Skip to content

[SPIRV] Addition of image_store.ll, signed_arithmetic_overflow.ll, spirv_global_varibale_declaration.ll #152289

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

aadeshps-mcw
Copy link
Contributor

--Test for verifying that decorations are correctly applied to global variables
--Test for signed arithmetic overflow intrinsics, which is for now expectedly failing
--Test for verifying emission of correct image type

@llvmbot
Copy link
Member

llvmbot commented Aug 6, 2025

@llvm/pr-subscribers-backend-spir-v

Author: Aadesh Premkumar (aadeshps-mcw)

Changes

--Test for verifying that decorations are correctly applied to global variables
--Test for signed arithmetic overflow intrinsics, which is for now expectedly failing
--Test for verifying emission of correct image type


Full diff: https://github.com/llvm/llvm-project/pull/152289.diff

3 Files Affected:

  • (added) llvm/test/CodeGen/SPIRV/image_store.ll (+22)
  • (added) llvm/test/CodeGen/SPIRV/llvm-intrinsics/signed_arithmetic_overflow.ll (+30)
  • (added) llvm/test/CodeGen/SPIRV/spirv_global_variable_decoration.ll (+29)
diff --git a/llvm/test/CodeGen/SPIRV/image_store.ll b/llvm/test/CodeGen/SPIRV/image_store.ll
new file mode 100644
index 0000000000000..73ea6ad038de7
--- /dev/null
+++ b/llvm/test/CodeGen/SPIRV/image_store.ll
@@ -0,0 +1,22 @@
+; RUN: llc -O0 -verify-machineinstrs -mtriple=spirv64-unknown-unknown %s -o - | FileCheck %s
+; RUN: %if spirv-tools %{ llc -O0 -mtriple=spirv64-unknown-unknown %s -o - -filetype=obj | spirv-val %}
+
+; Image types may be represented in two ways while translating to SPIR-V:
+; - OpenCL form, for example, '%opencl.image2d_ro_t',
+; - SPIR-V form, for example, '%spirv.Image._void_1_0_0_0_0_0_0',
+; but it is still one type which should be translated to one SPIR-V type.
+;
+; The test checks that the code below is successfully translated and only one
+; SPIR-V type for images is generated.
+
+; CHECK:     %[[#]] = OpTypeImage %[[#]] 2D
+; CHECK-NOT: %[[#]] = OpTypeImage %[[#]] 2D
+
+declare spir_func <4 x float> @_Z11read_imagef14ocl_image2d_ro11ocl_samplerDv2_ff(ptr addrspace(1), ptr addrspace(2), <2 x float>, float)
+
+define spir_kernel void @read_image(ptr addrspace(1) %srcimg, ptr addrspace(2) %sampler){
+entry:
+  %spirvimg.addr = alloca target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0), align 8
+  %val = call <4 x float> @_Z11read_imagef14ocl_image2d_ro11ocl_samplerDv2_ff(ptr addrspace(1) %srcimg, ptr addrspace(2) %sampler, <2 x float> zeroinitializer, float 0.0)
+  ret void
+}
diff --git a/llvm/test/CodeGen/SPIRV/llvm-intrinsics/signed_arithmetic_overflow.ll b/llvm/test/CodeGen/SPIRV/llvm-intrinsics/signed_arithmetic_overflow.ll
new file mode 100644
index 0000000000000..52f939faf0a9f
--- /dev/null
+++ b/llvm/test/CodeGen/SPIRV/llvm-intrinsics/signed_arithmetic_overflow.ll
@@ -0,0 +1,30 @@
+; RUN: llc -O0 -mtriple=spirv64-unknown-unknown %s -o - | FileCheck %s
+; RUN: %if spirv-tools %{ llc -O0 -mtriple=spirv64-unknown-unknown %s -filetype=obj -o - | spirv-val %}
+; XFAIL: *
+;@llvm.sadd.with.overflow and @llvm.ssub.with.overflow has not been implemented.
+
+define spir_func void @test_sadd_overflow(ptr %out_result, ptr %out_overflow, i32 %a, i32 %b) {
+entry:
+  %res = call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
+  %val = extractvalue { i32, i1 } %res, 0
+  %ofl = extractvalue { i32, i1 } %res, 1
+  store i32 %val, ptr %out_result
+  %zext_ofl = zext i1 %ofl to i8
+  store i8 %zext_ofl, ptr %out_overflow
+  ret void
+}
+
+declare { i32, i1 } @llvm.sadd.with.overflow.i32(i32, i32)
+
+define spir_func void @test_ssub_overflow(ptr %out_result, ptr %out_overflow, i32 %a, i32 %b) {
+entry:
+  %res = call { i32, i1 } @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
+  %val = extractvalue { i32, i1 } %res, 0
+  %ofl = extractvalue { i32, i1 } %res, 1
+  store i32 %val, ptr %out_result
+  %zext_ofl = zext i1 %ofl to i8
+  store i8 %zext_ofl, ptr %out_overflow
+  ret void
+}
+
+declare { i32, i1 } @llvm.ssub.with.overflow.i32(i32, i32)
diff --git a/llvm/test/CodeGen/SPIRV/spirv_global_variable_decoration.ll b/llvm/test/CodeGen/SPIRV/spirv_global_variable_decoration.ll
new file mode 100644
index 0000000000000..ce5ed087650cd
--- /dev/null
+++ b/llvm/test/CodeGen/SPIRV/spirv_global_variable_decoration.ll
@@ -0,0 +1,29 @@
+; RUN: llc -verify-machineinstrs -O0 -mtriple=spirv32-unknown-unknown %s -o - | FileCheck %s
+; RUN: %if spirv-tools %{ llc -O0 -mtriple=spirv32-unknown-unknown %s -o - -filetype=obj | spirv-val %}
+
+@v1 = addrspace(1) global i32 42, !spirv.Decorations !2
+@v2 = addrspace(1) global float 1.0, !spirv.Decorations !4
+
+define i32 @use_globals() {
+entry:
+  %v1_ptr = addrspacecast i32 addrspace(1)* @v1 to i32*
+  %v1_val = load i32, i32* %v1_ptr
+
+  %v2_ptr = addrspacecast float addrspace(1)* @v2 to float*
+  %v2_val = load float, float* %v2_ptr
+  %v2_int = fptosi float %v2_val to i32
+
+  %sum = add i32 %v1_val, %v2_int
+  ret i32 %sum
+}
+
+; CHECK: OpDecorate %[[#PId1:]] Constant
+; CHECK: OpDecorate %[[#PId2:]] Constant
+; CHECK: OpDecorate %[[#PId2]] Binding 1
+; CHECK: %[[#PId1]] = OpVariable %[[#]] CrossWorkgroup %[[#]]
+; CHECK: %[[#PId2]] = OpVariable %[[#]] CrossWorkgroup %[[#]]
+
+!1 = !{i32 22}                          ; Constant
+!2 = !{!1}                             ; @v1
+!3 = !{i32 33, i32 1}                  ; Binding 1
+!4 = !{!1, !3}                         ; @v2

@aadeshps-mcw aadeshps-mcw marked this pull request as draft August 6, 2025 11:30
--Test for verifying that decorations are correctly applied to global variables
--Test for signed arithmetic overflow intrinsics, which is for now expectedly failing
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants