Bug 1915216
| Summary: | VM failed to start with pmem enabled. | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux Advanced Virtualization | Reporter: | Dan Zheng <dzheng> |
| Component: | qemu-kvm | Assignee: | Virtualization Maintenance <virt-maint> |
| qemu-kvm sub component: | General | QA Contact: | Min Deng <mdeng> |
| Status: | CLOSED ERRATA | Docs Contact: | |
| Severity: | medium | ||
| Priority: | medium | CC: | bugproxy, ddepaula, dgibson, ehabkost, ehadley, gkurz, imammedo, mdeng, mprivozn, qzhang, virt-maint |
| Version: | 8.4 | Keywords: | FutureFeature, Patch, TestOnly, Triaged |
| Target Milestone: | rc | ||
| Target Release: | 8.5 | ||
| Hardware: | ppc64le | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | qemu-kvm-6.0.0-16.module+el8.5.0+10848+2dccc46d | Doc Type: | If docs needed, set a value |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2021-11-16 07:51:11 UTC | Type: | Feature Request |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | 1524652, 1568223 | ||
| Bug Blocks: | 1916117 | ||
This is spited from bz1568223. See https://bugzilla.redhat.com/show_bug.cgi?id=1568223#c44 for more details. ------- Comment From vajain21.com 2021-01-13 10:06 EDT------- I think PMDK/libpmem support hasnt yet landed in RHEL8.3 so qemu is most probably compiled with libpmem=off. This means that support for '<pmem/>' tag in domain-xml is not in place yet. Requesting you to please check removing the '<pmem/>' tag from the domain-xml and see if its working fine Thanks, IBM, Yes, just as you said, qemu has not ready to support <pmem/> which depends on bz1524652. Qemu will get re-compiled after 1524652 is fixed. So this bz is just an enhancement to make sure we may have this supported in future (it is supported on x86_64.) It works if '<pmem/>' is removed from domain xml. Dan Thanks Libvirt checks whether QEMU supports pmem: https://gitlab.com/libvirt/libvirt/-/commit/55b4fc78b6397caf7eff73a0dc7fbb8892e98d49 And it also adds pmem=on only if supported by QEMU: https://gitlab.com/libvirt/libvirt/-/commit/82576d8f35e If pmem is not supported then Libvirt errors out. The problem here is that QEMU reports 'pmem' attribute always and only at runtime it decides whether to error out (if compiled without libpmem) or no (compiled with libpmem). Therefore, I think this is a QEMU bug. I think we need something like this: https://lists.nongnu.org/archive/html/qemu-devel/2021-01/msg04558.html I tried the issue with build qemu-kvm-5.2.0-3.scrmod+el8.4.0+9533+d0097f55.wrb210120.ppc64le cli, 1.boot up a guest with /usr/libexec/qemu-kvm -name avocado-vt-vm1 -sandbox on -machine pseries -nodefaults -device VGA,bus=pci.0,addr=0x2 -m 20480 -smp 2,maxcpus=2,cores=1,threads=1,sockets=2 -cpu host -chardev socket,id=chardev_serial0,server,path=/tmp/SS,nowait -device spapr-vty,id=serial0,reg=0x30000000,chardev=chardev_serial0 -device pci-ohci,id=usbtest,bus=pci.0,addr=0x3 -device usb-tablet,id=usb-tablet1,bus=usbtest.0,port=1 -device usb-hub,id=usb-testdev,bus=usbtest.0,port=2 -device virtio-scsi-pci,id=virtio_scsi_pci0,bus=pci.0,addr=0x4 -blockdev node-name=file_image1,driver=file,auto-read-only=on,discard=unmap,aio=threads,filename=rhel840-ppc64le-virtio-scsi.qcow2,cache.direct=on,cache.no-flush=off -blockdev node-name=drive_image1,driver=qcow2,read-only=off,cache.direct=on,cache.no-flush=off,file=file_image1 -device scsi-hd,id=image1,drive=drive_image1,write-cache=on -vnc :1 -rtc base=utc,clock=host -boot menu=off,order=cdn,once=c,strict=off -enable-kvm -monitor stdio -object memory-backend-file,id=memnvdimm0,prealloc=yes,mem-path=/tmp/nvdimm-private-2,size=268697600,align=2097152,pmem=on,host-nodes=0,policy=preferre Actual results, The guest can't boot up obviously since it's not supported. qemu-kvm: -object memory-backend-file,id=memnvdimm0,prealloc=yes,mem-path=/tmp/nvdimm-private-2,size=268697600,align=2097152,pmem=on,host-nodes=0,policy=preferred: Lack of libpmem support while setting the 'pmem=on' of memory-backend-file. We can't ensure data persistence. Expected results, From qemu-kvm QE's perspective, if it's not supported, the guest shouldn't boot up,it's obvious. what's the expected results for libvirt then ? Thanks. (In reply to Min Deng from comment #6) > Expected results, > From qemu-kvm QE's perspective, if it's not supported, the guest shouldn't > boot up,it's obvious. what's the expected results for libvirt then ? > Yeah, from libvirt's POV the guest shouldn't be able to boot because it can't unsure <pmem/>. We really need that patch. I've resend the patch here: https://lists.nongnu.org/archive/html/qemu-devel/2021-01/msg06561.html Igor, Eduardo, I see that the patch from comment 8 isn't merged yet. Is it in either of your queues? If not, what should we do to expedite it? (In reply to David Gibson from comment #10) > Igor, Eduardo, > > I see that the patch from comment 8 isn't merged yet. Is it in either of > your queues? If not, what should we do to expedite it? Sorry for missing it. Queued upstream: https://lore.kernel.org/qemu-devel/20210216222357.GL4070@habkost.net ------- Comment From hannsj_uhl.com 2021-03-08 03:35 EDT------- (In reply to comment #16) > (In reply to David Gibson from comment #10) > > Igor, Eduardo, > > > > I see that the patch from comment 8 isn't merged yet. Is it in either of > > your queues? If not, what should we do to expedite it? > Sorry for missing it. Queued upstream: > https://lore.kernel.org/qemu-devel/20210216222357.GL4070@habkost.net . ... which I think is now upstream accepted as git commit https://git.qemu.org/gitweb.cgi?p=qemu.git;a=commit;h=def835f0da0d153b397071e6bb8f2b46f51f96b4 ("hostmem: Don't report pmem attribute if unsupported") ... Hi Danilo， Can you help to check how to proceed this bz because it's in DevMissed status. Thanks. By the way, bz1918555 has the same situation. Min Verified the bug on the following builds, qemu-kvm-6.0.0-16.module+el8.5.0+10848+2dccc46d.ppc64le Steps please refer to comment 6, Actual results, qemu-kvm: -object memory-backend-file,id=memnvdimm0,prealloc=yes,mem-path=/tmp/nvdimm-private-2,size=268697600,align=2097152,pmem=on,host-nodes=0,policy=preferred: Invalid parameter 'pmem' Expected results, Qemu will not report "pmem" any more. Base on above comments, the bug should be fixed, thanks a lot. Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (virt:av bug fix and enhancement update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2021:4684 |
Description of problem: VM failed to start with pmem enabled. Version-Release number of selected component (if applicable): libvirt-6.5.0-1.module+el8.3.0+7323+d54bb644.ppc64le qemu-kvm-5.0.0-2.module+el8.3.0+7379+0505d6ca.ppc64le kernel-4.18.0-221.el8.ppc64le How reproducible: 100% Steps to Reproduce: 1.With alignsize + pmem 1.1 Define vm with below: <memory model='nvdimm'> <source> <path>/tmp/nvdimm</path> <alignsize unit='KiB'>2048</alignsize> <pmem/> </source> <target> <size unit='KiB'>524288</size> <node>1</node> <label> <size unit='KiB'>128</size> </label> <readonly/> </target> <address type='dimm' slot='1'/> </memory> 1.2 Start vm # virsh start vm2 error: Failed to start domain vm2 error: internal error: process exited while connecting to monitor: 2020-07-20T12:09:31.096911Z qemu-kvm: -object memory-backend-file,id=memnvdimm0,prealloc=yes,mem-path=/tmp/nvdimm-private-2,size=268697600,align=2097152,pmem=on,host-nodes=0,policy=preferred: Lack of libpmem support while setting the 'pmem=on' of memory-backend-file. We can't ensure data persistence. 2. 3. Actual results: See above Expected results: VM should be started correctly. Additional info: