Bug 2041757

Hit the same issue for virtio-blk-pci device + a 100G LIO-ORG volume, the disk cannot be formated to ntfs normally, and it cannot be reproduced with a 10G volume:

command line as follows:

/usr/libexec/qemu-kvm -name guest=instance-00000007,debug-threads=on \
	-machine pc-i440fx-rhel7.6.0,usb=off,dump-guest-core=off \
	-accel kvm -cpu host,ss=on,vmx=on,pdcm=on,hypervisor=on,tsc-adjust=on,umip=on,pku=on,md-clear=on,stibp=on,arch-capabilities=on,xsaves=on,ibpb=on,ibrs=on,amd-stibp=on,amd-ssbd=on,rdctl-no=on,ibrs-all=on,skip-l1dfl-vmentry=on,mds-no=on,pschange-mc-no=on,tsx-ctrl=on,hle=off,rtm=off \
	-m 8192 \
	-overcommit mem-lock=off \
	-smp 4,sockets=4,dies=1,cores=1,threads=1 \
	-uuid 0905830f-496c-40b4-bd7e-7c00cb350fd0 \
	-smbios type=1,manufacturer="OpenStack Foundation",product="OpenStack Nova",version=24.0.1,serial=0905830f-496c-40b4-bd7e-7c00cb350fd0,uuid=0905830f-496c-40b4-bd7e-7c00cb350fd0,family="Virtual Machine" \
	-no-user-config -nodefaults \
	-rtc base=utc,driftfix=slew \
	-global kvm-pit.lost_tick_policy=delay \
	-no-hpet -no-shutdown -boot strict=on \
	-device piix3-usb-uhci,id=usb,bus=pci.0,addr=0x1.0x2 \
	-blockdev '{"driver":"file","filename":"/home/kvm_autotest_root/images/win2022-64-virtio.qcow2","node-name":"libvirt-2-storage","cache":{"direct":true,"no-flush":false},"auto-read-only":true,"discard":"unmap"}' \
	-blockdev '{"node-name":"libvirt-2-format","read-only":false,"cache":{"direct":true,"no-flush":false},"driver":"qcow2","file":"libvirt-2-storage"}' \
	-device virtio-blk-pci,bus=pci.0,addr=0x4,drive=libvirt-2-format,id=virtio-disk0,bootindex=1,write-cache=on \
	-blockdev '{"driver":"host_device","filename":"/dev/sdf","aio":"native","node-name":"libvirt-1-storage","cache":{"direct":true,"no-flush":false},"auto-read-only":true,"discard":"unmap"}' \
	-blockdev '{"node-name":"libvirt-1-format","read-only":false,"cache":{"direct":true,"no-flush":false},"driver":"raw","file":"libvirt-1-storage"}' \
	-device virtio-blk-pci,bus=pci.0,addr=0x5,drive=libvirt-1-format,id=virtio-disk1,write-cache=on,serial=e520a974-4bbf-4f0c-ace3-475d44be83bb \
	-netdev tap,id=hostnet0 -device virtio-net-pci,netdev=hostnet0,id=net0,mac=fa:16:3e:59:f6:7a,bus=pci.0,addr=0x3 \
	-device usb-tablet,id=input0,bus=usb.0,port=1 \
	-vnc :1 \
	-device cirrus-vga,id=video0,bus=pci.0,addr=0x2 \
	-device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x6 \
	-device vmcoreinfo \
	-cdrom /home/kvm_autotest_root/iso/windows/virtio-win-1.9.19-5.el9.iso \
        -msg timestamp=on


# lsscsi
[0:0:0:0]    disk    Generic- SD/MMC CRW       1.00  /dev/sda 
[1:0:0:0]    enclosu HPE      Smart Adapter    1.66  -        
[1:1:0:0]    disk    HPE      LOGICAL VOLUME   1.66  /dev/sdb 
[1:2:0:0]    storage HPE      P408i-a SR Gen10 1.66  -        
[2:0:0:0]    disk    LIO-ORG  disk1            4.0   /dev/sdc 
[2:0:0:1]    disk    LIO-ORG  disk2            4.0   /dev/sdd 
[3:0:0:0]    disk    LIO-ORG  disk1            4.0   /dev/sde 
[3:0:0:1]    disk    LIO-ORG  disk2            4.0   /dev/sdh 
[3:0:0:2]    disk    LIO-ORG  disk3            4.0   /dev/sdg 
[3:0:0:3]    disk    LIO-ORG  disk4            4.0   /dev/sdf 

Tested with virtio-scsi-pci + 100G LIO-ORG disk, we cannot reproduce this issue, disk format can be finished success.

Used version:
qemu-kvm-6.1.0-5.el9.x86_64
kernel-5.14.0-21.el9.x86_64
virtio-win-prewhql-215
seabios-bin-1.14.0-7.el9.noarch

Best Regards~
Peixiu

If only create 4G volume in #0 testing, it will pass if the whql version before virtio-win-prewhql-205.
I also tested virtio-blk-pci like #1, but not reproduce.

Is this a new test? If so, at what version did it last succeed?

Kevin, I'm tentatively assigning it to you, but let me know if you think there are better candidates or if your plate is already too full

scsi-block is for Paolo, usually, so moving it to him for now.

But do I understand correctly that this issue happens only on Windows? Then it could be a Windows driver bug, CCing Vadim as well.

This issue might be caused by the following commit  https://github.com/virtio-win/kvm-guest-drivers-windows/commit/c62a8a2c7bf73541fcfd6bc43bcf85147abf2620#diff-7878e1a52e8fb5a4d75b959b370267e82c07452c4dea6410c6cf32d68986207e
and should be fixed with https://bugzilla.redhat.com/show_bug.cgi?id=2022656

Peixiu, can you please check if adding ",max_sectors=64" parameter solves the problem?

Thanks,
Vadim.

Thank Vadim Rozenfeld.

it works  after adding option ",max_sectors=64" :

-device virtio-scsi-pci,id=virtio_scsi_pci0,bus=pcie-pci-bridge-0,addr=0x1,iothread=iothread1,max_sectors=64 \

So i have some questions :
1. It looks like related to specific HW, i can not reproduce it on the emulated iscsi disk.
When should add this option and what is the value?

2. There is no issue for rhel guest, why have issue on windows guest?

3. If we have to add option for it, do we need documentation ?

(In reply to Vadim Rozenfeld from comment #5)
> This issue might be caused by the following commit 
> https://github.com/virtio-win/kvm-guest-drivers-windows/commit/
> c62a8a2c7bf73541fcfd6bc43bcf85147abf2620#diff-
> 7878e1a52e8fb5a4d75b959b370267e82c07452c4dea6410c6cf32d68986207e
> and should be fixed with https://bugzilla.redhat.com/show_bug.cgi?id=2022656
> 
> Peixiu, can you please check if adding ",max_sectors=64" parameter solves
> the problem?
> 
For virtio-scsi part, thanks qinwang's testing.
For virtio-blk, it seems not support max_sectors param, hit follows error:
2022-01-20T09:37:22.133360Z qemu-kvm: -device virtio-blk-pci,bus=pci.0,addr=0x4,max_sectors=64,drive=libvirt-2-format,id=virtio-disk0,bootindex=1,write-cache=on: Property 'virtio-blk-pci.max_sectors' not found

And for virtio-blk, it works with 10G volume disk, but doesn't work with 100G volume disk, it seems it's an another issue, do we need to file another bz to track it? thanks~

> Thanks,
> Vadim.

(In reply to qing.wang from comment #6)
> Thank Vadim Rozenfeld.
> 
> it works  after adding option ",max_sectors=64" :
> 
> -device
> virtio-scsi-pci,id=virtio_scsi_pci0,bus=pcie-pci-bridge-0,addr=0x1,
> iothread=iothread1,max_sectors=64 \
> 
> So i have some questions :
> 1. It looks like related to specific HW, i can not reproduce it on the
> emulated iscsi disk.
> When should add this option and what is the value?

In case of the scsi-block the best thing is to check the max_segments
value of a specific block device 
cat /sys/block/sdb/queue/max_segments

and adjust the qemu ",max_sectors" parameter accordingly

> 
> 2. There is no issue for rhel guest, why have issue on windows guest?

The recent Windows platforms are capable to handle quite large disk IO 
DMA transfers (up to 512 entries in the SG List). That gives us a good
performance improvment for 256K and bigger blocks disk transfers. That 
was the main reason to enlarge the maximum transfer size and the number 
of physical breaks in vioscsi driver.

> 
> 3. If we have to add option for it, do we need documentation ?
Yes. It needs to be documented.

(In reply to Peixiu Hou from comment #7)
> (In reply to Vadim Rozenfeld from comment #5)
> > This issue might be caused by the following commit 
> > https://github.com/virtio-win/kvm-guest-drivers-windows/commit/
> > c62a8a2c7bf73541fcfd6bc43bcf85147abf2620#diff-
> > 7878e1a52e8fb5a4d75b959b370267e82c07452c4dea6410c6cf32d68986207e
> > and should be fixed with https://bugzilla.redhat.com/show_bug.cgi?id=2022656
> > 
> > Peixiu, can you please check if adding ",max_sectors=64" parameter solves
> > the problem?
> > 
> For virtio-scsi part, thanks qinwang's testing.
> For virtio-blk, it seems not support max_sectors param, hit follows error:
> 2022-01-20T09:37:22.133360Z qemu-kvm: -device
> virtio-blk-pci,bus=pci.0,addr=0x4,max_sectors=64,drive=libvirt-2-format,
> id=virtio-disk0,bootindex=1,write-cache=on: Property
> 'virtio-blk-pci.max_sectors' not found
> 
> And for virtio-blk, it works with 10G volume disk, but doesn't work with
> 100G volume disk, it seems it's an another issue, do we need to file another
> bz to track it? thanks~

Yes, please.
Technically, both viostor and vioscsi drivers have a registry parameter to limit the
maximum transfer length, but IIRC the default value is 256 PhysicalBreaks, which
can be too much in some cases of direct LUN configuration.  


> 
> > Thanks,
> > Vadim.

After investigating ,this bug should same as https://bugzilla.redhat.com/show_bug.cgi?id=2022656
But the blk issue looks like different.

-blockdev '{"driver":"host_device","filename":"/dev/sdf","aio":"native","node-name":"libvirt-1-storage","cache":{"direct":true,"no-flush":false},"auto-read-only":true,"discard":"unmap"}' \
-blockdev '{"node-name":"libvirt-1-format","read-only":false,"cache":{"direct":true,"no-flush":false},"driver":"raw","file":"libvirt-1-storage"}' \
-device virtio-blk-pci,bus=pci.0,addr=0x5,drive=libvirt-1-format,id=virtio-disk1,write-cache=on,serial=e520a974-4bbf-4f0c-ace3-475d44be83bb \

 phou, could you please help to open a bug to track it,thanks.

*** This bug has been marked as a duplicate of bug 2022656 ***

(In reply to qing.wang from comment #10)
> After investigating ,this bug should same as
> https://bugzilla.redhat.com/show_bug.cgi?id=2022656
> But the blk issue looks like different.
> 
> -blockdev
> '{"driver":"host_device","filename":"/dev/sdf","aio":"native","node-name":
> "libvirt-1-storage","cache":{"direct":true,"no-flush":false},"auto-read-
> only":true,"discard":"unmap"}' \
> -blockdev
> '{"node-name":"libvirt-1-format","read-only":false,"cache":{"direct":true,
> "no-flush":false},"driver":"raw","file":"libvirt-1-storage"}' \
> -device
> virtio-blk-pci,bus=pci.0,addr=0x5,drive=libvirt-1-format,id=virtio-disk1,
> write-cache=on,serial=e520a974-4bbf-4f0c-ace3-475d44be83bb \
> 
>  phou, could you please help to open a bug to track it,thanks.

Hi qinwang, Filed a new bug: https://bugzilla.redhat.com/show_bug.cgi?id=2043493, and will add a new test case based on the new bz, thanks a lot~