Sebastian Graf pushed to branch wip/T23083 at Glasgow Haskell Compiler / GHC
Commits:
e2c91bff by Gergő Érdi at 2023-08-03T02:55:46+01:00
Desugar bindings in the context of their evidence
Closes #23172
- - - - -
481f4a46 by Gergő Érdi at 2023-08-03T07:48:43+01:00
Add flag to `-f{no-}specialise-incoherents` to enable/disable specialisation of
incoherent instances
Fixes #23287
- - - - -
d751c583 by Profpatsch at 2023-08-04T12:24:26-04:00
base: Improve String & IsString documentation
- - - - -
01db1117 by Ben Gamari at 2023-08-04T12:25:02-04:00
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.
- - - - -
fdef003a by Ryan Scott at 2023-08-04T12:25:39-04:00
Look through TH splices in splitHsApps
This modifies `splitHsApps` (a key function used in typechecking function
applications) to look through untyped TH splices and quasiquotes. Not doing so
was the cause of #21077. This builds on !7821 by making `splitHsApps` match on
`HsUntypedSpliceTop`, which contains the `ThModFinalizers` that must be run as
part of invoking the TH splice. See the new `Note [Looking through Template
Haskell splices in splitHsApps]` in `GHC.Tc.Gen.Head`.
Along the way, I needed to make the type of `splitHsApps.set` slightly more
general to accommodate the fact that the location attached to a quasiquote is
a `SrcAnn NoEpAnns` rather than a `SrcSpanAnnA`.
Fixes #21077.
- - - - -
e77a0b41 by Ben Gamari at 2023-08-04T12:26:15-04:00
Bump deepseq submodule to 1.5.
And bump bounds
(cherry picked from commit 1228d3a4a08d30eaf0138a52d1be25b38339ef0b)
- - - - -
cebb5819 by Ben Gamari at 2023-08-04T12:26:15-04:00
configure: Bump minimal boot GHC version to 9.4
(cherry picked from commit d3ffdaf9137705894d15ccc3feff569d64163e8e)
- - - - -
83766dbf by Ben Gamari at 2023-08-04T12:26:15-04:00
template-haskell: Bump version to 2.21.0.0
Bumps exceptions submodule.
(cherry picked from commit bf57fc9aea1196f97f5adb72c8b56434ca4b87cb)
- - - - -
1211112a by Ben Gamari at 2023-08-04T12:26:15-04:00
base: Bump version to 4.19
Updates all boot library submodules.
(cherry picked from commit 433d99a3c24a55b14ec09099395e9b9641430143)
- - - - -
3ab5efd9 by Ben Gamari at 2023-08-04T12:26:15-04:00
testsuite: Normalise versions more aggressively
In backpack hashes can contain `+` characters.
(cherry picked from commit 024861af51aee807d800e01e122897166a65ea93)
- - - - -
d52be957 by Ben Gamari at 2023-08-04T12:26:15-04:00
testsuite: Declare bkpcabal08 as fragile
Due to spurious output changes described in #23648.
(cherry picked from commit c046a2382420f2be2c4a657c56f8d95f914ea47b)
- - - - -
e75a58d1 by Ben Gamari at 2023-08-04T12:26:15-04:00
gitlab-ci: Only mark linker_unload_native as broken in static jobs
This test passes on dynamically-linked Alpine.
(cherry picked from commit f356a7e8ec8ec3d6b2b30fd175598b9b80065d87)
- - - - -
8b176514 by Ben Gamari at 2023-08-04T12:26:15-04:00
testsuite: Update base-exports
- - - - -
4b647936 by Ben Gamari at 2023-08-04T12:26:15-04:00
testsuite/interface-stability: normalise versions
This eliminates spurious changes from version bumps.
- - - - -
0eb54c05 by Ben Gamari at 2023-08-04T12:26:51-04:00
linker/PEi386: Don't sign-extend symbol section number
Previously we incorrectly interpreted PE section numbers as signed
values. However, this isn't the case; rather, it's an unsigned 16-bit number
with a few special bit-patterns (0xffff and 0xfffe). This resulted in #22941
as the linker would conclude that the sections were invalid.
Fixing this required quite a bit of refactoring.
Closes #22941.
- - - - -
fd7ce39c by Ben Gamari at 2023-08-04T12:27:28-04:00
testsuite: Mark MulMayOflo_full as broken rather than skipping
To ensure that we don't accidentally fix it.
See #23742.
- - - - -
824092f2 by Ben Gamari at 2023-08-04T12:27:28-04:00
nativeGen/AArch64: Fix sign extension in MulMayOflo
Previously the 32-bit implementations of MulMayOflo would use the
a non-sensical sign-extension mode. Rewrite these to reflect what gcc 11
produces. Also similarly rework the 16- and 8-bit cases.
This now passes the MulMayOflo tests in ghc/test-primops> in all four
widths, including the precision tests.
Fixes #23721.
- - - - -
1b15dbc4 by Jan Hrček at 2023-08-04T12:28:08-04:00
Fix haddock markup in code example for coerce
- - - - -
46fd8ced by Vladislav Zavialov at 2023-08-04T12:28:44-04:00
Fix (~) and (@) infix operators in TH splices (#23748)
8168b42a "Whitespace-sensitive bang patterns" allows GHC to accept
the following infix operators:
a ~ b = ()
a @ b = ()
But not if TH is used to generate those declarations:
$([d| a ~ b = ()
a @ b = ()
|])
-- Test.hs:5:2: error: [GHC-55017]
-- Illegal variable name: ‘~’
-- When splicing a TH declaration: (~_0) a_1 b_2 = GHC.Tuple.Prim.()
This is easily fixed by modifying `reservedOps` in GHC.Utils.Lexeme
- - - - -
a1899d8f by Aaron Allen at 2023-08-04T12:29:24-04:00
[#23663] Show Flag Suggestions in GHCi
Makes suggestions when using `:set` in GHCi with a misspelled flag. This
mirrors how invalid flags are handled when passed to GHC directly. Logic
for producing flag suggestions was moved to GHC.Driver.Sesssion so it
can be shared.
resolves #23663
- - - - -
03f2debd by Rodrigo Mesquita at 2023-08-04T12:30:00-04:00
Improve ghc-toolchain validation configure warning
Fixes the layout of the ghc-toolchain validation warning produced by
configure.
- - - - -
de25487d by Alan Zimmerman at 2023-08-04T12:30:36-04:00
EPA make getLocA a synonym for getHasLoc
This is basically a no-op change, but allows us to make future changes
that can rely on the HasLoc instances
And I presume this means we can use more precise functions based on
class resolution, so the Windows CI build reports
Metric Decrease:
T12234
T13035
- - - - -
3ac423b9 by Ben Gamari at 2023-08-04T12:31:13-04:00
ghc-platform: Add upper bound on base
Hackage upload requires this.
- - - - -
8ba20b21 by Matthew Craven at 2023-08-04T17:22:59-04:00
Adjust and clarify handling of primop effects
Fixes #17900; fixes #20195.
The existing "can_fail" and "has_side_effects" primop attributes
that previously governed this were used in inconsistent and
confusingly-documented ways, especially with regard to raising
exceptions. This patch replaces them with a single "effect"
attribute, which has four possible values: NoEffect, CanFail,
ThrowsException, and ReadWriteEffect. These are described in
Note [Classifying primop effects].
A substantial amount of related documentation has been re-drafted
for clarity and accuracy.
In the process of making this attribute format change for literally
every primop, several existing mis-classifications were detected and
corrected. One of these mis-classifications was tagToEnum#, which
is now considered CanFail; this particular fix is known to cause a
regression in performance for derived Enum instances. (See #23782.)
Fixing this is left as future work.
New primop attributes "cheap" and "work_free" were also added,
and used in the corresponding parts of GHC.Core.Utils.
In view of their actual meaning and uses, `primOpOkForSideEffects`
and `exprOkForSideEffects` have been renamed to `primOpOkToDiscard`
and `exprOkToDiscard`, respectively.
Metric Increase:
T21839c
- - - - -
41bf2c09 by sheaf at 2023-08-04T17:23:42-04:00
Update inert_solved_dicts for ImplicitParams
When adding an implicit parameter dictionary to the inert set, we must
make sure that it replaces any previous implicit parameter dictionaries
that overlap, in order to get the appropriate shadowing behaviour, as in
let ?x = 1 in let ?x = 2 in ?x
We were already doing this for inert_cans, but we weren't doing the same
thing for inert_solved_dicts, which lead to the bug reported in #23761.
The fix is thus to make sure that, when handling an implicit parameter
dictionary in updInertDicts, we update **both** inert_cans and
inert_solved_dicts to ensure a new implicit parameter dictionary
correctly shadows old ones.
Fixes #23761
- - - - -
43578d60 by Matthew Craven at 2023-08-05T01:05:36-04:00
Bump bytestring submodule to 0.11.5.1
- - - - -
91353622 by Ben Gamari at 2023-08-05T01:06:13-04:00
Initial commit of Note [Thunks, blackholes, and indirections]
This Note attempts to summarize the treatment of thunks, thunk update,
and indirections.
This fell out of work on #23185.
- - - - -
8d686854 by sheaf at 2023-08-05T01:06:54-04:00
Remove zonk in tcVTA
This removes the zonk in GHC.Tc.Gen.App.tc_inst_forall_arg and its
accompanying Note [Visible type application zonk]. Indeed, this zonk
is no longer necessary, as we no longer maintain the invariant that
types are well-kinded without zonking; only that typeKind does not
crash; see Note [The Purely Kinded Type Invariant (PKTI)].
This commit removes this zonking step (as well as a secondary zonk),
and replaces the aforementioned Note with the explanatory
Note [Type application substitution], which justifies why the
substitution performed in tc_inst_forall_arg remains valid without
this zonking step.
Fixes #23661
- - - - -
19dea673 by Ben Gamari at 2023-08-05T01:07:30-04:00
Bump nofib submodule
Ensuring that nofib can be build using the same range of bootstrap
compilers as GHC itself.
- - - - -
aa07402e by Luite Stegeman at 2023-08-05T23:15:55+09:00
JS: Improve compatibility with recent emsdk
The JavaScript code in libraries/base/jsbits/base.js had some
hardcoded offsets for fields in structs, because we expected
the layout of the data structures to remain unchanged. Emsdk
3.1.42 changed the layout of the stat struct, breaking this
assumption, and causing code in .hsc files accessing the
stat struct to fail.
This patch improves compatibility with recent emsdk by
removing the assumption that data layouts stay unchanged:
1. offsets of fields in structs used by JavaScript code are
now computed by the configure script, so both the .js and
.hsc files will automatically use the new layout if anything
changes.
2. the distrib/configure script checks that the emsdk version
on a user's system is the same version that a bindist was
booted with, to avoid data layout inconsistencies
See #23641
- - - - -
b938950d by Luite Stegeman at 2023-08-07T06:27:51-04:00
JS: Fix missing local variable declarations
This fixes some missing local variable declarations that were
found by running the testsuite in strict mode.
Fixes #23775
- - - - -
86b6cb0a by Sebastian Graf at 2023-08-07T17:36:29+02:00
Cleanup a TODO introduced in 1f94e0f7
The change must have slipped through review of !4412
- - - - -
0b844408 by Sebastian Graf at 2023-08-07T17:36:29+02:00
More explicit strictness in GHC.Real
- - - - -
db9d796b by Sebastian Graf at 2023-08-07T17:37:56+02:00
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.)
- - - - -
2c45bbb1 by Sebastian Graf at 2023-08-07T17:37:57+02:00
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`.
- - - - -
79f6c285 by Sebastian Graf at 2023-08-07T17:37:57+02:00
Core.Ppr: Omit case binder for empty case alternatives
A minor improvement to pretty-printing
- - - - -
9821e904 by Sebastian Graf at 2023-08-07T17:37:58+02:00
Disable test RepPolyWrappedVar2 in JS backend
- - - - -
4b7c4370 by Sebastian Graf at 2023-08-07T17:37:58+02:00
Inlining literals into boring contexts is OK
- - - - -
97e09d06 by Sebastian Graf at 2023-08-07T17:46:11+02:00
Clarify floating of unsafeEqualityProofs (#23754)
- - - - -
23ccb5fe by Sebastian Graf at 2023-08-07T17:46:13+02:00
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.
- - - - -
40654a5b by Sebastian Graf at 2023-08-07T17:46:13+02:00
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.
- - - - -
6ffb3b84 by Sebastian Graf at 2023-08-07T17:46:13+02:00
Deactivate -fcatch-nonexhaustive-cases in ghc-bignum (#23345)
- - - - -
ba9747d6 by Sebastian Graf at 2023-08-07T17:46:13+02:00
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`.
- - - - -
d4b882fe by Sebastian Graf at 2023-08-07T17:46:13+02:00
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.
- - - - -
25 changed files:
- .gitlab-ci.yml
- .gitlab/generate-ci/gen_ci.hs
- .gitlab/jobs.yaml
- compiler/GHC/Builtin/PrimOps.hs
- compiler/GHC/Builtin/primops.txt.pp
- compiler/GHC/CmmToAsm/AArch64/CodeGen.hs
- compiler/GHC/Core.hs
- compiler/GHC/Core/Coercion.hs
- compiler/GHC/Core/Coercion/Opt.hs
- compiler/GHC/Core/FVs.hs
- compiler/GHC/Core/InstEnv.hs
- compiler/GHC/Core/Lint.hs
- compiler/GHC/Core/Opt/ConstantFold.hs
- compiler/GHC/Core/Opt/FloatIn.hs
- compiler/GHC/Core/Opt/SetLevels.hs
- compiler/GHC/Core/Opt/Simplify/Iteration.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
The diff was not included because it is too large.
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/869da8e25ef2ca038096cd668d60fad140b211f3...d4b882fecd65deeb8d108e498c9a16245c203856
--
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/869da8e25ef2ca038096cd668d60fad140b211f3...d4b882fecd65deeb8d108e498c9a16245c203856
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/20230807/8dc251c6/attachment-0001.html>