From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:59774) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dzSGR-00060r-VS for qemu-devel@nongnu.org; Tue, 03 Oct 2017 14:49:57 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1dzSGO-00061j-1Y for qemu-devel@nongnu.org; Tue, 03 Oct 2017 14:49:56 -0400 Received: from mx1.redhat.com ([209.132.183.28]:47592) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1dzSGN-0005w0-Ng for qemu-devel@nongnu.org; Tue, 03 Oct 2017 14:49:51 -0400 Received: from smtp.corp.redhat.com (int-mx03.intmail.prod.int.phx2.redhat.com [10.5.11.13]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 5CA5F5F2973 for ; Tue, 3 Oct 2017 18:49:50 +0000 (UTC) Date: Tue, 3 Oct 2017 15:49:46 -0300 From: Eduardo Habkost Message-ID: <20171003184946.GR17385@localhost.localdomain> References: <1507049162-27026-1-git-send-email-thuth@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1507049162-27026-1-git-send-email-thuth@redhat.com> Subject: Re: [Qemu-devel] [PATCH] qdev: Check for the availability of a hotplug controller before adding a device List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Thomas Huth Cc: qemu-devel@nongnu.org, Markus Armbruster , "Dr. David Alan Gilbert" , Igor Mammedov , Paolo Bonzini On Tue, Oct 03, 2017 at 06:46:02PM +0200, Thomas Huth wrote: > The qdev_unplug() function contains a g_assert(hotplug_ctrl) statement, > so QEMU crashes when the user tries to device_add + device_del a device > that does not have a corresponding hotplug controller. This could be > provoked for a couple of devices in the past (see commit 4c93950659487c7ad > or 84ebd3e8c7d4fe955 for example). So devices clearly need a hotplug > controller when they are suitable for device_add. > The code in qdev_device_add() already checks whether the bus has a proper > hotplug controller, but for devices that do not have a corresponding bus, > there is no appropriate check available. In that case we should check > whether the machine itself provides a suitable hotplug controller and > refuse to plug the device if none is available. > > Signed-off-by: Thomas Huth > --- > This is the follow-up patch from my earlier try "hw/core/qdev: Do not > allow hot-plugging without hotplug controller" ... AFAICS the function > qdev_device_add() is now the right spot to do the check. > > hw/core/qdev.c | 28 ++++++++++++++++++++-------- > include/hw/qdev-core.h | 1 + > qdev-monitor.c | 9 +++++++++ > 3 files changed, 30 insertions(+), 8 deletions(-) > > diff --git a/hw/core/qdev.c b/hw/core/qdev.c > index 606ab53..a953ec9 100644 > --- a/hw/core/qdev.c > +++ b/hw/core/qdev.c > @@ -253,19 +253,31 @@ void qdev_set_legacy_instance_id(DeviceState *dev, int alias_id, > dev->alias_required_for_version = required_for_version; > } > > +HotplugHandler *qdev_get_machine_hotplug_handler(DeviceState *dev) > +{ > + MachineState *machine; > + MachineClass *mc; > + Object *m_obj = qdev_get_machine(); > + > + if (object_dynamic_cast(m_obj, TYPE_MACHINE)) { > + machine = MACHINE(m_obj); > + mc = MACHINE_GET_CLASS(machine); > + if (mc->get_hotplug_handler) { > + return mc->get_hotplug_handler(machine, dev); > + } > + } > + > + return NULL; > +} > + > HotplugHandler *qdev_get_hotplug_handler(DeviceState *dev) > { > - HotplugHandler *hotplug_ctrl = NULL; > + HotplugHandler *hotplug_ctrl; > > if (dev->parent_bus && dev->parent_bus->hotplug_handler) { > hotplug_ctrl = dev->parent_bus->hotplug_handler; > - } else if (object_dynamic_cast(qdev_get_machine(), TYPE_MACHINE)) { > - MachineState *machine = MACHINE(qdev_get_machine()); > - MachineClass *mc = MACHINE_GET_CLASS(machine); > - > - if (mc->get_hotplug_handler) { > - hotplug_ctrl = mc->get_hotplug_handler(machine, dev); > - } > + } else { > + hotplug_ctrl = qdev_get_machine_hotplug_handler(dev); > } > return hotplug_ctrl; > } > diff --git a/include/hw/qdev-core.h b/include/hw/qdev-core.h > index 0891461..5aa536d 100644 > --- a/include/hw/qdev-core.h > +++ b/include/hw/qdev-core.h > @@ -285,6 +285,7 @@ DeviceState *qdev_try_create(BusState *bus, const char *name); > void qdev_init_nofail(DeviceState *dev); > void qdev_set_legacy_instance_id(DeviceState *dev, int alias_id, > int required_for_version); > +HotplugHandler *qdev_get_machine_hotplug_handler(DeviceState *dev); > HotplugHandler *qdev_get_hotplug_handler(DeviceState *dev); > void qdev_unplug(DeviceState *dev, Error **errp); > void qdev_simple_device_unplug_cb(HotplugHandler *hotplug_dev, > diff --git a/qdev-monitor.c b/qdev-monitor.c > index 8fd6df9..2891dde 100644 > --- a/qdev-monitor.c > +++ b/qdev-monitor.c > @@ -626,6 +626,15 @@ DeviceState *qdev_device_add(QemuOpts *opts, Error **errp) > return NULL; > } > > + /* In case we don't have a bus, there must be a machine hotplug handler */ > + if (qdev_hotplug && !bus && !qdev_get_machine_hotplug_handler(dev)) { > + error_setg(errp, "Device '%s' can not be hotplugged on this machine", > + driver); > + object_unparent(OBJECT(dev)); Isn't it better to check qdev_get_machine_hotplug_handler() earlier (before the qdev_set_parent_bus() and qdev_set_id() lines), so object_unparent() isn't necessary? (We probably don't need to call object_unparent() here, already, because bus is NULL. But moving the check before the "if (bus) qdev_set_parent_bus()" statement would make this more obvious). I would prefer to eventually make MachineClass::get_hotplug_handler() get a typename or DeviceClass* argument instead of DeviceState*, so we don't even create the device object. But I don't think it's a requirement for this bug fix. > + object_unref(OBJECT(dev)); > + return NULL; > + } > + > dev->opts = opts; > object_property_set_bool(OBJECT(dev), true, "realized", &err); > if (err != NULL) { > -- > 1.8.3.1 > -- Eduardo