Fix shift warnings from VC++ 2015 Update 2

Update 2 gives warnings when you shift an int and then assign to a
64-bit value. This behavior can lead to unintended truncation if you
shift too far. These warnings existed in VC++ 2013 as well, but VC++
2015 Update 2 issues the warnings more aggressively.

Sample code that is warned about:

  int64_t val = 1 << retry_num;

Sample fix:

  int64_t val = static_cast<int64_t>(1) << retry_num;

or (if retry_num is guaranteed to be small enough):

  int val = 1 << retry_num;

We could suppress the warnings, but I think they are good-ish warnings
so I'd rather not do that.

TBR=jam@chromium.org
BUG=440500

Review URL: https://codereview.chromium.org/1774003002

Cr-Commit-Position: refs/heads/master@{#379984}

1 files changed, 1 insertions, 1 deletions

diff --git a/google_apis/gaia/oauth2_token_service.cc b/google_apis/gaia/oauth2_token_service.cc
index e146ca9..def534a 100644
--- a/google_apis/gaia/oauth2_token_service.cc
+++ b/google_apis/gaia/oauth2_token_service.cc
@@ -313,7 +313,7 @@ void OAuth2TokenService::Fetcher::OnGetTokenFailure(
 int64_t OAuth2TokenService::Fetcher::ComputeExponentialBackOffMilliseconds(
     int retry_num) {
   DCHECK(retry_num < max_fetch_retry_num_);
-  int64_t exponential_backoff_in_seconds = 1 << retry_num;
+  int exponential_backoff_in_seconds = 1 << retry_num;
   // Returns a backoff with randomness < 1000ms
   return (exponential_backoff_in_seconds + base::RandDouble()) * 1000;
 }