MainLoop: improve error handling am: fd59221459 am: 8fbb6736ff am: 02b8923766android-vts-8.0_r9android-vts-8.0_r8android-vts-8.0_r7android-vts-8.0_r6android-vts-8.0_r2android-vts-8.0_r13android-vts-8.0_r12android-vts-8.0_r11android-vts-8.0_r10android-vts-8.0_r1android-security-8.0.0_r54android-security-8.0.0_r53android-security-8.0.0_r52android-cts-8.0_r9android-cts-8.0_r8android-cts-8.0_r7android-cts-8.0_r6android-cts-8.0_r5android-cts-8.0_r4android-cts-8.0_r3android-cts-8.0_r26android-cts-8.0_r25android-cts-8.0_r24android-cts-8.0_r23android-cts-8.0_r22android-cts-8.0_r21android-cts-8.0_r20android-cts-8.0_r2android-cts-8.0_r19android-cts-8.0_r18android-cts-8.0_r17android-cts-8.0_r16android-cts-8.0_r15android-cts-8.0_r14android-cts-8.0_r13android-cts-8.0_r12android-cts-8.0_r11android-cts-8.0_r10android-cts-8.0_r1android-8.0.0_r9android-8.0.0_r7android-8.0.0_r51android-8.0.0_r50android-8.0.0_r49android-8.0.0_r48android-8.0.0_r47android-8.0.0_r46android-8.0.0_r45android-8.0.0_r44android-8.0.0_r43android-8.0.0_r42android-8.0.0_r41android-8.0.0_r40android-8.0.0_r4android-8.0.0_r39android-8.0.0_r38android-8.0.0_r37android-8.0.0_r36android-8.0.0_r35android-8.0.0_r32android-8.0.0_r31android-8.0.0_r30android-8.0.0_r3android-8.0.0_r29android-8.0.0_r28android-8.0.0_r2android-8.0.0_r17android-8.0.0_r16android-8.0.0_r15android-8.0.0_r13android-8.0.0_r12android-8.0.0_r11android-8.0.0_r10android-8.0.0_r1security-oc-releaseoreo-vts-releaseoreo-security-releaseoreo-releaseoreo-r6-releaseoreo-r5-releaseoreo-r4-releaseoreo-r3-releaseoreo-r2-releaseoreo-devoreo-cts-release

am: 0568dbda87

Change-Id: Ie76fc927106cbbc692249ca67ddd19a6511016f7

3 files changed, 28 insertions, 3 deletions

diff --git a/main_loop.cpp b/main_loop.cpp
index 62bed31..8ad030a 100644
--- a/main_loop.cpp
+++ b/main_loop.cpp
@@ -30,6 +30,8 @@ namespace wifilogd {
 
 namespace {
 constexpr auto kMainBufferSizeBytes = 128 * 1024;
+// TODO(b/32840641): Tune the sleep time.
+constexpr auto kTransientErrorSleepTimeNsec = 100 * 1000;  // 100 usec
 }
 
 MainLoop::MainLoop(const std::string& socket_name)
@@ -53,8 +55,7 @@ void MainLoop::RunOnce() {
   std::tie(datagram_len, err) =
       os_->ReceiveDatagram(sock_fd_, input_buf.data(), input_buf.size());
   if (err) {
-    // TODO(b/32098735): Increment stats counter.
-    // TODO(b/32481888): Improve error handling.
+    ProcessError(err);
     return;
   }
 
@@ -67,5 +68,20 @@ void MainLoop::RunOnce() {
                                      Os::kInvalidFd);
 }
 
+// Private methods below.
+
+void MainLoop::ProcessError(Os::Errno err) {
+  if (err == EINTR || err == ENOMEM) {
+    // TODO(b/32098735): Increment stats counter.
+    os_->Nanosleep(kTransientErrorSleepTimeNsec);
+    return;
+  }
+
+  // Any other error is unexpected, and assumed to be non-recoverable.
+  // (If, e.g., our socket is in a bad state, then we won't be able to receive
+  // any new log messages.)
+  LOG(FATAL) << "Unexpected error: " << std::strerror(err);
+}
+
 }  // namespace wifilogd
 }  // namespace android
diff --git a/main_loop.h b/main_loop.h
index c5a4b48..53c42f8 100644
--- a/main_loop.h
+++ b/main_loop.h
@@ -39,6 +39,8 @@ class MainLoop {
   void RunOnce();
 
  private:
+  void ProcessError(Os::Errno err);
+
   std::unique_ptr<Os> os_;
   std::unique_ptr<CommandProcessor> command_processor_;
   // We use an int, rather than a unique_fd, because the file
diff --git a/tests/main_loop_unittest.cpp b/tests/main_loop_unittest.cpp
index db04027..b6c1141 100644
--- a/tests/main_loop_unittest.cpp
+++ b/tests/main_loop_unittest.cpp
@@ -109,9 +109,10 @@ TEST_F(MainLoopTest, RunOnceLimitsMaxSizeReportedToCommandProcessor) {
   main_loop_->RunOnce();
 }
 
-TEST_F(MainLoopTest, RunOnceDoesNotPassDataToCommandProcessorOnError) {
+TEST_F(MainLoopTest, RunOnceSleepsAndDoesNotPassDataToCommandProcessorOnError) {
   EXPECT_CALL(*os_, ReceiveDatagram(_, _, protocol::kMaxMessageSize))
       .WillOnce(Return(std::tuple<size_t, Os::Errno>{0, EINTR}));
+  EXPECT_CALL(*os_, Nanosleep(_));
   EXPECT_CALL(*command_processor_, ProcessCommand(_, _, _)).Times(0);
   main_loop_->RunOnce();
 }
@@ -131,5 +132,11 @@ TEST_F(MainLoopDeathTest, CtorFailureToFetchControlSocketCausesDeath) {
       "Failed to get control socket");
 }
 
+TEST_F(MainLoopDeathTest, RunOnceTerminatesOnUnexpectedError) {
+  ON_CALL(*os_, ReceiveDatagram(_, _, protocol::kMaxMessageSize))
+      .WillByDefault(Return(std::tuple<size_t, Os::Errno>{0, EFAULT}));
+  EXPECT_DEATH(main_loop_->RunOnce(), "Unexpected error");
+}
+
 }  // namespace wifilogd
 }  // namespace android