From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:58634) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1gBsPP-0003mY-He for qemu-devel@nongnu.org; Sun, 14 Oct 2018 22:15:04 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1gBsPL-0003vr-0U for qemu-devel@nongnu.org; Sun, 14 Oct 2018 22:15:03 -0400 Received: from cmccmta1.chinamobile.com ([221.176.66.79]:15732) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1gBsPK-0003sk-CD for qemu-devel@nongnu.org; Sun, 14 Oct 2018 22:14:58 -0400 References: <20181012083008.29027-1-maozhongyi@cmss.chinamobile.com> <20181012083008.29027-3-maozhongyi@cmss.chinamobile.com> <2eecebee-de1f-5985-9420-4adcd1767c29@redhat.com> <61bb828a-dc72-5bb2-44b0-2dc9bd737fc4@redhat.com> From: maozy Message-ID: Date: Mon, 15 Oct 2018 10:14:39 +0800 MIME-Version: 1.0 In-Reply-To: <61bb828a-dc72-5bb2-44b0-2dc9bd737fc4@redhat.com> Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: 8bit Subject: Re: [Qemu-devel] [PATCH v2 2/3] audio: use object link instead ofqdevproperty to pass wm8750 reference List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: =?UTF-8?Q?Philippe_Mathieu-Daud=c3=a9?= , qemu-devel@nongnu.org, Peter Maydell , Eduardo Habkost , Thomas Huth Cc: Jan Kiszka , Gerd Hoffmann Hi, Philippe On 10/13/18 2:40 AM, Philippe Mathieu-Daudé wrote: > Hi Mao, > > On 12/10/2018 14:30, Philippe Mathieu-Daudé wrote: >> Cc'ing Eduardo and Thomas. >> >> On 12/10/2018 13:51, maozy wrote: >>> Hi, Philippe >>> >>> On 10/12/18 5:53 PM, Philippe Mathieu-Daudé wrote: >>>> Hi Mao, >>>> >>>> On 12/10/2018 10:30, Mao Zhongyi wrote: >>>>> According to qdev-properties.h, properties of pointer type should >>>>> be avoided, it seems a link type property is a good substitution. >>>>> >>>>> Cc: Jan Kiszka >>>>> Cc: Peter Maydell >>>>> Cc: Gerd Hoffmann >>>>> To: qemu-arm@nongnu.org >>>>> >>>>> Signed-off-by: Mao Zhongyi >>>>> --- >>>>>   hw/arm/musicpal.c          |  3 ++- >>>>>   hw/audio/marvell_88w8618.c | 14 ++++++-------- >>>>>   2 files changed, 8 insertions(+), 9 deletions(-) >>>>> >>>>> diff --git a/hw/arm/musicpal.c b/hw/arm/musicpal.c >>>>> index 3dafb41b0b..ac266f9253 100644 >>>>> --- a/hw/arm/musicpal.c >>>>> +++ b/hw/arm/musicpal.c >>>>> @@ -1695,7 +1695,8 @@ static void musicpal_init(MachineState *machine) >>>>>       wm8750_dev = i2c_create_slave(i2c, TYPE_WM8750, MP_WM_ADDR); >>>>>       dev = qdev_create(NULL, "mv88w8618_audio"); >>>>>       s = SYS_BUS_DEVICE(dev); >>>>> -    qdev_prop_set_ptr(dev, TYPE_WM8750, wm8750_dev); >>>>> +    object_property_set_link(OBJECT(dev), OBJECT(wm8750_dev), >>>>> +                             TYPE_WM8750, NULL); >>>>>       qdev_init_nofail(dev); >>>>>       sysbus_mmio_map(s, 0, MP_AUDIO_BASE); >>>>>       sysbus_connect_irq(s, 0, pic[MP_AUDIO_IRQ]); >>>>> diff --git a/hw/audio/marvell_88w8618.c b/hw/audio/marvell_88w8618.c >>>>> index cf6ce6979b..baab4a3d53 100644 >>>>> --- a/hw/audio/marvell_88w8618.c >>>>> +++ b/hw/audio/marvell_88w8618.c >>>>> @@ -15,6 +15,7 @@ >>>>>   #include "hw/i2c/i2c.h" >>>>>   #include "hw/audio/wm8750.h" >>>>>   #include "audio/audio.h" >>>>> +#include "qapi/error.h" >>>>>     #define MP_AUDIO_SIZE           0x00001000 >>>>>   @@ -252,6 +253,11 @@ static void mv88w8618_audio_init(Object *obj) >>>>>       memory_region_init_io(&s->iomem, obj, &mv88w8618_audio_ops, s, >>>>>                             "audio", MP_AUDIO_SIZE); >>>>>       sysbus_init_mmio(dev, &s->iomem); >>>>> + >>>>> +    object_property_add_link(OBJECT(dev), "mv88w8618", TYPE_WM8750, >>>>> +                             (Object **) &s->wm, >>>>> +                             qdev_prop_allow_set_link_before_realize, >>>>> +                             0, &error_abort); >>>>>   } >>>>>     static void mv88w8618_audio_realize(DeviceState *dev, Error **errp) >>>>> @@ -279,11 +285,6 @@ static const VMStateDescription >>>>> mv88w8618_audio_vmsd = { >>>>>       } >>>>>   }; >>>>>   -static Property mv88w8618_audio_properties[] = { >>>>> -    DEFINE_PROP_PTR(TYPE_WM8750, mv88w8618_audio_state, wm), >>>>> -    {/* end of list */}, >>>>> -}; >>>>> - >>>>>   static void mv88w8618_audio_class_init(ObjectClass *klass, void *data) >>>>>   { >>>>>       DeviceClass *dc = DEVICE_CLASS(klass); >>>>> @@ -291,9 +292,6 @@ static void >>>>> mv88w8618_audio_class_init(ObjectClass *klass, void *data) >>>>>       dc->realize = mv88w8618_audio_realize; >>>>>       dc->reset = mv88w8618_audio_reset; >>>>>       dc->vmsd = &mv88w8618_audio_vmsd; >>>>> -    dc->props = mv88w8618_audio_properties; >>>>> -    /* Reason: pointer property "wm8750" */ >>>>> -    dc->user_creatable = false; >>>> >>>> Having a link property isn't it the same restriction? >>> >>> This task can found in >>> https://wiki.qemu.org/Contribute/BiteSizedTasks#Device_models >>> >>> Convert qdev pointer properties (defined with DEFINE_PROP_PTR) to QOM >>> links. Example: commit 873b4d3. >> >> I agree with the QOM conversion (the rest of this patch), what I'm >> wondering is if declaring this device now user_creatable is correct. I >> don't think we can set link property from command line, but maybe I'm >> wrong because I never investigate/tried to do it. >> >> Maybe devices having link property are automatically marked as >> user_creatable = false, then your change would be correct (except you >> should explicit that in the commit message). >> >> I'll post another mail on the list to ask about that. > > On https://lists.gnu.org/archive/html/qemu-devel/2018-10/msg02531.html > > Peter answered: > > I think whether a device with a link property is > user creatable might depend on what the property is > for and whether the device has a useful fallback for > "link not connected". > > So if you look the realize function, there is no check on the linked object: > > static void mv88w8618_audio_realize(DeviceState *dev, Error **errp) > { > mv88w8618_audio_state *s = MV88W8618_AUDIO(dev); > > wm8750_data_req_set(s->wm, mv88w8618_audio_callback, s); > } > > Then the called function dereference the WM8750State: > > void wm8750_data_req_set(DeviceState *dev, data_req_cb *data_req, > void *opaque) > { > WM8750State *s = WM8750(dev); > > s->data_req = data_req; > s->opaque = opaque; > } > > So this patch is incorrect. You should either keep the > 'dc->user_creatable = false' line, or add a check in the realize() > function, as here: > > static void bcm2836_realize(DeviceState *dev, Error **errp) > { > Object *obj; > Error *err = NULL; > > obj = object_property_get_link(OBJECT(dev), "ram", &err); > if (obj == NULL) { > error_setg(errp, "%s: required ram link not found: %s", > __func__, error_get_pretty(err)); > return; > } > ... > Sorry for the delay in response, I was away for the holiday weekend. I saw the mail you discussed and reviewed the code, I will fix it in the next version. Thank you! I really appreciate it. Thanks, Mao > Regards, > > Phil. > >> >> Regards, >> >> Phil. >> >>> >>> Thanks, >>> Mao >>> >>>> >>>>>   } >>>>>     static const TypeInfo mv88w8618_audio_info = { >>>>> >>>> >>> >>> >