[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#838660: linux/jessie: missing KVM fix causes migration issues on guests without rdtscp



Control: tags -1 + pending

Hi Aurelien,

On Fri, Sep 23, 2016 at 01:58:51PM +0200, Aurelien Jarno wrote:
> Source: linux
> Version: 3.16.36-1+deb8u1
> Severity: normal
> Tags: patch
> User: debian-admin@lists.debian.org
> Usertags: needed-by-DSA-Team
> 
> Dear Maintainer,
> 
> The following fix went to kernel 3.16.7-ckt25-1, as it was Cc:ed to
> stable@vger.kernel.org:
> 
> | commit 9dbe6cf941a6fe82933aef565e4095fb10f65023
> | Author: Paolo Bonzini <pbonzini@redhat.com>
> | Date:   Thu Nov 12 14:49:17 2015 +0100
> | 
> |     KVM: x86: expose MSR_TSC_AUX to userspace
> |     
> |     If we do not do this, it is not properly saved and restored across
> |     migration.  Windows notices due to its self-protection mechanisms,
> |     and is very upset about it (blue screen of death).
> |     
> |     Cc: Radim Krcmar <rkrcmar@redhat.com>
> |     Cc: stable@vger.kernel.org
> |     Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> 
> However it has been followed a bit later by the following fix, which has
> not been Cc:ed to stable@vger.kernel.org and thus not backported:
> 
> | commit 81b1b9ca6d5ca5f3ce91c0095402def657cf5db3
> | Author: Haozhong Zhang <haozhong.zhang@intel.com>
> | Date:   Mon Dec 14 23:13:38 2015 +0800
> | 
> |     KVM: VMX: Fix host initiated access to guest MSR_TSC_AUX
> |     
> |     The current handling of accesses to guest MSR_TSC_AUX returns error if
> |     vcpu does not support rdtscp, though those accesses are initiated by
> |     host. This can result in the reboot failure of some versions of
> |     QEMU. This patch fixes this issue by passing those host initiated
> |     accesses for further handling instead.
> |     
> |     Signed-off-by: Haozhong Zhang <haozhong.zhang@intel.com>
> |     Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> 
> This causes guest migrations issues when the host supports the rdtscp
> instruction, but it is not exposed to the guest. The migration itself
> works fine, but then the guest is frozen on the target host.
> 
> This patch does not apply cleanly to the 3.16 kernel, but can be
> backported relatively easily. It has already been backported in
> v4.2.8-ckt5 already, so the patch can be taken from there directly.

I have pushed the patch in the jessie branch in git.

Regards,
Salvatore


Reply to: