[Qemu-devel] migrate cancel bug in latest qemu upstream

[Qemu-devel] migrate cancel bug in latest qemu upstream

[Amos Kong]

Hi quintela, I found a migrate bug in Autotest result, not sure if
it's already known.

| commit fa3889162ff724d09eb6b6bd06ba54769100a77f
| Merge: 6e72a00 a4960ef
| Author: Anthony Liguori <aliguori@us.ibm.com>
| Date:   Sun Mar 10 20:39:17 2013 -0500


It can be caught by migrate.mig_cancel case in Autotest framework.
Manual reproduce method: cancel migrate before migration completes

(src monitor) migrate -d tcp:0:1234
(src monitor) migrate cancel

Destination vm exits expectedly
  qemu: warning: error while loading state section id 2

Problem:
 The migrate status of source vm is 'failed' (it should be 'canceled')
 (monitor hmp1) Response to 'info migrate'
 (monitor hmp1)    capabilities: xbzrle: off
 (monitor hmp1)    Migration status: failed
 (monitor hmp1)    total time: 0 milliseconds


Long cmdline: (can reproduce)
# qemu-system-x86_64 --enable-kvm -name 'vm0' -nodefaults -chardev
socket,id=serial_id_serial1,path=/tmp/serial-serial1-20130311-182651-kfslPrch,server,nowait
-device isa-serial,chardev=serial_id_serial1 -chardev
socket,id=seabioslog_id_20130311-182651-kfslPrch,path=/tmp/seabios-20130311-182651-kfslPrch,server,nowait
-device isa-debugcon,chardev=seabioslog_id_20130311-182651-kfslPrch,iobase=0x402
-device ich9-usb-uhci1,id=usb1 -drive
file='/tmp/kvm_autotest_root/images/RHEL-Server-6.3-64-virtio.qcow2',if=none,id=virtio0
-device virtio-blk-pci,drive=virtio0 -device
e1000,netdev=idkrCwff,mac='9a:87:88:89:8a:8b',id='idxiFFqe' -netdev
tap,id=idkrCwff -m 2048 -smp 4,maxcpus=4,cores=1,threads=1,sockets=4
-cpu 'Penryn' -M pc -device
usb-tablet,id=usb-tablet1,bus=usb1.0,port=1 -vnc :1 -vga std -rtc
base=utc,clock=host,driftfix=none -boot order=cdn,once=c,menu=off

Simple cmdline: (can reproduce)
# qemu-system-x86_64  --enable-kvm
 /tmp/kvm_autotest_root/images/RHEL-Server-6.3-64-virtio.qcow2 -m
 2048  -vnc :1  -monitor stdio -net none

Thanks, Amos

Re: [Qemu-devel] migrate cancel bug in latest qemu upstream

[Amos Kong]

On Tue, Mar 12, 2013 at 10:58:54AM +0800, Amos Kong wrote:
> Hi quintela, I found a migrate bug in Autotest result, not sure if
> it's already known.
> 
> | commit fa3889162ff724d09eb6b6bd06ba54769100a77f
> | Merge: 6e72a00 a4960ef
> | Author: Anthony Liguori <aliguori@us.ibm.com>
> | Date:   Sun Mar 10 20:39:17 2013 -0500
> 

Bug was introduced between v1.1.0..v1.2.0

----

This bug could not be reproduced in current latest qemu.git

| commit 225dc991b03f0f034aa348f5cf499de9d0979107
| Author: Cornelia Huck <cornelia.huck@de.ibm.com>
| Date:   Fri Mar 15 10:57:40 2013 +0100
| 
|     s390: Fix cpu refactoring fallout.
| 
 
> It can be caught by migrate.mig_cancel case in Autotest framework.
> Manual reproduce method: cancel migrate before migration completes
> 
> (src monitor) migrate -d tcp:0:1234
> (src monitor) migrate cancel
> 
> Destination vm exits expectedly
>   qemu: warning: error while loading state section id 2
> 
> Problem:
>  The migrate status of source vm is 'failed' (it should be 'canceled')
>  (monitor hmp1) Response to 'info migrate'
>  (monitor hmp1)    capabilities: xbzrle: off
>  (monitor hmp1)    Migration status: failed
>  (monitor hmp1)    total time: 0 milliseconds
> 
> 
> Long cmdline: (can reproduce)
> # qemu-system-x86_64 --enable-kvm -name 'vm0' -nodefaults -chardev
> socket,id=serial_id_serial1,path=/tmp/serial-serial1-20130311-182651-kfslPrch,server,nowait
> -device isa-serial,chardev=serial_id_serial1 -chardev
> socket,id=seabioslog_id_20130311-182651-kfslPrch,path=/tmp/seabios-20130311-182651-kfslPrch,server,nowait
> -device isa-debugcon,chardev=seabioslog_id_20130311-182651-kfslPrch,iobase=0x402
> -device ich9-usb-uhci1,id=usb1 -drive
> file='/tmp/kvm_autotest_root/images/RHEL-Server-6.3-64-virtio.qcow2',if=none,id=virtio0
> -device virtio-blk-pci,drive=virtio0 -device
> e1000,netdev=idkrCwff,mac='9a:87:88:89:8a:8b',id='idxiFFqe' -netdev
> tap,id=idkrCwff -m 2048 -smp 4,maxcpus=4,cores=1,threads=1,sockets=4
> -cpu 'Penryn' -M pc -device
> usb-tablet,id=usb-tablet1,bus=usb1.0,port=1 -vnc :1 -vga std -rtc
> base=utc,clock=host,driftfix=none -boot order=cdn,once=c,menu=off
> 
> Simple cmdline: (can reproduce)
> # qemu-system-x86_64  --enable-kvm
>  /tmp/kvm_autotest_root/images/RHEL-Server-6.3-64-virtio.qcow2 -m
>  2048  -vnc :1  -monitor stdio -net none
> 
> Thanks, Amos

Re: [Qemu-devel] migrate cancel bug in latest qemu upstream

[Amos Kong]

On Mon, Mar 18, 2013 at 09:20:03PM +0800, Amos Kong wrote:
> On Tue, Mar 12, 2013 at 10:58:54AM +0800, Amos Kong wrote:
> > Hi quintela, I found a migrate bug in Autotest result, not sure if
> > it's already known.
> > 
> > | commit fa3889162ff724d09eb6b6bd06ba54769100a77f
> > | Merge: 6e72a00 a4960ef
> > | Author: Anthony Liguori <aliguori@us.ibm.com>
> > | Date:   Sun Mar 10 20:39:17 2013 -0500
> > 
> 
> Bug was introduced between v1.1.0..v1.2.0
> 
> ----
> 
> This bug could not be reproduced in current latest qemu.git
> 
> | commit 225dc991b03f0f034aa348f5cf499de9d0979107
> | Author: Cornelia Huck <cornelia.huck@de.ibm.com>
> | Date:   Fri Mar 15 10:57:40 2013 +0100
> | 
> |     s390: Fix cpu refactoring fallout.
> | 


This bug was already fixed by this patch:

commit dba433c03a0f5dc22a459435dd89557886298921
Author: Paolo Bonzini <pbonzini@redhat.com>
Date:   Fri Feb 22 17:36:17 2013 +0100

    migration: simplify error handling
    
    Always use qemu_file_get_error to detect errors, since that is how
    QEMUFile itself drops I/O after an error occurs.  There is no need
    to propagate and check return values all the time.
    
    Also remove the "complete" member, since we know that it is set (via
    migrate_fd_cleanup) only when the state changes.
    
    Reviewed-by: Orit Wasserman <owasserm@redhat.com>
    Reviewed-by: Juan Quintela <quintela@redhat.com>
    Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
    Signed-off-by: Juan Quintela <quintela@redhat.com>

Re: [Qemu-devel] migrate cancel bug in latest qemu upstream

[Paolo Bonzini]

Il 19/03/2013 05:49, Amos Kong ha scritto:
> On Mon, Mar 18, 2013 at 09:20:03PM +0800, Amos Kong wrote:
>> On Tue, Mar 12, 2013 at 10:58:54AM +0800, Amos Kong wrote:
>>> Hi quintela, I found a migrate bug in Autotest result, not sure if
>>> it's already known.
>
> This bug was already fixed by this patch:
> 
> commit dba433c03a0f5dc22a459435dd89557886298921
> Author: Paolo Bonzini <pbonzini@redhat.com>
> Date:   Fri Feb 22 17:36:17 2013 +0100
> 
>     migration: simplify error handling
>     
>     Always use qemu_file_get_error to detect errors, since that is how
>     QEMUFile itself drops I/O after an error occurs.  There is no need
>     to propagate and check return values all the time.
>     
>     Also remove the "complete" member, since we know that it is set (via
>     migrate_fd_cleanup) only when the state changes.
>     
>     Reviewed-by: Orit Wasserman <owasserm@redhat.com>
>     Reviewed-by: Juan Quintela <quintela@redhat.com>
>     Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>     Signed-off-by: Juan Quintela <quintela@redhat.com>
> 

Should we add these patches to qemu-stable?

5da5aad (migration: simplify while loop, 2013-02-22)
891518a (migration: always use vm_stop_force_state, 2013-02-22)
7a2c172 (migration: move more error handling to migrate_fd_cleanup, 2013-02-22)
04943eb (migration: push qemu_savevm_state_cancel out of qemu_savevm_state_*, 2013-02-22)
93bf210 (qemu-file: pass errno from qemu_fflush via f->last_error, 2013-02-22)
47c8c17 (migration: use qemu_file_set_error to pass error codes back to qemu_savevm_state, 2013-02-22)
4eb9381 (qemu-file: temporarily expose qemu_file_set_error and qemu_fflush, 2013-02-22)
f582151 (migration: flush all data to fd when buffered_flush is called, 2013-02-22)
63dfbd7 (migration: use qemu_file_set_error, 2013-02-22)
dba433c (migration: simplify error handling, 2013-02-22)

Paolo