[NVPTX] Emit .pragma "nounroll" for loops marked with nounroll
Summary:
CUDA driver can unroll loops when jit-compiling PTX. To prevent CUDA
driver from unrolling a loop marked with llvm.loop.unroll.disable is not
unrolled by CUDA driver, we need to emit .pragma "nounroll" at the
header of that loop.
This patch also extracts getting unroll metadata from loop ID metadata
into a shared helper function.
Test Plan: test/CodeGen/NVPTX/nounroll.ll
Reviewers: eliben, meheff, jholewinski
Reviewed By: jholewinski
Subscribers: jholewinski, llvm-commits
Differential Revision: http://reviews.llvm.org/D7041
llvm-svn: 227703
diff --git a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
index 0a6bd84..86231a1 100644
--- a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
@@ -234,44 +234,27 @@
// Returns the loop hint metadata node with the given name (for example,
// "llvm.loop.unroll.count"). If no such metadata node exists, then nullptr is
// returned.
-static const MDNode *GetUnrollMetadata(const Loop *L, StringRef Name) {
+static const MDNode *GetUnrollMetadataForLoop(const Loop *L, StringRef Name) {
MDNode *LoopID = L->getLoopID();
if (!LoopID)
return nullptr;
-
- // First operand should refer to the loop id itself.
- assert(LoopID->getNumOperands() > 0 && "requires at least one operand");
- assert(LoopID->getOperand(0) == LoopID && "invalid loop id");
-
- for (unsigned i = 1, e = LoopID->getNumOperands(); i < e; ++i) {
- const MDNode *MD = dyn_cast<MDNode>(LoopID->getOperand(i));
- if (!MD)
- continue;
-
- const MDString *S = dyn_cast<MDString>(MD->getOperand(0));
- if (!S)
- continue;
-
- if (Name.equals(S->getString()))
- return MD;
- }
- return nullptr;
+ return GetUnrollMetadata(LoopID, Name);
}
// Returns true if the loop has an unroll(full) pragma.
static bool HasUnrollFullPragma(const Loop *L) {
- return GetUnrollMetadata(L, "llvm.loop.unroll.full");
+ return GetUnrollMetadataForLoop(L, "llvm.loop.unroll.full");
}
// Returns true if the loop has an unroll(disable) pragma.
static bool HasUnrollDisablePragma(const Loop *L) {
- return GetUnrollMetadata(L, "llvm.loop.unroll.disable");
+ return GetUnrollMetadataForLoop(L, "llvm.loop.unroll.disable");
}
// If loop has an unroll_count pragma return the (necessarily
// positive) value from the pragma. Otherwise return 0.
static unsigned UnrollCountPragmaValue(const Loop *L) {
- const MDNode *MD = GetUnrollMetadata(L, "llvm.loop.unroll.count");
+ const MDNode *MD = GetUnrollMetadataForLoop(L, "llvm.loop.unroll.count");
if (MD) {
assert(MD->getNumOperands() == 2 &&
"Unroll count hint metadata should have two operands.");
diff --git a/llvm/lib/Transforms/Utils/LoopUnroll.cpp b/llvm/lib/Transforms/Utils/LoopUnroll.cpp
index dbd032e..0fee692 100644
--- a/llvm/lib/Transforms/Utils/LoopUnroll.cpp
+++ b/llvm/lib/Transforms/Utils/LoopUnroll.cpp
@@ -549,3 +549,26 @@
return true;
}
+
+/// Given an llvm.loop loop id metadata node, returns the loop hint metadata
+/// node with the given name (for example, "llvm.loop.unroll.count"). If no
+/// such metadata node exists, then nullptr is returned.
+const MDNode *llvm::GetUnrollMetadata(const MDNode *LoopID, StringRef Name) {
+ // First operand should refer to the loop id itself.
+ assert(LoopID->getNumOperands() > 0 && "requires at least one operand");
+ assert(LoopID->getOperand(0) == LoopID && "invalid loop id");
+
+ for (unsigned i = 1, e = LoopID->getNumOperands(); i < e; ++i) {
+ const MDNode *MD = dyn_cast<MDNode>(LoopID->getOperand(i));
+ if (!MD)
+ continue;
+
+ const MDString *S = dyn_cast<MDString>(MD->getOperand(0));
+ if (!S)
+ continue;
+
+ if (Name.equals(S->getString()))
+ return MD;
+ }
+ return nullptr;
+}