[llvm] r366107 - Expand comment about how StringsToBuckets was computed, and add more entries

[Nico Weber via llvm-commits]

Author: nico
Date: Mon Jul 15 11:56:56 2019
New Revision: 366107

URL: http://llvm.org/viewvc/llvm-project?rev=366107&view=rev
Log:
Expand comment about how StringsToBuckets was computed, and add more entries

The construction was explained in
https://reviews.llvm.org/D44810?id=139526#inline-391999 but reading the code
shouldn't require hunting down old reviews to understand it.

The precomputed list was missing an entry for the empty list case, and
one entry at the very end. (The current last entry is the last one where
3 * BucketCount fits in a signed int, but the reference implementation
uses unsigneds as far as I can tell, so there's room for one more entry.)

No behavior change for inputs seen in practice.

Differential Revision: https://reviews.llvm.org/D64738

Modified:
    llvm/trunk/lib/DebugInfo/PDB/Native/PDBStringTableBuilder.cpp

Modified: llvm/trunk/lib/DebugInfo/PDB/Native/PDBStringTableBuilder.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/DebugInfo/PDB/Native/PDBStringTableBuilder.cpp?rev=366107&r1=366106&r2=366107&view=diff
==============================================================================
--- llvm/trunk/lib/DebugInfo/PDB/Native/PDBStringTableBuilder.cpp (original)
+++ llvm/trunk/lib/DebugInfo/PDB/Native/PDBStringTableBuilder.cpp Mon Jul 15 11:56:56 2019
@@ -60,7 +60,19 @@ static uint32_t computeBucketCount(uint3
   // strings.  Matching the reference algorithm exactly is not strictly
   // necessary for correctness, but it helps when comparing LLD's PDBs with
   // Microsoft's PDBs so as to eliminate superfluous differences.
+  // The reference implementation does (in nmt.h, NMT::grow()):
+  //   unsigned StringCount = 0;
+  //   unsigned BucketCount = 1;
+  //   fn insert() {
+  //     ++StringCount;
+  //     if (BucketCount * 3 / 4 < StringCount)
+  //       BucketCount = BucketCount * 3 / 2 + 1;
+  //   }
+  // This list contains all StringCount, BucketCount pairs where BucketCount was
+  // just incremented.  It ends before the first BucketCount entry where
+  // BucketCount * 3 would overflow a 32-bit unsigned int.
   static std::map<uint32_t, uint32_t> StringsToBuckets = {
+      {0, 1},
       {1, 2},
       {2, 4},
       {4, 7},
@@ -110,7 +122,8 @@ static uint32_t computeBucketCount(uint3
       {229865455, 459730910},
       {344798183, 689596366},
       {517197275, 1034394550},
-      {775795913, 1551591826}};
+      {775795913, 1551591826},
+      {1163693870, 2327387740}};
   auto Entry = StringsToBuckets.lower_bound(NumStrings);
   assert(Entry != StringsToBuckets.end());
   return Entry->second;