[llvm] [memprof] Speed up caller-callee pair extraction (PR #116184)

[Snehasish Kumar via llvm-commits]

================
@@ -1303,6 +1303,10 @@ Error IndexedMemProfReader::deserializeV3(const unsigned char *Start,
   FrameBase = Ptr;
   CallStackBase = Start + CallStackPayloadOffset;
 
+  // Compute the number of elements in the radix tree array.
+  RadixTreeSize = (RecordPayloadOffset - CallStackPayloadOffset) /
----------------
snehasish wrote:

Today the memprof on disk format is compact without padding. For example, the indexed profile format sections are aligned which introduces padding. Can we add a check here (or elsewhere) to note that we depend on this behaviour?

https://github.com/llvm/llvm-project/pull/116184