commit 929325dd01985ab055828f6526b42a58d9d486cd
author Primiano Tucci <primiano@google.com> Tue Mar 15 20:26:21 2022 +0000
committer Primiano Tucci <primiano@google.com> Thu Mar 17 11:39:12 2022 +0000
tree f61bb95cfed61bf7f32c3aaf9ea12f9c10f5a3d7
parent 3a26ff6de3b39577e92f51d254cfd1bc377c995c

atrace: don't write the clock sync marker if --only_userspace

This CL has no other intended side effect than reducing
spam in logs.
--only_userspace is effectively used only by Perfetto.
Today the unconditional invocation of writeClockSyncMarker()
writes consistently, because perfetto invokes atrace before
writing 1 to tracing_on, and that makes the write to
trace_marker fail.
The failure has always been there and has always been fine.
This CL just silences it. I am planning to bubble up non-fatal
errors coming from atrace to the UI, this would be a confusing
false positive.

Bug: 171085599
Test: manual: perfetto -t 1s bionic am foobar -o trace;
              check logcat (no clock sync errors)
Change-Id: Ic9b9cef42465bae0214e73c0dbce9ed2d288fdc6

cmds/atrace/atrace.cpp

diff --git a/cmds/atrace/atrace.cpp b/cmds/atrace/atrace.cpp
index c2a2bc3..9bd733d 100644
--- a/cmds/atrace/atrace.cpp
+++ b/cmds/atrace/atrace.cpp
@@ -1340,10 +1340,10 @@
         // contain entries from only one CPU can cause "begin" entries without a
         // matching "end" entry to show up if a task gets migrated from one CPU to
         // another.
-        if (!onlyUserspace)
+        if (!onlyUserspace) {
             ok = clearTrace();
-
-        writeClockSyncMarker();
+            writeClockSyncMarker();
+        }
         if (ok && !async && !traceStream) {
             // Sleep to allow the trace to be captured.
             struct timespec timeLeft;