[memprof] Compute CallStackId when deserializing IndexedAllocationInfo (#86421)

There are two ways to create in-memory instances of
IndexedAllocationInfo -- deserialization of the raw MemProf data and
that of the indexed MemProf data.

With:

  commit 74799f4
  Author: Kazu Hirata <[email protected]>
  Date:   Sat Mar 23 19:50:15 2024 -0700

we compute CallStackId for each call stack in IndexedAllocationInfo
while deserializing the raw MemProf data.

This patch does the same while deserilizing the indexed MemProf data.

As with the patch above, this patch does not add any use of
CallStackId yet.

Author: kazutakahirata

llvm/include/llvm/ProfileData/MemProf.h

@@ -635,6 +635,11 @@ class FrameLookupTrait {
 // Compute a CallStackId for a given call stack.
 CallStackId hashCallStack(ArrayRef<FrameId> CS);
 
+// Verify that each CallStackId is computed with hashCallStack.  This function
+// is intended to help transition from CallStack to CSId in
+// IndexedAllocationInfo.
+void verifyIndexedMemProfRecord(const IndexedMemProfRecord &Record);
+
 // Verify that each CallStackId is computed with hashCallStack.  This function
 // is intended to help transition from CallStack to CSId in
 // IndexedAllocationInfo.

llvm/lib/ProfileData/InstrProfReader.cpp

@@ -1261,6 +1261,14 @@ Error IndexedInstrProfReader::readHeader() {
         /*Buckets=*/Start + FrameTableOffset,
         /*Payload=*/Start + FramePayloadOffset,
         /*Base=*/Start, memprof::FrameLookupTrait()));
+
+#if EXPENSIVE_CHECKS
+    // Go through all the records and verify that CSId has been correctly
+    // populated.  Do this only under EXPENSIVE_CHECKS.  Otherwise, we
+    // would defeat the purpose of OnDiskIterableChainedHashTable.
+    for (const auto &Record : MemProfRecordTable->data())
+      verifyIndexedMemProfRecord(Record);
+#endif
   }
 
   // BinaryIdOffset field in the header is only valid when the format version

llvm/lib/ProfileData/MemProf.cpp

@@ -53,6 +53,7 @@ IndexedMemProfRecord::deserialize(const MemProfSchema &Schema,
           endian::readNext<FrameId, llvm::endianness::little, unaligned>(Ptr);
       Node.CallStack.push_back(Id);
     }
+    Node.CSId = hashCallStack(Node.CallStack);
     Node.Info.deserialize(Schema, Ptr);
     Ptr += PortableMemInfoBlock::serializedSize();
     Record.AllocSites.push_back(Node);
@@ -130,15 +131,19 @@ CallStackId hashCallStack(ArrayRef<FrameId> CS) {
   return CSId;
 }
 
+void verifyIndexedMemProfRecord(const IndexedMemProfRecord &Record) {
+  for (const auto &AS : Record.AllocSites) {
+    assert(AS.CSId == hashCallStack(AS.CallStack));
+    (void)AS;
+  }
+}
+
 void verifyFunctionProfileData(
     const llvm::MapVector<GlobalValue::GUID, IndexedMemProfRecord>
         &FunctionProfileData) {
   for (const auto &[GUID, Record] : FunctionProfileData) {
     (void)GUID;
-    for (const auto &AS : Record.AllocSites) {
-      assert(AS.CSId == hashCallStack(AS.CallStack));
-      (void)AS;
-    }
+    verifyIndexedMemProfRecord(Record);
   }
 }