Follow-up to #132003, in particular, see
https://github.com/llvm/llvm-project/pull/132003#issuecomment-2739701936.
This PR extends reduction support for `loop` directives. Consider the
following scenario:
```fortran
subroutine bar
implicit none
integer :: x, i
!$omp teams loop reduction(+: x)
DO i = 1, 5
call foo()
END DO
end subroutine
```
Note the following:
* According to the spec, the `reduction` clause will be attached to
`loop` during earlier stages in the compiler.
* Additionally, `loop` cannot be mapped to `distribute parallel for` due
to the call to a foreign function inside the loop's body.
* Therefore, `loop` must be mapped to `distribute`.
* However, `distribute` does not have `reduction` clauses.
* As a result, we have to move the `reduction`s from the `loop` to its
parent `teams` directive, which is what is done by this PR.
Fixes a bug in reductions when converting `teams loop` constructs with
`reduction` clauses.
According to the spec (v5.2, p340, 36):
```
The effect of the reduction clause is as if it is applied to all leaf
constructs that permit the clause, except for the following constructs:
* ....
* The teams construct, when combined with the loop construct.
```
Therefore, for a combined directive similar to: `!$omp teams loop
reduction(...)`, the earlier stages of the compiler assign the
`reduction` clauses only to the `loop` leaf and not to the `teams` leaf.
On the other hand, if we have a combined construct similar to: `!$omp
teams distribute parallel do`, the `reduction` clauses are assigned both
to the `teams` and the `do` leaves. We need to match this behavior when
we convert `teams` op with a nested `loop` op since the target set of
constructs/ops will be incorrect without moving the reductions up to the
`teams` op as well.
This PR introduces a pattern that does exactly this. Given the following
input:
```
omp.teams {
omp.loop reduction(@red_sym %red_op -> %red_arg : !fir.ref<i32>) {
omp.loop_nest ... {
...
}
}
}
```
this pattern updates the `omp.teams` op in-place to:
```
omp.teams reduction(@red_sym %red_op -> %teams_red_arg : !fir.ref<i32>) {
omp.loop reduction(@red_sym %teams_red_arg -> %red_arg : !fir.ref<i32>) {
omp.loop_nest ... {
...
}
}
}
```
Note the following:
* The nested `omp.loop` is not rewritten by this pattern, this happens
through `GenericLoopConversionPattern`.
* The reduction info are cloned from the nested `omp.loop` op to the
parent `omp.teams` op.
* The reduction operand of the `omp.loop` op is updated to be the
**new** reduction block argument of the `omp.teams` op.
Extends support for the `loop` directive by adding support for `parallel
loop` combined directive.
Parent PR: #127489. Only the latest commit is relevant.
This extends support for generic `loop` rewriting by:
1. Preventing nesting multiple worksharing loops inside each other. This
is checked by walking the `teams loop` region searching for any `loop`
directive whose `bind` modifier is `parallel`.
2. Preventing convert to worksharing loop if calls to unknow functions
are found in the `loop` directive's body.
We walk the `teams loop` body to identify either of the above 2
conditions, if either of them is found to be true, we map the `loop`
directive to `distribute`.
Extends generic `loop` directive support by supporting the `bind`
clause. Since semantic checking does the heavy lifting of verifying the
proper usage of the clause modifier, we can simply enable code-gen for
`teams loop bind(...)` without the need to differentiate between the
values the the clause can accept.
Move non-common files from FortranCommon to FortranSupport (analogous to
LLVMSupport) such that
* declarations and definitions that are only used by the Flang compiler,
but not by the runtime, are moved to FortranSupport
* declarations and definitions that are used by both ("common"), the
compiler and the runtime, remain in FortranCommon
* generic STL-like/ADT/utility classes and algorithms remain in
FortranCommon
This allows a for cleaner separation between compiler and runtime
components, which are compiled differently. For instance, runtime
sources must not use STL's `<optional>` which causes problems with CUDA
support. Instead, the surrogate header `flang/Common/optional.h` must be
used. This PR fixes this for `fast-int-sel.h`.
Declarations in include/Runtime are also used by both, but are
header-only. `ISO_Fortran_binding_wrapper.h`, a header used by compiler
and runtime, is also moved into FortranCommon.
This removes mentions of `target` from the generic `loop` rewrite pass
since there is not need for it anyway. It is enough to detect `loop`'s
nesting within `teams` or `parallel` directives.
Extends rewriting of `loop` directives by supporting `bind` clause for
standalone directives. This follows both the spec and the current state
of clang as follows:
* No `bind` or `bind(thread)`: the `loop` is rewritten to `simd`.
* `bind(parallel)`: the `loop` is rewritten to `do`.
* `bind(teams)`: the `loop` is rewritten to `distribute`.
This is a follow-up PR for
https://github.com/llvm/llvm-project/pull/122632, only the latest commit
in this PR is relevant to the PR.
Extends conversion support for `loop` directives. This PR handles
standalone `loop` constructs that do not have a `bind` clause attached
by rewriting them to equivalent `simd` constructs. The reasoning behind
that decision is documented in the rewrite function itself.
Introduces a new conversion pass that rewrites `omp.loop` ops to their
semantically equivalent op nests bases on the surrounding/binding
context of the `loop` op. Not all forms of `omp.loop` are supported yet.
See `isLoopConversionSupported` for more info on which forms are
supported.
