NFC.
gfx11_asm_vinterp.s already contained GFX12 run lines. Rename the
assembler and disassembler tests to be sorted based on real16 or fake16
instead of gfxip. Note, both GFX11 and GFX12 currently only have fake16
(fake16 in encoding, but not by name) upstream, so that is why the test
files have a -fake16 suffix.
One test input is changed, and that is the disassembler test for
unsupported bits in the instruction. It is now an input that is valid on
both GFX11 and GFX12. This was necessary because the size of the opcode
field changed.
This adds a new API built with the `ValueBoundsConstraintSet` to compute
the bounds of possibly scalable quantities. It uses knowledge of the
range of vscale (which is defined by the target architecture), to solve
for the bound as either a constant or an expression in terms of vscale.
The result is an `AffineMap` that will always take at most one
parameter, vscale, and returns a single result, which is the bound of
`value`.
The API is defined as follows:
```c++
FailureOr<ConstantOrScalableBound>
vector::ScalableValueBoundsConstraintSet::computeScalableBound(
Value value, std::optional<int64_t> dim,
unsigned vscaleMin, unsigned vscaleMax,
presburger::BoundType boundType,
bool closedUB = true,
StopConditionFn stopCondition = nullptr);
```
Note: `ConstantOrScalableBound` is a thin wrapper over the `AffineMap`
with a utility for converting the bound to a single quantity (i.e. a
size and scalable flag).
We believe this API could prove useful downstream in IREE (which uses a
similar analysis to hoist allocas, which currently fails for scalable
vectors).
Make sure all float/double comparison intrinsics specify what happens
with a NaN input. Update some existing descriptions of comparison
results to make them all consistent.
Also replace "yields" with "returns" throughout.
The existing implementation of tosa-layerwise-constant-fold only works
for constant values backed by DenseElementsAttr. For constants which
hold DenseResourceAttrs, the folder will end up asserting at runtime, as
it assumes that the backing data can always be accessed through
ElementsAttr::getValues.
This change reworks the logic so that types types used to perform
folding are based on whether the ElementsAttr can be converted to a
range of that particular type.
---------
Co-authored-by: Spenser Bauman <sabauma@mathworks.com>
Co-authored-by: Tina Jung <tinamaria.jung@amd.com>
This patch contains slight modifications to the reverted PR #85258 to
avoid issues with constructs containing multiple reduction clauses,
uncovered by a test on the gfortran testsuite.
This reverts commit 9f80444c2e669237a5c92013f1a42b91b5609012.
Add --skip-symbol and --skip-symbols options that allow to skip symbols
when executing other options that can change the symbol's name, binding
or visibility, similar to an existing option --keep-symbol that keeps a
symbol from being removed by other options.
Move 3 interface headers in `//mlir:IR` from `hdrs` to `srcs`.
Header files should not be added to multiple targets, but this is hard
to avoid because CMake is less strict with headers. But we should at
least avoid exposing them as headers by multiple targets because it
confuses tooling.
Add ones for every high bit that will cleared.
This will allow us to evaluate variables that have their bits known to
see if they have no risk of overflow despite the shift amount being
greater than the difference between the two types.
When building the OpenMP runtime with libomptarget support, the runtimes
configure step needs to have a dependency on various tools, in
particular opt, so that cmake configure checks yield the correct
results.
This did not work correctly, as the dependencies were only added if the
OPENMP_ENABLE_LIBOMPTARGET was set - but that variable is only set by
the openmp/CMakeLists.txt file, which isn't even parsed during the
initial cmake run (in fact, it is only parsed when executing the
runtimes configure step itself, but then it is too late).
Fixed by just adding those dependencies always.
In addition, the list of dependencies collected in ${extra_deps},
including those required for OpenMP, was only actually used when
configuring runtimes for the default set of targets - when the user
specifies a non-default LLVM_RUNTIME_TARGETS, those extra dependencies
were ignored (with the exception of ${hdrgen_deps}).
Fixed by passing the full ${extra_deps} in this case as well.
Fixes: https://github.com/llvm/llvm-project/issues/85933
Commit a7d5f73a03c81cab8df64dbd099e8acb40f5dfe1 introduced an
error in a target_compile_definitions on the SystemZ, causing
the build to break. Fixed by adding the missing "PRIVATE".
`copy_n` has been used to allow constant evaluation of `char_traits`. We
now have `__constexpr_memmove`, which `copy_n` just forwards to. We can
call `__constexpr_memmove` directly, avoiding a bunch of instantiations.
This reduces the time it takes to include `<string>` from 321ms to
285ms.
This is compatible with MSVC, `-machine:arm64x` is essentially an alias
to `-machine:arm64ec`. To make a type library that exposes both native
and EC symbols, an additional `-defArm64Native` argument is needed in
both cases.
Refactor the logic that checks if a module contains mixed
absolute/non-lowered LDS GVs.
The check now happens latter when the "worklists" are formed. This is
because in some cases (OpenMP) we can have non-lowered GVs in a lowered
module, and this is normal because those GVs are just unused and removed
from the list at some point before the end of `getUsesOfLDSByFunction`.
Doing the check later ensures that if a mixed module is spotted, then
it's a _real_ mixed module that needs rejection, not a module containing
an intentionally ignored GV.
We currently always lower shuffle to the struct-returning variant. I saw
some cases where this survived all the way through ptx, resulting in
increased register usage. The easiest fix is to simply lower to the
single-result version when the predicate is unused.
We don't have a concat_vector shuffle kind and improveShuffleKindFromMask won't alter the base type to match it as InsertSubvector.
But since this is how X86 will lower concat_vector anyhow, just recognise it explicitly.
Another step for #67803
When emitting the storage (or memory copy operations) for constant
initializers, the decision whether to split a constant structure or
array store into a sequence of field stores or to use `memcpy` is
based upon the optimization level and the size of the initializer.
In afe8b93ffdfef5d8879e1894b9d7dda40dee2b8d, we extended this by
allowing constants to be split when the array (or struct) type does
not match the type of data the address to the object (constant) is
expected to contain. This may happen when `emitStoresForConstant` is
called by `EmitAutoVarInit`, as the element type of the address gets
shrunk. When this occurs, let the initializer be split into a bunch
of stores only under `-ftrivial-auto-var-init=pattern`.
Fixes: https://github.com/llvm/llvm-project/issues/84178.
SIMachineFunctionInfo has a scan of the function body for inline asm
which may use AGPRs, or callees in SIMachineFunctionInfo. Move this
into the attributor, so it actually works interprocedurally.
Could probably avoid most of the test churn if this bothered to avoid
adding this on subtargets without AGPRs. We should also probably
try to delete the MIR scan in usesAGPRs but it seems to be trickier
to eliminate.
At the moment clang-format will return exit code 0 on incomplete
results. In scripts it would sometimes be useful if clang-format would
instead fail in those cases, signalling that there was something wrong
with the code being formatted.
---------
Co-authored-by: Björn Schäpers <github@hazardy.de>
Co-authored-by: Owen Pan <owenpiano@gmail.com>
When importing from LLVM IR the data layout of all pointer types
contains an index bitwidth that should be used for index computations.
This revision adds a getter to the DataLayout that provides access to
the already stored bitwidth. The function returns an optional since only
pointer-like types have an index bitwidth. Querying the bitwidth of a
non-pointer type returns std::nullopt.
The new function works for the built-in Index type and, using a type
interface, for the LLVMPointerType.
Reviving some of the progress on #71764. To recap, we explored removing
the V0 register copies to simplify the pass, but hit a limitation with
the register allocator due to our use of the vmv0 singleton reg class
and early-clobber constraints.
So since we will have to continue to track the definition of V0
ourselves, this patch simplifies it by storing it in a map. It will
allow us to move about copies to V0 in #71764 without having to do extra
bookkeeping.
With GPR pairs from Zdinx, we can't guarantee there are no subregisters
on integer instruction operands. I've been able to get these assertions
to fire after some other recent PRs.
I've added a FIXME to support this properly. I just wanted to prevent
the assertion failure for now.
No test case because my other patch #85982 that allowed me to fail the assert
hasn't been approved yet, and I don't know for that that patch is
required to hit this assert. It's just what exposed it for me. So I
think this patch is a good precaution regardless.
unless USE_DEPRECATED_GCC_INSTALL_PREFIX (temporary escape hatch) is
set. Setting GCC_INSTALL_PREFIX leads to a warning for Clang 18.1
(#77537) and will be completely removed for Clang 20.
Link:
discourse.llvm.org/t/add-gcc-install-dir-deprecate-gcc-toolchain-and-remove-gcc-install-prefix/65091
Link:
discourse.llvm.org/t/correct-cmake-parameters-for-building-clang-and-lld-for-riscv/72833
In preparation for the initial ClangIR upstreaming process, move clang
behind MLIR in the list of external projects. Otherwise, cmake will
attempt to build clang before MLIR.
