[Qemu-devel] [BUG] Migration broken by "e1000: port to vmstate"

[Qemu-devel] [BUG] Migration broken by "e1000: port to vmstate"

[Pierre Riteau]

e482dc3eaac43f88beea133843ae38c661262e97 breaks migration of a VM using an e1000 device (which is the default...).
Origin host is Debian Lenny 32-bits, destination host is Fedora 12 32-bit. Guest is running Debian Lenny 32-bit.
Symtoms: origin finishes migration correctly, but destination prints "load of migration failed" and the VM is stopped.

This is because the origin closed the connexion, so qemu_fill_buffer reads a buffer with len == 0, which is treated as an error later (f->has_error = 1).
Reverting e482dc3eaac43f88beea133843ae38c661262e97 fixes the problem.

-- 
Pierre Riteau -- http://perso.univ-rennes1.fr/pierre.riteau/

Re: [Qemu-devel] [BUG] Migration broken by "e1000: port to vmstate"

[Ryan Harper]

* Pierre Riteau <Pierre.Riteau@irisa.fr> [2009-11-24 09:35]:
> e482dc3eaac43f88beea133843ae38c661262e97 breaks migration of a VM using an e1000 device (which is the default...).
> Origin host is Debian Lenny 32-bits, destination host is Fedora 12 32-bit. Guest is running Debian Lenny 32-bit.
> Symtoms: origin finishes migration correctly, but destination prints "load of migration failed" and the VM is stopped.

What's your command line?

If I add -net user , or -net user -net nic,model=e1000 migration works.


-- 
Ryan Harper
Software Engineer; Linux Technology Center
IBM Corp., Austin, Tx
ryanh@us.ibm.com

[Qemu-devel] Re: [BUG] Migration broken by "e1000: port to vmstate"

[Juan Quintela]

Pierre Riteau <Pierre.Riteau@irisa.fr> wrote:
> e482dc3eaac43f88beea133843ae38c661262e97 breaks migration of a VM using an e1000 device (which is the default...).
> Origin host is Debian Lenny 32-bits, destination host is Fedora 12 32-bit. Guest is running Debian Lenny 32-bit.
> Symtoms: origin finishes migration correctly, but destination prints "load of migration failed" and the VM is stopped.
>
> This is because the origin closed the connexion, so qemu_fill_buffer reads a buffer with len == 0, which is treated as an error later (f->has_error = 1).
> Reverting e482dc3eaac43f88beea133843ae38c661262e97 fixes the problem.

Do you use --enable-kvm?  That could explain it.

I will take at that two bugs this week.  I tested everything on 64bit
hosts.

Later, Juan.

Re: [Qemu-devel] [BUG] Migration broken by "e1000: port to vmstate"

[Pierre Riteau]

On 24 nov. 2009, at 17:27, Ryan Harper wrote:

> * Pierre Riteau <Pierre.Riteau@irisa.fr> [2009-11-24 09:35]:
>> e482dc3eaac43f88beea133843ae38c661262e97 breaks migration of a VM using an e1000 device (which is the default...).
>> Origin host is Debian Lenny 32-bits, destination host is Fedora 12 32-bit. Guest is running Debian Lenny 32-bit.
>> Symtoms: origin finishes migration correctly, but destination prints "load of migration failed" and the VM is stopped.
> 
> What's your command line?
> 
> If I add -net user , or -net user -net nic,model=e1000 migration works.


Origin node: sudo qemu -m 1024 -hda /mnt/hda.img -net nic,macaddr=DE:AD:BE:EF:69:25 -net tap,script=/home/priteau/qemu-ifup -boot c
Destination node: sudo qemu -m 1024 -hda /mnt/hda.img -net nic,macaddr=DE:AD:BE:EF:69:25 -net tap,script=/home/priteau/qemu-ifup -boot c -incoming tcp:0:4444

-- 
Pierre Riteau -- http://perso.univ-rennes1.fr/pierre.riteau/

[Qemu-devel] Re: [BUG] Migration broken by "e1000: port to vmstate"

[Pierre Riteau]

On 24 nov. 2009, at 17:46, Juan Quintela wrote:

> Pierre Riteau <Pierre.Riteau@irisa.fr> wrote:
>> e482dc3eaac43f88beea133843ae38c661262e97 breaks migration of a VM using an e1000 device (which is the default...).
>> Origin host is Debian Lenny 32-bits, destination host is Fedora 12 32-bit. Guest is running Debian Lenny 32-bit.
>> Symtoms: origin finishes migration correctly, but destination prints "load of migration failed" and the VM is stopped.
>> 
>> This is because the origin closed the connexion, so qemu_fill_buffer reads a buffer with len == 0, which is treated as an error later (f->has_error = 1).
>> Reverting e482dc3eaac43f88beea133843ae38c661262e97 fixes the problem.
> 
> Do you use --enable-kvm?  That could explain it.

No I don't (although the bug appears with qemu-kvm as well, but I haven't tested the revert).

-- 
Pierre Riteau -- http://perso.univ-rennes1.fr/pierre.riteau/

[Qemu-devel] Re: [BUG] Migration broken by "e1000: port to vmstate"

[Jan Kiszka]

Juan Quintela wrote:
> Pierre Riteau <Pierre.Riteau@irisa.fr> wrote:
>> e482dc3eaac43f88beea133843ae38c661262e97 breaks migration of a VM using an e1000 device (which is the default...).
>> Origin host is Debian Lenny 32-bits, destination host is Fedora 12 32-bit. Guest is running Debian Lenny 32-bit.
>> Symtoms: origin finishes migration correctly, but destination prints "load of migration failed" and the VM is stopped.
>>
>> This is because the origin closed the connexion, so qemu_fill_buffer reads a buffer with len == 0, which is treated as an error later (f->has_error = 1).
>> Reverting e482dc3eaac43f88beea133843ae38c661262e97 fixes the problem.
> 
> Do you use --enable-kvm?  That could explain it.
> 

Confirmed here - the hard way, ie. after debugging a while on my own
patches. My scenario is default command line + -enable-kvm, migration to
file (exec:cat), then restore from that file. Restore fails while
reading the e1000 state with EOF.

> I will take at that two bugs this week.  I tested everything on 64bit
> hosts.

Any news on this?

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux

[Qemu-devel] Re: [BUG] Migration broken by "e1000: port to vmstate"

[Jan Kiszka]

Jan Kiszka wrote:
> Juan Quintela wrote:
>> Pierre Riteau <Pierre.Riteau@irisa.fr> wrote:
>>> e482dc3eaac43f88beea133843ae38c661262e97 breaks migration of a VM using an e1000 device (which is the default...).
>>> Origin host is Debian Lenny 32-bits, destination host is Fedora 12 32-bit. Guest is running Debian Lenny 32-bit.
>>> Symtoms: origin finishes migration correctly, but destination prints "load of migration failed" and the VM is stopped.
>>>
>>> This is because the origin closed the connexion, so qemu_fill_buffer reads a buffer with len == 0, which is treated as an error later (f->has_error = 1).
>>> Reverting e482dc3eaac43f88beea133843ae38c661262e97 fixes the problem.
>> Do you use --enable-kvm?  That could explain it.
>>
> 
> Confirmed here - the hard way, ie. after debugging a while on my own
> patches. My scenario is default command line + -enable-kvm, migration to
> file (exec:cat), then restore from that file. Restore fails while
> reading the e1000 state with EOF.

Forgot to add: IIRC, the EOF first popped up in the second "mac_reg"
section, ie. the VMSTATE_UINT32_SUB_ARRAY(..., MTA). May SUB_ARRAY is
buggy. e1000 seems to be its only user so far.

> 
>> I will take at that two bugs this week.  I tested everything on 64bit
>> hosts.
> 
> Any news on this?
> 
> Jan
> 

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux

[Qemu-devel] Re: [BUG] Migration broken by "e1000: port to vmstate"

[Juan Quintela]

Jan Kiszka <jan.kiszka@siemens.com> wrote:
> Juan Quintela wrote:
>> Pierre Riteau <Pierre.Riteau@irisa.fr> wrote:
>>> e482dc3eaac43f88beea133843ae38c661262e97 breaks migration of a VM using an e1000 device (which is the default...).
>>> Origin host is Debian Lenny 32-bits, destination host is Fedora 12 32-bit. Guest is running Debian Lenny 32-bit.
>>> Symtoms: origin finishes migration correctly, but destination prints "load of migration failed" and the VM is stopped.
>>>
>>> This is because the origin closed the connexion, so qemu_fill_buffer reads a buffer with len == 0, which is treated as an error later (f->has_error = 1).
>>> Reverting e482dc3eaac43f88beea133843ae38c661262e97 fixes the problem.
>> 
>> Do you use --enable-kvm?  That could explain it.
>> 
>
> Confirmed here - the hard way, ie. after debugging a while on my own
> patches. My scenario is default command line + -enable-kvm, migration to
> file (exec:cat), then restore from that file. Restore fails while
> reading the e1000 state with EOF.
>
>> I will take at that two bugs this week.  I tested everything on 64bit
>> hosts.
>
> Any news on this?

nope, migration backwards took more time that I wanted.

Will take a look tomorrow.

Sorry, Juan.

[Qemu-devel] live migration of 64-bit kvm guest (was: [BUG] Migration broken by "e1000: port to vmstate")

[Jan Kiszka]

Juan Quintela wrote:
> Jan Kiszka <jan.kiszka@siemens.com> wrote:
>> Juan Quintela wrote:
>>> Pierre Riteau <Pierre.Riteau@irisa.fr> wrote:
>>>> e482dc3eaac43f88beea133843ae38c661262e97 breaks migration of a VM using an e1000 device (which is the default...).
>>>> Origin host is Debian Lenny 32-bits, destination host is Fedora 12 32-bit. Guest is running Debian Lenny 32-bit.
>>>> Symtoms: origin finishes migration correctly, but destination prints "load of migration failed" and the VM is stopped.
>>>>
>>>> This is because the origin closed the connexion, so qemu_fill_buffer reads a buffer with len == 0, which is treated as an error later (f->has_error = 1).
>>>> Reverting e482dc3eaac43f88beea133843ae38c661262e97 fixes the problem.
>>> Do you use --enable-kvm?  That could explain it.
>>>
>> Confirmed here - the hard way, ie. after debugging a while on my own
>> patches. My scenario is default command line + -enable-kvm, migration to
>> file (exec:cat), then restore from that file. Restore fails while
>> reading the e1000 state with EOF.
>>
>>> I will take at that two bugs this week.  I tested everything on 64bit
>>> hosts.
>> Any news on this?
> 
> nope, migration backwards took more time that I wanted.
> 
> Will take a look tomorrow.
> 

I'm currently pulling hair while trying to migrate a 64-bit x86 guest in
kvm mode (via exec+snapshot file, but without block migration and any of
my patches). The target crashes after taking over.

Are you aware of any 64-bit migration issues related to kvm (as you said
there are more bugs)? 32 bit works fine here.

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux