[Git][ghc/ghc][wip/T23691] 30 commits: base: add COMPLETE pragma to BufferCodec PatSyn

Ben Gamari (@bgamari) gitlab at gitlab.haskell.org
Fri Jul 21 15:51:20 UTC 2023



Ben Gamari pushed to branch wip/T23691 at Glasgow Haskell Compiler / GHC


Commits:
22565506 by sheaf at 2023-07-17T21:12:59-04:00
base: add COMPLETE pragma to BufferCodec PatSyn

This implements CLC proposal #178, rectifying an oversight in the
implementation of CLC proposal #134 which could lead to spurious
pattern match warnings.

https://github.com/haskell/core-libraries-committee/issues/178
https://github.com/haskell/core-libraries-committee/issues/134

- - - - -
860f6269 by sheaf at 2023-07-17T21:13:00-04:00
exactprint: silence incomplete record update warnings

- - - - -
df706de3 by sheaf at 2023-07-17T21:13:00-04:00
Re-instate -Wincomplete-record-updates

Commit e74fc066 refactored the handling of record updates to use
the HsExpanded mechanism. This meant that the pattern matching inherent
to a record update was considered to be "generated code", and thus we
stopped emitting "incomplete record update" warnings entirely.

This commit changes the "data Origin = Source | Generated" datatype,
adding a field to the Generated constructor to indicate whether we
still want to perform pattern-match checking. We also have to do a bit
of plumbing with HsCase, to record that the HsCase arose from an
HsExpansion of a RecUpd, so that the error message continues to mention
record updates as opposed to a generic "incomplete pattern matches in case"
error.

Finally, this patch also changes the way we handle inaccessible code
warnings. Commit e74fc066 was also a regression in this regard, as we
were emitting "inaccessible code" warnings for case statements spuriously
generated when desugaring a record update (remember: the desugaring mechanism
happens before typechecking; it thus can't take into account e.g. GADT information
in order to decide which constructors to include in the RHS of the desugaring
of the record update).
We fix this by changing the mechanism through which we disable inaccessible
code warnings: we now check whether we are in generated code in
GHC.Tc.Utils.TcMType.newImplication in order to determine whether to
emit inaccessible code warnings.

Fixes #23520
Updates haddock submodule, to avoid incomplete record update warnings

- - - - -
1d05971e by sheaf at 2023-07-17T21:13:00-04:00
Propagate long-distance information in do-notation

The preceding commit re-enabled pattern-match checking inside record
updates. This revealed that #21360 was in fact NOT fixed by e74fc066.

This commit makes sure we correctly propagate long-distance information
in do blocks, e.g. in

```haskell
data T = A { fld :: Int } | B

f :: T -> Maybe T
f r = do
  a at A{} <- Just r
  Just $ case a of { A _ -> A 9 }
```

we need to propagate the fact that "a" is headed by the constructor "A"
to see that the case expression "case a of { A _ -> A 9 }" cannot fail.

Fixes #21360

- - - - -
bea0e323 by sheaf at 2023-07-17T21:13:00-04:00
Skip PMC for boring patterns

Some patterns introduce no new information to the pattern-match
checker (such as plain variable or wildcard patterns). We can thus
skip doing any pattern-match checking on them when the sole purpose
for doing so was introducing new long-distance information.

See Note [Boring patterns] in GHC.Hs.Pat.

Doing this avoids regressing in performance now that we do additional
pattern-match checking inside do notation.

- - - - -
ddcdd88c by Rodrigo Mesquita at 2023-07-17T21:13:36-04:00
Split GHC.Platform.ArchOS from ghc-boot into ghc-platform

Split off the `GHC.Platform.ArchOS` module from the `ghc-boot` package
into this reinstallable standalone package which abides by the PVP, in
part motivated by the ongoing work on `ghc-toolchain` towards runtime
retargetability.

- - - - -
b55a8ea7 by Sylvain Henry at 2023-07-17T21:14:27-04:00
JS: better implementation for plusWord64 (#23597)

- - - - -
889c2bbb by sheaf at 2023-07-18T06:37:32-04:00
Do primop rep-poly checks when instantiating

This patch changes how we perform representation-polymorphism checking
for primops (and other wired-in Ids such as coerce).
When instantiating the primop, we check whether each type variable
is required to instantiated to a concrete type, and if so we create a
new concrete metavariable (a ConcreteTv) instead of a simple MetaTv.
(A little subtlety is the need to apply the substitution obtained from
instantiating to the ConcreteTvOrigins, see
Note [substConcreteTvOrigin] in GHC.Tc.Utils.TcMType.)

This allows us to prevent representation-polymorphism in non-argument
position, as that is required for some of these primops.

We can also remove the logic in tcRemainingValArgs, except for
the part concerning representation-polymorphic unlifted newtypes.
The function has been renamed rejectRepPolyNewtypes; all it does now
is reject unsaturated occurrences of representation-polymorphic newtype
constructors when the representation of its argument isn't a concrete
RuntimeRep (i.e. still a PHASE 1 FixedRuntimeRep check).
The Note [Eta-expanding rep-poly unlifted newtypes] in GHC.Tc.Gen.Head
gives more explanation about a possible path to PHASE 2, which would be
in line with the treatment for primops taken in this patch.

We also update the Core Lint check to handle this new framework. This
means Core Lint now checks representation-polymorphism in continuation
position like needed for catch#.

Fixes #21906

-------------------------
Metric Increase:
    LargeRecord
-------------------------

- - - - -
00648e5d by Krzysztof Gogolewski at 2023-07-18T06:38:10-04:00
Core Lint: distinguish let and letrec in locations

Lint messages were saying "in the body of letrec" even for non-recursive
let.

I've also renamed BodyOfLetRec to BodyOfLet in stg, since there's no
separate letrec.

- - - - -
787bae96 by Krzysztof Gogolewski at 2023-07-18T06:38:50-04:00
Use extended literals when deriving Show

This implements GHC proposal
https://github.com/ghc-proposals/ghc-proposals/pull/596

Also add support for Int64# and Word64#; see testcase ShowPrim.

- - - - -
257f1567 by Jaro Reinders at 2023-07-18T06:39:29-04:00
Add StgFromCore and StgCodeGen linting

- - - - -
34d08a20 by Ben Gamari at 2023-07-19T03:33:22-04:00
Reg.Liveness: Strictness

- - - - -
c5deaa27 by Ben Gamari at 2023-07-19T03:33:22-04:00
Reg.Liveness: Don't repeatedly construct UniqSets

- - - - -
b947250b by Ben Gamari at 2023-07-19T03:33:22-04:00
compiler/Types: Ensure that fromList-type operations can fuse

In #20740 I noticed that mkUniqSet does not fuse. In practice, allowing
it to do so makes a considerable difference in allocations due to the
backend.

Metric Decrease:
    T12707
    T13379
    T3294
    T4801
    T5321FD
    T5321Fun
    T783

- - - - -
6c88c2ba by Sven Tennie at 2023-07-19T03:33:59-04:00
x86 Codegen: Implement MO_S_MulMayOflo for W16

- - - - -
5f1154e0 by Sven Tennie at 2023-07-19T03:33:59-04:00
x86 CodeGen: MO_S_MulMayOflo better error message for rep > W64

It's useful to see which value made the pattern match fail. (If it ever
occurs.)

- - - - -
e8c9a95f by Sven Tennie at 2023-07-19T03:33:59-04:00
x86 CodeGen: Implement MO_S_MulMayOflo for W8

This case wasn't handled before. But, the test-primops test suite showed
that it actually might appear.

- - - - -
a36f9dc9 by Sven Tennie at 2023-07-19T03:33:59-04:00
Add test for %mulmayoflo primop

The test expects a perfect implementation with no false positives.

- - - - -
38a36248 by Matthew Pickering at 2023-07-19T03:34:36-04:00
lint-ci-config: Generate jobs-metadata.json

We also now save the jobs-metadata.json and jobs.yaml file as artifacts
as:

* It might be useful for someone who is modifying CI to copy jobs.yaml
  if they are having trouble regenerating locally.
* jobs-metadata.json is very useful for downstream pipelines to work out
  the right job to download.

Fixes #23654

- - - - -
1535a671 by Vladislav Zavialov at 2023-07-19T03:35:12-04:00
Initialize 9.10.1-notes.rst

Create new release notes for the next GHC release (GHC 9.10)

- - - - -
3bd4d5b5 by sheaf at 2023-07-19T03:35:53-04:00
Prioritise Parent when looking up class sub-binder

When we look up children GlobalRdrElts of a given Parent, we sometimes
would rather prioritise those GlobalRdrElts which have the right Parent,
and sometimes prioritise those that have the right NameSpace:

  - in export lists, we should prioritise NameSpace
  - for class/instance binders, we should prioritise Parent

See Note [childGREPriority] in GHC.Types.Name.Reader.

fixes #23664

- - - - -
9c8fdda3 by Alan Zimmerman at 2023-07-19T03:36:29-04:00
EPA: Improve annotation management in getMonoBind

Ensure the LHsDecl for a FunBind has the correct leading comments and
trailing annotations.

See the added note for details.

- - - - -
ff884b77 by Matthew Pickering at 2023-07-19T11:42:02+01:00
Remove unused files in .gitlab

These were left over after 6078b429

- - - - -
29ef590c by Matthew Pickering at 2023-07-19T11:42:52+01:00
gen_ci: Add hie.yaml file

This allows you to load `gen_ci.hs` into HLS, and now it is a huge
module, that is quite useful.

- - - - -
808b55cf by Matthew Pickering at 2023-07-19T12:24:41+01:00
ci: Make "fast-ci" the default validate configuration

We are trying out a lighter weight validation pipeline where by default
we just test on 5 platforms:

* x86_64-deb10-slow-validate
* windows
* x86_64-fedora33-release
* aarch64-darwin
* aarch64-linux-deb10

In order to enable the "full" validation pipeline you can apply the
`full-ci` label which will enable all the validation pipelines.

All the validation jobs are still run on a marge batch.

The goal is to reduce the overall CI capacity so that pipelines start
faster for MRs and marge bot batches are faster.

Fixes #23694

- - - - -
0b23db03 by Alan Zimmerman at 2023-07-20T05:28:47-04:00
EPA: Simplify GHC/Parser.y sL1

This is the next patch in a series simplifying location management in
GHC/Parser.y

This one simplifies sL1, to use the HasLoc instances introduced in
!10743 (closed)

- - - - -
3ece9856 by Ben Gamari at 2023-07-21T07:30:45-04:00
nativeGen: Explicitly set flags of text sections on Windows

The binutils documentation (for COFF) claims,

> If no flags are specified, the default flags depend upon the section
> name. If the section name is not recognized, the default will be for the
> section to be loaded and writable.

We previously assumed that this would do the right thing for split
sections (e.g. a section named `.text$foo` would be correctly inferred
to be a text section). However, we have observed that this is not the
case (at least under the clang toolchain used on Windows): when
split-sections is enabled, text sections are treated by the assembler as
data (matching the "default" behavior specified by the documentation).

Avoid this by setting section flags explicitly. This should fix split
sections on Windows.

Fixes #22834.

- - - - -
db7f7240 by Ben Gamari at 2023-07-21T07:30:45-04:00
nativeGen: Set explicit section types on all platforms

- - - - -
b444c16f by Finley McIlwaine at 2023-07-21T07:31:28-04:00
Insert documentation into parsed signature modules

Causes haddock comments in signature modules to be properly
inserted into the AST (just as they are for regular modules)
if the `-haddock` flag is given.

Also adds a test that compares `-ddump-parsed-ast` output
for a signature module to prevent further regressions.

Fixes #23315

- - - - -
1b769e1a by Ben Gamari at 2023-07-21T11:51:12-04:00
rts/win32: Ensure reliability of IO manager shutdown

When the Win32 threaded IO manager shuts down, `ioManagerDie` sends an
`IO_MANAGER_DIE` event to the IO manager thread using the
`io_manager_event` event object. Finally, it will closes the event object,
and invalidate `io_manager_event`.

Previously, `readIOManagerEvent` would see that `io_manager_event` is
invalid and return `0`, suggesting that everything is right with the
world. This meant that if `ioManagerDie` invalidated the handle before
the event manager was blocked on the event we would end up in a
situation where the event manager would never realize it was asked to
shut down.

Fix this by ensuring that `readIOManagerEvent` instead returns
`IO_MANAGER_DIE` when we detect that the event object has been
invalidated by `ioManagerDie`.

Fixes #23691.

- - - - -


29 changed files:

- .gitlab-ci.yml
- − .gitlab/gen-ci.cabal
- .gitlab/generate-ci/gen_ci.hs
- .gitlab/hie.yaml → .gitlab/generate-ci/hie.yaml
- .gitlab/jobs.yaml
- compiler/GHC/Builtin/PrimOps.hs
- compiler/GHC/Builtin/PrimOps/Ids.hs
- compiler/GHC/Builtin/Types.hs
- compiler/GHC/Builtin/primops.txt.pp
- compiler/GHC/Cmm/MachOp.hs
- compiler/GHC/CmmToAsm/Ppr.hs
- compiler/GHC/CmmToAsm/Reg/Liveness.hs
- compiler/GHC/CmmToAsm/X86/CodeGen.hs
- compiler/GHC/Core.hs
- compiler/GHC/Core/DataCon.hs
- compiler/GHC/Core/Lint.hs
- compiler/GHC/Core/Opt/Simplify/Iteration.hs
- compiler/GHC/Core/TyCo/Subst.hs
- compiler/GHC/Core/Type.hs
- compiler/GHC/Driver/Config/Core/Lint.hs
- compiler/GHC/Hs/Expr.hs
- compiler/GHC/Hs/Pat.hs
- compiler/GHC/Hs/Utils.hs
- compiler/GHC/HsToCore.hs
- compiler/GHC/HsToCore/Arrows.hs
- compiler/GHC/HsToCore/Errors/Ppr.hs
- compiler/GHC/HsToCore/Errors/Types.hs
- compiler/GHC/HsToCore/Expr.hs
- compiler/GHC/HsToCore/GuardedRHSs.hs


The diff was not included because it is too large.


View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/fc15f1da5a8739d7a4fe7391976bf01143dba919...1b769e1af03880686b4a21408e04ec73a97c8a97

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/fc15f1da5a8739d7a4fe7391976bf01143dba919...1b769e1af03880686b4a21408e04ec73a97c8a97
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/20230721/60dc1001/attachment-0001.html>


More information about the ghc-commits mailing list