From: Matt Arsenault <Matthew.Arsenault@amd.com>
Date: Thu, 25 Aug 2016 07:31:40 +0000 (+0000)
Subject: amdgcn: Fix return type of get_num_groups
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=958fce3192ac088dae3371bdedb43ffd6d4a5402;p=platform%2Fupstream%2Fllvm.git

amdgcn: Fix return type of get_num_groups

llvm-svn: 279723
---

diff --git a/libclc/amdgcn/lib/SOURCES b/libclc/amdgcn/lib/SOURCES
index f7e2671..1ff5fd1 100644
--- a/libclc/amdgcn/lib/SOURCES
+++ b/libclc/amdgcn/lib/SOURCES
@@ -5,4 +5,5 @@ workitem/get_group_id.cl
 workitem/get_global_size.ll
 workitem/get_local_id.cl
 workitem/get_local_size.ll
+workitem/get_num_groups.ll
 workitem/get_work_dim.cl
diff --git a/libclc/amdgcn/lib/workitem/get_num_groups.ll b/libclc/amdgcn/lib/workitem/get_num_groups.ll
new file mode 100644
index 0000000..5b4b1cc
--- /dev/null
+++ b/libclc/amdgcn/lib/workitem/get_num_groups.ll
@@ -0,0 +1,21 @@
+declare i32 @llvm.r600.read.ngroups.x() nounwind readnone
+declare i32 @llvm.r600.read.ngroups.y() nounwind readnone
+declare i32 @llvm.r600.read.ngroups.z() nounwind readnone
+
+define i64 @get_num_groups(i32 %dim) nounwind readnone alwaysinline {
+  switch i32 %dim, label %default [i32 0, label %x_dim i32 1, label %y_dim i32 2, label %z_dim]
+x_dim:
+  %x = call i32 @llvm.r600.read.ngroups.x()
+  %x.ext = zext i32 %x to i64
+  ret i64 %x.ext
+y_dim:
+  %y = call i32 @llvm.r600.read.ngroups.y()
+  %y.ext = zext i32 %y to i64
+  ret i64 %y.ext
+z_dim:
+  %z = call i32 @llvm.r600.read.ngroups.z()
+  %z.ext = zext i32 %z to i64
+  ret i64 %z.ext
+default:
+  ret i64 1
+}
diff --git a/libclc/amdgpu/lib/SOURCES b/libclc/amdgpu/lib/SOURCES
index fac2fc3..9379ad7 100644
--- a/libclc/amdgpu/lib/SOURCES
+++ b/libclc/amdgpu/lib/SOURCES
@@ -16,4 +16,3 @@ image/write_imagef.cl
 image/write_imagei.cl
 image/write_imageui.cl
 image/write_image_impl.ll
-workitem/get_num_groups.ll
diff --git a/libclc/r600/lib/SOURCES b/libclc/r600/lib/SOURCES
index d55ff30..284052c 100644
--- a/libclc/r600/lib/SOURCES
+++ b/libclc/r600/lib/SOURCES
@@ -4,4 +4,5 @@ workitem/get_group_id.cl
 workitem/get_global_size.ll
 workitem/get_local_id.cl
 workitem/get_local_size.ll
+workitem/get_num_groups.ll
 workitem/get_work_dim.cl
diff --git a/libclc/amdgpu/lib/workitem/get_num_groups.ll b/libclc/r600/lib/workitem/get_num_groups.ll
similarity index 98%
rename from libclc/amdgpu/lib/workitem/get_num_groups.ll
rename to libclc/r600/lib/workitem/get_num_groups.ll
index a708f42..d85a797 100644
--- a/libclc/amdgpu/lib/workitem/get_num_groups.ll
+++ b/libclc/r600/lib/workitem/get_num_groups.ll
@@ -14,5 +14,5 @@ z_dim:
   %z = call i32 @llvm.r600.read.ngroups.z() nounwind readnone
   ret i32 %z
 default:
-  ret i32 0
+  ret i32 1
 }