| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously `-dtag-inference-checks`, `-dcheck-prim-bounds`, and
`-falignment-sanitization` all aborted by calling `barf` from the
mutator. However, this can lead to deadlocks in the threaded RTS. For
instance, in the case of `-dcheck-prim-bounds` the following can happen
1. the mutator takes a capability and begins execution
2. the bounds check fails, calling `barf`
3. `barf` calls `rtsFatalInternalErrorFn`, which in turn calls `endEventLogging`
4. `endEventLogging` calls `flushEventLog`, which it turn initiates a
sync to request that all capabilities flush their local event logs
5. we deadlock as the the capability held by the crashing mutator can
never join the sync
To avoid this we now have a more principled means of aborting: we return
to the scheduler setting the thread's return value to ThreadAborting.
The scheduler will see this and call `barf`.
Fixes #22038.
|
| |
|
|
|
|
|
| |
This fixes #22065. We were failing to retain a quantifier that
was mentioned in the kind of another retained quantifier.
Easy to fix.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
The -x option is used to manually specify which phase a file should be
started to be compiled from (even if it lacks the correct extension). I
just failed to implement this when refactoring the driver.
In particular Cabal calls GHC with `-E -cpp -x hs Foo.cpphs` to
preprocess source files using GHC.
I added a test to exercise this case.
Fixes #22044
|
| |
|
|
|
|
|
| |
An extra field was added to the TSO structure in 6d1700b6 but the
decoding logic in ghc-heap was not updated for this new field.
Fixes #22046
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is an improvement to HPC authored by Richard Wallace
(https://github.com/purefn) and myself. I have received permission from
him to attempt to upstream it. This improvement was originally
implemented as a patch to HPC via input-output-hk/haskell.nix:
https://github.com/input-output-hk/haskell.nix/pull/1464
Paraphrasing Richard, HPC currently requires all inputs as command line arguments.
With large projects this can result in an argument list too long error.
I have only seen this error in Nix, but I assume it can occur is a plain Unix environment.
This MR adds the standard response file syntax support to HPC. For
example you can now pass a file to the command line which contains the
arguments.
```
hpc @response_file_1 @response_file_2 ...
The contents of a Response File must have this format:
COMMAND ...
example:
report my_library.tix --include=ModuleA --include=ModuleB
```
Updates hpc submodule
Co-authored-by: Richard Wallace <rwallace@thewallacepack.net>
Fixes #22050
|
| |
|
|
|
|
| |
Previously the wording was a tad unclear. Fix this.
Closes #21114.
|
| |
|
|
| |
As noted in #21988, some users rely on this.
|
| |
|
|
| |
Closes #21922.
|
| |
|
|
|
| |
Includes merge of `main` into `ghc-head` as well as some Haddock users
guide fixes.
|
| |
|
|
| |
This relocates it from docs/ to doc/
|
| | |
|
| | |
|
| |
|
|
| |
Fixes #22052.
|
| |
|
|
|
| |
I don't have good reason to believe the optimization level should affect
if sharing works or not here. So limit the test to the normal way.
|
| | |
|
| |
|
|
|
|
|
|
|
| |
Fixes #21492
MonoidalMap is inlined and used to implement Variables, as before.
The top-level value "jobs" is reimplemented as a regular Map, since it
doesn't use the monoidal union anyway.
|
| |
|
|
|
|
|
|
|
| |
Our aliasification logic would previously turn builtin LLVM variables
into aliases, which apparently confuses LLVM. This manifested in
initializers failing to be emitted, resulting in many profiling failures
with the LLVM backend.
Fixes #22019.
|
| | |
|
| |
|
|
|
|
|
|
|
| |
This reverts commit 5765e13370634979eb6a0d9f67aa9afa797bee46.
The job was not tested before being merged and fails CI
(https://gitlab.haskell.org/ghc/ghc/-/jobs/1139392)
Ticket #22005
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
For the code
{-# LANGUAGE OverloadedRecordUpdate #-}
operatorUpdate f = f{(+) = 1}
There are no exact print annotations for the parens around the +
symbol, nor does normal ppr print them.
This MR fixes that.
Closes #21805
Updates haddock submodule
|
| |
|
|
| |
machines
|
| | |
|
| |
|
|
|
|
|
|
|
| |
Apple's ABI documentation [1] says: "The platforms reserve register x18.
Don’t use this register." While this wasn't problematic in previous
Darwin releases, macOS 13 appears to start zeroing this register
periodically. See #21964.
[1] https://developer.apple.com/documentation/xcode/writing-arm64-code-for-apple-platforms
|
| |
|
|
|
|
|
|
|
| |
Previously we would attempt to install documentation even if the `doc`
directory doesn't exist (e.g. due to `--docs=none`). This would result
in the surprising side-effect of the entire contents of the bindist
being installed in the destination documentation directory. Fix this.
Fixes #21976.
|
| | |
|
| |
|
|
|
|
|
| |
It appears that the CI refactoring carried out in
5ff690b8474c74e9c968ef31e568c1ad0fe719a1 failed to carry over some
critical configuration: setting the build/host/target platforms and
forcing use of a non-broken linker.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch adds a new command-line flag:
-fplugin-library=<file-path>;<unit-id>;<module>;<args>
used like this:
-fplugin-library=path/to/plugin.so;package-123;Plugin.Module;["Argument","List"]
It allows a plugin to be loaded directly from a shared library. With
this approach, GHC doesn't compile anything for the plugin and doesn't
load any .hi file for the plugin and its dependencies. As such GHC
doesn't need to support two environments (one for plugins, one for
target code), which was the more ambitious approach tracked in #14335.
Fix #20964
Co-authored-by: Josh Meredith <joshmeredith2008@gmail.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
avoids warning seen on 9.4.1:
src/Settings/Builders/RunTest.hs:264:53: warning: [-Wtype-defaults]
• Defaulting the following constraints to type ‘Integer’
(Show a0)
arising from a use of ‘show’
at src/Settings/Builders/RunTest.hs:264:53-84
(Num a0)
arising from a use of ‘stageNumber’
at src/Settings/Builders/RunTest.hs:264:59-83
• In the second argument of ‘(++)’, namely
‘show (stageNumber (C.stage ctx))’
In the second argument of ‘($)’, namely
‘"config.stage=" ++ show (stageNumber (C.stage ctx))’
In the expression:
arg $ "config.stage=" ++ show (stageNumber (C.stage ctx))
|
264 | , arg "-e", arg $ "config.stage=" ++ show (stageNumber (C.stage ctx))
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
compilation tested locally
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Adding a 'Status' field with a few values:
- Deprecated
- Experimental
- InternalUseOnly
- Noting if included in 'GHC2021', 'Haskell2010' or 'Haskell98'
Those values are pulled from the existing descriptions or elsewhere in
the documentation.
While at it, include the :implied by: where appropriate, to provide
more detail.
Fixes #21475
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Fix for #21651
Fixes three bugs:
- writes to eventManager should be atomic. It is accessed concurrently by ioManagerCapabilitiesChanged and closeFdWith.
- The race in closeFdWith described in the ticket.
- A race in getSystemEventManager where it accesses the 'IOArray' in
'eventManager' before 'ioManagerCapabilitiesChanged' has written to
'eventManager', causing an Array Index exception. The fix here is to
'yield' and retry.
|
| |
|
|
|
| |
This bug does not affect windows, which does not use the
base module GHC.Event.Thread.
|
| |
|
|
|
|
|
|
| |
The size_up_alloc function mistakenly considered any type that isn't
lifted to not allocate anything, which is wrong. What we want instead
is to check the type isn't boxed. This accounts for (BoxedRep Unlifted).
Fixes #21939
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
FreeBSD's libiconv includes an implementation of the
iconv_* functions in libc. Unfortunately these can
only be resolved using dlvsym, which is how the RTS linker
usually resolves such functions. To fix this we include an ad-hoc
special case for iconv_*.
Fixes #20354.
|
| |
|
|
|
|
|
|
|
|
| |
* Remove hack when printing OccNames. No longer needed since e3dcc0d5
* Remove unused `pprCmms` and `instance Outputable Instr`
* Simplify `pprCLabel` (no need to pass platform)
* Remove evil `Show`/`Eq` instances for `SDoc`. They were needed by
ImmLit, but that can take just a String instead.
* Remove instance `Outputable CLabel` - proper output of labels
needs a platform, and is done by the `OutputableP` instance
|
| |
|
|
|
|
|
|
| |
Previously hadrian's bindist Makefile would modify package
registrations placed by `install` via a shell pipeline and `mv`.
However, the use of `mv` means that if umask is set then the user may
otherwise end up with package registrations which are inaccessible.
Fix this by ensuring that the mode is 0644.
|
| |
|
|
|
|
|
| |
It turns out that `cp -P` on Darwin does not always copy a symlink as
a symlink. In order to get these semantics one must pass `-RP`. It's not
entirely clear whether this is valid under POSIX, but it is nevertheless
what Apple does.
|
| |
|
|
|
|
|
|
|
| |
In general we want to ensure that the tested environment is as similar
as possible to the environment the user will use. In the case of Darwin,
this means we want to use the system's BSD command-line utilities, not
coreutils.
This would have caught #21974.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
In #19143 I noticed that newArray# failed to initialize the card table
of newly-allocated arrays. However, embarrassingly, I then only fixed
the issue in newArrayArray# and, in so doing, introduced the potential
for an integer underflow on zero-length arrays (#21962).
Here I fix the issue in newArray#, this time ensuring that we do not
underflow in pathological cases.
Fixes #19143.
|
| |
|
|
| |
Here we add a simple qemu-based test for cross-compilers.
|
| | |
|
| |
|
|
| |
Begins to address #11958.
|
| | |
|
| |
|
|
|
|
|
|
| |
GHC is happy to parse `(f) x y = x + y` when it should be a parse error
based on the Haskell report. Seems harmless enough so we won't fix it
but it's documented now.
Fixes #19788
|
| |
|
|
|
|
| |
This out-of-line primop has no Haskell wrapper and hasn't been used
anywhere in the tree. Furthermore, the code gets in the way of !7632, so
it should be garbage collected.
|
| |
|
|
| |
This addresses one part of #21710.
|
| | |
|
| |
|
|
|
|
|
| |
As noted in #21506, it is now necessary to remove extended attributes
from `/lib` as well as `/bin` to avoid SIP issues on Darwin.
Fixes #21506.
|
