From fa07e8a945193f8f1cf94de5e11f0a2d8df9a820 Mon Sep 17 00:00:00 2001
From: Danila Kutenin <danilak@google.com>
Date: Wed, 26 Mar 2025 14:51:54 +0000
Subject: [PATCH] Fix data corruption when output of snappy compression is more
 than 4GB

This fixes #201
---
 snappy.cc          |  7 ++++---
 snappy_unittest.cc | 13 +++++++++++++
 2 files changed, 17 insertions(+), 3 deletions(-)

diff --git a/snappy.cc b/snappy.cc
index 877b65a..7a8920f 100644
--- a/snappy.cc
+++ b/snappy.cc
@@ -1499,7 +1499,7 @@ class SnappyDecompressor {
   // If ip < ip_limit_min_maxtaglen_ it's safe to read kMaxTagLength from
   // buffer.
   const char* ip_limit_min_maxtaglen_;
-  uint32_t peeked_;                  // Bytes peeked from reader (need to skip)
+  uint64_t peeked_;                  // Bytes peeked from reader (need to skip)
   bool eof_;                         // Hit end of input without an error?
   char scratch_[kMaximumTagLength];  // See RefillTag().
 
@@ -1726,7 +1726,7 @@ bool SnappyDecompressor::RefillTag() {
   assert(needed <= sizeof(scratch_));
 
   // Read more bytes from reader if needed
-  uint32_t nbuf = ip_limit_ - ip;
+  uint64_t nbuf = ip_limit_ - ip;
   if (nbuf < needed) {
     // Stitch together bytes from ip and reader to form the word
     // contents.  We store the needed bytes in "scratch_".  They
@@ -1739,7 +1739,7 @@ bool SnappyDecompressor::RefillTag() {
       size_t length;
       const char* src = reader_->Peek(&length);
       if (length == 0) return false;
-      uint32_t to_add = std::min<uint32_t>(needed - nbuf, length);
+      uint64_t to_add = std::min<uint64_t>(needed - nbuf, length);
       std::memcpy(scratch_ + nbuf, src, to_add);
       nbuf += to_add;
       reader_->Skip(to_add);
@@ -1802,6 +1802,7 @@ size_t Compress(Source* reader, Sink* writer, CompressionOptions options) {
   int token = 0;
   size_t written = 0;
   size_t N = reader->Available();
+  assert(N <= 0xFFFFFFFFu);
   const size_t uncompressed_size = N;
   char ulength[Varint::kMax32];
   char* p = Varint::Encode32(ulength, N);
diff --git a/snappy_unittest.cc b/snappy_unittest.cc
index e57b13d..923a0ec 100644
--- a/snappy_unittest.cc
+++ b/snappy_unittest.cc
@@ -27,6 +27,7 @@
 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 #include <algorithm>
+#include <cinttypes>
 #include <cmath>
 #include <cstdlib>
 #include <random>
@@ -484,6 +485,18 @@ TEST(Snappy, MaxBlowup) {
   Verify(input);
 }
 
+// Issue #201, when output is more than 4GB, we had a data corruption bug.
+// We cannot run this test always because of CI constraints.
+TEST(Snappy, DISABLED_MoreThan4GB) {
+  std::mt19937 rng;
+  std::uniform_int_distribution<int> uniform_byte(0, 255);
+  std::string input;
+  input.resize((1ull << 32) - 1);
+  for (uint64_t i = 0; i < ((1ull << 32) - 1); ++i)
+    input[i] = static_cast<char>(uniform_byte(rng));
+  Verify(input);
+}
+
 TEST(Snappy, RandomData) {
   std::minstd_rand0 rng(snappy::GetFlag(FLAGS_test_random_seed));
   std::uniform_int_distribution<int> uniform_0_to_3(0, 3);