mgr', 'wip/tsan/timer' and 'wip/tsan/stats' into wip/tsan/all

- - - - -


30 changed files:

- .gitlab-ci.yml
- compiler/.hlint.yaml
- compiler/GHC.hs
- compiler/GHC/Builtin/PrimOps.hs
- compiler/GHC/Builtin/Types.hs
- compiler/GHC/Builtin/Types/Prim.hs
- compiler/GHC/Builtin/Utils.hs
- + compiler/GHC/Builtin/bytearray-ops.txt.pp
- compiler/GHC/Builtin/primops.txt.pp
- compiler/GHC/ByteCode/Asm.hs
- compiler/GHC/ByteCode/InfoTable.hs
- compiler/GHC/ByteCode/Linker.hs
- compiler/GHC/Cmm/Graph.hs
- compiler/GHC/Cmm/LayoutStack.hs
- compiler/GHC/Cmm/Pipeline.hs
- compiler/GHC/Cmm/ProcPoint.hs
- compiler/GHC/Cmm/Type.hs
- compiler/GHC/CmmToAsm.hs
- compiler/GHC/CmmToAsm/BlockLayout.hs
- compiler/GHC/CmmToAsm/CFG/Dominators.hs
- compiler/GHC/CmmToAsm/PPC/CodeGen.hs
- compiler/GHC/CmmToAsm/Reg/Graph/Spill.hs
- compiler/GHC/CmmToAsm/SPARC/CodeGen.hs
- compiler/GHC/CmmToAsm/X86/CodeGen.hs
- compiler/GHC/CmmToAsm/X86/Cond.hs
- compiler/GHC/CmmToLlvm/CodeGen.hs
- compiler/GHC/Core.hs
- compiler/GHC/Core/Coercion.hs
- compiler/GHC/Core/Coercion/Axiom.hs
- − compiler/GHC/Core/ConLike.hs-boot


The diff was not included because it is too large.


View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/07e82ba52228580cfbd90ff031e657acbecc715b...30476e1b940bbe494a08b8cb2b20a895dc8f4d75

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/07e82ba52228580cfbd90ff031e657acbecc715b...30476e1b940bbe494a08b8cb2b20a895dc8f4d75
You're receiving this email because of your account on gitlab.haskell.org.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-commits/attachments/20201108/f4eb3ebd/attachment-0001.html>


More information about the ghc-commits mailing list