Question of xl dump-core

Question of xl dump-core

[wj zhou]

Hello all,

Sorry to disturb you, but I really want to figure it out.
The xen core of redhat 6 with pod is unable to be used with crash.

I installed a hvm of redhat 6 by xen 4.7.0-rc2.
And the memory is set as below:
memory=1024
maxmem=4096

"xl dump-core" is executed, and the core is produced successfully.
I got the following message:
xc: info: exceeded nr_pages (261111) losing pages

Unfortunately, I got some errors when executing crash with it.
The below is the log of crash.

<cut>
crash 7.0.9-4.el7
...

please wait... (gathering kmem slab cache data)
crash: read error: kernel virtual address: ffff88010b532e00  type:
"kmem_cache buffer"

crash: unable to initialize kmem slab cache subsystem

please wait... (gathering module symbol data)
crash: read error: physical address: 1058a1000  type: "page table"
<cut>

I knew balloon is not supported by redhat 6, so pod is also not supported.
But I wonder the reason why the above error happens.
Balloon and pod are also not supported by redhat 7, but it won't
happen in redhat 7 hvm.
I'm very appreciated if someone can help me.

-- 
Regards
Zhou

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

Re: Question of xl dump-core

[Konrad Rzeszutek Wilk]

On Tue, Jun 14, 2016 at 08:21:16AM +0800, wj zhou wrote:
> Hello all,

Hey,

CC-ing Daniel, and Dave.
> 
> Sorry to disturb you, but I really want to figure it out.
> The xen core of redhat 6 with pod is unable to be used with crash.
> 
> I installed a hvm of redhat 6 by xen 4.7.0-rc2.
> And the memory is set as below:
> memory=1024
> maxmem=4096
> 
> "xl dump-core" is executed, and the core is produced successfully.
> I got the following message:
> xc: info: exceeded nr_pages (261111) losing pages
> 
> Unfortunately, I got some errors when executing crash with it.
> The below is the log of crash.
> 
> <cut>
> crash 7.0.9-4.el7

http://people.redhat.com/anderson says that the latest is 7.1.5.
Can you try that version?

> ...
> 
> please wait... (gathering kmem slab cache data)
> crash: read error: kernel virtual address: ffff88010b532e00  type:
> "kmem_cache buffer"
> 
> crash: unable to initialize kmem slab cache subsystem
> 
> please wait... (gathering module symbol data)
> crash: read error: physical address: 1058a1000  type: "page table"
> <cut>
> 
> I knew balloon is not supported by redhat 6, so pod is also not supported.

?

> But I wonder the reason why the above error happens.
> Balloon and pod are also not supported by redhat 7, but it won't
> happen in redhat 7 hvm.
> I'm very appreciated if someone can help me.

Well, does it work if you maxmem != memory ?

> 
> -- 
> Regards
> Zhou
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel

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

Re: Question of xl dump-core

[wj zhou]

Hi,

Thanks a lot for your reply!

On Tue, Jun 14, 2016 at 11:02 PM, Konrad Rzeszutek Wilk
<konrad.wilk@oracle.com> wrote:
> On Tue, Jun 14, 2016 at 08:21:16AM +0800, wj zhou wrote:
>> Hello all,
>
> Hey,
>
> CC-ing Daniel, and Dave.
>>
>> Sorry to disturb you, but I really want to figure it out.
>> The xen core of redhat 6 with pod is unable to be used with crash.
>>
>> I installed a hvm of redhat 6 by xen 4.7.0-rc2.
>> And the memory is set as below:
>> memory=1024
>> maxmem=4096
>>
>> "xl dump-core" is executed, and the core is produced successfully.
>> I got the following message:
>> xc: info: exceeded nr_pages (261111) losing pages
>>
>> Unfortunately, I got some errors when executing crash with it.
>> The below is the log of crash.
>>
>> <cut>
>> crash 7.0.9-4.el7
>
> http://people.redhat.com/anderson says that the latest is 7.1.5.
> Can you try that version?
>

I have just tried the latest crash version, and got the same error message.
Since the following info exists, I think there is something wrong in xen.
xc: info: exceeded nr_pages (261111) losing pages

>> ...
>>
>> please wait... (gathering kmem slab cache data)
>> crash: read error: kernel virtual address: ffff88010b532e00  type:
>> "kmem_cache buffer"
>>
>> crash: unable to initialize kmem slab cache subsystem
>>
>> please wait... (gathering module symbol data)
>> crash: read error: physical address: 1058a1000  type: "page table"
>> <cut>
>>
>> I knew balloon is not supported by redhat 6, so pod is also not supported.
>
> ?
>

In redhat 6 hvm, there is no balloon module.

>> But I wonder the reason why the above error happens.
>> Balloon and pod are also not supported by redhat 7, but it won't
>> happen in redhat 7 hvm.
>> I'm very appreciated if someone can help me.
>
> Well, does it work if you maxmem != memory ?
>

Yes, it works in redhat 7 hvm when maxmem>memory.

But I found something strange.
"free -m" performs quite different between redhat6 and redhat7 hvm.
In redhat 6 the value by "free -m" equals maxmem(4096).
However, in redhat 7 it equals memory(1024),

-- 
Regards
Zhou

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

Re: Question of xl dump-core

[Daniel Kiper]

On Wed, Jun 15, 2016 at 09:46:45AM +0800, wj zhou wrote:
> Hi,
>
> Thanks a lot for your reply!
>
> On Tue, Jun 14, 2016 at 11:02 PM, Konrad Rzeszutek Wilk
> <konrad.wilk@oracle.com> wrote:
> > On Tue, Jun 14, 2016 at 08:21:16AM +0800, wj zhou wrote:
> >> Hello all,
> >
> > Hey,
> >
> > CC-ing Daniel, and Dave.
> >>
> >> Sorry to disturb you, but I really want to figure it out.
> >> The xen core of redhat 6 with pod is unable to be used with crash.
> >>
> >> I installed a hvm of redhat 6 by xen 4.7.0-rc2.
> >> And the memory is set as below:
> >> memory=1024
> >> maxmem=4096
> >>
> >> "xl dump-core" is executed, and the core is produced successfully.
> >> I got the following message:
> >> xc: info: exceeded nr_pages (261111) losing pages
> >>
> >> Unfortunately, I got some errors when executing crash with it.
> >> The below is the log of crash.
> >>
> >> <cut>
> >> crash 7.0.9-4.el7
> >
> > http://people.redhat.com/anderson says that the latest is 7.1.5.
> > Can you try that version?
> >
>
> I have just tried the latest crash version, and got the same error message.
> Since the following info exists, I think there is something wrong in xen.
> xc: info: exceeded nr_pages (261111) losing pages
>
> >> ...
> >>
> >> please wait... (gathering kmem slab cache data)
> >> crash: read error: kernel virtual address: ffff88010b532e00  type:
> >> "kmem_cache buffer"
> >>
> >> crash: unable to initialize kmem slab cache subsystem
> >>
> >> please wait... (gathering module symbol data)
> >> crash: read error: physical address: 1058a1000  type: "page table"
> >> <cut>
> >>
> >> I knew balloon is not supported by redhat 6, so pod is also not supported.
> >
> > ?
> >
>
> In redhat 6 hvm, there is no balloon module.
>
> >> But I wonder the reason why the above error happens.
> >> Balloon and pod are also not supported by redhat 7, but it won't
> >> happen in redhat 7 hvm.
> >> I'm very appreciated if someone can help me.
> >
> > Well, does it work if you maxmem != memory ?
> >
>
> Yes, it works in redhat 7 hvm when maxmem>memory.
>
> But I found something strange.
> "free -m" performs quite different between redhat6 and redhat7 hvm.
> In redhat 6 the value by "free -m" equals maxmem(4096).
> However, in redhat 7 it equals memory(1024),

Sadly I am not able to play with this issue right know.
However, I could review patches. So, If you wish to do
that I think that you should first of all investigate
xl issue described earlier and then look at crash code.

Daniel

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