[Git][ghc/ghc][wip/T21134] 71 commits: IPE data compression

Ben Gamari (@bgamari) gitlab at gitlab.haskell.org
Fri Jun 30 16:03:57 UTC 2023 


Ben Gamari pushed to branch wip/T21134 at Glasgow Haskell Compiler / GHC


Commits:
cb9e1ce4 by Finley McIlwaine at 2023-06-18T21:16:45-06:00
IPE data compression

IPE data resulting from the `-finfo-table-map` flag may now be
compressed by configuring the GHC build with the
`--enable-ipe-data-compression` flag. This results in about a 20%
reduction in the size of IPE-enabled build results.

The compression library, zstd, may optionally be statically linked by
configuring with the `--enabled-static-libzstd` flag (on non-darwin
platforms)

libzstd version 1.4.0 or greater is required.

- - - - -
0cbc3ae0 by Gergő Érdi at 2023-06-19T09:11:38-04:00
Add `IfaceWarnings` to represent the `ModIface`-storable parts
of a `Warnings GhcRn`.

Fixes #23516

- - - - -
3e80c2b4 by Arnaud Spiwack at 2023-06-20T03:19:41-04:00
Avoid desugaring non-recursive lets into recursive lets

This prepares for having linear let expressions in the frontend.

When desugaring lets, SPECIALISE statements create more copies of a
let binding. Because of the rewrite rules attached to the bindings,
there are dependencies between the generated binds.

Before this commit, we simply wrapped all these in a mutually
recursive let block, and left it to the simplified to sort it out.

With this commit: we are careful to generate the bindings in
dependency order, so that we can wrap them in consecutive lets (if the
source is non-recursive).

- - - - -
9fad49e0 by Ben Gamari at 2023-06-20T03:20:19-04:00
rts: Do not call exit() from SIGINT handler

Previously `shutdown_handler` would call `stg_exit` if the scheduler was
Oalready found to be in `SCHED_INTERRUPTING` state (or higher). However,
`stg_exit` is not signal-safe as it calls `exit` (which calls `atexit`
handlers). The only safe thing to do in this situation is to call
`_exit`, which terminates with minimal cleanup.

Fixes #23417.

- - - - -
7485f848 by Bodigrim at 2023-06-20T03:20:57-04:00
Bump Cabal submodule

This requires changing the recomp007 test because now cabal passes
`this-unit-id` to executable components, and that unit-id contains a
hash which includes the ABI of the dependencies. Therefore changing the
dependencies means that -this-unit-id changes and recompilation is
triggered.

The spririt of the test is to test GHC's recompilation logic assuming
that `-this-unit-id` is constant, so we explicitly pass `-ipid` to
`./configure` rather than letting `Cabal` work it out.

- - - - -
1464a2a8 by mangoiv at 2023-06-20T03:21:34-04:00
[feat] add a hint to `HasField` error message
- add a hint that indicates that the record that the record dot is used
  on might just be missing a field
- as the intention of the programmer is not entirely clear, it is only
  shown if the type is known
- This addresses in part issue #22382

- - - - -
b65e78dd by Ben Gamari at 2023-06-20T16:56:43-04:00
rts/ipe: Fix unused lock warning

- - - - -
6086effd by Ben Gamari at 2023-06-20T16:56:44-04:00
rts/ProfilerReportJson: Fix memory leak

- - - - -
1e48c434 by Ben Gamari at 2023-06-20T16:56:44-04:00
rts: Various warnings fixes

- - - - -
471486b9 by Ben Gamari at 2023-06-20T16:56:44-04:00
rts: Fix printf format mismatch

- - - - -
80603fb3 by Ben Gamari at 2023-06-20T16:56:44-04:00
rts: Fix incorrect #include <sys/poll.h>

According to Alpine's warnings and poll(2), <poll.h> should be
preferred.

- - - - -
ff18e6fd by Ben Gamari at 2023-06-20T16:56:44-04:00
nonmoving: Fix unused definition warrnings

- - - - -
6e7fe8ee by Ben Gamari at 2023-06-20T16:56:44-04:00
Disable futimens on Darwin.

See #22938

- - - - -
b7706508 by Ben Gamari at 2023-06-20T16:56:44-04:00
rts: Fix incorrect CPP guard

- - - - -
94f00e9b by Ben Gamari at 2023-06-20T16:56:44-04:00
hadrian: Ensure that -Werror is passed when compiling the RTS.

Previously the `+werror` transformer would only pass `-Werror` to GHC,
which does not ensure that the same is passed to the C compiler when
building the RTS. Arguably this is itself a bug but for now we will just
work around this by passing `-optc-Werror` to GHC.

I tried to enable `-Werror` in all C compilations but the boot libraries
are something of a portability nightmare.

- - - - -
5fb54bf8 by Ben Gamari at 2023-06-20T16:56:44-04:00
rts: Disable `#pragma GCC`s on clang compilers

Otherwise the build fails due to warnings. See #23530.

- - - - -
cf87f380 by Ben Gamari at 2023-06-20T16:56:44-04:00
rts: Fix capitalization of prototype

- - - - -
17f250d7 by Ben Gamari at 2023-06-20T16:56:44-04:00
rts: Fix incorrect format specifier

- - - - -
0ff1c501 by Josh Meredith at 2023-06-20T16:57:20-04:00
JS: remove js_broken(22576) in favour of the pre-existing wordsize(32) condition (#22576)

- - - - -
3d1d42b7 by Finley McIlwaine at 2023-06-21T12:04:58-04:00
Memory usage fixes for Haddock

- Do not include `mi_globals` in the `NoBackend` backend. It was only included
  for Haddock, but Haddock does not actually need it. This causes a 200MB
  reduction in max residency when generating haddocks on the Agda codebase
  (roughly 1GB to 800MB).

- Make haddock_{parser,renamer}_perf tests more accurate by forcing docs to
  be written to interface files using `-fwrite-interface`

Bumps haddock submodule.

Metric Decrease:
    haddock.base

- - - - -
8185b1c2 by Finley McIlwaine at 2023-06-21T12:04:58-04:00
Fix associated data family doc structure items

Associated data families were being given their own export DocStructureItems,
which resulted in them being documented separately from their classes in
haddocks. This commit fixes it.

- - - - -
4d356ea3 by Sylvain Henry at 2023-06-21T12:04:59-04:00
JS: implement TH support

- Add ghc-interp.js bootstrap script for the JS interpreter
- Interactively link and execute iserv code from the ghci package
- Incrementally load and run JS code for splices into the running iserv

Co-authored-by: Luite Stegeman <stegeman at gmail.com>

- - - - -
3249cf12 by Sylvain Henry at 2023-06-21T12:04:59-04:00
Don't use getKey

- - - - -
f84ff161 by Sylvain Henry at 2023-06-21T12:04:59-04:00
Stg: return imported FVs

This is used to determine what to link when using the interpreter.
For now it's only used by the JS interpreter but it could easily be used
by the native interpreter too (instead of extracting names from compiled BCOs).

- - - - -
fab2ad23 by Sylvain Henry at 2023-06-21T12:04:59-04:00
Fix some recompilation avoidance tests

- - - - -
a897dc13 by Sylvain Henry at 2023-06-21T12:04:59-04:00
TH_import_loop is now broken as expected

- - - - -
dbb4ad51 by Sylvain Henry at 2023-06-21T12:04:59-04:00
JS: always recompile when TH is enabled (cf #23013)

- - - - -
711b1d24 by Bartłomiej Cieślar at 2023-06-21T12:59:27-04:00
Add support for deprecating exported items (proposal #134)

This is an implementation of the deprecated exports proposal #134.
The proposal introduces an ability to introduce warnings to exports.
This allows for deprecating a name only when it is exported from a specific
module, rather than always depreacting its usage. In this example:

    module A ({-# DEPRECATED "do not use" #-} x) where
    x = undefined
    ---
    module B where
    import A(x)

`x` will emit a warning when it is explicitly imported.

Like the declaration warnings, export warnings are first accumulated within
the `Warnings` struct, then passed into the ModIface, from which they are
then looked up and warned about in the importing module in the `lookup_ie`
helpers of the `filterImports` function (for the explicitly imported names)
and in the `addUsedGRE(s)` functions where they warn about regular usages
of the imported name.

In terms of the AST information, the custom warning is stored in the
extension field of the variants of the `IE` type (see Trees that Grow for
more information).

The commit includes a bump to the haddock submodule added in MR #28

Signed-off-by: Bartłomiej Cieślar <bcieslar2001 at gmail.com>

- - - - -
c1865854 by Ben Gamari at 2023-06-21T12:59:30-04:00
configure: Bump version to 9.8

Bumps Haddock submodule

- - - - -
4e1de71c by Ben Gamari at 2023-06-21T21:07:48-04:00
configure: Bump version to 9.9

Bumps haddock submodule.

- - - - -
5b6612bc by Ben Gamari at 2023-06-23T03:56:49-04:00
rts: Work around missing prototypes errors

Darwin's toolchain inexpliciably claims that `write_barrier` and friends
have declarations without prototypes, despite the fact that (a) they are
definitions, and (b) the prototypes appear only a few lines above. Work
around this by making the definitions proper prototypes.

- - - - -
43b66a13 by Matthew Pickering at 2023-06-23T03:57:26-04:00
ghcup-metadata: Fix date modifier (M = minutes, m = month)

Fixes #23552

- - - - -
564164ef by Luite Stegeman at 2023-06-24T10:27:29+09:00
Support large stack frames/offsets in GHCi bytecode interpreter

Bytecode instructions like PUSH_L (push a local variable) contain
an operand that refers to the stack slot. Before this patch, the
operand type was SmallOp (Word16), limiting the maximum stack
offset to 65535 words. This could cause compiler panics in some
cases (See #22888).

This patch changes the operand type for stack offsets from
SmallOp to Op, removing the stack offset limit.

Fixes #22888

- - - - -
8d6574bc by Sylvain Henry at 2023-06-26T13:15:06-04:00
JS: support levity-polymorphic datatypes (#22360,#22291)

- thread knowledge about levity into PrimRep instead of panicking
- JS: remove assumption that unlifted heap objects are rts objects (TVar#, etc.)

Doing this also fixes #22291 (test added).

There is a small performance hit (~1% more allocations).

Metric Increase:
    T18698a
    T18698b

- - - - -
5578bbad by Matthew Pickering at 2023-06-26T13:15:43-04:00
MR Review Template: Mention "Blocked on Review" label

In order to improve our MR review processes we now have the label
"Blocked on Review" which allows people to signal that a MR is waiting
on a review to happen.

See: https://mail.haskell.org/pipermail/ghc-devs/2023-June/021255.html

- - - - -
4427e9cf by Matthew Pickering at 2023-06-26T13:15:43-04:00
Move MR template to Default.md

This makes it more obvious what you have to modify to affect the default
template rather than looking in the project settings.

- - - - -
522bd584 by Arnaud Spiwack at 2023-06-26T13:16:33-04:00
Revert "Avoid desugaring non-recursive lets into recursive lets"

This (temporary) reverts commit 3e80c2b40213bebe302b1bd239af48b33f1b30ef.

Fixes #23550

- - - - -
c59fbb0b by Torsten Schmits at 2023-06-26T19:34:20+02:00
Propagate breakpoint information when inlining across modules

Tracking ticket: #23394

MR: !10448

* Add constructor `IfaceBreakpoint` to `IfaceTickish`
* Store breakpoint data in interface files
* Store `BreakArray` for the breakpoint's module, not the current module, in BCOs
* Store module name in BCOs instead of `Unique`, since the `Unique` from an `Iface` doesn't match the modules in GHCi's
  state
* Allocate module name in `ModBreaks`, like `BreakArray`
* Lookup breakpoint by module name in GHCi
* Skip creating breakpoint instructions when no `ModBreaks` are available, rather than injecting `ModBreaks` in the
  linker when breakpoints are enabled, and panicking when `ModBreaks` is missing

- - - - -
6f904808 by Greg Steuck at 2023-06-27T16:53:07-04:00
Remove undefined FP_PROG_LD_BUILD_ID from configure.ac's

- - - - -
e89aa072 by Andrei Borzenkov at 2023-06-27T16:53:44-04:00
Remove arity inference in type declarations (#23514)

Arity inference in type declarations was introduced
as a workaround for the lack of @k-binders.

They were added in 4aea0a72040, so I simplified all
of this by simply removing arity inference altogether.

This is part of GHC Proposal #425 "Invisible binders in type
declarations".

- - - - -
459dee1b by Torsten Schmits at 2023-06-27T16:54:20-04:00
Relax defaulting of RuntimeRep/Levity when printing

Fixes #16468

MR: !10702

Only default RuntimeRep to LiftedRep when variables are bound by the toplevel forall

- - - - -
151f8f18 by Torsten Schmits at 2023-06-27T16:54:57-04:00
Remove duplicate link label in linear types docs

- - - - -
ecdc4353 by Rodrigo Mesquita at 2023-06-28T12:24:57-04:00
Stop configuring unused Ld command in `settings`

GHC has no direct dependence on the linker. Rather, we depend upon the C
compiler for linking and an object-merging program (which is typically
`ld`) for production of GHCi objects and merging of C stubs into final
object files.

Despite this, for historical reasons we still recorded information about
the linker into `settings`. Remove these entries from `settings`,
`hadrian/cfg/system.config`, as well as the `configure` logic
responsible for this information.

Closes #23566.

- - - - -
bf9ec3e4 by Bryan Richter at 2023-06-28T12:25:33-04:00
Remove extraneous debug output

- - - - -
7eb68dd6 by Bryan Richter at 2023-06-28T12:25:33-04:00
Work with unset vars in -e mode

- - - - -
49c27936 by Bryan Richter at 2023-06-28T12:25:33-04:00
Pass positional arguments in their positions

By quoting $cmd, the default "bash -i" is a single argument to run, and
no file named "bash -i" actually exists to be run.

- - - - -
887dc4fc by Bryan Richter at 2023-06-28T12:25:33-04:00
Handle unset value in -e context

- - - - -
5ffc7d7b by Rodrigo Mesquita at 2023-06-28T21:07:36-04:00
Configure CPP into settings

There is a distinction to be made between the Haskell Preprocessor and
the C preprocessor. The former is used to preprocess Haskell files,
while the latter is used in C preprocessing such as Cmm files.

In practice, they are both the same program (usually the C compiler) but
invoked with different flags.

Previously we would, at configure time, configure the haskell
preprocessor and save the configuration in the settings file, but,
instead of doing the same for CPP, we had hardcoded in GHC that the CPP
program was either `cc -E` or `cpp`.

This commit fixes that asymmetry by also configuring CPP at configure
time, and tries to make more explicit the difference between HsCpp and
Cpp (see Note [Preprocessing invocations]).

Note that we don't use the standard CPP and CPPFLAGS to configure Cpp,
but instead use the non-standard --with-cpp and --with-cpp-flags.
The reason is that autoconf sets CPP to "$CC -E", whereas we expect the
CPP command to be configured as a standalone executable rather than a
command. These are symmetrical with --with-hs-cpp and
--with-hs-cpp-flags.

Cleanup: Hadrian no longer needs to pass the CPP configuration for CPP
         to be C99 compatible through -optP, since we now configure that
         into settings.

Closes #23422

- - - - -
5efa9ca5 by Ben Gamari at 2023-06-28T21:08:13-04:00
hadrian: Always canonicalize topDirectory

Hadrian's `topDirectory` is intended to provide an absolute path to the
root of the GHC tree. However, if the tree is reached via a symlink this

One question here is whether the `canonicalizePath` call is expensive
enough to warrant caching. In a quick microbenchmark I observed that
`canonicalizePath "."` takes around 10us per call; this seems
sufficiently low not to worry.

Alternatively, another approach here would have been to rather move the
canonicalization into `m4/fp_find_root.m4`. This would have avoided
repeated canonicalization but sadly path canonicalization is a hard
problem in POSIX shell.

Addresses #22451.

- - - - -
b3e1436f by aadaa_fgtaa at 2023-06-28T21:08:53-04:00
Optimise ELF linker (#23464)

- cache last elements of `relTable`, `relaTable` and `symbolTables` in `ocInit_ELF`
- cache shndx table in ObjectCode
- run `checkProddableBlock` only with debug rts

- - - - -
30525b00 by Ben Gamari at 2023-06-28T21:09:30-04:00
compiler: Introduce MO_{ACQUIRE,RELEASE}_FENCE

- - - - -
b787e259 by Ben Gamari at 2023-06-28T21:09:30-04:00
compiler: Drop MO_WriteBarrier

rts: Drop write_barrier

- - - - -
7550b4a5 by Ben Gamari at 2023-06-28T21:09:30-04:00
rts: Drop load_store_barrier()

This is no longer used.

- - - - -
d5f2875e by Ben Gamari at 2023-06-28T21:09:31-04:00
rts: Drop last instances of prim_{write,read}_barrier

- - - - -
965ac2ba by Ben Gamari at 2023-06-28T21:09:31-04:00
rts: Eliminate remaining uses of load_load_barrier

- - - - -
0fc5cb97 by Sven Tennie at 2023-06-28T21:09:31-04:00
compiler: Drop MO_ReadBarrier

- - - - -
7a7d326c by Ben Gamari at 2023-06-28T21:09:31-04:00
rts: Drop load_load_barrier

This is no longer used.

- - - - -
9f63da66 by Sven Tennie at 2023-06-28T21:09:31-04:00
Delete write_barrier function

- - - - -
bb0ed354 by Ben Gamari at 2023-06-28T21:09:31-04:00
rts: Make collectFreshWeakPtrs definition a prototype

x86-64/Darwin's toolchain inexplicably warns that collectFreshWeakPtrs
needs to be a prototype.

- - - - -
ef81a1eb by Sven Tennie at 2023-06-28T21:10:08-04:00
Fix number of free double regs

D1..D4 are defined for aarch64 and thus not free.

- - - - -
c335fb7c by Ryan Scott at 2023-06-28T21:10:44-04:00
Fix typechecking of promoted empty lists

The `'[]` case in `tc_infer_hs_type` is smart enough to handle arity-0 uses of
`'[]` (see the newly added `T23543` test case for an example), but the `'[]`
case in `tc_hs_type` was not. We fix this by changing the `tc_hs_type` case to
invoke `tc_infer_hs_type`, as prescribed in `Note [Future-proofing the type
checker]`.

There are some benign changes to test cases' expected output due to the new
code path using `forall a. [a]` as the kind of `'[]` rather than `[k]`.

Fixes #23543.

- - - - -
fcf310e7 by Rodrigo Mesquita at 2023-06-28T21:11:21-04:00
Configure MergeObjs supports response files rather than Ld

The previous configuration script to test whether Ld supported response
files was
* Incorrect (see #23542)
* Used, in practice, to check if the *merge objects tool* supported
  response files.

This commit modifies the macro to run the merge objects tool (rather
than Ld), using a response file, and checking the result with $NM

Fixes #23542

- - - - -
78b2f3cc by Sylvain Henry at 2023-06-28T21:12:02-04:00
JS: fix JS stack printing (#23565)

- - - - -
9f01d14b by Matthew Pickering at 2023-06-29T04:13:41-04:00
Add -fpolymorphic-specialisation flag (off by default at all optimisation levels)

Polymorphic specialisation has led to a number of hard to diagnose
incorrect runtime result bugs (see #23469, #23109, #21229, #23445) so
this commit introduces a flag `-fpolymorhphic-specialisation` which
allows users to turn on this experimental optimisation if they are
willing to buy into things going very wrong.

Ticket #23469

- - - - -
b1e611d5 by Ben Gamari at 2023-06-29T04:14:17-04:00
Rip out runtime linker/compiler checks

We used to choose flags to pass to the toolchain at runtime based on the
platform running GHC, and in this commit we drop all of those runtime
linker checks

Ultimately, this represents a change in policy: We no longer adapt at
runtime to the toolchain being used, but rather make final decisions
about the toolchain used at /configure time/
(we have deleted Note [Run-time linker info] altogether!).

This works towards the goal of having all toolchain configuration logic
living in the same place, which facilities the work towards a
runtime-retargetable GHC (see #19877).

As of this commit, the runtime linker/compiler logic was moved to
autoconf, but soon it, and the rest of the existing toolchain
configuration logic, will live in the standalone ghc-toolchain program
(see !9263)

In particular, what used to be done at runtime is now as follows:
* The flags -Wl,--no-as-needed for needed shared libs are configured
  into settings
* The flag -fstack-check is configured into settings
* The check for broken tables-next-to-code was outdated
* We use the configured c compiler by default as the assembler program
* We drop `asmOpts` because we already configure -Qunused-arguments flag
  into settings (see !10589)

Fixes #23562

Co-author: Rodrigo Mesquita (@alt-romes)

- - - - -
8b35e8ca by Ben Gamari at 2023-06-29T18:46:12-04:00
Define FFI_GO_CLOSURES

The libffi shipped with Apple's XCode toolchain does not contain a
definition of the FFI_GO_CLOSURES macro, despite containing references
to said macro. Work around this by defining the macro, following the
model of a similar workaround in OpenJDK [1].

[1] https://github.com/openjdk/jdk17u-dev/pull/741/files

- - - - -
d7ef1704 by Ben Gamari at 2023-06-29T18:46:12-04:00
base: Fix incorrect CPP guard

This was guarded on `darwin_HOST_OS` instead of `defined(darwin_HOST_OS)`.

- - - - -
7c7d1f66 by Ben Gamari at 2023-06-29T18:46:48-04:00
rts/Trace: Ensure that debugTrace arguments are used

As debugTrace is a macro we must take care to ensure that
the fact is clear to the compiler lest we see warnings.

- - - - -
cb92051e by Ben Gamari at 2023-06-29T18:46:48-04:00
rts: Various warnings fixes

- - - - -
dec81dd1 by Ben Gamari at 2023-06-29T18:46:48-04:00
hadrian: Ignore warnings in unix and semaphore-compat

- - - - -
5125cba0 by Ben Gamari at 2023-06-30T12:03:49-04:00
Drop latent mentions of -split-objs

Closes #21134.

- - - - -


30 changed files:

- .gitlab-ci.yml
- .gitlab/ci.sh
- .gitlab/darwin/toolchain.nix
- .gitlab/gen_ci.hs
- .gitlab/jobs.yaml
- .gitlab/merge_request_templates/merge-request.md → .gitlab/merge_request_templates/Default.md
- compiler/GHC.hs
- compiler/GHC/Builtin/Names.hs
- compiler/GHC/Builtin/Types.hs
- compiler/GHC/ByteCode/Asm.hs
- compiler/GHC/ByteCode/Instr.hs
- compiler/GHC/ByteCode/Linker.hs
- compiler/GHC/ByteCode/Types.hs
- compiler/GHC/Cmm/MachOp.hs
- compiler/GHC/Cmm/Parser.y
- compiler/GHC/Cmm/Utils.hs
- compiler/GHC/CmmToAsm/AArch64/CodeGen.hs
- compiler/GHC/CmmToAsm/AArch64/Instr.hs
- compiler/GHC/CmmToAsm/AArch64/Ppr.hs
- compiler/GHC/CmmToAsm/PPC/CodeGen.hs
- compiler/GHC/CmmToAsm/Reg/Graph/TrivColorable.hs
- compiler/GHC/CmmToAsm/Wasm/FromCmm.hs
- compiler/GHC/CmmToAsm/X86/CodeGen.hs
- compiler/GHC/CmmToC.hs
- compiler/GHC/CmmToLlvm/CodeGen.hs
- compiler/GHC/Core/FVs.hs
- compiler/GHC/Core/Lint.hs
- compiler/GHC/Core/Map/Expr.hs
- compiler/GHC/Core/Opt/OccurAnal.hs
- compiler/GHC/Core/Opt/Simplify/Iteration.hs


The diff was not included because it is too large.


View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/191fd7b77ed375a70be5e06fc85ddde5047c45e8...5125cba0507700eae7e024baddd12dfca60eb553

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/191fd7b77ed375a70be5e06fc85ddde5047c45e8...5125cba0507700eae7e024baddd12dfca60eb553
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/20230630/8c5ae94b/attachment-0001.html>

More information about the ghc-commits mailing list