[Libclc-dev] [PATCH 2/2] amdgcn: rewrite barrier() using fence and clang __builtin_amdgcn_s_barrier

[Jan Vesely via Libclc-dev]

Specs require using fences when barrier() is invoked:
"The barrier function will either flush any variables stored in local memory
or queue a memory fence to ensure correct ordering of memory operations to local memory."
and
"The barrier function will queue a memory fence to ensure correct ordering
of memory operations to global memory."

Signed-off-by: Jan Vesely <jan.vesely at rutgers.edu>
---
 amdgcn/lib/SOURCES                         |  2 +-
 amdgcn/lib/synchronization/barrier.cl      |  7 +++++++
 amdgcn/lib/synchronization/barrier_impl.ll | 32 ------------------------------
 3 files changed, 8 insertions(+), 33 deletions(-)
 create mode 100644 amdgcn/lib/synchronization/barrier.cl
 delete mode 100644 amdgcn/lib/synchronization/barrier_impl.ll

diff --git a/amdgcn/lib/SOURCES b/amdgcn/lib/SOURCES
index 24f5949..a1f9483 100644
--- a/amdgcn/lib/SOURCES
+++ b/amdgcn/lib/SOURCES
@@ -1,7 +1,7 @@
 math/ldexp.cl
 mem_fence/fence.cl
 mem_fence/waitcnt.ll
-synchronization/barrier_impl.ll
+synchronization/barrier.cl
 workitem/get_global_offset.cl
 workitem/get_group_id.cl
 workitem/get_global_size.ll
diff --git a/amdgcn/lib/synchronization/barrier.cl b/amdgcn/lib/synchronization/barrier.cl
new file mode 100644
index 0000000..e2f3c13
--- /dev/null
+++ b/amdgcn/lib/synchronization/barrier.cl
@@ -0,0 +1,7 @@
+#include <clc/clc.h>
+
+_CLC_DEF void barrier(cl_mem_fence_flags flags)
+{
+	mem_fence(flags);
+	__builtin_amdgcn_s_barrier();
+}
diff --git a/amdgcn/lib/synchronization/barrier_impl.ll b/amdgcn/lib/synchronization/barrier_impl.ll
deleted file mode 100644
index 1809edd..0000000
--- a/amdgcn/lib/synchronization/barrier_impl.ll
+++ /dev/null
@@ -1,32 +0,0 @@
-declare i32 @__clc_clk_local_mem_fence() #1
-declare i32 @__clc_clk_global_mem_fence() #1
-declare void @llvm.amdgcn.s.barrier() #0
-
-define void @barrier(i32 %flags) #2 {
-barrier_local_test:
-  %CLK_LOCAL_MEM_FENCE = call i32 @__clc_clk_local_mem_fence()
-  %0 = and i32 %flags, %CLK_LOCAL_MEM_FENCE
-  %1 = icmp ne i32 %0, 0
-  br i1 %1, label %barrier_local, label %barrier_global_test
-
-barrier_local:
-  call void @llvm.amdgcn.s.barrier()
-  br label %barrier_global_test
-
-barrier_global_test:
-  %CLK_GLOBAL_MEM_FENCE = call i32 @__clc_clk_global_mem_fence()
-  %2 = and i32 %flags, %CLK_GLOBAL_MEM_FENCE
-  %3 = icmp ne i32 %2, 0
-  br i1 %3, label %barrier_global, label %done
-
-barrier_global:
-  call void @llvm.amdgcn.s.barrier()
-  br label %done
-
-done:
-  ret void
-}
-
-attributes #0 = { nounwind convergent }
-attributes #1 = { nounwind alwaysinline }
-attributes #2 = { nounwind convergent alwaysinline }
-- 
2.9.4