debuggerd: wait for dump completion on crashes.
When a process crashes, both ActivityManager and init will try to kill
its process group when they notice. The recent change to minimize the
amount of time a process is paused results in crash dumps being killed
before they finish as a result of this. Since anything that needs to be
low-latency is probably not going to be too happy if it crashes, just
wait for completion whenever we're processing a real crash.
Bug: http://b/70343110
Test: debuggerd_test
Change-Id: I894bb06efd264b1ba005df06f7326a72f4b767bb
diff --git a/debuggerd/handler/debuggerd_handler.cpp b/debuggerd/handler/debuggerd_handler.cpp
index 02bc4b8..96f3c7c 100644
--- a/debuggerd/handler/debuggerd_handler.cpp
+++ b/debuggerd/handler/debuggerd_handler.cpp
@@ -395,9 +395,6 @@
// crash_dump is ptracing us, fork off a copy of our address space for it to use.
create_vm_process();
- input_read.reset();
- input_write.reset();
-
// Don't leave a zombie child.
int status;
if (TEMP_FAILURE_RETRY(waitpid(crash_dump_pid, &status, 0)) == -1) {
@@ -406,6 +403,14 @@
} else if (WIFSTOPPED(status) || WIFSIGNALED(status)) {
async_safe_format_log(ANDROID_LOG_FATAL, "libc", "crash_dump helper crashed or stopped");
}
+
+ if (thread_info->siginfo->si_signo != DEBUGGER_SIGNAL) {
+ // For crashes, we don't need to minimize pause latency.
+ // Wait for the dump to complete before having the process exit, to avoid being murdered by
+ // ActivityManager or init.
+ TEMP_FAILURE_RETRY(read(input_read, &buf, sizeof(buf)));
+ }
+
return 0;
}