Re: [Xen-devel] [PATCH] x86/hvm: implement save/restore for posted interrupts

["Zhang, Yang Z" <yang.z.zhang@xxxxxxxxx>]

Olaf Hering wrote on 2014-10-02:
> On Wed, Oct 01, Jan Beulich wrote:
> 
> > But I though it's not so much that you need a test system than you
> > describing to Intel folks the precise conditions needed for
> > reproduction.
> 
> Finally I got access to a system, perhaps I can use it for a while.
> 
> It has sles11sp3 installed. So I installed the latest kernel and xen packages 
> from
> sles11sp3, and used this domU.cfg:
> 
> 
> name="sles11sp3_full_bug866902_nfsroot"
> uuid="071cdcce-4c2c-48ba-b09a-6984ac91692b"
> memory=512
> vcpus=1
> on_poweroff="destroy"
> on_reboot="restart"
> on_crash="coredump-destroy"
> localtime=0
> keymap="de"
> 
> builder="hvm"
> device_model="/usr/lib/xen/bin/qemu-dm"
> kernel="/usr/lib/xen/boot/hvmloader"
> boot="d"
> disk=[
> 
> 'file:/nfsmnt/bug866902/vdisk-sles11sp3_full_bug866902_nfsroot-disk0,hda,w'
> ,
> 
> 'file:/nfsmnt/iso/SLES-11-SP3-DVD-x86_64-GM-DVD1.iso,hdc:cdrom,r'
> ]
> vif=[ 'bridge=br0,type=netfront,mac=00:16:3e:4e:7c:4b', ]
> stdvga=0
> vnc=1
> vncunused=1
> viridian=0
> acpi=1
> pae=1
> apic=1
> 
> serial="pty"
> 
> 
> Then these commands to boot the HVM guest into the provided iso image:
> 
> xm new ~/ohering/sles11sp3_full_bug866902_nfsroot.cfg
> xm start sles11sp3_full_bug866902_nfsroot xm vnc
> sles11sp3_full_bug866902_nfsroot &
> 
>  -> select "Rescue system"
>  -> set cmdline to "netsetup=dhcp sysrq_always_enabled"
>  -> RETURN
>  -> wait until login prompt appears, login as root with no password  -> run
> dmesg, its nearly empty
> 
> xm sysrq sles11sp3_full_bug866902_nfsroot s
> 
>  -> run dmesg, should show the sysrequest  -> remember dmesg timestamp
> 
> xm save -f sles11sp3_full_bug866902_nfsroot
> /dev/shm/sles11sp3_full_bug866902_nfsroot.dump
> xm restore /dev/shm/sles11sp3_full_bug866902_nfsroot.dump
> xm sysrq sles11sp3_full_bug866902_nfsroot s xm vnc
> sles11sp3_full_bug866902_nfsroot &
> 
>  -> run dmesg, there is suspend/resume output but just a single sysrequest  ->
> expected are two sysrequest entries
> 
> xm destroy sles11sp3_full_bug866902_nfsroot
> 
> 
> If I use openSUSE-13.1-DVD-x86_64.iso instead of the sles11sp3.iso the guest
> is frozen, no cursor blinking, no input possible. Not sure why that happens.
> 
> 
> Then I tried todays staging branch + v5 of my --prefix series:

What's the v5 patch?

> 
> insserv -r xendomains
> insserv -r xencommons
> rpm -Uvh xen-upstream.rpm
> cp -avLt /boot --parents /opt/xen/staging-upstream/boot/xen.gz
> vi /boot/grub/menu.lst
> /opt/xen/staging-upstream/etc/init.d/xencommons start
> /opt/xen/staging-upstream/sbin/xl -vvvv create -ddddd -V
> ~/ohering/sles11sp3_full_bug866902_nfsroot.cfg
> 
>  -> select "Rescue system"
>  -> set cmdline to "netsetup=dhcp sysrq_always_enabled"
>  -> RETURN
>  -> wait until login prompt appears, login as root with no password  -> run
> dmesg, its nearly empty
> 
> /opt/xen/staging-upstream/sbin/xl -vvvv sysrq
> sles11sp3_full_bug866902_nfsroot s
> 
>  -> run dmesg, should show the sysrequest  -> remember dmesg timestamp
> 
> /opt/xen/staging-upstream/sbin/xl -vvvv save -f
> sles11sp3_full_bug866902_nfsroot
> /dev/shm/sles11sp3_full_bug866902_nfsroot.dump
> /opt/xen/staging-upstream/sbin/xl -vvvv restore
> /dev/shm/sles11sp3_full_bug866902_nfsroot.dump
> /opt/xen/staging-upstream/sbin/xl -vvvv sysrq
> sles11sp3_full_bug866902_nfsroot s /opt/xen/staging-upstream/sbin/xl -vvvv
> vnc sles11sp3_full_bug866902_nfsroot &
> 
>  -> run dmesg, there is suspend/resume output but just a single sysrequest  ->
> expected are two sysrequest entries
> 
> /opt/xen/staging-upstream/sbin/xl -vvvv destroy
> sles11sp3_full_bug866902_nfsroot
> 
> 
> So this is my repro case. Hopefully this will work for folks at Intel.

Can you reproduce it with APICv disabled? You can add apicv=0 into your grub to 
disable it.

> 
> Tomorrow I will try if any of the suggested patches will fix xen-upsteam.
> 
> 
> One thing I'm not sure about is the firmware of such IvyBridge system.
> The firmware of the system I can use for testing dumps alot to serial during
> startup. Not sure if its EFI or not. At least it boots into grub1, so I 
> suspect it has
> a legacy BIOS. If an EFI system is used, would that change anything? The 
> initial
> report I got was appearently also with legacy BIOS. Did Intel test with EFI by
> any chance?
> 
> What I have is some Intel Romley box with "Intel(R) Xeon(R) CPU E5-2697
> v2 @ 2.70GHz", dmesg has:
> <7>DMI: Intel Corporation S2600CP/S2600CP, BIOS
> RMLSDP.86I.R3.27.D685.1305151734 05/15/2013
> 
> 
> Olaf

Best regards,
Yang


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel