[libc] Improve memcpy copy loop

Rewriting loop so the terminating condition does not depend on the loop body

Differential Revision: https://reviews.llvm.org/D91976

diff --git a/libc/src/string/memory_utils/memcpy_utils.h b/libc/src/string/memory_utils/memcpy_utils.h
index a0e5ccc..aa27b3c 100644 (file)
--- a/libc/src/string/memory_utils/memcpy_utils.h
+++ b/libc/src/string/memory_utils/memcpy_utils.h
@@ -90,9 +90,10 @@ static void CopyAlignedBlocks(char *__restrict dst, const char *__restrict src,
   CopyBlock<kBlockSize>(dst, src); // Copy first block
 
   // Copy aligned blocks
-  size_t offset = kBlockSize - offset_from_last_aligned<kBlockSize>(dst);
-  for (; offset + kBlockSize < count; offset += kBlockSize)
-    CopyBlock<kBlockSize>(dst + offset, src + offset);
+  const size_t ofla = offset_from_last_aligned<kBlockSize>(dst);
+  const size_t limit = count + ofla - kBlockSize;
+  for (size_t offset = kBlockSize; offset < limit; offset += kBlockSize)
+    CopyBlock<kBlockSize>(dst - ofla + offset, src - ofla + offset);
 
   CopyLastBlock<kBlockSize>(dst, src, count); // Copy last block
 }

diff --git a/libc/test/src/string/memory_utils/memcpy_utils_test.cpp b/libc/test/src/string/memory_utils/memcpy_utils_test.cpp
index 7e32fb4..93c0c48 100644 (file)
--- a/libc/test/src/string/memory_utils/memcpy_utils_test.cpp
+++ b/libc/test/src/string/memory_utils/memcpy_utils_test.cpp
@@ -162,7 +162,23 @@ TEST(MemcpyUtilsTest, CopyBlockOverlap) {
 
 TEST(MemcpyUtilsTest, CopyAlignedBlocks) {
   auto &trace = GetTrace();
-  // Destination is aligned already.
+  // Destination is aligned and multiple of alignment.
+  //   "1111"
+  trace.Clear();
+  CopyAlignedBlocks<4>(I(0), I(0), 4);
+  EXPECT_STREQ(trace.Write(), "2222");
+  EXPECT_STREQ(trace.Read(), "2222");
+
+  // Destination is aligned and multiple of alignment.
+  //   "11110000"
+  // + "00001111"
+  // = "11111111"
+  trace.Clear();
+  CopyAlignedBlocks<4>(I(0), I(0), 8);
+  EXPECT_STREQ(trace.Write(), "11111111");
+  EXPECT_STREQ(trace.Read(), "11111111");
+
+  // Destination is aligned already overlap at end.
   //   "1111000000000"
   // + "0000111100000"
   // + "0000000011110"
@@ -173,7 +189,7 @@ TEST(MemcpyUtilsTest, CopyAlignedBlocks) {
   EXPECT_STREQ(trace.Write(), "1111111112221");
   EXPECT_STREQ(trace.Read(), "1111111112221");
 
-  // Misaligned destination
+  // Misaligned destination.
   //   "01111000000000"
   // + "00001111000000"
   // + "00000000111100"
@@ -183,6 +199,16 @@ TEST(MemcpyUtilsTest, CopyAlignedBlocks) {
   CopyAlignedBlocks<4>(I(1), I(0), 13);
   EXPECT_STREQ(trace.Write(), "01112111112211");
   EXPECT_STREQ(trace.Read(), "1112111112211");
+
+  // Misaligned destination aligned at end.
+  //   "011110000000"
+  // + "000011110000"
+  // + "000000001111"
+  // = "011121111111"
+  trace.Clear();
+  CopyAlignedBlocks<4>(I(1), I(0), 11);
+  EXPECT_STREQ(trace.Write(), "011121111111");
+  EXPECT_STREQ(trace.Read(), "11121111111");
 }
 
 TEST(MemcpyUtilsTest, MaxReloads) {