This patch fixes missing -ddump-cmm-raw output when compiling .cmm,
which is useful for debugging cmm related codegen issues.

(cherry picked from commit 6346c669)
(cherry picked from commit e5a1dd43)

This patch removes special consideration for ArchWasm32 in
cmmDoCmmSwitchPlans, which means the compiler will now disable
cmmImplementSwitchPlans for wasm unreg backend, just like unreg
backend of other targets. We enabled it in the past to workaround some
compile-time panic in older versions of LLVM, but those panics are no
longer present, hence no need to keep this workaround.

(cherry picked from commit bf0737c0)
(cherry picked from commit 8bd28cf2)

When -fsplit-sections is passed to GHC, pass -ffunction-sections
-fdata-sections to gcc/clang when building C/C++. Previously,
-fsplit-sections was only respected by the NCG/LLVM backends, but not
the unregisterised backend; the GHC driver did not pass
-fdata-sections and -ffunction-sections to the C compiler, which
resulted in excessive executable sizes.

Fixes #23381.

-------------------------
Metric Decrease:
    size_hello_artifact
    size_hello_unicode
-------------------------

(cherry picked from commit 02b1f91e)
(cherry picked from commit 74cdf564)

This patch makes the STG->Cmm backend avoid saving foreign call target
to local when there are no caller-save GlobalRegs.

Since 321941a8, when we lower a
foreign call, we unconditionally save the foreign call target to a
temporary local first, then rely on cmmSink to clean it up later,
which only happens with -fcmm-sink (implied by -O) and not in
unoptimized code.

And this is troublesome for the wasm backend NCG, which needs to infer
a foreign call target symbol's type signature from the Cmm call site.
Previously, the NCG has been emitting incorrect type signatures for
unoptimized code, which happens to work with `wasm-ld` most of the
time, but this is never future-proof against upstream toolchain
updates, and it causes horrible breakages when LTO objects are
included in linker input. Hence this patch.

(cherry picked from commit 8dd8a076)
(cherry picked from commit 2b2a3db8)

This commit enables the generic cmm optimizations in other NCGs to be
run in the wasm backend as well, followed by a late cmm control-flow
optimization pass. The added optimizations do catch some corner cases
not handled by the pre-NCG cmm pipeline and are useful in generating
smaller CFGs.

(cherry picked from commit c6ce242e)

This commit explicitly disables the ncgPIC flag for the wasm32 target.
The wasm backend doesn't support PIC for the time being.

(cherry picked from commit 87e34888)

This commit moves GHC.CmmToAsm.cmmToCmm to a standalone module,
GHC.Cmm.GenericOpt. The main motivation is enabling this logic to be
run in the wasm backend NCG code, which is defined in other modules
that's imported by GHC.CmmToAsm, causing a cyclic dependency issue.

(cherry picked from commit 6534da24)

There are objections raised on the MR (!13237) and the interface change
makes me rather uncomfortable.

This reverts commit fb82ee70.

This commit was wrong, as noted in the `master` revert cfeb70d3.
It appears to have ultimately been superceded by 0fe2b410
which is already present in `ghc-9.8`.

This reverts commit 44e119c9.

Simon Peyton Jones authored 8 months ago and Ben Gamari committed 7 months ago

(cherry picked from commit 09d24d82)

Simon Peyton Jones authored 8 months ago and Ben Gamari committed 7 months ago

(cherry picked from commit 0dfaeb66)

Simon Peyton Jones authored 8 months ago and Ben Gamari committed 7 months ago

This MR fixes #25325

See GHC.Tc.Types.Constraint, Note [Insoluble Wanteds], especially (IW2)

There is a small change in the error message for T14172, but it looks
entirely acceptable to me.

(cherry picked from commit 083703a1)

sheaf authored 8 months ago and Ben Gamari committed 7 months ago

For consistency with the NCG as well as with Clang and GCC, we make
the SSE4.2 feature flag imply +popcnt when using the LLVM backend.

Fixes #25353

(cherry picked from commit 06ae8507)

sheaf authored 8 months ago and Ben Gamari committed 7 months ago

LLVM expects the former instead of the latter since version 3.4.

Fixes #25019

(cherry picked from commit 694489ed)

Simon Peyton Jones authored 1 year ago and Ben Gamari committed 7 months ago

GHC was outright crashing before this fix: #24604

(cherry picked from commit faa30b41)

Cheng Shao authored 1 year ago and Ben Gamari committed 7 months ago

This patch makes the driver always merge objects with `ld -r` when
possible, and only fall back to calling `ar -L` when merge objects
command is unavailable. This completely reverts !8887 and !12313,
given more fixes in Cabal seems to be needed to avoid breaking certain
configurations and the maintainence cost is exceeding the behefits in
this case :/

(cherry picked from commit 631cefec)

Torsten Schmits authored 9 months ago and Ben Gamari committed 7 months ago

Backport note: I have ported `uncacheModule` instead of removing it as
is done in 0d170eaf since the necessary
type change is likely easier for possible GHC API users to accommodate
than outright removal.

(cherry picked from commit fff55592)

Rodrigo Mesquita authored 1 year ago and Ben Gamari committed 7 months ago

This commit makes the symbol cache that was used by the external
interpreter available for the internal interpreter too.

This follows from the analysis in #23415 that suggests the internal
interpreter could benefit from this cache too, and that there is no good
reason not to have the cache for it too. It also makes it a bit more
uniform to have the symbol cache range over both the internal and
external interpreter.

This commit also refactors the cache into a function which is used by
both `lookupSymbol` and also by `lookupSymbolInDLL`, extending the
caching logic to `lookupSymbolInDLL` too.

Alexis King authored 1 year ago and Ben Gamari committed 7 months ago

See the primary Note [Looking up symbols in the relevant objects] for a
more in-depth explanation.

When dynamically loading a Haskell symbol (typical when running a splice or
GHCi expression), before this commit we would search for the symbol in
all dynamic libraries that were loaded. However, this could be very
inefficient when too many packages are loaded (which can happen if there are
many package dependencies) because the time to lookup the would be
linear in the number of packages loaded.

This commit drastically improves symbol loading performance by
introducing a mapping from units to the handles of corresponding loaded
dlls. These handles are returned by dlopen when we load a dll, and can
then be used to look up in a specific dynamic library.

Looking up a given Name is now much more precise because we can get
lookup its unit in the mapping and lookup the symbol solely in the
handles of the dynamic libraries loaded for that unit.

In one measurement, the wait time before the expression was executed
went from +-38 seconds down to +-2s.

This commit also includes Note [Symbols may not be found in pkgs_loaded],
explaining the fallback to the old behaviour in case no dll can be found
in the unit mapping for a given Name.

Fixes #23415

Co-authored-by: Rodrigo Mesquita <(@alt-romes)>
(cherry picked from commit e008a19a)

When checking if a jump has more than one destination account for the
possibility of some jumps not being representable by a BlockId.

We do so by having isJumpishInstr return a `Maybe BlockId` where Nothing
represents non-BlockId jump destinations.

Fixes #24507

(cherry picked from commit 5bd8ed53)

Hannes Siebenhandl authored 1 year ago and Ben Gamari committed 8 months ago

Uses responseFile syntax.

The issue arises when GHC is installed on windows into a location that
has a space, for example the user name is 'Fake User'.
The $topdir will also contain a space, consequentially.
When we resolve the top dir in the string `-I$topdir/mingw/include`,
then `words` will turn this single argument into `-I/C/Users/Fake` and
`User/.../mingw/include` which trips up the flag argument parser of
various tools such as gcc or clang.
We avoid this by escaping the $topdir before replacing it in
`initSettngs`.
Additionally, we allow to escape spaces and quotation marks for
arguments in `settings` file.

Add regression test case to count the number of options after variable
expansion and argument escaping took place.
Additionally, we check that escaped spaces and double quotation marks are
correctly parsed.

(cherry picked from commit 31bf85ee)
(cherry picked from commit d74ffbbb)

Simon Peyton Jones authored 1 year ago and Ben Gamari committed 8 months ago

Omitting this caused type inference to fail in #24146.
This was an accidental omision in my refactoring of the
equality solver.

Check for ptr/int arguments rather than 64-bit width arguments when counting
integer register arguments.
The old approach broke when we stopped using exclusively W64-sized types to represent
sub-word sized integers.

Fixes #24314

(cherry picked from commit de589554)

Vladislav Zavialov authored 1 year ago and Ben Gamari committed 8 months ago

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

(cherry picked from commit 46fd8ced)

Fix a bug in the x86 ncg where results would be wrong when the desired output
register and one of the input registers were the same global.

Also adds a tiny optimization to make use of the memory addressing
support when convenient.

Fixes #24496

(cherry picked from commit 82ccb801)

The implementation of `checkHomeUnitsClosed` was traversing every single path
in the unit dependency graph - this grows exponentially and quickly grows to be
infeasible on larger unit dependency graphs.

Instead we replace this with a faster implementation which follows from the
specificiation of the closure property - there is a closure error if there are
units which are both are both (transitively) depended upon by home units and
(transitively) depend on home units, but are not themselves home units.

To compute the set of units required for closure, we first compute the closure
of the unit dependency graph, then the transpose of this closure, and find all
units that are reachable from the home units in the transpose of the closure.

(cherry picked from commit a933aff3)

Luite Stegeman authored 1 year ago and Ben Gamari committed 8 months ago

(cherry picked from commit 0c4a9686)

The optimization introducing h$appendToHsString wasn't kicking in
anymore (while it did in 9.8.1) because of the changes introduced in #23270 (7e0c8b3b).
This patch reenables the optimization by matching on case-expression, as
done in Cmm for unpackCString# standard thunks.

The test is also T24495 added in the next commits (two commits for ease
of backporting to 9.8).

(cherry picked from commit b36ee57b)

Fixes #24601.

(cherry picked from commit f8f384a8)

Matthew Pickering authored 1 year ago and Ben Gamari committed 8 months ago

(cherry picked from commit 28009fbc)

Add a new method to the Instruction class to check if we can eliminate a
jump in favour of fallthrough control flow.

Fixes #24507

(cherry picked from commit 0fe2b410)

Simon Peyton Jones authored 1 year ago and Ben Gamari committed 8 months ago

Previously, the logic for checking if a data constructor needs a wrapper or not
would take into account whether the constructor's fields have explicit
strictness (e.g., `data T = MkT !Int`), but the logic would _not_ take into
account whether `StrictData` was enabled. This meant that something like `type
data T = MkT Int` would incorrectly generate a wrapper for `MkT` if
`StrictData` was enabled, leading to the horrible errors seen in #24620. To fix
this, we disable generating wrappers for `type data` constructors altogether.

Fixes #24620.

Co-authored-by: Ryan Scott <ryan.gl.scott@gmail.com>
(cherry picked from commit 5e4f4ba8)

Simon Peyton Jones authored 1 year ago and Ben Gamari committed 8 months ago

(cherry picked from commit e5687186)

Simon Peyton Jones authored 1 year ago and Ben Gamari committed 8 months ago

Just for debugging, no effect on normal code

(cherry picked from commit 58408c77)

Simon Peyton Jones authored 1 year ago and Ben Gamari committed 8 months ago

This small patch fixes #24726, by tracking in-scope variables
properly in -drule-check.  Not hard to do!

(cherry picked from commit be1e60ee)

Torsten Schmits authored 1 year ago and Ben Gamari committed 8 months ago

(cherry picked from commit bc672166)

(cherry picked from commit daeda834)

We need to recompile if this flag is changed because later modules might depend on the
simplified core for this module if -fprefer-bytecode is enabled.

Fixes #24656

(cherry picked from commit dddc9dff)

We need to ensure that the output of `cvtp` is parenthesized (at precedence
`sigPrec`) so that any pattern signatures with a surrounding pattern signature
can parse correctly.

Fixes #24837.

(cherry picked from commit a3cd3a1d)

When evaluating PUSH_G try to tag the reference we are pushing if it's a
constructor. This is potentially helpful for performance and required to
fix #24870.

(cherry picked from commit 1bfa9111)