| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This adds an I/O event manager backend based on the io_uring interface
available in modern Linux kernels. The relevant code is in
/libraries/base/GHC/Event/IoUring.hsc. It also contains the following
changes to support the new backend:
* Port over the io_uring bindings from @bgamari into base.
* Add a new HAVE_IO_URING check similar to HAVE_EPOLL, HAVE_KQUEUE etc.
All changes are no-opped if this flag is not set.
See also: #18390
|
| |
|
| |
|
| |
|
|
|
|
| |
isInfoTableLabel does not take Cmm info table into account. This patch is required for data section layout of wasm32 NCG to work.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Ticket #22162 pointed out that the build directory was leaking into the
ABI hash of a module because the BufPos depended on the location of the
build tree.
BufPos is only used in GHC.Parser.PostProcess.Haddock, and the
information doesn't need to be propagated outside the context of a
module.
Fixes #22162
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Clang 13+ supports annotating a return statement using the musttail
attribute, which guarantees that it lowers to a tail call if compilation
succeeds.
This patch takes advantage of that feature for the unregisterised code
generator. The configure script tests availability of the musttail
attribute, if it's available, the Cmm tail calls will become C tail
calls that avoids the mini interpreter trampoline overhead. Nothing is
affected if the musttail attribute is not supported.
Clang documentation:
https://clang.llvm.org/docs/AttributeReference.html#musttail
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When doing performance debugging on #22084 / !8901, I found that the
algorithm in SpecConstr.decreaseSpecCount was so aggressive that if
there were /more/ specialisations available for an outer function,
that could more or less kill off specialisation for an /inner/
function. (An example was in nofib/spectral/fibheaps.)
This patch makes it a bit more aggressive, by dividing by 2, rather
than by the number of outer specialisations.
This makes the program bigger, temporarily:
T19695(normal) ghc/alloc +11.3% BAD
because we get more specialisation. But lots of other programs
compile a bit faster and the geometric mean in perf/compiler
is 0.0%.
Metric Increase:
T19695
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We were religiously keeping exit join points throughout, which
had some bad effects (#21148, #22084).
This MR does two things:
* Arranges that exit join points are inhibited from inlining
only in /one/ Simplifier pass (right after Exitification).
See Note [Be selective about not-inlining exit join points]
in GHC.Core.Opt.Exitify
It's not a big deal, but it shaves 0.1% off compile times.
* Inline used-once non-recursive join points very aggressively
Given join j x = rhs in
joinrec k y = ....j x....
where this is the only occurrence of `j`, we want to inline `j`.
(Unless sm_keep_exits is on.)
See Note [Inline used-once non-recursive join points] in
GHC.Core.Opt.Simplify.Utils
This is just a tidy-up really. It doesn't change allocation, but
getting rid of a binding is always good.
Very effect on nofib -- some up and down.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch teachs the code generation logic of -fno-code about
-fprefer-byte-code, so that if we need to generate code for a module
which prefers byte code, then we generate byte code rather than object
code.
We keep track separately which modules need object code and which byte
code and then enable the relevant code generation for each. Typically
the option will be enabled globally so one of these sets should be empty
and we will just turn on byte code or object code generation.
We also fix the bug where we would generate code for a module which
enables Template Haskell despite the fact it was unecessary.
Fixes #22016
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit adds three new flags
* -fwrite-if-simplified-core: Writes the whole core program into an interface
file
* -fbyte-code-and-object-code: Generate both byte code and object code
when compiling a file
* -fprefer-byte-code: Prefer to use byte-code if it's available when
running TH splices.
The goal for including the core bindings in an interface file is to be able to restart the compiler pipeline
at the point just after simplification and before code generation. Once compilation is
restarted then code can be created for the byte code backend.
This can significantly speed up
start-times for projects in GHCi. HLS already implements its own version of these extended interface
files for this reason.
Preferring to use byte-code means that we can avoid some potentially
expensive code generation steps (see #21700)
* Producing object code is much slower than producing bytecode, and normally you
need to compile with `-dynamic-too` to produce code in the static and dynamic way, the
dynamic way just for Template Haskell execution when using a dynamically linked compiler.
* Linking many large object files, which happens once per splice, can be quite
expensive compared to linking bytecode.
And you can get GHC to compile the necessary byte code so
`-fprefer-byte-code` has access to it by using
`-fbyte-code-and-object-code`.
Fixes #21067
|
|
|
|
|
| |
We want to put implicit binds into fat interface files, so the easiest
thing to do seems to be to treat them uniformly with other binders.
|
|
|
|
|
| |
as `escapeJsonString` is used in `renderJSON`, so the `JSString`
constructor is meant to carry the unescaped string.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If a rewrite rule and a rewrite rule compete in the simplifier, this
patch makes sure that the rewrite rule "win". That is, in general
a bit fragile, but it's a huge help when making specialisation work
reliably, as #21851 and #22097 showed.
The change is fairly straightforwad, and documented in
Note [Rewrite rules and inlining]
in GHC.Core.Opt.Simplify.Iteration.
Compile-times change, up and down a bit -- in some cases because
we get better specialisation. But the payoff (more reliable
specialisation) is large.
Metrics: compile_time/bytes allocated
-----------------------------------------------
T10421(normal) +3.7% BAD
T10421a(normal) +5.5%
T13253(normal) +1.3%
T14052(ghci) +1.8%
T15304(normal) -1.4%
T16577(normal) +3.1% BAD
T17516(normal) +2.3%
T17836(normal) -1.9%
T18223(normal) -1.8%
T8095(normal) -1.3%
T9961(normal) +2.5% BAD
geo. mean +0.0%
minimum -1.9%
maximum +5.5%
Nofib results are (bytes allocated)
+-------------------------------++----------+
| ||tsv (rel) |
+===============================++==========+
| imaginary/paraffins || +0.27% |
| imaginary/rfib || -0.04% |
| real/anna || +0.02% |
| real/fem || -0.04% |
| real/fluid || +1.68% |
| real/gamteb || -0.34% |
| real/gg || +1.54% |
| real/hidden || -0.01% |
| real/hpg || -0.03% |
| real/infer || -0.03% |
| real/prolog || +0.02% |
| real/veritas || -0.47% |
| shootout/fannkuch-redux || -0.03% |
| shootout/k-nucleotide || -0.02% |
| shootout/n-body || -0.06% |
| shootout/spectral-norm || -0.01% |
| spectral/cryptarithm2 || +1.25% |
| spectral/fibheaps || +18.33% |
| spectral/last-piece || -0.34% |
+===============================++==========+
| geom mean || +0.17% |
There are extensive notes in !8897 about the regressions.
Briefly
* fibheaps: there was a very delicately balanced inlining that
tipped over the wrong way after this change.
* cryptarithm2 and paraffins are caused by #22274, which is
a separate issue really. (I.e. the right fix is *not* to
make inlining "win" over rules.)
So I'm accepting these changes
Metric Increase:
T10421
T16577
T9961
|
|
|
|
|
|
|
|
|
|
| |
There are three functions named failOp* in the parser:
failOpNotEnabledImportQualifiedPost
failOpImportQualifiedTwice
failOpFewArgs
Only the last one has anything to do with operators. The other two
were named this way either by mistake or due to a misunderstanding of
what "op" stands for. This small patch corrects this.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This implements this Core Libraries Proposal:
https://github.com/haskell/core-libraries-committee/issues/85
In particular, it:
1. Exposes the `symbolSing` method of `KnownSymbol`,
2. Exports the abstract `SSymbol` type used in `symbolSing`, and
3. Defines an API for interacting with `SSymbol`.
This also makes corresponding changes for `natSing`/`KnownNat`/`SNat` and
`charSing`/`KnownChar`/`SChar`. This fixes #15183 and addresses part (2)
of #21568.
|
|
|
|
|
| |
We simply need to align to the word size for pointer tagging to work. On
32-bit targets, aligned((8)) is wasteful.
|
|
|
|
|
| |
- Fix changelog for Unicode 15.0.0
- Fix the checksums of the downloaded Unicode files, in base's tool: "ucd2haskell".
|
| |
|
| |
|
| |
|
|
|
|
|
| |
`viewCons` used to ignore `Many` constructor completely, returning `VNothing`.
`viewSnoc` violated internal invariant of `Many` being a non-empty list.
|
| |
|
|
|
|
| |
Avoids some uses of `head` and `tail`, and some panics when an argument is null.
|
|
|
|
|
|
|
|
| |
A small refactoring in our Core Opt pipeline and some new functions for
transfering argument boxities from one signature to another to facilitate
`Note [Don't change boxity without worker/wrapper]`.
Fixes #21754.
|
|
|
|
|
|
| |
Combined wxth XZ_OPT=9, this blew the memory capacity of CI runners.
This reverts commit a5f9c35f5831ef5108e87813a96eac62803852ab.
|
|
|
|
|
|
|
|
| |
Justification in #22231. Short form: In a demand like `1C1(C1(L))`
it was too easy to confuse which `1` belongs to which `C`. Now
that should be more obvious.
Fixes #22231
|
|
|
|
|
| |
A small step towards #22185 to avoid partial functions + safe implementation
of `startsWithUnderscore`.
|