From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:39539) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1fO2Ml-0006UY-79 for qemu-devel@nongnu.org; Wed, 30 May 2018 10:46:20 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1fO2Mj-00017q-Rp for qemu-devel@nongnu.org; Wed, 30 May 2018 10:46:19 -0400 From: Shameerali Kolothum Thodi Date: Wed, 30 May 2018 14:46:00 +0000 Message-ID: <5FC3163CFD30C246ABAA99954A238FA8386F17D9@FRAEML521-MBX.china.huawei.com> References: <20180516152026.2920-1-shameerali.kolothum.thodi@huawei.com> <20180516152026.2920-4-shameerali.kolothum.thodi@huawei.com> <1f8e3cd4-1e54-5350-3804-c77bbbfbd3e4@redhat.com> In-Reply-To: <1f8e3cd4-1e54-5350-3804-c77bbbfbd3e4@redhat.com> Content-Language: en-US Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Subject: Re: [Qemu-devel] [RFC v2 3/6] hw/arm/virt: Add pc-dimm mem hotplug framework List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Auger Eric , "qemu-devel@nongnu.org" , "qemu-arm@nongnu.org" Cc: "drjones@redhat.com" , "imammedo@redhat.com" , "peter.maydell@linaro.org" , "alex.williamson@redhat.com" , Zhaoshenglong , Jonathan Cameron , Linuxarm Hi Eric, > -----Original Message----- > From: Auger Eric [mailto:eric.auger@redhat.com] > Sent: Monday, May 28, 2018 3:22 PM > To: Shameerali Kolothum Thodi ; > qemu-devel@nongnu.org; qemu-arm@nongnu.org > Cc: drjones@redhat.com; imammedo@redhat.com; peter.maydell@linaro.org; > alex.williamson@redhat.com; Zhaoshenglong ; > Jonathan Cameron ; Linuxarm > > Subject: Re: [RFC v2 3/6] hw/arm/virt: Add pc-dimm mem hotplug framework >=20 > Hi Shameer, >=20 > On 05/16/2018 05:20 PM, Shameer Kolothum wrote: > > This will be used in subsequent patches to model a chunk of > > memory as pc-dimm(cold plug) if the valid iova regions are > > non-contiguous. This is not yet a full hotplug support. > Please can you give more details about this restriction? > > > > Signed-off-by: Shameer Kolothum > > --- > > default-configs/aarch64-softmmu.mak | 1 + > > hw/arm/virt.c | 82 > +++++++++++++++++++++++++++++++++++++ > > include/hw/arm/virt.h | 2 + > > 3 files changed, 85 insertions(+) > > > > diff --git a/default-configs/aarch64-softmmu.mak b/default-configs/aarc= h64- > softmmu.mak > > index 9ddccf8..7a82ed8 100644 > > --- a/default-configs/aarch64-softmmu.mak > > +++ b/default-configs/aarch64-softmmu.mak > > @@ -8,3 +8,4 @@ CONFIG_DDC=3Dy > > CONFIG_DPCD=3Dy > > CONFIG_XLNX_ZYNQMP=3Dy > > CONFIG_XLNX_ZYNQMP_ARM=3Dy > > +CONFIG_MEM_HOTPLUG=3Dy > > diff --git a/hw/arm/virt.c b/hw/arm/virt.c > > index 05fcb62..be3ad14 100644 > > --- a/hw/arm/virt.c > > +++ b/hw/arm/virt.c > > @@ -1552,9 +1552,82 @@ static const CPUArchIdList > *virt_possible_cpu_arch_ids(MachineState *ms) > > return ms->possible_cpus; > > } > > > > +static void virt_dimm_plug(HotplugHandler *hotplug_dev, > > + DeviceState *dev, Error **errp) > > +{ > > + VirtMachineState *vms =3D VIRT_MACHINE(hotplug_dev); > > + PCDIMMDevice *dimm =3D PC_DIMM(dev); > > + PCDIMMDeviceClass *ddc =3D PC_DIMM_GET_CLASS(dimm); > > + MemoryRegion *mr; > > + uint64_t align; > > + Error *local_err =3D NULL; > > + > > + mr =3D ddc->get_memory_region(dimm, &local_err); > > + if (local_err) { > > + goto out; > > + } > > + > > + align =3D memory_region_get_alignment(mr); > I see that on PC machine class they have an "enforce_aligned_dimm" > option. Also looks memory_region_get_alignment(mr) can return 0. >=20 > if (memory_region_get_alignment(mr) && pcmc->enforce_aligned_dimm) { Ok. =20 >=20 > > + pc_dimm_memory_plug(dev, &vms->hotplug_memory, mr, align, > &local_err); > > + if (local_err) { > > + goto out; > > + } > useless block and error_propagate does nothing if local_err is NULL so > we are fine. Ok. I will remove this from this patch. > > + > > +out: > > + error_propagate(errp, local_err); > > +} > > + > > +static void virt_dimm_unplug(HotplugHandler *hotplug_dev, > > + DeviceState *dev, Error **errp) > > +{ > > + VirtMachineState *vms =3D VIRT_MACHINE(hotplug_dev); > > + PCDIMMDevice *dimm =3D PC_DIMM(dev); > > + PCDIMMDeviceClass *ddc =3D PC_DIMM_GET_CLASS(dimm); > > + MemoryRegion *mr; > > + Error *local_err =3D NULL; > > + > > + mr =3D ddc->get_memory_region(dimm, &local_err); > don't you want to avoid executing the following if local_err? True. > > + pc_dimm_memory_unplug(dev, &vms->hotplug_memory, mr); > needs a rebase on top of "pc-dimm: no need to pass the memory region" > in master pc_dimm_memory_unplug we now have > MemoryRegion *mr =3D ddc->get_memory_region(dimm, &error_abort); Right. I will follow that then. > Thanks >=20 > Eric > > + object_unparent(OBJECT(dev)); > > + > > + error_propagate(errp, local_err); > > +} > > + > > +static void virt_machinedevice_plug_cb(HotplugHandler *hotplug_dev, > > + DeviceState *dev, Error **errp) > > +{ > > + if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) { > > + virt_dimm_plug(hotplug_dev, dev, errp); > > + } else { > > + error_setg(errp, "device plug request for not supported device= " > > + " type: %s", object_get_typename(OBJECT(dev))); > > + } > > +} > > + > > +static void virt_machinedevice_unplug_cb(HotplugHandler *hotplug_dev, > > + DeviceState *dev, Error **errp= ) > > +{ > > + if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) { > > + virt_dimm_unplug(hotplug_dev, dev, errp); > > + } else { > > + error_setg(errp, "device unplug for not supported device" > > + " type: %s", object_get_typename(OBJECT(dev))); > > + } > > +} > > + > > +static HotplugHandler *virt_get_hotplug_handler(MachineState *machine, > > + DeviceState *dev) > > +{ > > + if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) { > > + return HOTPLUG_HANDLER(machine); > > + } > > + return NULL; > > +} > > + > > static void virt_machine_class_init(ObjectClass *oc, void *data) > > { > > MachineClass *mc =3D MACHINE_CLASS(oc); > > + HotplugHandlerClass *hc =3D HOTPLUG_HANDLER_CLASS(oc); > > > > mc->init =3D machvirt_init; > > /* Start max_cpus at the maximum QEMU supports. We'll further rest= rict > > @@ -1573,6 +1646,11 @@ static void virt_machine_class_init(ObjectClass > *oc, void *data) > > mc->cpu_index_to_instance_props =3D virt_cpu_index_to_props; > > mc->default_cpu_type =3D ARM_CPU_TYPE_NAME("cortex-a15"); > > mc->get_default_cpu_node_id =3D virt_get_default_cpu_node_id; > > + > > + mc->get_hotplug_handler =3D virt_get_hotplug_handler; > This requires a rebase on top of Igor's and David's series. >=20 Sure. Thanks, Shameer > Thanks >=20 > Eric > > + hc->plug =3D virt_machinedevice_plug_cb; > > + hc->unplug =3D virt_machinedevice_unplug_cb; > > + > > } > > > > static const TypeInfo virt_machine_info =3D { > > @@ -1582,6 +1660,10 @@ static const TypeInfo virt_machine_info =3D { > > .instance_size =3D sizeof(VirtMachineState), > > .class_size =3D sizeof(VirtMachineClass), > > .class_init =3D virt_machine_class_init, > > + .interfaces =3D (InterfaceInfo[]) { > > + { TYPE_HOTPLUG_HANDLER }, > > + { } > > + }, > > }; > > > > static void machvirt_machine_init(void) > > diff --git a/include/hw/arm/virt.h b/include/hw/arm/virt.h > > index fc24f3a..a39f29e 100644 > > --- a/include/hw/arm/virt.h > > +++ b/include/hw/arm/virt.h > > @@ -35,6 +35,7 @@ > > #include "qemu/notify.h" > > #include "hw/boards.h" > > #include "hw/arm/arm.h" > > +#include "hw/mem/pc-dimm.h" > > > > #define NUM_GICV2M_SPIS 64 > > #define NUM_VIRTIO_TRANSPORTS 32 > > @@ -108,6 +109,7 @@ typedef struct { > > uint32_t gic_phandle; > > uint32_t msi_phandle; > > int psci_conduit; > > + MemoryHotplugState hotplug_memory; > > } VirtMachineState; > > > > #define TYPE_VIRT_MACHINE MACHINE_TYPE_NAME("virt") > >