Shift by the right number of bytes when grabbing the high 32 bits.

When size_t is not 64 bits (8 bytes) but instead N < 8 bytes, we
want to grab the top N bytes from the 64-bit result.

This code was meant to do this:
hash64 >> (sizeof(uint64) - sizeof(std::size_t))

But this translates to:
hash64 >> (8 - N)

And 8-N is the number of bytes we want to shift, not the number of
bits, where as operator>> excepts a number of bits.

Fix this to correct the algorithm for platforms with small size_t.

R=ajwong
BUG=

Review URL: https://codereview.chromium.org/156103003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@253557 0039d316-1c4b-4281-b951-d872f2087c98

1 files changed, 2 insertions, 2 deletions

diff --git a/base/containers/hash_tables.h b/base/containers/hash_tables.h
index 365b586..6f37c49 100644
--- a/base/containers/hash_tables.h
+++ b/base/containers/hash_tables.h
@@ -140,7 +140,7 @@ inline std::size_t HashInts32(uint32 value1, uint32 value2) {
 
   hash64 = hash64 * odd_random + shift_random;
   std::size_t high_bits = static_cast<std::size_t>(
-      hash64 >> (sizeof(uint64) - sizeof(std::size_t)));
+      hash64 >> (8 * (sizeof(uint64) - sizeof(std::size_t))));
   return high_bits;
 }
 
@@ -175,7 +175,7 @@ inline std::size_t HashInts64(uint64 value1, uint64 value2) {
 
   hash64 = hash64 * odd_random + shift_random;
   std::size_t high_bits = static_cast<std::size_t>(
-      hash64 >> (sizeof(uint64) - sizeof(std::size_t)));
+      hash64 >> (8 * (sizeof(uint64) - sizeof(std::size_t))));
   return high_bits;
 }