[Git][ghc/ghc][wip/fendor/ghc-iface-sharing-avoid-reserialisation] 25 commits: Fix a QuickLook bug

Hannes Siebenhandl (@fendor) gitlab at gitlab.haskell.org
Sat Jun 15 12:24:54 UTC 2024



Hannes Siebenhandl pushed to branch wip/fendor/ghc-iface-sharing-avoid-reserialisation at Glasgow Haskell Compiler / GHC


Commits:
2747cd34 by Simon Peyton Jones at 2024-06-12T12:51:37-04:00
Fix a QuickLook bug

This MR fixes the bug exposed by #24676.  The problem was that
quickLookArg was trying to avoid calling tcInstFun unnecessarily; but
it was in fact necessary.  But that in turn forced me into a
significant refactoring, putting more fields into EValArgQL.

Highlights: see Note [Quick Look overview] in GHC.Tc.Gen.App

* Instantiation variables are now distinguishable from ordinary
  unification variables, by level number = QLInstVar. This is
  treated like "level infinity".  See Note [The QLInstVar TcLevel]
  in GHC.Tc.Utils.TcType.

* In `tcApp`, we don't track the instantiation variables in a set Delta
  any more; instead, we just tell them apart by their level number.

* EValArgQL now much more clearly captures the "half-done" state
  of typechecking an argument, ready for later resumption.
  See Note [Quick Look at value arguments] in GHC.Tc.Gen.App

* Elminated a bogus (never used) fast-path in
  GHC.Tc.Utils.Instantiate.instCallConstraints
  See Note [Possible fast path for equality constraints]

Many other small refactorings.

- - - - -
1b1523b1 by George Thomas at 2024-06-12T12:52:18-04:00
Fix non-compiling extensible record `HasField` example
- - - - -
97b141a3 by Zubin Duggal at 2024-06-12T12:52:55-04:00
haddock: Fix hyperlinker source urls (#24907)

This fixes a bug introduced by f56838c36235febb224107fa62334ebfe9941aba Links to
external modules in the hyperlinker are uniformly generated using splicing the
template given to us instead of attempting to construct the url in an ad-hoc manner.

- - - - -
954f864c by Zubin Duggal at 2024-06-12T12:52:55-04:00
haddock: Add name anchor to external source urls from documentation page

URLs for external source links from documentation pages were missing a splice
location for the name.

Fixes #24912

- - - - -
b0b64177 by Simon Peyton Jones at 2024-06-12T12:53:31-04:00
Prioritise nominal equalities

The main payload of this patch is

* Prioritise nominal equalities in the constraint solver. This
  ameliorates the incompleteness of solving for representational
  constraints over newtypes: see #24887.

   See (EX2) in Note [Decomposing newtype equalities] in
   GHC.Tc.Solver.Equality

In doing this patch I tripped over some other things that I refactored:

* Move `isCoVarType` from `GHC.Core.Type` to `GHC.Core.Predicate`
  where it seems more at home.

* Clarify the "rewrite role" of a constraint.  I was very puzzled
  about what the role of, say `(Eq a)` might be, but see the new
  Note [The rewrite-role of a constraint].

  In doing so I made predTypeEqRel crash when given a non-equality.
  Usually it expects an equality; but it was being mis-used for
  the above rewrite-role stuff.

- - - - -
cb7c1b83 by Liam Goodacre at 2024-06-12T12:54:09-04:00
compiler: missing-deriving-strategies suggested fix

Extends the missing-deriving-strategies warning with a suggested fix
that includes which deriving strategies were assumed.

For info about the warning, see comments for
`TcRnNoDerivStratSpecified`, `TcRnNoDerivingClauseStrategySpecified`, &
`TcRnNoStandaloneDerivingStrategySpecified`.

For info about the suggested fix, see
`SuggestExplicitDerivingClauseStrategies` &
`SuggestExplicitStandalanoDerivingStrategy`.

docs: Rewords missing-deriving-strategies to mention the suggested fix.

Resolves #24955

- - - - -
4e36d3a3 by Jan Hrček at 2024-06-12T12:54:48-04:00
Further haddocks improvements in Language.Haskell.Syntax.Pat.Pat

- - - - -
558353f4 by Cheng Shao at 2024-06-12T12:55:24-04:00
rts: use page sized mblocks on wasm

This patch changes mblock size to page size on wasm. It allows us to
simplify our wasi-libc fork, makes it much easier to test third party
libc allocators like emmalloc/mimalloc, as well as experimenting with
threaded RTS in wasm.

- - - - -
b3cc5366 by Matthew Pickering at 2024-06-12T23:06:57-04:00
compiler: Make ghc-experimental not wired in

If you need to wire in definitions, then place them in ghc-internal and
reexport them from ghc-experimental.

Ticket #24903

- - - - -
700eeab9 by Hécate Kleidukos at 2024-06-12T23:07:37-04:00
base: Use a more appropriate unicode arrow for the ByteArray diagram

This commit rectifies the usage of a unicode arrow in favour of one that
doesn't provoke mis-alignment.

- - - - -
cca7de25 by Matthew Pickering at 2024-06-12T23:08:14-04:00
ghcup-metadata: Fix debian version ranges

This was caught by `ghcup-ci` failing and attempting to install a deb12
bindist on deb11.

```
configure: WARNING: m4/prep_target_file.m4: Expecting YES/NO but got  in ArSupportsDashL_STAGE0. Defaulting to False.
bin/ghc-toolchain-bin: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.34' not found (required by bin/ghc-toolchain-bin)
bin/ghc-toolchain-bin: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.34' not found (required by /tmp/tmp.LBhwvFbVoy/foobarbaz/.ghcup/tmp/ghcup-708d9668d5d82287/ghc-9.11.20240609-x86_64-unknown-linux/bin/../lib/x86_64-linux-ghc-9.11.20240609/libHSunix-2.8.5.1-inplace-ghc9.11.20240609.so)
bin/ghc-toolchain-bin: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.33' not found (required by /tmp/tmp.LBhwvFbVoy/foobarbaz/.ghcup/tmp/ghcup-708d9668d5d82287/ghc-9.11.20240609-x86_64-unknown-linux/bin/../lib/x86_64-linux-ghc-9.11.20240609/libHSunix-2.8.5.1-inplace-ghc9.11.20240609.so)
```

Fixes #24974

- - - - -
7b23ce8b by Pierre Le Marre at 2024-06-13T15:35:04-04:00
ucd2haskell: remove Streamly dependency + misc

- Remove dead code.
- Remove `streamly` dependency.
- Process files with `bytestring`.
- Replace Unicode files parsers with the corresponding ones from the
  package `unicode-data-parser`.
- Simplify cabal file and rename module
- Regenerate `ghc-internal` Unicode files with new header

- - - - -
4570319f by Jacco Krijnen at 2024-06-13T15:35:41-04:00
Document how to run haddocks tests (#24976)

Also remove ghc 9.7 requirement

- - - - -
fb629e24 by amesgen at 2024-06-14T00:28:20-04:00
compiler: refactor lower_CmmExpr_Ptr

- - - - -
def46c8c by amesgen at 2024-06-14T00:28:20-04:00
compiler: handle CmmRegOff in lower_CmmExpr_Ptr

- - - - -
ce76bf78 by Simon Peyton Jones at 2024-06-14T00:28:56-04:00
Small documentation update in Quick Look

- - - - -
19bcfc9b by Simon Peyton Jones at 2024-06-14T14:44:19-04:00
Add hack for #24623

..Th bug in #24623 is randomly triggered by this MR!..

- - - - -
7a08a025 by Simon Peyton Jones at 2024-06-14T14:44:19-04:00
Various fixes to type-tidying

This MR was triggered by #24868, but I found a number of bugs
and infelicities in type-tidying as I went along.  Highlights:

* Fix to #24868 is in GHC.Tc.Errors.report_unsolved: avoid
  using the OccNames of /bound/ variables when tidying /free/
  variables; see the call to `tidyAvoiding`.  That avoid the
  gratuitous renaming which was the cause of #24868. See
     Note [tidyAvoiding] in GHC.Core.TyCo.Tidy

* Refactor and document the tidying of open types.
  See GHC.Core.TyCo.Tidy
     Note [Tidying open types]
     Note [Tidying is idempotent]

* Tidy the coercion variable in HoleCo. That's important so
  that tidied types have tidied kinds.

* Some small renaming to make things consistent.  In particular
  the "X" forms return a new TidyEnv.  E.g.
     tidyOpenType  :: TidyEnv -> Type -> Type
     tidyOpenTypeX :: TidyEnv -> Type -> (TidyEnv, Type)

- - - - -
2eac0288 by Simon Peyton Jones at 2024-06-14T14:44:19-04:00
Wibble

- - - - -
e5d24cc2 by Simon Peyton Jones at 2024-06-14T14:44:20-04:00
Wibbles

- - - - -
246bc3a4 by Simon Peyton Jones at 2024-06-14T14:44:56-04:00
Localise a case-binder in SpecConstr.mkSeqs

This small change fixes #24944

See (SCF1) in Note [SpecConstr and strict fields]

- - - - -
a5994380 by Sylvain Henry at 2024-06-15T03:20:29-04:00
PPC: display foreign label in panic message (cf #23969)

- - - - -
bd95553a by Rodrigo Mesquita at 2024-06-15T03:21:06-04:00
cmm: Parse MO_BSwap primitive operation

Parsing this operation allows it to be tested using `test-primops` in a
subsequent MR.

- - - - -
fca4b83b by Fendor at 2024-06-15T14:24:32+02:00
Improve sharing of duplicated values in `ModIface`, fixes #24723

As a `ModIface` often contains duplicated values that are not
necessarily shared, we improve sharing by serialising the `ModIface`
to an in-memory byte array. Serialisation uses deduplication tables, and
deserialisation implicitly shares duplicated values.

This helps reducing the peak memory usage while compiling in
`--make` mode. The peak memory usage is especially smaller when
generating interface files with core expressions
(`-fwrite-if-simplified-core`).

On agda, this reduces the peak memory usage:

* `2.2 GB` to `1.9 GB` for a ghci session.

On `lib:Cabal`, we report:

* `570 MB` to `500 MB` for a ghci session
* `790 MB` to `667 MB` for compiling `lib:Cabal` with ghc

There is a small impact on execution time, around 2% on the agda code
base.

- - - - -
146d298e by Fendor at 2024-06-15T14:24:32+02:00
Avoid unneccessarily re-serialising the `ModIface`

To reduce memory usage of `ModIface`, we serialise `ModIface` to an
in-memory byte array, which implicitly shares duplicated values.

This serialised byte array can be reused to avoid work when we actually
write the `ModIface` to disk.
We introduce a new field to `ModIface` which allows us to save the byte
array, and write it direclty to disk if the `ModIface` wasn't changed
after the initial serialisation.

This requires us to change absolute offsets, for example to jump to the
deduplication table for `Name` or `FastString` with relative offsets, as
the deduplication byte array doesn't contain header information, such as
fingerprints.
To allow us to dump the binary blob to disk, we need to replace all
absolute offsets with relative ones.

We introduce additional helpers for `ModIface` binary serialisation, which
construct relocatable binary blobs. We say the binary blob is relocatable,
if the binary representation can be moved and does not contain any
absolute offsets.

Further, we introduce new primitives for `Binary` that allow to create
relocatable binaries, such as `forwardGetRel` and `forwardPutRel`.

-------------------------
Metric Decrease:
    MultiLayerModulesDefsGhcWithCore
Metric Increase:
    MultiComponentModules
    MultiLayerModules
    T10421
    T12150
    T12234
    T12425
    T13035
    T13253-spj
    T13701
    T13719
    T14697
    T15703
    T16875
    T18140
    T18304
    T18698a
    T18730
    T18923
    T20049
    T24582
    T5837
    T6048
    T9198
    T9961
    mhu-perf
-------------------------

These metric increases may look bad, but they are all completely benign,
we simply allocate 1 MB per module for `shareIface`. As this allocation
is quite quick, it has a negligible impact on run-time performance.
In fact, the performance difference wasn't measurable on my local
machine. Reducing the size of the pre-allocated 1 MB buffer avoids these
test failures, but also requires us to reallocate the buffer if the
interface file is too big. These reallocations *did* have an impact on
performance, which is why I have opted to accept all these metric
increases, as the number of allocated bytes is merely a guidance.

This 1MB allocation increase causes a lot of tests to fail that
generally have a low allocation number. E.g., increasing from 40MB to
41MB is a 2.5% increase.
In particular, the tests T12150, T13253-spj, T18140, T18304, T18698a,
T18923, T20049, T24582, T5837, T6048, and T9961 only fail on i386-darwin
job, where the number of allocated bytes seems to be lower than in other
jobs.
The test T16875 fails on i386-linux-debian10 for the same reason.

- - - - -


30 changed files:

- .gitlab/rel_eng/mk-ghcup-metadata/mk_ghcup_metadata.py
- compiler/GHC.hs
- compiler/GHC/Builtin/Names.hs
- compiler/GHC/Cmm/Parser.y
- compiler/GHC/CmmToAsm/PPC/Ppr.hs
- compiler/GHC/CmmToAsm/Wasm/FromCmm.hs
- compiler/GHC/Core.hs
- compiler/GHC/Core/DataCon.hs
- compiler/GHC/Core/Lint.hs
- compiler/GHC/Core/Make.hs
- compiler/GHC/Core/Opt/SpecConstr.hs
- compiler/GHC/Core/Predicate.hs
- compiler/GHC/Core/Rules.hs
- compiler/GHC/Core/SimpleOpt.hs
- compiler/GHC/Core/TyCo/FVs.hs
- compiler/GHC/Core/TyCo/Ppr.hs
- compiler/GHC/Core/TyCo/Tidy.hs
- compiler/GHC/Core/Type.hs
- compiler/GHC/Core/Utils.hs
- compiler/GHC/Driver/Config/Core/Lint.hs
- compiler/GHC/Iface/Binary.hs
- compiler/GHC/Iface/Ext/Binary.hs
- compiler/GHC/Iface/Ext/Fields.hs
- compiler/GHC/Iface/Load.hs
- compiler/GHC/Iface/Make.hs
- compiler/GHC/Iface/Recomp.hs
- compiler/GHC/Iface/Rename.hs
- compiler/GHC/Iface/Tidy/StaticPtrTable.hs
- compiler/GHC/Rename/Module.hs
- compiler/GHC/Rename/Names.hs


The diff was not included because it is too large.


View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/f5f0961718fc023c2f293aab56cd1f3c927c4455...146d298ecaa3b343b35b2a56713d1b022ce77728

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/f5f0961718fc023c2f293aab56cd1f3c927c4455...146d298ecaa3b343b35b2a56713d1b022ce77728
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/20240615/d97228ff/attachment-0001.html>


More information about the ghc-commits mailing list