Use larger write sizes for MCFillFragment.

This brings the pr26208 testcase down to 3.2 seconds. Not checking it in
since it does create a 4GB .o file.

llvm-svn: 258149

1 files changed, 14 insertions, 3 deletions

diff --git a/llvm/lib/MC/MCAssembler.cpp b/llvm/lib/MC/MCAssembler.cpp
index a88e3df..6965b1b 100644
--- a/llvm/lib/MC/MCAssembler.cpp
+++ b/llvm/lib/MC/MCAssembler.cpp
@@ -488,9 +488,20 @@ static void writeFragment(const MCAssembler &Asm, const MCAsmLayout &Layout,
   case MCFragment::FT_Fill: {
     ++stats::EmittedFillFragments;
     const MCFillFragment &FF = cast<MCFillFragment>(F);
-
-    for (uint64_t I = 0, E = FF.getSize(); I != E; ++I)
-      OW->write8(FF.getValue());
+    uint8_t V = FF.getValue();
+    const unsigned MaxChunkSize = 16;
+    char Data[MaxChunkSize];
+    memcpy(Data, &V, 1);
+    for (unsigned I = 1; I < MaxChunkSize; ++I)
+      Data[I] = Data[0];
+
+    uint64_t Size = FF.getSize();
+    for (unsigned ChunkSize = MaxChunkSize; ChunkSize; ChunkSize /= 2) {
+      StringRef Ref(Data, ChunkSize);
+      for (uint64_t I = 0, E = Size / ChunkSize; I != E; ++I)
+        OW->writeBytes(Ref);
+      Size = Size % ChunkSize;
+    }
     break;
   }