[PATCH] D33503: Make BinaryStreamReader::readCString a bit faster.

[Zachary Turner via Phabricator via llvm-commits]

zturner created this revision.

Previously it would do a character by character search for a null terminator, to account for the fact that an arbitrary stream need not store its data contiguously so you couldn't just do a `memchr`.  However, the stream API has a function which will return the longest contiguous chunk without doing a copy, and by using this function we can do a `memchr` on the individual chunks.  For certain types of streams like data from object files etc, this is guaranteed to find the null terminator with only a single `memchr`, but even with discontiguous streams such as `MappedBlockStream`, it's rare that any given string will cross a block boundary, so even those will almost always be satisfied with a single `memchr`.

This optimization is worth a 10-12% reduction in link time (4.2 seconds -> 3.75 seconds)


https://reviews.llvm.org/D33503

Files:
  llvm/lib/Support/BinaryStreamReader.cpp


Index: llvm/lib/Support/BinaryStreamReader.cpp
===================================================================
--- llvm/lib/Support/BinaryStreamReader.cpp
+++ llvm/lib/Support/BinaryStreamReader.cpp
@@ -42,29 +42,30 @@
 }
 
 Error BinaryStreamReader::readCString(StringRef &Dest) {
-  // TODO: This could be made more efficient by using readLongestContiguousChunk
-  // and searching for null terminators in the resulting buffer.
-
-  uint32_t Length = 0;
-  // First compute the length of the string by reading 1 byte at a time.
   uint32_t OriginalOffset = getOffset();
-  const char *C;
+  uint32_t FoundOffset = 0;
   while (true) {
-    if (auto EC = readObject(C))
+    uint32_t ThisOffset = getOffset();
+    ArrayRef<uint8_t> Buffer;
+    if (auto EC = readLongestContiguousChunk(Buffer))
       return EC;
-    if (*C == '\0')
+    StringRef S(reinterpret_cast<const char *>(Buffer.begin()), Buffer.size());
+    size_t Pos = S.find_first_of('\0');
+    if (LLVM_LIKELY(Pos != StringRef::npos)) {
+      FoundOffset = Pos + ThisOffset;
       break;
-    ++Length;
+    }
   }
-  // Now go back and request a reference for that many bytes.
-  uint32_t NewOffset = getOffset();
+  assert(FoundOffset >= OriginalOffset);
+
   setOffset(OriginalOffset);
+  size_t Length = FoundOffset - OriginalOffset;
 
   if (auto EC = readFixedString(Dest, Length))
     return EC;
 
-  // Now set the offset back to where it was after we calculated the length.
-  setOffset(NewOffset);
+  // Now set the offset back to after the null terminator.
+  setOffset(FoundOffset + 1);
   return Error::success();
 }
 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D33503.100114.patch
Type: text/x-patch
Size: 1608 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170524/224863b6/attachment.bin>