Cabal + Make based build #34
Draft
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
angerman
force-pushed
the
wip/make-build-pre-zig
branch
from
74fdffe to
a3d41c0
Compare
This commit makes all the FFIInfo needed in CCALL instructions serializable. Previously, when doing STG to BCO lowering, we would allocate a libffi ffi_cif struct and keep its remote pointer as FFIInfo; but actually we can just keep the type signature as FFIInfo and defer the actual allocation to link-time.
This commit removes the redundant NewBreakModule message from ghci: it just allocates two strings! This functionality can be implemented with existing MallocStrings in one iserv call.
This commit makes breakpoint module name and unit id serializable, in BRK_FUN instructions as well as ModBreaks. We can simply keep the module name and unit ids, and defer the buffer allocation to link time.
This commit removes the now unused newModule function from GHC.
This commit adds BCONPtrFS as a BCO non-pointer literal kind, which has the same semantics of BCONPtrStr, except it contains a FastString instead of a ByteString. By using BCONPtrFS to represent top level string literals that are already FastString in the compilation pipeline, we enjoy the FastString interning logic and avoid allocating a bunch of redundant ByteStrings for the same FastStrings, especially when we lower the BRK_FUN instruction.
This fixes #26017.
* `integerRecipMod x 1` now returns `(# 1 | #)` for all x; previously it
incorrectly returned `(# | () #)`, indicating failure.
* `integerPowMod 0 e m` now returns `(# | () #)` for e<0 and m>1, indicating
failure; previously it incorrectly returned `(# 0 | #)`.
It was fairly complex to do so and it doesn't seem to improve anything. Nofib allocations were unaffected as well. See also Historical Note [Floating dictionaries out of cases]
This commit adds support for a number of primops directly to the interpreter. This avoids the indirection of going through the primop wrapper for those primops speeding interpretation of optimized code up massively. Code involving IntSet runs about 25% faster with optimized core and these changes. For core without breakpoints it's even more pronouced and I saw reductions in runtime by up to 50%. Running GHC itself in the interpreter was sped up by ~15% through this change. Additionally this comment does a few other related changes: testsuite: * Run foundation test in ghci and ghci-opt ways to test these primops. * Vastly expand the foundation test to cover all basic primops by comparing result with the result of calling the wrapper. Interpreter: * When pushing arguments for interpreted primops extend each argument to at least word with when pushing. This avoids some issues with big endian. We can revisit this if it causes performance issues. * Restructure the stack chunk check logic. There are now macros for read accesses which might cross stack chunk boundries and macros which omit the checks which are used when we statically know we access an address in the current stack chunk.
This commit ensures we do a shape test before unifying. This ensures
we don't try to unify a TyVarTv with a non-tyvar, e.g.
alpha[tyv] := Int
On the way, we refactor simpleUnifyCheck:
1. Move the checkTopShape check into simpleUnifyCheck
2. Refactors simpleUnifyCheck to return a value of the new type
SimpleUnifyResult type. Now, simpleUnifyCheck returns "can unify",
"cannot unify" or "dunno" (with "cannot unify" being the new result
it can return). Now:
- touchabilityTest is included; it it fails we return "cannot unify"
- checkTopShape now returns "cannot unify" instead of "dunno" upon failure
3. Move the call to simpleUnifyCheck out of checkTouchableTyVarEq.
After that, checkTouchableTyVarEq becames a simple call to
checkTyEqRhs, so we inline it.
This allows the logic in canEqCanLHSFinish_try_unification to be simplified.
In particular, we now avoid calling 'checkTopShape' twice.
Two further changes suggested by Simon were also implemented:
- In canEqCanLHSFinish, if checkTyEqRhs returns PuFail with
'do_not_prevent_rewriting', we now **continue with this constraint**.
This allows us to use the constraint for rewriting.
- checkTyEqRhs now has a top-level check to avoid flattening a tyfam app
in a top-level equality of the form alpha ~ F tys, as this is
going around in circles. This simplifies the implementation without
any change in behaviour.
Fixes #25950
Fixes #26030
This commit tweaks the FixedRuntimeRepError case of pprTcSolverReportMsg,
to avoid including an explanation which refers to a type variable that
appears nowhere else.
For example, the old error message could look like the following:
The pattern binding does not have a fixed runtime representation.
Its type is:
T :: TYPE R
Cannot unify ‘R’ with the type variable ‘c0’
because the former is not a concrete ‘RuntimeRep’.
With this commit, we now omit the last two lines, because the concrete
type variable (here 'c0') does not appear in the type displayed to the
user (here 'T :: TYPE R').
If ghcversion.h can't be found, don't try to include it. This happens when there is no rts package in the package db and when -ghcversion-file argument isn't passed. Co-authored-by: Syvlain Henry <sylvain@haskus.fr>
In accordance with GHC Proposal #581 "Namespace-specified imports", section 2.3 "Deprecate use of pattern in import/export lists", the `pattern` namespace specifier is now deprecated. Test cases: T25900 T25900_noext
Addresses #25886.
Previously the linker's "proddable blocks" check relied on a simple linked list of spans. This resulted in extremely poor complexity while linking objects with lots of small sections (e.g. objects built with split sections). Rework the mechanism to instead use a simple interval set implemented via binary search. Fixes #26009.
The `LOAD_LIBRARY_SEARCH_USER_DIRS` and `LOAD_LIBRARY_SEARCH_DEFAULT_DIRS` were introduced in Windows Vista and have been available every since. As we no longer support Windows XP we can drop this check. Addresses #26009.
This is a useful helper which can be used for non-strings as well.
Previously the happy path of PEi386 used `break` in a double-`for` loop resulting in redundant calls to `LoadLibraryEx`. Fixes #26052.
Previously every DLL-imported symbol would result in a call to `LoadLibraryEx`. This ended up constituting over 40% of the runtime of `ghc --interactive -e 42` on Windows. Avoid this by maintaining a hash-set of loaded DLL names, skipping the call if we have already loaded the requested DLL. Addresses #26009.
Use --target=foo to select a target installed in $topdir/targets/foo/
It requires zig 0.14 at least to compile xxhash.h which uses evex512. Wrappers are needed because zig needs its first argument (`cc` or `c++`) always. Trying to pass it as CFLAGS fails with: ghc-toolchain, GHC's response files used to call the linker (this arg isn't supported by zig in the response file), configure scripts that call the C compiler without passing the specified CFLAGS. Use the wrappers like this: CC=/path/to/zig-cc CXX=/path/to/zig-c++ make
We need to use wrappers for, which is not pretty, as neither ghc nor cabal can deal with `zig ...`
E.g. make _build/stage0/bin/cabal make test CABAL=$PWD/_build/stage0/bin/cabal This must not be automated, the build tools should be distinctly separate from what we are building, no intertwining.
angerman
force-pushed
the
wip/make-build-pre-zig
branch
from
15d5708 to
4ef9539
Compare
Replaced separate logging for stderr with `tee` command to capture both stdout and stderr in the same log file for better clarity and debugging.
…ecent enough Cabal!)" This reverts commit 0b68a56.
…nough Cabal!)" This reverts commit 8337b88.
(cherry picked from commit ffa530c)
(cherry picked from commit b1ca8d4)
(cherry picked from commit 0efa3c5)
angerman
force-pushed
the
wip/make-build-pre-zig
branch
4 times, most recently
from
201b6dc to
f7375b8
Compare
angerman
force-pushed
the
wip/make-build-pre-zig
branch
from
f7375b8 to
2345fd5
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This builds upon wip/make-build, and integrates some of the andrea/make-build parts.