kvm-devel

Re: [kvm-devel] RHEL5 smp guests on RHE5.1 hosts hang with kvm-52

[Farkas L. <lf...@bp...>]

david ahern wrote:
> (Changed the subject to correspond with email.)
> 
> I am having the same problem on the 64-bit host running RHEL5.1 as well, it just takes more reboots. Same symptoms as I mentioned for the 32-bit host. kernel side stack traces for each qemu thread for one of the lockups is attached; the file contains traces for each thread at 3 sample times in case it helps get some insight.

it's seems i'm not the only one:-)

-- 
  Levente                               "Si vis pacem para bellum!"

Re: [kvm-devel] [ANNOUNCE] kvm-51 release

[Avi K. <av...@qu...>]

david ahern wrote:
> With kvm-52 my 32-bit host running RHEL5.1 can start an RHEL 5 SMP guest only once. Second and subsequent attempts hang. Removing kvm and kvm_intel modules have no affect; I need to reboot the host to get an SMP guest to start. My similarly configured 64-bit host does not seem to have this problem.
>
>
>   

What happens if you boot a host, wait a while (say a couple of hours),
and then start a guest?

(I'm suspecting unsynchronized tscs)

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.

Re: [kvm-devel] [ANNOUNCE] kvm-51 release

[david a. <da...@ci...>]

I let the host stay up for 90 minutes before loading kvm and starting a VM. On the first reboot it hangs at 'Starting udev'.

I added 'noapic' to the kernel boot options, and it boots fine. (Turns out I only added that to grub.conf in images that run a particular ap for which I am running performance tests.)

I would like to know why I need the noapic option to get around this and the networking problem. Are there performance hits as a side effect?

david

Avi Kivity wrote:
> david ahern wrote:
>> With kvm-52 my 32-bit host running RHEL5.1 can start an RHEL 5 SMP guest only once. Second and subsequent attempts hang. Removing kvm and kvm_intel modules have no affect; I need to reboot the host to get an SMP guest to start. My similarly configured 64-bit host does not seem to have this problem.
>>
>>
>>   
> 
> What happens if you boot a host, wait a while (say a couple of hours),
> and then start a guest?
> 
> (I'm suspecting unsynchronized tscs)
> 

Re: [kvm-devel] [ANNOUNCE] kvm-51 release

[Farkas L. <lf...@bp...>]

Avi Kivity wrote:
> david ahern wrote:
>> In RHEL 5.1 <linux/notifier.h> defines:
>>
>> #define CPU_TASKS_FROZEN    0x0010
>>
>> #define CPU_ONLINE_FROZEN   (CPU_ONLINE | CPU_TASKS_FROZEN)
>> #define CPU_DEAD_FROZEN     (CPU_DEAD | CPU_TASKS_FROZEN)
>>
>> which means in kvm-51/kernel/external-module-compat.h the '#ifndef
>> CPU_TASKS_FROZEN' needs to have a case. For my purposes, I just moved
>> up the endif around what was defined.
>>   
> 
> I committed a change which renders this unnecessary.  Will be part of
> kvm-52.
> 
>> With that change, kvm-51 compiles. I am still seeing 32-bit SMP guests
>> hang on boot for both 32-bit and 64-bit hosts (again running RHEL5.1).
>>   
> 
> I still don't.  Can you test the attached patch?

can you tell us which cpu, memory, host and guest os are you using? may
be we can guess the differences.


ps. i can't test it until tomorrow since i dare not restart it remotely
since most of the time host are crash and if i'm not at there all guest
hand until tomorrow.

-- 
  Levente                               "Si vis pacem para bellum!"

Re: [kvm-devel] [ANNOUNCE] kvm-51 release

[Avi K. <av...@qu...>]

Farkas Levente wrote:
>>
>>> With that change, kvm-51 compiles. I am still seeing 32-bit SMP guests
>>> hang on boot for both 32-bit and 64-bit hosts (again running RHEL5.1).
>>>   
>>>       
>> I still don't.  Can you test the attached patch?
>>     
>
> can you tell us which cpu, memory, host and guest os are you using? may
> be we can guess the differences.
>
>
>   
processor       : 1
vendor_id       : GenuineIntel
cpu family      : 6
model           : 15
model name      : Intel(R) Xeon(R) CPU            5150  @ 2.66GHz
stepping        : 6
cpu MHz         : 1998.000
cache size      : 4096 KB
physical id     : 0
siblings        : 2
core id         : 1
cpu cores       : 2
fpu             : yes
fpu_exception   : yes
cpuid level     : 10
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge 
mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm syscall 
nx lm constant_tsc pni monitor ds_cpl vmx est tm2 cx16 xtpr lahf_lm
bogomips        : 5319.99
clflush size    : 64
cache_alignment : 64
address sizes   : 36 bits physical, 48 bits virtual
power management:


(total 2 cores)

             total       used       free     shared    buffers     cached
Mem:      16437164    7874104    8563060          0     112444    7488552
-/+ buffers/cache:     273108   16164056
Swap:      2031608          0    2031608


Host is clean centos-5 x86_64 w/ kvm.git

Guest is clean FC6 x86_64

-- 
error compiling committee.c: too many arguments to function

Re: [kvm-devel] [ANNOUNCE] kvm-51 release

[Farkas L. <lf...@bp...>]

Avi Kivity wrote:
> Farkas Levente wrote:
>>>
>>>> With that change, kvm-51 compiles. I am still seeing 32-bit SMP guests
>>>> hang on boot for both 32-bit and 64-bit hosts (again running RHEL5.1).
>>>>         
>>> I still don't.  Can you test the attached patch?
>>>     
>>
>> can you tell us which cpu, memory, host and guest os are you using? may
>> be we can guess the differences.
> Host is clean centos-5 x86_64 w/ kvm.git

same host just different hardware:-(

> Guest is clean FC6 x86_64

can you recheck it with 1 clean centos-5 x86_64 and one centos-5 i386
both have 2GB ram and in your case 2 vcpu and try to start the guest
parallel?

if it's still works for you then the only difference will be the
hardware or that i try 4 vcpu (since our host has 4 cpu).

-- 
  Levente                               "Si vis pacem para bellum!"

Re: [kvm-devel] [ANNOUNCE] kvm-51 release

[Avi K. <av...@qu...>]

Farkas Levente wrote:
>
>> Guest is clean FC6 x86_64
>>     
>
> can you recheck it with 1 clean centos-5 x86_64 and one centos-5 i386
> both have 2GB ram and in your case 2 vcpu and try to start the guest
> parallel?
>
> if it's still works for you then the only difference will be the
> hardware or that i try 4 vcpu (since our host has 4 cpu).
>
>   

Since David Ahern reported success with my patch, I'll wait for you to 
try it or kvm-52.

-- 
error compiling committee.c: too many arguments to function