Copilot commented on code in PR #3368:
URL: https://github.com/apache/brpc/pull/3368#discussion_r3524170962
##########
test/brpc_socket_unittest.cpp:
##########
@@ -1123,6 +1126,48 @@ void CheckKeepalive(int fd,
ASSERT_EQ(expected_keepalive_count, keepalive_count);
}
+struct SocketBufferValues {
+ int recv_buffer;
+ int send_buffer;
+
+ SocketBufferValues() : recv_buffer(0), send_buffer(0) {}
+};
+
+void GetSocketBufferValues(int fd, SocketBufferValues* values) {
+ socklen_t len = sizeof(values->recv_buffer);
+ ASSERT_EQ(0, getsockopt(fd, SOL_SOCKET, SO_RCVBUF,
+ &values->recv_buffer, &len));
+ len = sizeof(values->send_buffer);
+ ASSERT_EQ(0, getsockopt(fd, SOL_SOCKET, SO_SNDBUF,
+ &values->send_buffer, &len));
+}
+
+void GetExpectedSocketBufferValues(int buffer_size,
+ SocketBufferValues* expected) {
+ SocketBufferValues default_values;
+ butil::fd_guard default_fd(socket(AF_INET, SOCK_STREAM, 0));
+ ASSERT_GT(default_fd, 0);
+ GetSocketBufferValues(default_fd, &default_values);
+
+ butil::fd_guard reference_fd(socket(AF_INET, SOCK_STREAM, 0));
+ ASSERT_GT(reference_fd, 0);
+ ASSERT_EQ(0, setsockopt(reference_fd, SOL_SOCKET, SO_RCVBUF, &buffer_size,
+ sizeof(buffer_size)));
+ ASSERT_EQ(0, setsockopt(reference_fd, SOL_SOCKET, SO_SNDBUF, &buffer_size,
+ sizeof(buffer_size)));
+ GetSocketBufferValues(reference_fd, expected);
+
+ ASSERT_NE(default_values.recv_buffer, expected->recv_buffer);
+ ASSERT_NE(default_values.send_buffer, expected->send_buffer);
Review Comment:
`GetExpectedSocketBufferValues` asserts that the system default
SO_RCVBUF/SO_SNDBUF values must differ from the values after `setsockopt`. On
hosts where the default already matches (or where kernel limits clamp both to
the same value), these `ASSERT_NE` checks can fail even though the code under
test is correct, making the test environment-dependent/flaky. Consider removing
the distinctness assertions (or turning them into a non-fatal log) and relying
on the later comparisons against `expected` instead.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
