diff options
author | Sree Kuchibhotla <sreek@google.com> | 2018-07-25 10:39:16 -0700 |
---|---|---|
committer | Sree Kuchibhotla <sreek@google.com> | 2018-07-25 10:39:16 -0700 |
commit | dd45987a5bf976079d9d89fa127740bd6f516a16 (patch) | |
tree | fda914253ce0f1b38afd0daf3b8ca17e5d7844ff | |
parent | 8f39834dd1d68c6fed18d2af2eb5f7acdb8549ab (diff) |
fix comment
-rw-r--r-- | src/core/lib/iomgr/resource_quota.cc | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/src/core/lib/iomgr/resource_quota.cc b/src/core/lib/iomgr/resource_quota.cc index 67d05aa202..47b7856e95 100644 --- a/src/core/lib/iomgr/resource_quota.cc +++ b/src/core/lib/iomgr/resource_quota.cc @@ -141,8 +141,8 @@ struct grpc_resource_quota { /* Mutex to protect max_threads and num_threads */ /* Note: We could have used gpr_atm for max_threads and num_threads and avoid * having this mutex; but in that case, each invocation of the function - * grpc_resource_user_alloc_threads() will have to do atleast two atomic loads - * (for max_threads and num_threads) followed by a CAS (on num_threads). + * grpc_resource_user_alloc_threads() would have had to do at least two atomic + * loads (for max_threads and num_threads) followed by a CAS (on num_threads). * Moreover, we expect grpc_resource_user_alloc_threads() to be often called * concurrently thereby increasing the chances of failing the CAS operation. * This additional complexity is not worth the tiny perf gain we may (or may |