On Fri, 19 Feb 2021 at 08:46, Akihiko Odaki <akihiko.od...@gmail.com> wrote:
>
> ui/cocoa used to call exit immediately after calling
> qemu_system_shutdown_request, which prevents QEMU from actually
> perfoming system shutdown. Just sleep forever, and wait QEMU to call
> exit and kill the Cocoa thread.
>
> Signed-off-by: Akihiko Odaki <akihiko.od...@gmail.com>
> ---
> ui/cocoa.m | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/ui/cocoa.m b/ui/cocoa.m
> index 13fba8103e1..65bb74134ca 100644
> --- a/ui/cocoa.m
> +++ b/ui/cocoa.m
> @@ -1121,7 +1121,7 @@ - (void)applicationWillTerminate:(NSNotification
> *)aNotification
> COCOA_DEBUG("QemuCocoaAppController: applicationWillTerminate\n");
>
> qemu_system_shutdown_request(SHUTDOWN_CAUSE_HOST_UI);
> - exit(0);
> + [NSThread sleepForTimeInterval:INFINITY];
I think that as it stands this change is probably a bit risky,
because the QEMU action for "what do I do on a shutdown request"
is not necessarily "exit" -- the QMP 'set-action' command allows
a user to configure QEMU to just pause on a shutdown, which will
result in this loop going forever (or until OSX gets bored and
forcibly terminates the process).
It would also be useful to have a comment, something like:
/*
* Sleep here, because returning will cause OSX to kill us
* immediately; the QEMU main loop will handle the shutdown
* request and terminate the process.
*/
?
> }
Gerd: I notice that the only ui frontends that try to do this
SHUTDOWN_CAUSE_HOST_UI thing are the Cocoa UI and the SDL2 UI.
The GTK UI does
qmp_quit()
instead; the SDL2 UI does
shutdown_action = SHUTDOWN_ACTION_POWEROFF;
qemu_system_shutdown_request(SHUTDOWN_CAUSE_HOST_UI);
(presumably to avoid the "maybe the user told us to 'pause'"
issue I mention above). None of the other UI frontends have
any shutdown related handling. Shouldn't we be consistent
about how we do this ?
thanks
-- PMM
