From: Xiao Guangrong <guangrong.xiao@linux.intel.com> To: "Michael S. Tsirkin" <mst@redhat.com> Cc: pbonzini@redhat.com, imammedo@redhat.com, gleb@kernel.org, mtosatti@redhat.com, stefanha@redhat.com, rth@twiddle.net, ehabkost@redhat.com, dan.j.williams@intel.com, kvm@vger.kernel.org, qemu-devel@nongnu.org Subject: Re: [PATCH v6 19/33] dimm: keep the state of the whole backend memory Date: Mon, 9 Nov 2015 19:13:42 +0800 [thread overview] Message-ID: <56407FE6.1010906@linux.intel.com> (raw) In-Reply-To: <20151109110439.GB3324@redhat.com> On 11/09/2015 07:04 PM, Michael S. Tsirkin wrote: > On Fri, Oct 30, 2015 at 01:56:13PM +0800, Xiao Guangrong wrote: >> QEMU keeps the state of memory of dimm device during live migration, >> however, it is not enough for nvdimm device as its memory does not >> contain its label data, so that we should protect the whole backend >> memory instead >> >> Signed-off-by: Xiao Guangrong <guangrong.xiao@linux.intel.com> > > It looks like there's now a difference between > host_memory_backend_get_memory and get_memory_region, > whereas previously they were exactly interchangeable. Yes, it is. host_memory_backend_get_memory() is used to get the whole backend memory region. get_memory_region() is used to get the memory region which will be mapped to guest's address space. They are on the same page for pc-dimm, but it's different for nvdimm since part of backend memory used as label data which is not directly mapped to guest. > > This needs some thought, in particular the theoretically > generic dimm.c has to do tricks to accomodate nvdimm. > > The missing piece for NVDIMM is the 128k label space at the end, > right? Can't nvdimm specific code just register that as a > separate RAM chunk in order to migrate it? Yes, it is. I thought this before, then we need to have addition callbackes: - plug(), which is called when the dimm device is plugged, then nvdimm has the chance to do it own migration things. - unplug(), let nvdimm undo the thing of migration. It needs more codes but the logic seems more clear. If you like this way, i will do. > >> --- >> hw/mem/dimm.c | 14 ++++++++++++-- >> 1 file changed, 12 insertions(+), 2 deletions(-) >> >> diff --git a/hw/mem/dimm.c b/hw/mem/dimm.c >> index 498d380..44447d1 100644 >> --- a/hw/mem/dimm.c >> +++ b/hw/mem/dimm.c >> @@ -134,9 +134,16 @@ void dimm_memory_plug(DeviceState *dev, MemoryHotplugState *hpms, >> } >> >> memory_region_add_subregion(&hpms->mr, addr - hpms->base, mr); >> - vmstate_register_ram(mr, dev); >> numa_set_mem_node_id(addr, memory_region_size(mr), dimm->node); >> >> + /* >> + * save the state only for @mr is not enough as it does not contain >> + * the label data of NVDIMM device, so that we keep the state of >> + * whole hostmem instead. >> + */ >> + vmstate_register_ram(host_memory_backend_get_memory(dimm->hostmem, errp), >> + dev); >> + >> out: >> error_propagate(errp, local_err); >> } >> @@ -145,10 +152,13 @@ void dimm_memory_unplug(DeviceState *dev, MemoryHotplugState *hpms, >> MemoryRegion *mr) >> { >> DIMMDevice *dimm = DIMM(dev); >> + MemoryRegion *backend_mr; >> + >> + backend_mr = host_memory_backend_get_memory(dimm->hostmem, &error_abort); >> >> numa_unset_mem_node_id(dimm->addr, memory_region_size(mr), dimm->node); >> memory_region_del_subregion(&hpms->mr, mr); >> - vmstate_unregister_ram(mr, dev); >> + vmstate_unregister_ram(backend_mr, dev); >> } >> >> int qmp_dimm_device_list(Object *obj, void *opaque) >> -- >> 1.8.3.1 >
WARNING: multiple messages have this Message-ID (diff)
From: Xiao Guangrong <guangrong.xiao@linux.intel.com> To: "Michael S. Tsirkin" <mst@redhat.com> Cc: ehabkost@redhat.com, kvm@vger.kernel.org, gleb@kernel.org, mtosatti@redhat.com, qemu-devel@nongnu.org, stefanha@redhat.com, imammedo@redhat.com, pbonzini@redhat.com, dan.j.williams@intel.com, rth@twiddle.net Subject: Re: [Qemu-devel] [PATCH v6 19/33] dimm: keep the state of the whole backend memory Date: Mon, 9 Nov 2015 19:13:42 +0800 [thread overview] Message-ID: <56407FE6.1010906@linux.intel.com> (raw) In-Reply-To: <20151109110439.GB3324@redhat.com> On 11/09/2015 07:04 PM, Michael S. Tsirkin wrote: > On Fri, Oct 30, 2015 at 01:56:13PM +0800, Xiao Guangrong wrote: >> QEMU keeps the state of memory of dimm device during live migration, >> however, it is not enough for nvdimm device as its memory does not >> contain its label data, so that we should protect the whole backend >> memory instead >> >> Signed-off-by: Xiao Guangrong <guangrong.xiao@linux.intel.com> > > It looks like there's now a difference between > host_memory_backend_get_memory and get_memory_region, > whereas previously they were exactly interchangeable. Yes, it is. host_memory_backend_get_memory() is used to get the whole backend memory region. get_memory_region() is used to get the memory region which will be mapped to guest's address space. They are on the same page for pc-dimm, but it's different for nvdimm since part of backend memory used as label data which is not directly mapped to guest. > > This needs some thought, in particular the theoretically > generic dimm.c has to do tricks to accomodate nvdimm. > > The missing piece for NVDIMM is the 128k label space at the end, > right? Can't nvdimm specific code just register that as a > separate RAM chunk in order to migrate it? Yes, it is. I thought this before, then we need to have addition callbackes: - plug(), which is called when the dimm device is plugged, then nvdimm has the chance to do it own migration things. - unplug(), let nvdimm undo the thing of migration. It needs more codes but the logic seems more clear. If you like this way, i will do. > >> --- >> hw/mem/dimm.c | 14 ++++++++++++-- >> 1 file changed, 12 insertions(+), 2 deletions(-) >> >> diff --git a/hw/mem/dimm.c b/hw/mem/dimm.c >> index 498d380..44447d1 100644 >> --- a/hw/mem/dimm.c >> +++ b/hw/mem/dimm.c >> @@ -134,9 +134,16 @@ void dimm_memory_plug(DeviceState *dev, MemoryHotplugState *hpms, >> } >> >> memory_region_add_subregion(&hpms->mr, addr - hpms->base, mr); >> - vmstate_register_ram(mr, dev); >> numa_set_mem_node_id(addr, memory_region_size(mr), dimm->node); >> >> + /* >> + * save the state only for @mr is not enough as it does not contain >> + * the label data of NVDIMM device, so that we keep the state of >> + * whole hostmem instead. >> + */ >> + vmstate_register_ram(host_memory_backend_get_memory(dimm->hostmem, errp), >> + dev); >> + >> out: >> error_propagate(errp, local_err); >> } >> @@ -145,10 +152,13 @@ void dimm_memory_unplug(DeviceState *dev, MemoryHotplugState *hpms, >> MemoryRegion *mr) >> { >> DIMMDevice *dimm = DIMM(dev); >> + MemoryRegion *backend_mr; >> + >> + backend_mr = host_memory_backend_get_memory(dimm->hostmem, &error_abort); >> >> numa_unset_mem_node_id(dimm->addr, memory_region_size(mr), dimm->node); >> memory_region_del_subregion(&hpms->mr, mr); >> - vmstate_unregister_ram(mr, dev); >> + vmstate_unregister_ram(backend_mr, dev); >> } >> >> int qmp_dimm_device_list(Object *obj, void *opaque) >> -- >> 1.8.3.1 >
next prev parent reply other threads:[~2015-11-09 11:20 UTC|newest] Thread overview: 182+ messages / expand[flat|nested] mbox.gz Atom feed top 2015-10-30 5:55 [PATCH v6 00/33] implement vNVDIMM Xiao Guangrong 2015-10-30 5:55 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:55 ` [PATCH v6 01/33] acpi: add aml_derefof Xiao Guangrong 2015-10-30 5:55 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:55 ` [PATCH v6 02/33] acpi: add aml_sizeof Xiao Guangrong 2015-10-30 5:55 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:55 ` [PATCH v6 03/33] acpi: add aml_create_field Xiao Guangrong 2015-10-30 5:55 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:55 ` [PATCH v6 04/33] acpi: add aml_concatenate Xiao Guangrong 2015-10-30 5:55 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:55 ` [PATCH v6 05/33] acpi: add aml_object_type Xiao Guangrong 2015-10-30 5:55 ` [Qemu-devel] " Xiao Guangrong 2015-11-09 11:35 ` Michael S. Tsirkin 2015-11-09 11:35 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 12:40 ` Igor Mammedov 2015-11-09 12:40 ` [Qemu-devel] " Igor Mammedov 2015-10-30 5:56 ` [PATCH v6 06/33] acpi: add aml_method_serialized Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-11-09 11:14 ` Michael S. Tsirkin 2015-11-09 11:14 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 11:18 ` Xiao Guangrong 2015-11-09 11:18 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 07/33] util: introduce qemu_file_get_page_size() Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 13:26 ` Vladimir Sementsov-Ogievskiy 2015-10-30 13:26 ` Vladimir Sementsov-Ogievskiy 2015-10-31 7:26 ` Xiao Guangrong 2015-10-31 7:26 ` [Qemu-devel] " Xiao Guangrong 2015-10-31 9:37 ` Vladimir Sementsov-Ogievskiy 2015-10-31 9:37 ` Vladimir Sementsov-Ogievskiy 2015-10-31 14:06 ` Xiao Guangrong 2015-10-31 14:06 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 15:54 ` Eduardo Habkost 2015-10-30 15:54 ` [Qemu-devel] " Eduardo Habkost 2015-10-31 8:09 ` Xiao Guangrong 2015-10-31 8:09 ` [Qemu-devel] " Xiao Guangrong 2015-10-31 14:11 ` Eduardo Habkost 2015-10-31 14:11 ` [Qemu-devel] " Eduardo Habkost 2015-10-31 17:03 ` Xiao Guangrong 2015-10-31 17:03 ` [Qemu-devel] " Xiao Guangrong 2015-11-09 20:00 ` Michael S. Tsirkin 2015-11-09 20:00 ` [Qemu-devel] " Michael S. Tsirkin 2015-10-31 8:25 ` Xiao Guangrong 2015-10-31 8:25 ` [Qemu-devel] " Xiao Guangrong 2015-11-09 10:33 ` Michael S. Tsirkin 2015-11-09 10:33 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 10:55 ` Xiao Guangrong 2015-11-09 10:55 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 08/33] exec: allow memory to be allocated from any kind of path Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 14:04 ` Vladimir Sementsov-Ogievskiy 2015-10-30 14:04 ` Vladimir Sementsov-Ogievskiy 2015-10-31 7:44 ` Xiao Guangrong 2015-10-31 13:55 ` Eduardo Habkost 2015-10-31 13:55 ` Eduardo Habkost 2015-10-31 15:56 ` Xiao Guangrong 2015-10-31 15:56 ` Xiao Guangrong 2015-11-09 10:39 ` Michael S. Tsirkin 2015-11-09 10:39 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 11:01 ` Xiao Guangrong 2015-11-09 11:01 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 09/33] exec: allow file_ram_alloc to work on file Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 14:25 ` Vladimir Sementsov-Ogievskiy 2015-10-30 14:25 ` Vladimir Sementsov-Ogievskiy 2015-10-31 7:53 ` Xiao Guangrong 2015-10-31 7:53 ` [Qemu-devel] " Xiao Guangrong 2015-11-09 10:13 ` Michael S. Tsirkin 2015-11-09 10:13 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 10:48 ` Xiao Guangrong 2015-11-09 10:48 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 10/33] hostmem-file: clean up memory allocation Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 14:54 ` Vladimir Sementsov-Ogievskiy 2015-10-30 14:54 ` Vladimir Sementsov-Ogievskiy 2015-10-30 5:56 ` [PATCH v6 11/33] hostmem-file: use whole file size if possible Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 15:27 ` Vladimir Sementsov-Ogievskiy 2015-10-30 15:27 ` Vladimir Sementsov-Ogievskiy 2015-10-31 7:59 ` Xiao Guangrong 2015-10-31 7:59 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 17:30 ` Eduardo Habkost 2015-10-30 17:30 ` Eduardo Habkost 2015-10-31 8:46 ` Xiao Guangrong 2015-10-31 8:46 ` Xiao Guangrong 2015-10-31 13:45 ` Eduardo Habkost 2015-10-31 13:45 ` Eduardo Habkost 2015-10-31 16:59 ` Xiao Guangrong 2015-10-31 16:59 ` [Qemu-devel] " Xiao Guangrong 2015-11-09 10:17 ` Michael S. Tsirkin 2015-11-09 10:17 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 10:49 ` Xiao Guangrong 2015-11-09 10:49 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 12/33] pc-dimm: remove DEFAULT_PC_DIMMSIZE Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-11-09 10:40 ` Michael S. Tsirkin 2015-11-09 10:40 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 11:03 ` Xiao Guangrong 2015-11-09 11:03 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 13/33] pc-dimm: make pc_existing_dimms_capacity static and rename it Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 15:38 ` Vladimir Sementsov-Ogievskiy 2015-10-30 15:38 ` Vladimir Sementsov-Ogievskiy 2015-10-30 5:56 ` [PATCH v6 14/33] pc-dimm: drop the prefix of pc-dimm Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 16:10 ` Vladimir Sementsov-Ogievskiy 2015-10-30 16:10 ` Vladimir Sementsov-Ogievskiy 2015-10-31 8:18 ` Xiao Guangrong 2015-10-30 17:06 ` Eric Blake 2015-10-31 8:23 ` Xiao Guangrong 2015-11-09 20:05 ` Michael S. Tsirkin 2015-11-09 20:05 ` [Qemu-devel] " Michael S. Tsirkin 2015-10-30 5:56 ` [PATCH v6 15/33] stubs: rename qmp_pc_dimm_device_list.c Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 16/33] pc-dimm: rename pc-dimm.c and pc-dimm.h Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-31 10:28 ` Vladimir Sementsov-Ogievskiy 2015-10-31 10:28 ` Vladimir Sementsov-Ogievskiy 2015-10-30 5:56 ` [PATCH v6 17/33] dimm: abstract dimm device from pc-dimm Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-31 10:41 ` Vladimir Sementsov-Ogievskiy 2015-10-31 10:41 ` [Qemu-devel] " Vladimir Sementsov-Ogievskiy 2015-10-30 5:56 ` [PATCH v6 18/33] dimm: get mapped memory region from DIMMDeviceClass->get_memory_region Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-31 10:52 ` Vladimir Sementsov-Ogievskiy 2015-10-31 10:52 ` Vladimir Sementsov-Ogievskiy 2015-10-31 14:15 ` Xiao Guangrong 2015-10-31 14:15 ` [Qemu-devel] " Xiao Guangrong 2015-11-02 9:18 ` Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 19/33] dimm: keep the state of the whole backend memory Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-31 11:05 ` Vladimir Sementsov-Ogievskiy 2015-10-31 11:05 ` Vladimir Sementsov-Ogievskiy 2015-10-31 14:19 ` Xiao Guangrong 2015-11-09 11:04 ` Michael S. Tsirkin 2015-11-09 11:04 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 11:13 ` Xiao Guangrong [this message] 2015-11-09 11:13 ` Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 20/33] dimm: introduce realize callback Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-31 11:22 ` Vladimir Sementsov-Ogievskiy 2015-10-31 11:22 ` [Qemu-devel] " Vladimir Sementsov-Ogievskiy 2015-10-30 5:56 ` [PATCH v6 21/33] nvdimm: implement NVDIMM device abstract Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 22/33] docs: add NVDIMM ACPI documentation Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 23/33] nvdimm acpi: init the resource used by NVDIMM ACPI Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 24/33] nvdimm acpi: build ACPI NFIT table Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 25/33] nvdimm acpi: build ACPI nvdimm devices Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-11-08 17:38 ` Michael S. Tsirkin 2015-11-08 17:38 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 6:07 ` Xiao Guangrong 2015-11-09 6:07 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 26/33] nvdimm acpi: save arg3 for NVDIMM device _DSM method Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 27/33] nvdimm acpi: support function 0 Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 10:14 ` Stefan Hajnoczi 2015-10-30 10:14 ` [Qemu-devel] " Stefan Hajnoczi 2015-10-30 13:01 ` Xiao Guangrong 2015-10-30 13:01 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 28/33] nvdimm acpi: support Get Namespace Label Size function Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 29/33] nvdimm acpi: support Get Namespace Label Data function Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 30/33] nvdimm acpi: support Set " Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 31/33] nvdimm: allow using whole backend memory as pmem Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 32/33] nvdimm acpi: support _FIT method Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-11-08 17:50 ` Michael S. Tsirkin 2015-11-08 17:50 ` [Qemu-devel] " Michael S. Tsirkin 2015-11-09 6:37 ` Xiao Guangrong 2015-11-09 6:37 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 5:56 ` [PATCH v6 33/33] nvdimm: add maintain info Xiao Guangrong 2015-10-30 5:56 ` [Qemu-devel] " Xiao Guangrong 2015-10-30 10:15 ` [PATCH v6 00/33] implement vNVDIMM Stefan Hajnoczi 2015-10-30 10:15 ` [Qemu-devel] " Stefan Hajnoczi
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=56407FE6.1010906@linux.intel.com \ --to=guangrong.xiao@linux.intel.com \ --cc=dan.j.williams@intel.com \ --cc=ehabkost@redhat.com \ --cc=gleb@kernel.org \ --cc=imammedo@redhat.com \ --cc=kvm@vger.kernel.org \ --cc=mst@redhat.com \ --cc=mtosatti@redhat.com \ --cc=pbonzini@redhat.com \ --cc=qemu-devel@nongnu.org \ --cc=rth@twiddle.net \ --cc=stefanha@redhat.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.