[Git][ghc/ghc][wip/T22011] 39 commits: Improve the situation with the stimes cycle

Ben Gamari (@bgamari) gitlab at gitlab.haskell.org
Wed Jul 19 15:21:48 UTC 2023 


Ben Gamari pushed to branch wip/T22011 at Glasgow Haskell Compiler / GHC


Commits:
9ce44336 by meooow25 at 2023-07-05T11:42:37-04:00
Improve the situation with the stimes cycle

Currently the Semigroup stimes cycle is resolved in GHC.Base by
importing stimes implementations from a hs-boot file. Resolve the cycle
using hs-boot files for required classes (Num, Integral) instead. Now
stimes can be defined directly in GHC.Base, making inlining and
specialization possible.

This leads to some new boot files for `GHC.Num` and `GHC.Real`, the
methods for those are only used to implement `stimes` so it doesn't
appear that these boot files will introduce any new performance traps.

Metric Decrease:
    T13386
    T8095
Metric Increase:
    T13253
    T13386
    T18698a
    T18698b
    T19695
    T8095

- - - - -
9edcb1fb by Jaro Reinders at 2023-07-05T11:43:24-04:00
Refactor Unique to be represented by Word64

In #22010 we established that Int was not always sufficient to store all
the uniques we generate during compilation on 32-bit platforms. This
commit addresses that problem by using Word64 instead of Int for
uniques.

The core of the change is in GHC.Core.Types.Unique and
GHC.Core.Types.Unique.Supply. However, the representation of uniques is
used in many other places, so those needed changes too. Additionally, the RTS
has been extended with an atomic_inc64 operation.

One major change from this commit is the introduction of the Word64Set and
Word64Map data types. These are adapted versions of IntSet and IntMap
from the containers package. These are planned to be upstreamed in the
future.

As a natural consequence of these changes, the compiler will be a bit
slower and take more space on 32-bit platforms. Our CI tests indicate
around a 5% residency increase.

Metric Increase:
    CoOpt_Read
    CoOpt_Singletons
    LargeRecord
    ManyAlternatives
    ManyConstructors
    MultiComponentModules
    MultiComponentModulesRecomp
    MultiLayerModulesTH_OneShot
    RecordUpdPerf
    T10421
    T10547
    T12150
    T12227
    T12234
    T12425
    T12707
    T13035
    T13056
    T13253
    T13253-spj
    T13379
    T13386
    T13719
    T14683
    T14697
    T14766
    T15164
    T15703
    T16577
    T16875
    T17516
    T18140
    T18223
    T18282
    T18304
    T18698a
    T18698b
    T18923
    T1969
    T19695
    T20049
    T21839c
    T3064
    T3294
    T4801
    T5030
    T5321FD
    T5321Fun
    T5631
    T5642
    T5837
    T6048
    T783
    T8095
    T9020
    T9198
    T9233
    T9630
    T9675
    T9872a
    T9872b
    T9872b_defer
    T9872c
    T9872d
    T9961
    TcPlugin_RewritePerf
    UniqLoop
    WWRec
    hard_hole_fits

- - - - -
6b9db7d4 by Brandon Chinn at 2023-07-05T11:44:03-04:00
Fix docs for __GLASGOW_HASKELL_FULL_VERSION__ macro
- - - - -
40f4ef7c by Torsten Schmits at 2023-07-05T18:06:19-04:00
Substitute free variables captured by breakpoints in SpecConstr

Fixes #23267

- - - - -
2b55cb5f by sheaf at 2023-07-05T18:07:07-04:00
Reinstate untouchable variable error messages

This extra bit of information was accidentally being discarded after
a refactoring of the way we reported problems when unifying a type
variable with another type. This patch rectifies that.

- - - - -
53ed21c5 by Rodrigo Mesquita at 2023-07-05T18:07:47-04:00
configure: Drop Clang command from settings

Due to 01542cb7227614a93508b97ecad5b16dddeb6486 we no longer use the
`runClang` function, and no longer need to configure into settings the
Clang command. We used to determine options at runtime to pass clang when
it was used as an assembler, but now that we configure at configure time
we no longer need to.

- - - - -
6fdcf969 by Torsten Schmits at 2023-07-06T12:12:09-04:00
Filter out nontrivial substituted expressions in substTickish

Fixes #23272

- - - - -
41968fd6 by Sylvain Henry at 2023-07-06T12:13:02-04:00
JS: testsuite: use req_c predicate instead of js_broken

- - - - -
74a4dd2e by Sylvain Henry at 2023-07-06T12:13:02-04:00
JS: implement some file primitives (lstat,rmdir) (#22374)

- Implement lstat and rmdir.
- Implement base_c_s_is* functions (testing a file type)
- Enable passing tests

- - - - -
7e759914 by Sylvain Henry at 2023-07-07T02:39:38-04:00
JS: cleanup utils (#23314)

- Removed unused code
- Don't export unused functions
- Move toTypeList to Closure module

- - - - -
f617655c by Sylvain Henry at 2023-07-07T02:39:38-04:00
JS: rename VarType/Vt into JSRep

- - - - -
19216ca5 by Sylvain Henry at 2023-07-07T02:39:38-04:00
JS: remove custom PrimRep conversion (#23314)

We use the usual conversion to PrimRep and then we convert these
PrimReps to JSReps.

- - - - -
d3de8668 by Sylvain Henry at 2023-07-07T02:39:38-04:00
JS: don't use isRuntimeRepKindedTy in JS FFI

- - - - -
8d1b75cb by Matthew Pickering at 2023-07-07T02:40:15-04:00
ghcup-metadata: Also updates ghcup-nightlies-0.0.7.yaml file

Fixes #23600

- - - - -
e524fa7f by Matthew Pickering at 2023-07-07T02:40:15-04:00
ghcup-metadata: Use dynamically linked alpine bindists

In theory these will work much better on alpine to allow people to build
statically linked applications there. We don't need to distribute a
statically linked application ourselves in order to allow that.

Fixes #23602

- - - - -
b9e7beb9 by Ben Gamari at 2023-07-07T11:32:22-04:00
Drop circle-ci-job.sh

- - - - -
9955eead by Ben Gamari at 2023-07-07T11:32:22-04:00
testsuite: Allow preservation of unexpected output

Here we introduce a new flag to the testsuite driver,
--unexpected-output-dir=<dir>, which allows the user to ask
the driver to preserve unexpected output from tests. The intent is for
this to be used in CI to allow users to more easily fix unexpected
platform-dependent output.

- - - - -
48f80968 by Ben Gamari at 2023-07-07T11:32:22-04:00
gitlab-ci: Preserve unexpected output

Here we enable use of the testsuite driver's `--unexpected-output-dir`
flag by CI, preserving the result as an artifact for use by users.

- - - - -
76983a0d by Matthew Pickering at 2023-07-07T11:32:58-04:00
driver: Fix -S with .cmm files

There was an oversight in the driver which assumed that you would always
produce a `.o` file when compiling a .cmm file.

Fixes #23610

- - - - -
6df15e93 by Mike Pilgrem at 2023-07-07T11:33:40-04:00
Update Hadrian's stack.yaml
- - - - -
1dff43cf by Ben Gamari at 2023-07-08T05:05:37-04:00
compiler: Rework ShowSome

Previously the field used to filter the sub-declarations to show
was rather ad-hoc and was only able to show at most one sub-declaration.

- - - - -
8165404b by Ben Gamari at 2023-07-08T05:05:37-04:00
testsuite: Add test to catch changes in core libraries

This adds testing infrastructure to ensure that changes in core
libraries (e.g. `base` and `ghc-prim`) are caught in CI.

- - - - -
ec1c32e2 by Melanie Phoenix at 2023-07-08T05:06:14-04:00
Deprecate Data.List.NonEmpty.unzip

- - - - -
5d2442b8 by Ben Gamari at 2023-07-08T05:06:51-04:00
Drop latent mentions of -split-objs

Closes #21134.

- - - - -
a9bc20cb by Oleg Grenrus at 2023-07-08T05:07:31-04:00
Add warn_and_run test kind

This is a compile_and_run variant which also captures the GHC's
stderr. The warn_and_run name is best I can come up with,
as compile_and_run is taken.

This is useful specifically for testing warnings.  We want to test that
when warning triggers, and it's not a false positive, i.e. that the
runtime behaviour is indeed "incorrect".

As an example a single test is altered to use warn_and_run

- - - - -
c7026962 by Ben Gamari at 2023-07-08T05:08:11-04:00
configure: Don't use ld.gold on i386

ld.gold appears to produce invalid static constructor tables on i386.
While ideally we would add an autoconf check to check for this
brokenness, sadly such a check isn't easy to compose. Instead to
summarily reject such linkers on i386.

Somewhat hackily closes #23579.

- - - - -
054261dd by Bodigrim at 2023-07-08T19:32:47-04:00
Add since annotations for Data.Foldable1

- - - - -
550af505 by Sylvain Henry at 2023-07-08T19:33:28-04:00
JS: support -this-unit-id for programs in the linker (#23613)

- - - - -
d284470a by Bodigrim at 2023-07-08T19:34:08-04:00
Bump text submodule

- - - - -
8e11630e by jade at 2023-07-10T16:58:40-04:00
Add a hint to enable ExplicitNamespaces for type operator imports (Fixes/Enhances #20007)

As suggested in #20007 and implemented in !8895, trying to import type operators
will suggest a fix to use the 'type' keyword, without considering whether
ExplicitNamespaces is enabled. This patch will query whether ExplicitNamespaces
is enabled and add a hint to suggest enabling ExplicitNamespaces if it isn't enabled,
alongside the suggestion of adding the 'type' keyword.

- - - - -
61b1932e by sheaf at 2023-07-10T16:59:26-04:00
tyThingLocalGREs: include all DataCons for RecFlds

The GREInfo for a record field should include the collection of all
the data constructors of the parent TyCon that have this record field.
This information was being incorrectly computed in the tyThingLocalGREs
function for a DataCon, as we were not taking into account other
DataCons with the same parent TyCon.

Fixes #23546

- - - - -
e6627cbd by Alan Zimmerman at 2023-07-10T17:00:05-04:00
EPA: Simplify GHC/Parser.y comb3

A follow up to !10743

- - - - -
ee20da34 by Bodigrim at 2023-07-10T17:01:01-04:00
Document that compareByteArrays# is available since ghc-prim-0.5.2.0

- - - - -
4926af7b by Matthew Pickering at 2023-07-10T17:01:38-04:00
Revert "Bump text submodule"

This reverts commit d284470a77042e6bc17bdb0ab0d740011196958a.

This commit requires that we bootstrap with ghc-9.4, which we do not
require until #23195 has been completed.

Subsequently this has broken nighty jobs such as the rocky8 job which in
turn has broken nightly releases.

- - - - -
d1c92bf3 by Ben Gamari at 2023-07-11T08:07:02-04:00
compiler: Fingerprint more code generation flags

Previously our recompilation check was quite inconsistent in its
coverage of non-optimisation code generation flags. Specifically, we
failed to account for most flags that would affect the behavior of
generated code in ways that might affect the result of a program's
execution (e.g. `-feager-blackholing`, `-fstrict-dicts`)

Closes #23369.

- - - - -
eb623149 by Ben Gamari at 2023-07-11T08:07:02-04:00
compiler: Record original thunk info tables on stack

Here we introduce a new code generation option, `-forig-thunk-info`,
which ensures that an `stg_orig_thunk_info` frame is pushed before every
update frame. This can be invaluable when debugging thunk cycles and
similar.

See Note [Original thunk info table frames] for details.

Closes #23255.

- - - - -
4731f44e by Jaro Reinders at 2023-07-11T08:07:40-04:00
Fix wrong MIN_VERSION_GLASGOW_HASKELL macros

I forgot to change these after rebasing.

- - - - -
542073cc by Ben Gamari at 2023-07-11T09:15:00-04:00
rts/RtsSymbols: Add AArch64 outline atomic operations

Fixes #22012 by adding the symbols described in
https://github.com/llvm/llvm-project/blob/main/llvm/docs/Atomics.rst#libcalls-atomic.

Ultimately this would be better addressed by #22011, but this is a first
step in the right direction and fixes the immediate symptom.

Generated via:

```python
ops = []
ORDERINGS = [ 'relax', 'acq', 'rel', 'acq_rel' ]
for order in ORDERINGS:
    for n in [1,2,4,8]:
        for op in ['swp', 'ldadd', 'ldclr', 'ldeor', 'ldset']:
            ops.append(f'__aarch64_{op}{n}_{order}')

    for m in [1,2,4,8,16]:
        ops.append(f'__aarch64_cas{n}_{order}')

print('\n'.join(f'    SymE_NeedsProto({op}),' for op in sorted(ops)))
```

- - - - -
c6951512 by Ben Gamari at 2023-07-19T11:21:41-04:00
rts: Add generator for RtsSymbols from libgcc

- - - - -


30 changed files:

- .gitlab-ci.yml
- .gitlab/ci.sh
- − .gitlab/circle-ci-job.sh
- .gitlab/generate-ci/gen_ci.hs
- .gitlab/jobs.yaml
- compiler/GHC/Builtin/Uniques.hs
- compiler/GHC/Builtin/primops.txt.pp
- compiler/GHC/Cmm/CLabel.hs
- compiler/GHC/Cmm/CommonBlockElim.hs
- compiler/GHC/Cmm/Dataflow/Collections.hs
- compiler/GHC/Cmm/Dataflow/Label.hs
- compiler/GHC/Cmm/Dominators.hs
- compiler/GHC/Cmm/LRegSet.hs
- compiler/GHC/Cmm/Sink.hs
- compiler/GHC/CmmToAsm/CFG.hs
- compiler/GHC/CmmToAsm/CFG/Dominators.hs
- compiler/GHC/CmmToAsm/Wasm/Asm.hs
- compiler/GHC/CmmToAsm/Wasm/FromCmm.hs
- compiler/GHC/CmmToAsm/Wasm/Types.hs
- compiler/GHC/CmmToLlvm/Base.hs
- compiler/GHC/Core.hs
- compiler/GHC/Core/Opt/Monad.hs
- compiler/GHC/Core/Opt/Pipeline.hs
- compiler/GHC/Core/Opt/Simplify/Iteration.hs
- compiler/GHC/Core/Opt/Simplify/Monad.hs
- compiler/GHC/Core/Opt/SpecConstr.hs
- compiler/GHC/Core/Subst.hs
- compiler/GHC/Data/Bag.hs
- compiler/GHC/Data/Graph/UnVar.hs
- + compiler/GHC/Data/Word64Map.hs


The diff was not included because it is too large.


View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/b7a3b3b2fcdc7eefe80df6994c347310f29e18ef...c695151212668d5ff3914f7575f23af05e4dcd32

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/b7a3b3b2fcdc7eefe80df6994c347310f29e18ef...c695151212668d5ff3914f7575f23af05e4dcd32
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/20230719/0eb44237/attachment-0001.html>

More information about the ghc-commits mailing list