shutdown: Preserve shutdown cause through replay

With the recent addition of ShutdownCause, we want to be able to pass
a cause through any shutdown request, and then faithfully replay that
cause when later replaying the same sequence.  The easiest way is to
expand the reply event mechanism to track a series of values for
EVENT_SHUTDOWN, one corresponding to each value of ShutdownCause.

We are free to change the replay stream as needed, since there are
already no guarantees about being able to use a replay stream by
any other version of qemu than the one that generated it.

The cause is not actually fed back until the next patch changes the
signature for requesting a shutdown; a TODO marks that upcoming change.

Yes, this uses the gcc/clang extension of a ranged case label,
but this is not the first time we've used non-C99 constructs.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Pavel Dovgalyuk <pavel.dovgaluk@ispras.ru>
Message-Id: <20170515214114.15442-4-eblake@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
This commit is contained in:
Eric Blake 2017-05-15 16:41:12 -05:00 committed by Markus Armbruster
parent aedbe19297
commit 802f045a5f
4 changed files with 9 additions and 6 deletions

2
vl.c
View file

@ -1821,8 +1821,8 @@ void qemu_system_killed(int signal, pid_t pid)
void qemu_system_shutdown_request(void)
{
trace_qemu_system_shutdown_request();
replay_shutdown_request();
/* TODO - add a parameter to allow callers to specify reason */
replay_shutdown_request(SHUTDOWN_CAUSE_HOST_ERROR);
shutdown_requested = SHUTDOWN_CAUSE_HOST_ERROR;
qemu_notify_event();
}