On 07/24/2018 09:25 AM, Stefan Hajnoczi wrote:
The simple trace backend spawns a write-out thread which is used to
asynchronously flush the in-memory ring buffer to disk.
fork(2) does not clone all threads, only the thread that invoked
fork(2). As a result there is no write-out thread in the child process!
This causes a hang during shutdown when atexit(3) handler installed by
the simple trace backend waits for the non-existent write-out thread.
This patch uses pthread_atfork(3) to terminate the write-out thread
before fork and restart it in both the parent and child after fork.
This solves a hang in qemu-iotests 147 due to qemu-nbd --fork usage.
Reported-by: Cornelia Huck <coh...@redhat.com>
Tested-by: Cornelia Huck <coh...@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefa...@redhat.com>
Message-id: 20180717101944.11691-1-stefa...@redhat.com
Suggested-by: Paolo Bonzini <pbonz...@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefa...@redhat.com>
---
+static void restart_writeout_thread(void)
+{
+ trace_writeout_running = true;
+ trace_writeout_thread = trace_thread_create(writeout_thread);
+ if (!trace_writeout_thread) {
+ warn_report("unable to initialize simple trace backend");
+ }
+
+ /* This relies on undefined behavior in the fork() child (it's fine in the
+ * fork() parent). g_mutex_unlock() on a mutex acquired by another thread
+ * is undefined (see glib documentation).
+ */
+ g_mutex_unlock(&trace_lock);
Dan's point about stopping tracing prior to fork, then restarting it
from scratch in both the parent and in specific children, would also get
rid of this risky non-portable behavior of trying to manipulate a mutex
acquired by the parent process' thread.
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3266
Virtualization: qemu.org | libvirt.org