[llvm] 23cd70d - [PDB] Fix out-of-bounds acces when sorting GSI buckets

[Alexandre Ganea via llvm-commits]

Author: Alexandre Ganea
Date: 2020-07-10T10:55:27-04:00
New Revision: 23cd70d71c10dc0b31ac37a733349f9de2e9b84c

URL: https://github.com/llvm/llvm-project/commit/23cd70d71c10dc0b31ac37a733349f9de2e9b84c
DIFF: https://github.com/llvm/llvm-project/commit/23cd70d71c10dc0b31ac37a733349f9de2e9b84c.diff

LOG: [PDB] Fix out-of-bounds acces when sorting GSI buckets

When building in Debug on Windows-MSVC after b7402edce315, a lot of tests were failing because we were dereferencing an element past the end of HashRecords. This happened towards the end of the table, in unused slots.

Added: 
    

Modified: 
    llvm/lib/DebugInfo/PDB/Native/GSIStreamBuilder.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/DebugInfo/PDB/Native/GSIStreamBuilder.cpp b/llvm/lib/DebugInfo/PDB/Native/GSIStreamBuilder.cpp
index ce248f34762d..4e58489f1401 100644
--- a/llvm/lib/DebugInfo/PDB/Native/GSIStreamBuilder.cpp
+++ b/llvm/lib/DebugInfo/PDB/Native/GSIStreamBuilder.cpp
@@ -232,8 +232,10 @@ void GSIHashStreamBuilder::finalizeBuckets(
   // The algorithm used here corresponds to the function
   // caseInsensitiveComparePchPchCchCch in the reference implementation.
   parallelForEachN(0, IPHR_HASH, [&](size_t I) {
-    auto B = &HashRecords[BucketStarts[I]];
-    auto E = &HashRecords[BucketCursors[I]];
+    auto B = HashRecords.begin() + BucketStarts[I];
+    auto E = HashRecords.begin() + BucketCursors[I];
+    if (B == E)
+      return;
     auto BucketCmp = [Records](const PSHashRecord &LHash,
                                const PSHashRecord &RHash) {
       const BulkPublic &L = Records[uint32_t(LHash.Off)];