a63f915771
ACCMP atomics do not support type conversion. Specifically, I have
encountered semantically incorrect code for atomic reads.
Example:
```
program main
implicit none
real(8) :: n
integer :: x
x = 1.0
!$acc atomic capture
n = x
x = n
!$acc end atomic
end program main
```
We have this error when compiling it with flang-new: `error:
loc("rep.f90":6:9): expected three operations in atomic.capture region
(one terminator, and two atomic ops)`
Yet, in the following generated FIR code, we observe three issues.
1. `fir.convert` intrudes into the capture region.
2. An incorrect temporary (`%2`) is being updated instead of `n`.
3. If we allow `n` in place of `%2`, the operand types of `atomic.read`
do not match. Introducing a `!fir.ref<i32> -> !fir.ref<f64>` conversion
on `x` is inaccurate because we need to convert the value of `x`.
```
%2 = "fir.alloca"() <{in_type = i32, operandSegmentSizes = array<i32: 0, 0>}> : () -> !fir.ref<i32>
%3 = "fir.alloca"() <{bindc_name = "n", in_type = f64, operandSegmentSizes = array<i32: 0, 0>, uniq_name = "_QFEn"}> : () -> !fir.ref<f64>
%4:2 = "hlfir.declare"(%3) <{operandSegmentSizes = array<i32: 1, 0, 0, 0>, uniq_name = "_QFEn"}> : (!fir.ref<f64>) -> (!fir.ref<f64>, !fir.ref<f64>)
%5 = "fir.alloca"() <{bindc_name = "x", in_type = i32, operandSegmentSizes = array<i32: 0, 0>, uniq_name = "_QFEx"}> : () -> !fir.ref<i32>
%6:2 = "hlfir.declare"(%5) <{operandSegmentSizes = array<i32: 1, 0, 0, 0>, uniq_name = "_QFEx"}> : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
%7 = "arith.constant"() <{value = 1 : i32}> : () -> i32
"hlfir.assign"(%7, %6#0) : (i32, !fir.ref<i32>) -> ()
%8 = "fir.load"(%4#0) : (!fir.ref<f64>) -> f64
%9 = "fir.convert"(%8) : (f64) -> i32
"fir.store"(%9, %2) : (i32, !fir.ref<i32>) -> ()
%10 = "fir.load"(%6#0) : (!fir.ref<i32>) -> i32
%11 = "fir.convert"(%10) : (i32) -> f64
"acc.atomic.capture"() ({
"acc.atomic.read"(%2, %6#1) <{element_type = f64}> : (!fir.ref<i32>, !fir.ref<i32>) -> ()
%12 = "fir.convert"(%11) : (f64) -> i32
"acc.atomic.write"(%2, %12) : (!fir.ref<i32>, i32) -> ()
"acc.terminator"() : () -> ()
}) : () -> ()
```
This PR updates `flang/lib/Lower/DirectivesCommon.h` to solve the issues
by taking the following approaches (from top to bottom):
1. Move `fir.convert` for `atomic.write` out of the capture region.
2. Remove the `!fir.ref<i32> -> !fir.ref<f64>` conversion found in
`genOmpAccAtomicRead`.
3. Eliminate unnecessary `genExprAddr` calls on the RHS, which create an
invalid temporary for `x = 1.0`.
4. When generating a capture operation, refer to the original LHS
instead of the type-casted RHS.
Here, we have to allow for the cases where the operand types of
`atomic.read` differ from one another. Thus, this PR also removes the
`AllTypesMatch` trait from both `acc.atomic.read` and `omp.atomic.read`.
The example code is converted as follows:
```
%0 = fir.alloca f64 {bindc_name = "n", uniq_name = "_QFEn"}
%1:2 = hlfir.declare %0 {uniq_name = "_QFEn"} : (!fir.ref<f64>) -> (!fir.ref<f64>, !fir.ref<f64>)
%2 = fir.alloca i32 {bindc_name = "x", uniq_name = "_QFEx"}
%3:2 = hlfir.declare %2 {uniq_name = "_QFEx"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
%c1_i32 = arith.constant 1 : i32
hlfir.assign %c1_i32 to %3#0 : i32, !fir.ref<i32>
%4 = fir.load %1#0 : !fir.ref<f64>
%5 = fir.convert %4 : (f64) -> i32
acc.atomic.capture {
acc.atomic.read %1#1 = %3#1 : !fir.ref<f64>, !fir.ref<i32>, i32
acc.atomic.write %3#1 = %5 : !fir.ref<i32>, i32
}
```
Fixes #112911.
93 lines
5.0 KiB
Fortran
93 lines
5.0 KiB
Fortran
! RUN: bbc -fopenacc -emit-hlfir %s -o - | FileCheck %s
|
|
|
|
subroutine atomic_update_array1(r, n, x)
|
|
implicit none
|
|
integer :: n
|
|
real :: r(n), x
|
|
integer :: i
|
|
|
|
!$acc data copy(r)
|
|
|
|
!$acc parallel loop
|
|
do i = 1, n
|
|
!$acc atomic update
|
|
r(i) = r(i) + x
|
|
!$acc end atomic
|
|
end do
|
|
|
|
!$acc end data
|
|
end subroutine
|
|
|
|
! CHECK-LABEL: func.func @_QPatomic_update_array1(
|
|
! CHECK-SAME: %[[ARG0:.*]]: !fir.ref<!fir.array<?xf32>> {fir.bindc_name = "r"}, %[[ARG1:.*]]: !fir.ref<i32> {fir.bindc_name = "n"}, %[[ARG2:.*]]: !fir.ref<f32> {fir.bindc_name = "x"}) {
|
|
! CHECK: %[[DECL_ARG2:.*]]:2 = hlfir.declare %[[ARG2]] dummy_scope %{{[0-9]+}} {uniq_name = "_QFatomic_update_array1Ex"} : (!fir.ref<f32>, !fir.dscope) -> (!fir.ref<f32>, !fir.ref<f32>)
|
|
! CHECK: %[[DECL_ARG0:.*]]:2 = hlfir.declare %[[ARG0]](%{{.*}}) dummy_scope %{{[0-9]+}} {uniq_name = "_QFatomic_update_array1Er"} : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, !fir.dscope) -> (!fir.box<!fir.array<?xf32>>, !fir.ref<!fir.array<?xf32>>)
|
|
! CHECK: %[[ARRAY_REF:.*]] = hlfir.designate %[[DECL_ARG0]]#0 (%{{.*}}) : (!fir.box<!fir.array<?xf32>>, i64) -> !fir.ref<f32>
|
|
! CHECK: %[[LOAD_X:.*]] = fir.load %[[DECL_ARG2]]#0 : !fir.ref<f32>
|
|
! CHECK: acc.atomic.update %[[ARRAY_REF]] : !fir.ref<f32> {
|
|
! CHECK: ^bb0(%[[ARG:.*]]: f32):
|
|
! CHECK: %[[ATOMIC:.*]] = arith.addf %[[ARG]], %[[LOAD_X]] fastmath<contract> : f32
|
|
! CHECK: acc.yield %[[ATOMIC]] : f32
|
|
! CHECK: }
|
|
|
|
|
|
subroutine atomic_read_array1(r, n, x)
|
|
implicit none
|
|
integer :: n
|
|
real :: r(n), x
|
|
|
|
!$acc atomic read
|
|
x = r(n)
|
|
end subroutine
|
|
|
|
! CHECK-LABEL: func.func @_QPatomic_read_array1(
|
|
! CHECK-SAME: %[[ARG0:.*]]: !fir.ref<!fir.array<?xf32>> {fir.bindc_name = "r"}, %[[ARG1:.*]]: !fir.ref<i32> {fir.bindc_name = "n"}, %[[ARG2:.*]]: !fir.ref<f32> {fir.bindc_name = "x"}) {
|
|
! CHECK: %[[DECL_X:.*]]:2 = hlfir.declare %[[ARG2]] dummy_scope %{{[0-9]+}} {uniq_name = "_QFatomic_read_array1Ex"} : (!fir.ref<f32>, !fir.dscope) -> (!fir.ref<f32>, !fir.ref<f32>)
|
|
! CHECK: %[[DECL_R:.*]]:2 = hlfir.declare %[[ARG0]](%{{.*}}) dummy_scope %{{[0-9]+}} {uniq_name = "_QFatomic_read_array1Er"} : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, !fir.dscope) -> (!fir.box<!fir.array<?xf32>>, !fir.ref<!fir.array<?xf32>>)
|
|
! CHECK: %[[DES:.*]] = hlfir.designate %[[DECL_R]]#0 (%{{.*}}) : (!fir.box<!fir.array<?xf32>>, i64) -> !fir.ref<f32>
|
|
! CHECK: acc.atomic.read %[[DECL_X]]#1 = %[[DES]] : !fir.ref<f32>, !fir.ref<f32>, f32
|
|
|
|
subroutine atomic_write_array1(r, n, x)
|
|
implicit none
|
|
integer :: n
|
|
real :: r(n), x
|
|
|
|
!$acc atomic write
|
|
x = r(n)
|
|
end subroutine
|
|
|
|
! CHECK-LABEL: func.func @_QPatomic_write_array1(
|
|
! CHECK-SAME: %[[ARG0:.*]]: !fir.ref<!fir.array<?xf32>> {fir.bindc_name = "r"}, %[[ARG1:.*]]: !fir.ref<i32> {fir.bindc_name = "n"}, %[[ARG2:.*]]: !fir.ref<f32> {fir.bindc_name = "x"}) {
|
|
! CHECK: %[[DECL_X:.*]]:2 = hlfir.declare %[[ARG2]] dummy_scope %{{[0-9]+}} {uniq_name = "_QFatomic_write_array1Ex"} : (!fir.ref<f32>, !fir.dscope) -> (!fir.ref<f32>, !fir.ref<f32>)
|
|
! CHECK: %[[DECL_R:.*]]:2 = hlfir.declare %[[ARG0]](%{{.*}}) dummy_scope %{{[0-9]+}} {uniq_name = "_QFatomic_write_array1Er"} : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, !fir.dscope) -> (!fir.box<!fir.array<?xf32>>, !fir.ref<!fir.array<?xf32>>)
|
|
! CHECK: %[[DES:.*]] = hlfir.designate %[[DECL_R]]#0 (%{{.*}}) : (!fir.box<!fir.array<?xf32>>, i64) -> !fir.ref<f32>
|
|
! CHECK: %[[LOAD:.*]] = fir.load %[[DES]] : !fir.ref<f32>
|
|
! CHECK: acc.atomic.write %[[DECL_X]]#1 = %[[LOAD]] : !fir.ref<f32>, f32
|
|
|
|
subroutine atomic_capture_array1(r, n, x, y)
|
|
implicit none
|
|
integer :: n, i
|
|
real :: r(n), x, y
|
|
|
|
!$acc atomic capture
|
|
r(i) = r(i) + x
|
|
y = r(i)
|
|
!$acc end atomic
|
|
end subroutine
|
|
|
|
! CHECK-LABEL: func.func @_QPatomic_capture_array1(
|
|
! CHECK-SAME: %[[ARG0:.*]]: !fir.ref<!fir.array<?xf32>> {fir.bindc_name = "r"}, %[[ARG1:.*]]: !fir.ref<i32> {fir.bindc_name = "n"}, %[[ARG2:.*]]: !fir.ref<f32> {fir.bindc_name = "x"}, %[[ARG3:.*]]: !fir.ref<f32> {fir.bindc_name = "y"}) {
|
|
! CHECK: %[[DECL_X:.*]]:2 = hlfir.declare %[[ARG2]] dummy_scope %{{[0-9]+}} {uniq_name = "_QFatomic_capture_array1Ex"} : (!fir.ref<f32>, !fir.dscope) -> (!fir.ref<f32>, !fir.ref<f32>)
|
|
! CHECK: %[[DECL_Y:.*]]:2 = hlfir.declare %[[ARG3]] dummy_scope %{{[0-9]+}} {uniq_name = "_QFatomic_capture_array1Ey"} : (!fir.ref<f32>, !fir.dscope) -> (!fir.ref<f32>, !fir.ref<f32>)
|
|
! CHECK: %[[DECL_R:.*]]:2 = hlfir.declare %[[ARG0]](%{{.*}}) dummy_scope %{{[0-9]+}} {uniq_name = "_QFatomic_capture_array1Er"} : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, !fir.dscope) -> (!fir.box<!fir.array<?xf32>>, !fir.ref<!fir.array<?xf32>>)
|
|
! CHECK: %[[R_I:.*]] = hlfir.designate %[[DECL_R]]#0 (%{{.*}}) : (!fir.box<!fir.array<?xf32>>, i64) -> !fir.ref<f32>
|
|
! CHECK: %[[LOAD:.*]] = fir.load %[[DECL_X]]#0 : !fir.ref<f32>
|
|
! CHECK: acc.atomic.capture {
|
|
! CHECK: acc.atomic.update %[[R_I]] : !fir.ref<f32> {
|
|
! CHECK: ^bb0(%[[ARG:.*]]: f32):
|
|
! CHECK: %[[ADD:.*]] = arith.addf %[[ARG]], %[[LOAD]] fastmath<contract> : f32
|
|
! CHECK: acc.yield %[[ADD]] : f32
|
|
! CHECK: }
|
|
! CHECK: acc.atomic.read %[[DECL_Y]]#1 = %[[R_I]] : !fir.ref<f32>, !fir.ref<f32>, f32
|
|
! CHECK: }
|