Xen project Mailing List

On Fri, Mar 28, 2014 at 3:31 AM, Ian Campbell <Ian.Campbell@xxxxxxxxxx> wrote:

(my MUA crashed, apparently without sending this first, sorry if it
comes through twice).

On Thu, 2014-03-27 at 10:58 -0700, Doug Breshears wrote:
>
>

> Made some progress but still hanging up. See below.

I'm a bit stumped so I'm copying xen-devel and some qemu-xen savvy
people (lets ignore qemu-trad for now, since that was just for
comparison)

Stefano, Anthony, the summary (from the end of Doug's mail) is:

So, to summarize, the Xen 4.3.1 HVM (booting Windows XP) fails
to run reliably.
In standard mode it hands on "Setup is searching you
computer..."
In traditional mode it hangs on "Setup is starting
windows" (Much farther along, after loading its drivers)

Most of the info is in the mail quoted below, the full thread is at
http://lists.xen.org/archives/html/xen-users/2014-03/msg00162.html.

Doug, can you confirm which Service Pack of WinXP you are using? IIRC
SP3 was considerably different (i.e. the kernel was upgraded) to SP2 and
earlier. I think most people use SP3 and this is what is known to work
with Xen (i.e. what we test in our automated tests).

Thanks,
Ian.

>
> On Thu, Mar 27, 2014 at 4:50 AM, Ian Campbell
> <Ian.Campbell@xxxxxxxxxx> wrote:
> On Wed, 2014-03-26 at 21:13 -0700, Doug Breshears wrote:
>
> > -----------------------------------
> >
> > As a side note I attempted to "cat /proc/cpuinfo |
> grep svm"
> > and it failed to find anything.
> > However "xl dmesg" does..?
> > I assume it is a "kernel thing" since the debian OS
> before it
> > displayed it. What I don't know is if it is
> important :)
> > I just wanted to air my concerns just in case..
>
>
> I think this is OK. Xen takes over the the SVM functionality
> in hardware
> so it is not exposed to dom0.
>
> > Tried using local disk and ISO rather than ATAoE but
> nothing
> > changed.
>
>
> I think this was a good thing to try -- eliminate variables
> etc.
>
> > Tried disabling grsec and got a change in the
> qemu-dm log.
> > From the one line..
> >
> > qemu: terminating on signal 1 from pid 11465
>
>
> This certainly can't be a good thing, I'm surprised things
> worked well
> enough to get to hanging in the windows installer if qemu has
> been
> killed.
>
>
> This was only after I "destroyed" the domain. It was an empty log
> before that.
>
>
> > To the following...
>
> This looks much more normal. I take it things still didn't
> work in this
> case? With the exact same failure mode?
>
> You are using Xen 4.3.x which I think means this is using
> "qemu-xen" (upstream qemu) rather than
> "qemu-xen-traditional" (old Xen
> fork). Can you confirm? (ps should show you the process and
> the path
> with make it clear). You can also force this using the
> device_model_version directive in your cfg file, it might be
> worth
> testing both explicitly.
>
>
>
>
> #device_model_version = "qemu-xen-traditional"
> device_model_version = "qemu-xen"
>
>
> 6674 root 0:00 /usr/lib/xen/bin/qemu-system-i386 -xen-domid 12
> -chardev
> socket,id=libxl-cmd,path=/var/run/xen/qmp-libxl-12,server,nowait -mon
> chardev=libxl-cmd,mode=control -name irrigctrlview -vnc
> 0.0.0.0:12,to=99 -global isa-fdc.driveA= -vga cirrus -global
> vga.vram_size_mb=8 -boot order=dc -usb -usbdevice tablet -no-acpi
> -device rtl8139,id=nic0,netdev=net0,mac=00:16:3e:1c:83:81 -netdev
> type=tap,id=net0,ifname=vif12.0-emu,script=no,downscript=no -M xenfv
> -m 1016 -drive
> file=/dev/etherd/e400.18,if=ide,index=0,media=disk,format=raw,cache=writeback-drive file=/dev/etherd/e300.1,if=ide,index=2,media=cdrom,format=raw,cache=writeback,id=ide-5632
>
>
>
>
> device_model_version = "qemu-xen-traditional"
> #device_model_version = "qemu-xen"
> 7184 root 0:02 /usr/lib/xen/bin/qemu-dm -d 13 -domain-name
> irrigctrlview -vnc 0.0.0.0:12 -vncunused -videoram 8 -boot dc -usb
> -usbdevice tablet -vcpu_avail 0x01 -net
> nic,vlan=0,macaddr=00:16:3e:1c:83:81,model=rtl8139 -net
> tap,vlan=0,ifname=vif13.0-emu,bridge=br0,script=no,downscript=no -M
> xenfv
>
>
>
>
>
>
> #device_model_version = "qemu-xen-traditional"
> #device_model_version = "qemu-xen"
>
>
> 6161 root 0:01 /usr/lib/xen/bin/qemu-system-i386 -xen-domid 11
> -chardev
> socket,id=libxl-cmd,path=/var/run/xen/qmp-libxl-11,server,nowait -mon
> chardev=libxl-cmd,mode=control -name irrigctrlview -vnc
> 0.0.0.0:12,to=99 -global isa-fdc.driveA= -vga cirrus -global
> vga.vram_size_mb=8 -boot order=dc -usb -usbdevice tablet -no-acpi
> -device rtl8139,id=nic0,netdev=net0,mac=00:16:3e:1c:83:81 -netdev
> type=tap,id=net0,ifname=vif11.0-emu,script=no,downscript=no -M xenfv
> -m 1016 -drive
> file=/dev/etherd/e400.18,if=ide,index=0,media=disk,format=raw,cache=writeback -drive file=/dev/etherd/e300.1,if=ide,index=2,media=cdrom,format=raw,cache=writeback,id=ide-5632
>
>
>
>
>
>
> Yes, it looks like the "qemu-xen" is the default since the process
> listings seems to be nearly identical.
> However the traditional does not show much of anything.. No disks is
> the thing that stands out to me.
>
> Can you also please post the logs from "xl -vvv create
> irrview.cfg" and
> any other logs relating to this domain from under /var/log/xen
> in case
> one of them gives us a hint. Also a full "xl dmesg" and
> "dmesg" might
> have something.
>
> [...]
> > -------------- Start irrview.cfg ------------------
> > builder="hvm"
> > #bootloader="pygrub"
> > name = "irrigctrlview"
> > memory = 1024
> > vcpus = 1
> > #pae = 1
> > acpi = 1
> > #apic = 1
> > vif = [ 'ip=192.168.8.38, mac=00:16:3E:1C:83:81,
> bridge=br0']
>
>
> Clutching a straw: Can you remove and spaces from within the
> quotes.
>
> > disk = [
> > 'phy:/dev/etherd/e400.18,xvda,w',
> > 'phy:/dev/etherd/e300.1,xvdc:cdrom,r'
>
>
> Another thing to try: Use hda and hdc.
>
> Ian.
>
>
>
>
>
>
> FIRST
> ===================
> I removed the spaces in the disk section
> It fails in exactly the same manner as before (Starts but hangs on
> "Setup searching hardware..")
>
>
>
> -- RUN --
> http://pastebin.com/yqBXar4b
>
>
>
> -- xl dmesg --
> http://pastebin.com/aaKKpVr8
>
>
>
>
>
>
>
> SECOND
>
> ======================
> I changed the header to add "Traditional Model"
> This one fails to even start..
>
>
> -- CONFIG --
> device_model_version = "qemu-xen-traditional"
> device_model_override = "/usr/lib/xen/bin/qemu-system-i386"
> bios = "seabios"
>
>
> -- RUN --
> jshvmhost1:/etc/xen# xl -vvv create irrigview.cfg
> Parsing config from irrigview.cfg
> failed to free memory for the domain
> xc: debug: hypercall buffer: total allocations:6 total releases:6
> xc: debug: hypercall buffer: current allocations:0 maximum
> allocations:1
> xc: debug: hypercall buffer: cache current size:1
> xc: debug: hypercall buffer: cache hits:5 misses:1 toobig:0
>
>
>
>
> THIRD
> ===========================
> Changed the header a little...
> This one still fails to start but got farther...
>
>
> -- CONFIG --
> device_model_version = "qemu-xen-traditional"
> device_model_override = "/usr/lib/xen/bin/qemu-system-i386"
> #bios = "seabios"
>
>
> -- RUN --
> http://pastebin.com/gVj1mJrm
>
>
>
>
>
>
>
> FOURTH
> ==============================
> Changed again, removing the override line..
> This one gets past the search, it loads all the drivers then says
> "Setup is starting Windows" and hangs.
>
>
> -- CONFIG --
> device_model_version = "qemu-xen-traditional"
> #device_model_override = "/usr/lib/xen/bin/qemu-system-i386"
> #bios = "seabios"
>
>
> -- RUN --
> Verbose Output: http://pastebin.com/0W4qmSRV
>
>
>
> -- Here is the qemu-dm log output --
>
>
>
> jshvmhost1:/var/log/xen# cat qemu-dm-irrigctrlview.log
> domid: 8
> -videoram option does not work with cirrus vga device model. Videoram
> set to 4M.
> Using file /dev/etherd/e400.18 in read-write mode
> Using file /dev/etherd/e300.1 in read-only mode
> Watching /local/domain/0/device-model/8/logdirty/cmd
> Watching /local/domain/0/device-model/8/command
> Watching /local/domain/8/cpu
> qemu_map_cache_init nr_buckets = 10000 size 4194304
> shared page at pfn feffd
> buffered io page at pfn feffb
> Guest uuid = 6d821cff-a03b-4900-a6f8-1b2cc2459fca
> populating video RAM at ff000000
> mapping video RAM from ff000000
> Register xen platform.
> Done register platform.
> platform_fixed_ioport: changed ro/rw state of ROM memory area. now is
> rw state.
> xs_read(/local/domain/0/device-model/8/xen_extended_power_mgmt): read
> error
> xs_read(): vncpasswd get
> error. /vm/6d821cff-a03b-4900-a6f8-1b2cc2459fca/vncpasswd.
> medium change watch on `hdc' (index: 1): /dev/etherd/e300.1
> I/O request not ready: 0, ptr: 0, port: 0, data: 0, count: 0, size: 0
> Log-dirty: no command yet.
> vcpu-set: watch node error.
> [xenstore_process_vcpu_set_event]: /local/domain/8/cpu has no CPU!
> xs_read(/local/domain/8/log-throttling): read error
> qemu: ignoring not-understood drive `/local/domain/8/log-throttling'
> medium change watch on `/local/domain/8/log-throttling' - unknown
> device, ignored
> cirrus vga map change while on lfb mode
> mapping vram to f0000000 - f0400000
> platform_fixed_ioport: changed ro/rw state of ROM memory area. now is
> rw state.
> platform_fixed_ioport: changed ro/rw state of ROM memory area. now is
> ro state.
>
>
>
>
> So, to summarize, the Xen 4.3.1 HVM (booting Windows XP) fails to run
> reliably.
> In standard mode it hands on "Setup is searching you computer..."
> In traditional mode it hangs on "Setup is starting windows" (Much
> farther along, after loading its drivers)
>
>
> Any suggestions?

Doug Breshears

JSH Farms Inc.

541-571-3332

Re: [Xen-users] Xen 4.3.1 HVM hangs after boot