[Git][ghc/ghc][wip/ghc-toolchain-fixes] 63 commits: Cleanup a TODO introduced in 1f94e0f7

Matthew Pickering (@mpickering) gitlab at gitlab.haskell.org
Tue Aug 15 08:45:27 UTC 2023



Matthew Pickering pushed to branch wip/ghc-toolchain-fixes at Glasgow Haskell Compiler / GHC


Commits:
1a98d673 by Sebastian Graf at 2023-08-09T16:24:29-04:00
Cleanup a TODO introduced in 1f94e0f7

The change must have slipped through review of !4412

- - - - -
2274abc8 by Sebastian Graf at 2023-08-09T16:24:29-04:00
More explicit strictness in GHC.Real

- - - - -
ce8aa54c by Sebastian Graf at 2023-08-09T16:24:30-04:00
exprIsTrivial: Factor out shared implementation

The duplication between `exprIsTrivial` and `getIdFromTrivialExpr_maybe` has
been bugging me for a long time.

This patch introduces an inlinable worker function `trivial_expr_fold` acting
as the single, shared decision procedure of triviality. It "returns" a
Church-encoded `Maybe (Maybe Id)`, so when it is inlined, it fuses to similar
code as before.
(Better code, even, in the case of `getIdFromTrivialExpr` which presently
allocates a `Just` constructor that cancels away after this patch.)

- - - - -
d004a36d by Sebastian Graf at 2023-08-09T16:24:30-04:00
Simplify: Simplification of arguments in a single function

The Simplifier had a function `simplArg` that wasn't called in `rebuildCall`,
which seems to be the main way to simplify args. Hence I consolidated the code
path to call `simplArg`, too, renaming to `simplLazyArg`.

- - - - -
8c73505e by Sebastian Graf at 2023-08-09T16:24:30-04:00
Core.Ppr: Omit case binder for empty case alternatives

A minor improvement to pretty-printing

- - - - -
d8d993f1 by Sebastian Graf at 2023-08-09T16:24:30-04:00
Disable tests RepPolyWrappedVar2 and RepPolyUnsafeCoerce1 in JS backend

... because those coerce between incompatible/unknown PrimReps.

- - - - -
f06e87e4 by Sebastian Graf at 2023-08-09T16:24:30-04:00
Inlining literals into boring contexts is OK

- - - - -
4a6b7c87 by Sebastian Graf at 2023-08-09T16:24:30-04:00
Clarify floating of unsafeEqualityProofs (#23754)

- - - - -
b0f4752e by Sebastian Graf at 2023-08-09T16:24:30-04:00
Kill SetLevel.notWorthFloating.is_triv (#23270)

We have had it since b84ba676034, when it operated on annotated expressions.
Nowadays it operates on vanilla `CoreExpr` though, so we should just call
`exprIsTrivial`; thus handling empty cases and string literals correctly.

- - - - -
7e0c8b3b by Sebastian Graf at 2023-08-09T16:24:30-04:00
ANFise string literal arguments (#23270)

This instates the invariant that a trivial CoreExpr translates to an atomic
StgExpr. Nice.

Annoyingly, in -O0 we sometimes generate
```
foo = case "blah"# of sat { __DEFAULT -> unpackCString# sat }
```
which makes it a bit harder to spot that we can emit a standard
`stg_unpack_cstring` thunk.

Fixes #23270.

- - - - -
357f2738 by Sebastian Graf at 2023-08-09T16:24:30-04:00
Deactivate -fcatch-nonexhaustive-cases in ghc-bignum (#23345)

- - - - -
59202c80 by Sebastian Graf at 2023-08-09T16:24:30-04:00
CorePrep: Eliminate EmptyCase and unsafeEqualityProof in CoreToStg instead

We eliminate EmptyCase by way of `coreToStg (Case e _ _ []) = coreToStg e` now.
The main reason is that it plays far better in conjunction with eta expansion
(as we aim to do for arguments in CorePrep, #23083), because we can discard
any arguments, `(case e of {}) eta == case e of {}`, whereas in `(e |> co) eta`
it's impossible to discard the argument.

We do also give the same treatment to unsafeCoerce proofs and treat them as
trivial iff their RHS is trivial.

It is also both much simpler to describe than the previous mechanism of emitting
an unsafe coercion and simpler to implement, removing quite a bit of commentary
and `CorePrepProv`.

In the ghc/alloc perf test `LargeRecord`, we introduce an additional Simplifier
iteration due to #17910. E.g., FloatOut produces a binding
```
lvl_s6uK [Occ=Once1] :: GHC.Types.Int
[LclId]
lvl_s6uK = GHC.Types.I# 2#

lvl_s6uL [Occ=Once1] :: GHC.Types.Any
[LclId]
lvl_s6uL
  = case Unsafe.Coerce.unsafeEqualityProof ... of
    { Unsafe.Coerce.UnsafeRefl v2_i6tr -> lvl_s6uK `cast` (... v2_i6tr ...)
    }
```
That occurs once and hence is pre-inlined unconditionally in the next Simplifier
pass. It's non-trivial to find a way around that, but not really harmful
otherwise. Hence we accept a 1.2% increase on some architectures.

Metric Increase:
    LargeRecord

- - - - -
00d31188 by Sebastian Graf at 2023-08-09T16:24:30-04:00
CorePrep: Eta expand arguments (#23083)

Previously, we'd only eta expand let bindings and lambdas,
now we'll also eta expand arguments such as in T23083:
```hs
g f h = f (h `seq` (h $))
```
Unless `-fpedantic-bottoms` is set, we'll now transform to
```hs
g f h = f (\eta -> h eta)
```
in CorePrep.

See the new `Note [Eta expansion of arguments in CorePrep]` for the details.

We only do this optimisation with -O2 because we saw 2-3% ghc/alloc regressions
in T4801 and T5321FD.

Fixes #23083.

- - - - -
bf885d7a by Matthew Craven at 2023-08-09T16:25:07-04:00
Bump bytestring submodule to 0.11.5, again

Fixes #23789.

The bytestring commit used here is unreleased;
a release can be made when necessary.

- - - - -
7acbf0fd by Sven Tennie at 2023-08-10T19:17:11-04:00
Serialize CmmRetInfo in .rodata

The handling of case was missing.

- - - - -
0c3136f2 by Sven Tennie at 2023-08-10T19:17:11-04:00
Reference StgRetFun payload by its struct field address

This is easier to grasp than relative pointer offsets.

- - - - -
f68ff313 by Sven Tennie at 2023-08-10T19:17:11-04:00
Better variable name: u -> frame

The 'u' was likely introduced by copy'n'paste.

- - - - -
0131bb7f by Sven Tennie at 2023-08-10T19:17:11-04:00
Make checkSTACK() public

Such that it can also be used in tests.

- - - - -
7b6e1e53 by Sven Tennie at 2023-08-10T19:17:11-04:00
Publish stack related fields in DerivedConstants.h

These will be used in ghc-heap to decode these parts of the stack.

- - - - -
907ed054 by Sven Tennie at 2023-08-10T19:17:11-04:00
ghc-heap: Decode StgStack and its stack frames

Previously, ghc-heap could only decode heap closures.

The approach is explained in detail in note
[Decoding the stack].

- - - - -
6beb6ac2 by Sven Tennie at 2023-08-10T19:17:11-04:00
Remove RetFunType from RetFun stack frame representation

It's a technical detail. The single usage is replaced by a predicate.

- - - - -
006bb4f3 by Sven Tennie at 2023-08-10T19:17:11-04:00
Better parameter name

The call-site uses the term "offset", too.

- - - - -
d4c2c1af by Sven Tennie at 2023-08-10T19:17:11-04:00
Make closure boxing pure

There seems to be no need to do something complicated. However, the
strictness of the closure pointer matters, otherwise a thunk gets
decoded.

- - - - -
8d8426c9 by Sven Tennie at 2023-08-10T19:17:11-04:00
Document entertainGC in test

It wasn't obvious why it's there and what its role is.

Also, increase the "entertainment level" a bit.

I checked in STG and Cmm dumps that this really generates closures (and
is not e.g. constant folded away.)

- - - - -
cc52c358 by Finley McIlwaine at 2023-08-10T19:17:47-04:00
Add -dipe-stats flag

This is useful for seeing which info tables have information.

- - - - -
261c4acb by Finley McIlwaine at 2023-08-10T19:17:47-04:00
Add -finfo-table-map-with-fallback -finfo-table-map-with-stack

The -fno-info-table-map-with-stack flag omits STACK info tables from the info
table map, and the -fno-info-table-map-with-fallback flag omits info tables
with defaulted source locations from the map. In a test on the Agda codebase
the build results were about 7% smaller when both of those types of tables
were omitted.

Adds a test that verifies that passing each combination of these flags
results in the correct output for -dipe-stats, which is disabled for the js
backend since profiling is not implemented.

This commit also refactors a lot of the logic around extracting info tables
from the Cmm results and building the info table map.

This commit also fixes some issues in the users guide rst source to fix
warnings that were noticed while debugging the documentation for these flags.

Fixes #23702

- - - - -
d7047e0d by Jaro Reinders at 2023-08-14T04:41:42-04:00
Add changelog entry for specialised Enum Int64/Word64 instances

- - - - -
52f5e8fb by cydparser at 2023-08-14T04:42:20-04:00
Fix -ddump-to-file and -ddump-timings interaction (#20316)

- - - - -
1274c5d6 by cydparser at 2023-08-14T04:42:20-04:00
Update release notes (#20316)

- - - - -
8e699b23 by Matthew Pickering at 2023-08-14T10:44:47-04:00
base: Add changelog entry for CLC #188

This proposal modified the implementations of copyBytes, moveBytes and
fillBytes (as detailed in the proposal)

https://github.com/haskell/core-libraries-committee/issues/188

- - - - -
026f040a by Matthew Pickering at 2023-08-14T10:45:23-04:00
packaging: Build manpage in separate directory to other documentation

We were installing two copies of the manpage:

* One useless one in the `share/doc` folder, because we copy the doc/
  folder into share/
* The one we deliberately installed into `share/man` etc

The solution is to build the manpage into the `manpage` directory when
building the bindist, and then just install it separately.

Fixes #23707

- - - - -
524c60c8 by Bartłomiej Cieślar at 2023-08-14T13:46:33-04:00
Report deprecated fields bound by record wildcards when used

This commit ensures that we emit the appropriate warnings when
a deprecated record field bound by a record wildcard is used.

For example:

    module A where
    data Foo = Foo {x :: Int, y :: Bool, z :: Char}

    {-# DEPRECATED x "Don't use x" #-}
    {-# WARNING y "Don't use y" #-}

    module B where
    import A

    foo (Foo {..}) = x

This will cause us to emit a "Don't use x" warning, with location the
location of the record wildcard. Note that we don't warn about `y`,
because it is unused in the RHS of `foo`.

Fixes #23382

- - - - -
d6130065 by Matthew Pickering at 2023-08-14T13:47:11-04:00
Add zstd suffix to jobs which rely on zstd

This was causing some confusion as the job was named simply
"x86_64-linux-deb10-validate", which implies a standard configuration
rather than any dependency on libzstd.

- - - - -
e24e44fc by Matthew Pickering at 2023-08-14T13:47:11-04:00
ci: Always run project-version job

This is needed for the downstream test-primops pipeline to workout what
the version of a bindist produced by a pipeline is.

- - - - -
f17b9d62 by Matthew Pickering at 2023-08-14T13:47:11-04:00
gen_ci: Rework how jobs-metadata.json is generated

* We now represent a job group a triple of Maybes, which makes it easier
  to work out when jobs are enabled/disabled on certain pipelines.

```
data JobGroup a = StandardTriple { v :: Maybe (NamedJob a)
                                 , n :: Maybe (NamedJob a)
                                 , r :: Maybe (NamedJob a) }
```

* `jobs-metadata.json`  generation is reworked using the following
  algorithm.
  - For each pipeline type, find all the platforms we are doing builds
    for.
  - Select one build per platform
  - Zip together the results

This way we can choose different pipelines for validate/nightly/release
which makes the metadata also useful for validate pipelines. This
feature is used by the test-primops downstream CI in order to select the
right bindist for testing validate pipelines.

This makes it easier to inspect which jobs are going to be enabled on a
particular pipeline.

- - - - -
f9a5563d by Matthew Pickering at 2023-08-14T13:47:11-04:00
gen_ci: Rules rework

In particular we now distinguish between whether we are dealing with a
Nightly/Release pipeline (which labels don't matter for) and a validate
pipeline where labels do matter.

The overall goal here is to allow a disjunction of labels for validate
pipelines, for example,

> Run a job if we have the full-ci label or test-primops label

Therefore the "ValidateOnly" rules are treated as a set of disjunctions
rather than conjunctions like before.

What this means in particular is that if we want to ONLY run a job if a
label is set, for example, "FreeBSD" label then we have to override the
whole label set.

Fixes #23772

- - - - -
d54b0c1d by Matthew Pickering at 2023-08-14T13:47:11-04:00
ci: set -e for lint-ci-config scripts

- - - - -
994a9b35 by Matthew Pickering at 2023-08-14T13:47:11-04:00
ci: Fix job metadata generation

- - - - -
d09e9fd7 by Matthew Pickering at 2023-08-15T09:42:11+01:00
ghc-toolchain: Match CPP args with configure script

At the moment we need ghc-toolchain to precisely match the output as
provided by the normal configure script. The normal configure script
(FP_HSCPP_CMD_WITH_ARGS) branches on whether we are using clang or gcc
so we match that logic exactly in ghc-toolchain.

The old implementation (which checks if certain flags are supported) is
better but for now we have to match to catch any potential errors in the
configuration.

Ticket: #23720

- - - - -
654d3466 by Matthew Pickering at 2023-08-15T09:42:11+01:00
configure: Fix `-Wl,--no-as-needed` check

The check was failing because the args supplied by $$1 were quoted which
failed because then the C compiler thought they were an input file.

Fixes #23720

- - - - -
a526f006 by Matthew Pickering at 2023-08-15T09:42:11+01:00
configure: Add flag which turns ghc-toolchain check into error

We want to catch these errors in CI, but first we need to a flag which
turns this check into an error.

- - - - -
b25201a6 by Matthew Pickering at 2023-08-15T09:44:53+01:00
ci: Enable --enable-strict-ghc-toolchain-check for all CI jobs

This will cause any CI job to fail if we have a mismatch between what
ghc-toolchain reports and what ./configure natively reports. Fixing
these kinds of issues is highest priority for 9.10 release.

- - - - -
c027eb08 by Rodrigo Mesquita at 2023-08-15T09:44:54+01:00
Pass user-specified options to ghc-toolchain

The current user interface to configuring target toolchains is `./configure`.

In !9263 we added a new tool to configure target toolchains called
`ghc-toolchain`, but the blessed way of creating these toolchains is
still through configure.

However, we were not passing the user-specified options given with the
`./configure` invocation to the ghc-toolchain tool.

This commit remedies that by storing the user options and environment
variables in USER_* variables, which then get passed to GHC-toolchain.

The exception to the rule is the windows bundled toolchain, which
overrides the USER_* variables with whatever flags the windows bundled
toolchain requires to work.
We consider the bundled toolchain to be effectively the user specifying
options, since the actual user delegated that configuration work.

Closes #23678

- - - - -
e8653745 by Rodrigo Mesquita at 2023-08-15T09:44:54+01:00
ghc-toolchain: Parse javascript and ghcjs as a Arch and OS

- - - - -
39f2885e by Rodrigo Mesquita at 2023-08-15T09:44:54+01:00
ghc-toolchain: Fix ranlib option

- - - - -
7005cd12 by Rodrigo Mesquita at 2023-08-15T09:44:54+01:00
Check Link Works with -Werror

- - - - -
b028f79d by Matthew Pickering at 2023-08-15T09:44:54+01:00
Only check for no_compact_unwind support on darwin

While writing ghc-toolchain we noticed that the
FP_PROG_LD_NO_COMPACT_UNWIND check is subtly wrong. Specifically, we
pass -Wl,-no_compact_unwind to cc. However, ld.gold interprets this as
-n o_compact_unwind, which is a valid argument.

Fixes #23676

- - - - -
5516bf1d by Matthew Pickering at 2023-08-15T09:44:54+01:00
Add some javascript special cases to ghc-toolchain

On javascript there isn't a choice of toolchain but some of the
configure checks were not accurately providing the correct answer.

1. The linker was reported as gnu LD because the --version output
   mentioned gnu LD.
2. The --target flag makes no sense on javascript but it was just
   ignored by the linker, so we add a special case to stop ghc-toolchain
   thinking that emcc supports --target when used as a linker.

- - - - -
3963ff97 by Matthew Pickering at 2023-08-15T09:44:54+01:00
check for emcc in gnu_LD check

- - - - -
fc2c08e9 by Matthew Pickering at 2023-08-15T09:44:54+01:00
Add ldOverrideWhitelist to only default to ldOverride on windows/linux

On some platforms - ie darwin, javascript etc we really do not want to
allow the user to use any linker other than the default one as this
leads to all kinds of bugs. Therefore it is a bit more prudant to add a
whitelist which specifies on which platforms it might be possible to use
a different linker.

- - - - -
42a6c3af by Matthew Pickering at 2023-08-15T09:44:54+01:00
Fix plaform glob in FPTOOLS_SET_C_LD_FLAGS

A normal triple may look like

x86_64-unknown-linux

but when cross-compiling you get $target set to a quad such as..

aarch64-unknown-linux-gnu

Which should also match this check.

- - - - -
a446787f by Matthew Pickering at 2023-08-15T09:44:54+01:00
ghc-toolchain: Pass ld-override onto ghc-toolchain

- - - - -
279b67a6 by Matthew Pickering at 2023-08-15T09:44:54+01:00
ld override: Make whitelist override user given option

- - - - -
e8b78cb3 by Matthew Pickering at 2023-08-15T09:44:55+01:00
ghc-toolchain: Add format mode to normalise differences before diffing.

The "format" mode takes an "--input" and "--ouput" target file and
formats it.

This is intended to be useful on windows where the
configure/ghc-toolchain target files can't be diffed very easily because
the path separators are different.

- - - - -
848ff358 by Matthew Pickering at 2023-08-15T09:44:55+01:00
ghc-toolchain: Bump ci-images commit to get new ghc-wasm-meta

We needed to remove -Wno-unused-command-line-argument from the arguments
passed in order for the configure check to report correctly.

See https://gitlab.haskell.org/ghc/ghc/-/merge_requests/10976#note_516335

- - - - -
21cf8090 by Matthew Pickering at 2023-08-15T09:44:55+01:00
configure: MergeObjsCmd - distinguish between empty string and unset variable

If `MergeObjsCmd` is explicitly set to the empty string then we should
assume that MergeObjs is just not supported.

This is especially important for windows where we set MergeObjsCmd to ""
in m4/fp_setup_windows_toolchain.m4.

- - - - -
700692b4 by Matthew Pickering at 2023-08-15T09:44:55+01:00
configure: Add proper check to see if object merging works

- - - - -
b5d34266 by Matthew Pickering at 2023-08-15T09:44:55+01:00
ghc-toolchain: If MergeObjsCmd is not set, replace setting with Nothing

If the user explicitly chooses to not set a MergeObjsCmd then it is
correct to use Nothing for tgtMergeObjs field in the Target file.

- - - - -
d974ccad by Matthew Pickering at 2023-08-15T09:44:55+01:00
HsCppArgs: Augment the HsCppOptions

This is important when we pass -I when setting up the windows toolchain.

- - - - -
896ae50d by Matthew Pickering at 2023-08-15T09:44:55+01:00
Set USER_CPP_ARGS when setting up windows toolchain

- - - - -
91601649 by Rodrigo Mesquita at 2023-08-15T09:44:55+01:00
Improve handling of Cc as a fallback

- - - - -
6c26b983 by Rodrigo Mesquita at 2023-08-15T09:44:55+01:00
ghc-toolchain: Configure Cpp and HsCpp correctly when user specifies flags

In ghc-toolchain, we were only /not/ configuring required flags when the
user specified any flags at all for the  of the HsCpp and Cpp tools.

Otherwise, the linker takes into consideration the user specified flags
to determine whether to search for a better linker implementation, but
already configured the remaining GHC and platform-specific flags
regardless of the user options.

Other Tools consider the user options as a baseline for further
configuration (see `findProgram`), so #23689 is not applicable.

Closes #23689

- - - - -
fd6d34e4 by Matthew Pickering at 2023-08-15T09:44:55+01:00
CPP_ARGS: Put new options after user specified options

This matches up with the behaviour of ghc-toolchain, so that the output
of both matches.

- - - - -


26 changed files:

- .gitlab-ci.yml
- .gitlab/generate-ci/gen_ci.hs
- .gitlab/generate-ci/generate-job-metadata
- .gitlab/generate-ci/generate-jobs
- .gitlab/jobs.yaml
- compiler/GHC/Cmm.hs
- compiler/GHC/Cmm/CLabel.hs
- compiler/GHC/Core.hs
- compiler/GHC/Core/Coercion.hs
- compiler/GHC/Core/Coercion/Opt.hs
- compiler/GHC/Core/FVs.hs
- compiler/GHC/Core/Lint.hs
- compiler/GHC/Core/Opt/Arity.hs
- compiler/GHC/Core/Opt/SetLevels.hs
- compiler/GHC/Core/Opt/Simplify/Iteration.hs
- compiler/GHC/Core/Opt/Simplify/Utils.hs
- compiler/GHC/Core/Ppr.hs
- compiler/GHC/Core/TyCo/FVs.hs
- compiler/GHC/Core/TyCo/Rep.hs
- compiler/GHC/Core/TyCo/Subst.hs
- compiler/GHC/Core/TyCo/Tidy.hs
- compiler/GHC/Core/Type.hs
- compiler/GHC/Core/Unfold.hs
- compiler/GHC/Core/Utils.hs
- compiler/GHC/CoreToIface.hs
- compiler/GHC/CoreToStg.hs


The diff was not included because it is too large.


View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/8e61993439926f17458cc95f57feda5a5d058ebb...fd6d34e4854773c3d37c8f9cf3182cc0ac5d6d2b

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/8e61993439926f17458cc95f57feda5a5d058ebb...fd6d34e4854773c3d37c8f9cf3182cc0ac5d6d2b
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/20230815/adc5b989/attachment-0001.html>


More information about the ghc-commits mailing list