Skip to contentSkip to conversation

kvm: zapping shadow pages for mmio generation wraparound

DISCUSSION Zlatko Calusic 3 participants 5 messages Jul 22, 2013 – Dec 21, 2014 KVM

lore
Share & tools
lore

Zlatko Calusic

author
#0 lore raw

I'm getting a lot of these in kernel 3.11.0-rc1:

Jul 19 19:53:06 ps kernel: kvm: zapping shadow pages for mmio generation wraparound
Jul 19 22:37:47 ps kernel: kvm: zapping shadow pages for mmio generation wraparound
Jul 20 00:07:24 ps kernel: kvm: zapping shadow pages for mmio generation wraparound
Jul 20 00:11:04 ps kernel: kvm: zapping shadow pages for mmio generation wraparound
Jul 20 00:39:14 ps kernel: kvm: zapping shadow pages for mmio generation wraparound
Jul 21 00:25:32 ps kernel: kvm: zapping shadow pages for mmio generation wraparound

There seems to be one line in the kernel log for every reboot of a WinXP guest, and I'm getting a bit bored with it. I don't understand what those messages mean? What can I do to stop them flooding the log?

Please Cc: (not subscribed)

Regards,

Signature · 1 line
Zlatko
maintainer
#1 lore raw

Il 22/07/2013 22:57, Zlatko Calusic ha scritto:

Quoted 13 lines

I'm getting a lot of these in kernel 3.11.0-rc1:

Jul 19 19:53:06 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 19 22:37:47 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 20 00:07:24 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 20 00:11:04 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 20 00:39:14 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 21 00:25:32 ps kernel: kvm: zapping shadow pages for mmio generation
Quoted 7 lines

wraparound

There seems to be one line in the kernel log for every reboot of a WinXP guest, and I'm getting a bit bored with it. I don't understand what those messages mean? What can I do to stop them flooding the log?

Please Cc: (not subscribed)

The log message would let us correlate any reports with a new optimization (added in 3.11). We will probably downgrade it to KERN_DEBUG before 3.11 goes out (or otherwise cause it to disappear), in the meanwhile it's harmless.

Paolo

author
#2 lore raw
Quoted 20 lines · Paolo Bonzini

Il 22/07/2013 22:57, Zlatko Calusic ha scritto:

Quoted 19 lines

I'm getting a lot of these in kernel 3.11.0-rc1:

Jul 19 19:53:06 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 19 22:37:47 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 20 00:07:24 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 20 00:11:04 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 20 00:39:14 ps kernel: kvm: zapping shadow pages for mmio generation
wraparound
Jul 21 00:25:32 ps kernel: kvm: zapping shadow pages for mmio generation

wraparound

There seems to be one line in the kernel log for every reboot of a WinXP guest, and I'm getting a bit bored with it. I don't understand what those messages mean? What can I do to stop them flooding the log?

Quoted 7 lines
Quoted 1 line

Please Cc: (not subscribed)

The log message would let us correlate any reports with a new optimization (added in 3.11). We will probably downgrade it to KERN_DEBUG before 3.11 goes out (or otherwise cause it to disappear), in the meanwhile it's harmless.

Ah, OK then. Thanks for the info. And certainly thanks for the great virtualization platform!

Best regards,

Signature · 1 line
Zlatko
#3 lore raw
In reply to Zlatko Calusic

On Tue, 23 Jul 2013 13:50:21 +0200, Paolo Bonzini wrote:
...

Quoted 7 lines
Quoted 2 lines
Jul 21 00:25:32 ps kernel: kvm: zapping shadow pages for mmio generation

wraparound

The log message would let us correlate any reports with a new optimization (added in 3.11). We will probably downgrade it to KERN_DEBUG before 3.11 goes out (or otherwise cause it to disappear), in the meanwhile it's harmless.

Hm. I just upgraded one of our heavy-kvm-usage host from 3.10 to 3.16, and it started showing these messages (hence I started searching and found this thread from 2013).

Has it not been decided to downgrade this message to KERN_DEBUG for some reason?

What does this message mean anyway?
Why this very rare case (according to comments in arch/x86/kvm/mmu.c) happens on this machine?

Thanks,

/mjt

maintainer
#4 lore raw
Quoted 2 lines · Michael Tokarev

Has it not been decided to downgrade this message to KERN_DEBUG for some reason?

I've just never gotten round to do it. I'll do it next week.

Quoted 3 lines

What does this message mean anyway?
Why this very rare case (according to comments in arch/x86/kvm/mmu.c) happens on this machine?

It is rare, but we make it happen once per VM to try and avoid that it bitrots.

Paolo