| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch addresses #20988 by refactoring the way the
occurrence analyser deals with lambdas.
Previously it used collectBinders to split off a group of binders,
and deal with them together. Now I deal with them one at a time
in occAnalLam, which allows me to skip casts easily. See
Note [Occurrence analysis for lambda binders]
about "lambda-groups"
This avoidance of splitting out a list of binders has some good
consequences. Less code, more efficient, and I think, more clear.
The Simplifier needed a similar change, now that lambda-groups
can inlude casts. It turned out that I could simplify the code
here too, in particular elminating the sm_bndrs field of StrictBind.
Simpler, more efficient.
Compile-time metrics improve slightly; here are the ones that are
+/- 0.5% or greater:
Baseline
Test Metric value New value Change
--------------------------------------------------------------------
T11303b(normal) ghc/alloc 40,736,702 40,543,992 -0.5%
T12425(optasm) ghc/alloc 90,443,459 90,034,104 -0.5%
T14683(normal) ghc/alloc 2,991,496,696 2,956,277,288 -1.2%
T16875(normal) ghc/alloc 34,937,866 34,739,328 -0.6%
T17977b(normal) ghc/alloc 37,908,550 37,709,096 -0.5%
T20261(normal) ghc/alloc 621,154,237 618,312,480 -0.5%
T3064(normal) ghc/alloc 190,832,320 189,952,312 -0.5%
T3294(normal) ghc/alloc 1,604,674,178 1,604,608,264 -0.0%
T5321FD(normal) ghc/alloc 270,540,489 251,888,480 -6.9% GOOD
T5321Fun(normal) ghc/alloc 300,707,814 281,856,200 -6.3% GOOD
WWRec(normal) ghc/alloc 588,460,916 585,536,400 -0.5%
geo. mean -0.3%
Metric Decrease:
T5321FD
T5321Fun
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This test has been the scourge of contributors for a long time.
It has caused many failed CI runs and wasted hours debugging a test
which barely does anything. The fact is does nothing is the reason for
the flakiness and it's very sensitive to small changes in initialisation costs,
in particular adding wired-in things can cause this test to fluctuate
quite a bit.
Therefore we admit defeat and just bump the threshold up to 10% to catch
very large regressions but otherwise don't care what this test does.
Fixes #19414
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, `DeriveGeneric` would look up the fixity of a data constructor
using `getFixityEnv`, but this is subtly incorrect for data constructors
defined in external modules. This sort of situation can happen with
`StandaloneDeriving`, as noticed in #20994. In fact, the same bug has occurred
in the past in #9830, and while that bug was fixed for `deriving Read` and
`deriving Show`, the fix was never extended to `DeriveGeneric` due to an
oversight. This patch corrects that oversight.
Fixes #20994.
|
|
|
|
|
| |
deb9 is now end-of-life so we are dropping support for producing
bindists.
|
| |
|
|
|
|
| |
Fixes #20984.
|
|
|
|
| |
Fixes bootstrap with GHC 9.0 after 5a6efd218734dbb5c1350531680cd3f4177690f1
|
| |
|
|
|
|
|
|
|
|
| |
As requested by Simon after review of !7342.
I also took liberty to define the `Functor` instance by hand, as the derived one
subverts the invariants maintained by the pattern synonym (as already stated in
`Note [The one-shot state monad trick]`).
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This gives users the choice to enable __compact_unwind sections
when linking. These were previously hardcoded to be removed.
This can be used to solved the problem "C++ does not catch
exceptions when used with Haskell-main and linked by ghc",
https://gitlab.haskell.org/ghc/ghc/-/issues/11829
It does not change the default behavior, because I can not
estimate the impact this would have.
When Apple first introduced the compact unwind ABI, a number of
open source projects have taken the easy route of disabling it,
avoiding errors or even just warnings shortly after its
introduction.
Since then, about a decade has passed, so it seems quite possible
that Apple itself, and presumably many programs with it, have
successfully switched to the new format, to the point where the
old __eh_frame section support is in disrepair. Perhaps we should
get along with the program, but for now we can test the waters
with this flag, and use it to fix packages that need it.
|
|
|
|
|
|
|
| |
Fixes #15531 by ensuring that we know the corresponding C type for all
marshalling wrappers.
Closes #15531.
|
|
|
|
| |
See #20939
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
The test reads a 16 bit value from an array of 8 bit values. Naturally,
that leads to different values read on big-endian architectures than
on little-endian. In this case the value read is 0x8081 on big-endian
and 0x8180 on little endian. This patch changes the argument of the `and`
machop to mask bit 7 which is the only bit different. The test still checks
that bit 15 is zero, which was the original issue in #20638.
Fixes #20906.
|
|
|
|
|
|
|
| |
Newer lld versions may include vendor info in --version output and
thus the version string may not start with ‘LLD’.
Fixes #20907
|
|
|
|
| |
Closed #20904
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Here we refactor WinIO's IO completion scheme, squashing a memory leak
and fixing #18382.
To fix #18382 we drop the special thread status introduced for IoPort
blocking, BlockedOnIoCompletion, as well as drop the non-threaded RTS's
special dead-lock detection logic (which is redundant to the GC's
deadlock detection logic), as proposed in #20947.
Previously WinIO relied on foreign import ccall "wrapper" to create an
adjustor thunk which can be attached to the OVERLAPPED structure passed
to the operating system. It would then use foreign import ccall
"dynamic" to back out the original continuation from the adjustor. This
roundtrip is significantly more expensive than the alternative, using a
StablePtr. Furthermore, the implementation let the adjustor leak,
meaning that every IO request would leak a page of memory.
Fixes T18382.
|
|
|
|
|
|
| |
I found it weird that most of the combinators weren't actually strict. Making
`pure` strict in the state should hopefully give Nested CPR an easier time to
unbox the nested state.
|
|
|
|
| |
Issue was reported on #13306
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The simplest way to do this seemed to be to persist the whole type in
the extension field from the typechecker so that the few relevant places
* Desugaring can work out the return type by splitting this type rather
than calling `dsExpr` (slightly more efficient).
* hsExprType can just return the correct type.
* Zonking has to now zonk the type as well
The other option we considered was wiring in StaticPtr but that is
actually quite tricky because StaticPtr refers to StaticPtrInfo which
has field selectors (which we can't easily wire in).
Fixes #20150
|
|
|
|
|
|
| |
like we do for the non-Hadrian wrapper script.
Otherwise if $bindir/ghc is a different ghc version then versioned ghci will incorrectly run the other ghc version instead.
(Normally this would only happen if there are parallel ghc versions installed in bindir.)
All the other wrapper scripts already have versioned executablename
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously (9ebda74ec5331911881d734b21fbb31c00a0a22f) `environ` was
added to `RtsSymbols` to ensure that environment was correctly
propagated when statically linking. However, this introduced #20577
since platforms are inconsistent in whether they provide a prototype for
`environ`. I fixed this by providing a prototype but while doing so
dropped symbol-table entry, presumably thinking that it was redundant
due to the entry in the mingw-specific table.
Here I reintroduce the symbol table entry for `environ` and move libc
symbols shared by Windows and Linux into a new macro,
`RTS_LIBC_SYMBOLS`, avoiding this potential confusion.
|
| |
|
| |
|
| |
|
|
|
| |
This reverts commit c0b854e929f82c680530e944e12fad24f9e14f8e
|
| |
|
| |
|
| |
|
|
|
|
| |
See #20802.
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
Avoid requiring to pass DynFlags to mkDataConRep/buildDataCon. When we
load an interface file, these functions don't use the flags.
This is preliminary work to decouple the loader from the type-checker
for #14335.
|
|
|
|
| |
Fixes #20928
|
|
|
|
|
|
| |
(Fixes #10616 and #10617)
Co-authored-by: Roland Senn <rsx@bluewin.ch>
|
| |
|
|
|
|
|
| |
They are likely broken for the same reason as FreeBSD where the tests
are already disabled.
|
|
|
|
|
|
|
| |
The test currently times out waiting for end of stdin in getContents.
The expected output indicates that nothing should come for the test to
pass as written. It is unclear how the test was supposed to pass, but
this looks like a sufficient hack to make it work.
|
|
|
|
|
|
|
|
|
| |
Diagnostics for outofmem test on OpenBSD includes the amount of memory
that it failed to allocate. This seems like an irrelevant detail that
could change over time and isn't required for determining if test
passed.
Typical elided text is '(requested 2148532224 bytes)'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
ASSERT should be used in situations where something very bad will happen
later on if a certain invariant doesn't hold. The idea is that IF we
catch the assertion earlier then it will be easier to work out what's
going on at that point rather than at some indeterminate point in the
future of the program.
The assertions in Stats.c do not obey this philsophy and it is quite
annoying if you are running a debug build (or a ticky compiler) and one
of these assertions fails right at the end of your program, before the
ticky report is printed out so you don't get any profiling information.
Given that nothing terrible happens if these assertions are not true, or
at least the terrible thing will happen in very close proximity to the
assertion failure, these assertions use the new WARN macro which prints
the assertion failure to stdout but does not exit the program.
Of course, it would be better to fix these metrics to not trigger the
assertion in the first place but if they did fail again in the future it
is frustrating to be bamboozled in this manner.
Fixes #20899
|
|
|
|
| |
Part of #20889
|
|
|
|
|
|
|
|
|
| |
The documentation states that the interactive flags should be use for
any interactive expressions. The interactive flags are used when
typechecking these expressions but not when printing. The session flags
(modified by :set) are only used when loading a module.
Fixes #20909
|
|
|
|
| |
This makes it more similar to pprTrace, pprPanic etc.
|